[patch v3 00/36] Hexagon: Add support for Qualcomm Hexagon architecture

[patch v3 00/36] Hexagon: Add support for Qualcomm Hexagon architecture

[Richard Kuo]

This is the third version of the patchset to the Linux kernel to support
the Qualcomm Hexagon architecture.  These patches have been organized like
the last time, in a somewhat logically grouped fashion.

A summary of changes from the last patchset:
* removed __delay
* cleaned up checksum header to pull generic one
* implemented vDSO so signal trampolines are no longer on stack
* fixed cpu_idle()
* removed init_IRQ() wrapper
* cleaned up (most of) the wrongness in time.c
* cleaned up smp.c to use more generic IPI handlers
* filled in ins/outs routines in IO and other misc cleanup
* removed redundant vunmap from ioremap
* removed more highmem cruft
* moved rwsem to asm-generic
* bonus ifdefs added to generic checksum.h

Also, I have been compiling with sparse and it seems to pass.

These patches are available on Code Aurora:

https://www.codeaurora.org/patches/quic/hlk/hexagon-linux-20110908.tar.gz

Thanks again for everyone's feedback!

-Richard Kuo

-- 

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 01/36] Hexagon: Add generic headers

[Richard Kuo]

[-- Attachment #1: generic-asm-headers.diff --]
[-- Type: text/plain, Size: 7615 bytes --]

Reworked to use the Kbuild generic headers mechanism.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/Kbuild        |   60 +++++++++++++++++++++++++
 arch/hexagon/include/asm/bitsperlong.h |   26 ++++++++++
 arch/hexagon/include/asm/mutex.h       |    8 +++
 arch/hexagon/include/asm/setup.h       |   29 ++++++++++++
 arch/hexagon/include/asm/user.h        |   79 +++++++++++++++++++++++++++++++++
 5 files changed, 202 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/Kbuild
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/Kbuild	2011-09-07 13:00:30.870646698 -0500
@@ -0,0 +1,60 @@
+include include/asm-generic/Kbuild.asm
+
+header-y += registers.h
+header-y += ucontext.h
+
+generic-y += auxvec.h
+generic-y += bug.h
+generic-y += bugs.h
+generic-y += cpumask.h
+generic-y += cputime.h
+generic-y += current.h
+generic-y += device.h
+generic-y += div64.h
+generic-y += emergency-restart.h
+generic-y += errno.h
+generic-y += fb.h
+generic-y += fcntl.h
+generic-y += ftrace.h
+generic-y += hardirq.h
+generic-y += hw_irq.h
+generic-y += ioctl.h
+generic-y += ioctls.h
+generic-y += iomap.h
+generic-y += ipcbuf.h
+generic-y += ipc.h
+generic-y += irq_regs.h
+generic-y += kdebug.h
+generic-y += kmap_types.h
+generic-y += local64.h
+generic-y += local.h
+generic-y += local.h
+generic-y += mman.h
+generic-y += msgbuf.h
+generic-y += pci.h
+generic-y += percpu.h
+generic-y += poll.h
+generic-y += posix_types.h
+generic-y += resource.h
+generic-y += rwsem.h
+generic-y += scatterlist.h
+generic-y += sections.h
+generic-y += segment.h
+generic-y += sembuf.h
+generic-y += shmbuf.h
+generic-y += shmparam.h
+generic-y += siginfo.h
+generic-y += socket.h
+generic-y += sockios.h
+generic-y += statfs.h
+generic-y += stat.h
+generic-y += termbits.h
+generic-y += termios.h
+generic-y += topology.h
+generic-y += types.h
+generic-y += ucontext.h
+generic-y += unaligned.h
+generic-y += xor.h
+
+
+
Index: linux-hexagon-kernel/arch/hexagon/include/asm/bitsperlong.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/bitsperlong.h	2011-09-03 20:14:59.484980719 -0500
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef __ASM_HEXAGON_BITSPERLONG_H
+#define __ASM_HEXAGON_BITSPERLONG_H
+
+#define __BITS_PER_LONG 32
+
+#include <asm-generic/bitsperlong.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/mutex.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/mutex.h	2011-09-03 20:14:59.484980719 -0500
@@ -0,0 +1,8 @@
+/*
+ * Pull in the generic implementation for the mutex fastpath.
+ *
+ * TODO: implement optimized primitives instead, or leave the generic
+ * implementation in place, or pick the atomic_xchg() based generic
+ * implementation. (see asm-generic/mutex-xchg.h for details)
+ */
+#include <asm-generic/mutex-xchg.h>
Index: linux-hexagon-kernel/arch/hexagon/include/asm/setup.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/setup.h	2011-09-03 20:14:59.484980719 -0500
@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SETUP_H
+#define _ASM_SETUP_H
+
+#include <linux/init.h>
+#include <asm-generic/setup.h>
+
+extern char external_cmdline_buffer;
+
+void __init setup_arch_memory(void);
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/user.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/user.h	2011-09-03 20:14:59.484980719 -0500
@@ -0,0 +1,79 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef HEXAGON_ASM_USER_H
+#define HEXAGON_ASM_USER_H
+
+#include <asm-generic/user.h>
+
+/*
+ * Layout for registers passed in elf core dumps to userspace.
+ *
+ * Basically a rearranged subset of "pt_regs".
+ *
+ * Interested parties:  libc, gdb...
+ */
+
+struct user_regs_struct {
+	unsigned long r0;
+	unsigned long r1;
+	unsigned long r2;
+	unsigned long r3;
+	unsigned long r4;
+	unsigned long r5;
+	unsigned long r6;
+	unsigned long r7;
+	unsigned long r8;
+	unsigned long r9;
+	unsigned long r10;
+	unsigned long r11;
+	unsigned long r12;
+	unsigned long r13;
+	unsigned long r14;
+	unsigned long r15;
+	unsigned long r16;
+	unsigned long r17;
+	unsigned long r18;
+	unsigned long r19;
+	unsigned long r20;
+	unsigned long r21;
+	unsigned long r22;
+	unsigned long r23;
+	unsigned long r24;
+	unsigned long r25;
+	unsigned long r26;
+	unsigned long r27;
+	unsigned long r28;
+	unsigned long r29;
+	unsigned long r30;
+	unsigned long r31;
+	unsigned long gp;
+	unsigned long ugp;
+	unsigned long sa0;
+	unsigned long lc0;
+	unsigned long sa1;
+	unsigned long lc1;
+	unsigned long m0;
+	unsigned long m1;
+	unsigned long p3_0;
+	unsigned long pc;
+	unsigned long cause;
+	unsigned long badva;
+};
+
+#endif

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 02/36] Hexagon: Core arch-specific header files

[Richard Kuo]

[-- Attachment #1: basic-asm-headers.diff --]
[-- Type: text/plain, Size: 22885 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/byteorder.h  |   28 ++++
 arch/hexagon/include/asm/elf.h        |  228 ++++++++++++++++++++++++++++++++++
 arch/hexagon/include/asm/fpu.h        |    4 
 arch/hexagon/include/asm/futex.h      |  137 ++++++++++++++++++++
 arch/hexagon/include/asm/linkage.h    |   25 +++
 arch/hexagon/include/asm/module.h     |   26 +++
 arch/hexagon/include/asm/param.h      |   26 +++
 arch/hexagon/include/asm/perf_event.h |   24 +++
 arch/hexagon/include/asm/string.h     |   32 ++++
 arch/hexagon/include/asm/suspend.h    |   27 ++++
 arch/hexagon/include/asm/swab.h       |   24 +++
 arch/hexagon/include/asm/timex.h      |   37 +++++
 12 files changed, 618 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/byteorder.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/byteorder.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,28 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_BYTEORDER_H
+#define _ASM_BYTEORDER_H
+
+#if defined(__GNUC__) && !defined(__STRICT_ANSI__) || defined(__KERNEL__)
+#  define __BYTEORDER_HAS_U64__
+#endif
+
+#include <linux/byteorder/little_endian.h>
+
+#endif /* _ASM_BYTEORDER_H */
Index: linux-hexagon-kernel/arch/hexagon/include/asm/elf.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/elf.h	2011-09-08 19:23:10.620454090 -0500
@@ -0,0 +1,228 @@
+/*
+ * ELF definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef __ASM_ELF_H
+#define __ASM_ELF_H
+
+#include <asm/ptrace.h>
+#include <asm/user.h>
+
+/*
+ * This should really be in linux/elf-em.h.
+ */
+#define EM_HEXAGON	164   /* QUALCOMM Hexagon */
+
+struct elf32_hdr;
+
+/*
+ * ELF header e_flags defines.
+ */
+
+/*  should have stuff like "CPU type" and maybe "ABI version", etc  */
+
+/* Hexagon relocations */
+  /* V2 */
+#define R_HEXAGON_NONE           0
+#define R_HEXAGON_B22_PCREL      1
+#define R_HEXAGON_B15_PCREL      2
+#define R_HEXAGON_B7_PCREL       3
+#define R_HEXAGON_LO16           4
+#define R_HEXAGON_HI16           5
+#define R_HEXAGON_32             6
+#define R_HEXAGON_16             7
+#define R_HEXAGON_8              8
+#define R_HEXAGON_GPREL16_0      9
+#define R_HEXAGON_GPREL16_1     10
+#define R_HEXAGON_GPREL16_2     11
+#define R_HEXAGON_GPREL16_3     12
+#define R_HEXAGON_HL16          13
+  /* V3 */
+#define R_HEXAGON_B13_PCREL     14
+  /* V4 */
+#define R_HEXAGON_B9_PCREL      15
+  /* V4 (extenders) */
+#define R_HEXAGON_B32_PCREL_X   16
+#define R_HEXAGON_32_6_X        17
+  /* V4 (extended) */
+#define R_HEXAGON_B22_PCREL_X   18
+#define R_HEXAGON_B15_PCREL_X   19
+#define R_HEXAGON_B13_PCREL_X   20
+#define R_HEXAGON_B9_PCREL_X    21
+#define R_HEXAGON_B7_PCREL_X    22
+#define R_HEXAGON_16_X          23
+#define R_HEXAGON_12_X          24
+#define R_HEXAGON_11_X          25
+#define R_HEXAGON_10_X          26
+#define R_HEXAGON_9_X           27
+#define R_HEXAGON_8_X           28
+#define R_HEXAGON_7_X           29
+#define R_HEXAGON_6_X           30
+  /* V2 PIC */
+#define R_HEXAGON_32_PCREL      31
+#define R_HEXAGON_COPY          32
+#define R_HEXAGON_GLOB_DAT      33
+#define R_HEXAGON_JMP_SLOT      34
+#define R_HEXAGON_RELATIVE      35
+#define R_HEXAGON_PLT_B22_PCREL 36
+#define R_HEXAGON_GOTOFF_LO16   37
+#define R_HEXAGON_GOTOFF_HI16   38
+#define R_HEXAGON_GOTOFF_32     39
+#define R_HEXAGON_GOT_LO16      40
+#define R_HEXAGON_GOT_HI16      41
+#define R_HEXAGON_GOT_32        42
+#define R_HEXAGON_GOT_16        43
+
+/*
+ * ELF register definitions..
+ */
+typedef unsigned long elf_greg_t;
+
+/*  May need an ELF_NGREG later down the line  */
+typedef struct user_regs_struct elf_gregset_t;
+
+/*  Placeholder  */
+typedef unsigned long elf_fpregset_t;
+
+/*
+ * Bypass the whole "regsets" thing for now and use the define.
+ */
+
+#define ELF_CORE_COPY_REGS(DEST, REGS)	\
+do {					\
+	DEST.r0 = REGS->r00;		\
+	DEST.r1 = REGS->r01;		\
+	DEST.r2 = REGS->r02;		\
+	DEST.r3 = REGS->r03;		\
+	DEST.r4 = REGS->r04;		\
+	DEST.r5 = REGS->r05;		\
+	DEST.r6 = REGS->r06;		\
+	DEST.r7 = REGS->r07;		\
+	DEST.r8 = REGS->r08;		\
+	DEST.r9 = REGS->r09;		\
+	DEST.r10 = REGS->r10;		\
+	DEST.r11 = REGS->r11;		\
+	DEST.r12 = REGS->r12;		\
+	DEST.r13 = REGS->r13;		\
+	DEST.r14 = REGS->r14;		\
+	DEST.r15 = REGS->r15;		\
+	DEST.r16 = REGS->r16;		\
+	DEST.r17 = REGS->r17;		\
+	DEST.r18 = REGS->r18;		\
+	DEST.r19 = REGS->r19;		\
+	DEST.r20 = REGS->r20;		\
+	DEST.r21 = REGS->r21;		\
+	DEST.r22 = REGS->r22;		\
+	DEST.r23 = REGS->r23;		\
+	DEST.r24 = REGS->r24;		\
+	DEST.r25 = REGS->r25;		\
+	DEST.r26 = REGS->r26;		\
+	DEST.r27 = REGS->r27;		\
+	DEST.r28 = REGS->r28;		\
+	DEST.r29 = pt_psp(REGS);	\
+	DEST.r30 = REGS->r30;		\
+	DEST.r31 = REGS->r31;		\
+	DEST.gp = REGS->gp;		\
+	DEST.ugp = REGS->ugp;		\
+	DEST.sa0 = REGS->sa0;		\
+	DEST.lc0 = REGS->lc0;		\
+	DEST.sa1 = REGS->sa1;		\
+	DEST.lc1 = REGS->lc1;		\
+	DEST.m0 = REGS->m0;		\
+	DEST.m1 = REGS->m1;		\
+	DEST.p3_0 = REGS->preds;	\
+	DEST.pc = pt_elr(REGS);	\
+	DEST.cause = pt_cause(REGS);	\
+	DEST.badva = pt_badva(REGS);	\
+} while (0);
+
+
+
+/*
+ * This is used to ensure we don't load something for the wrong architecture.
+ * Checks the machine and ABI type.
+ */
+#define elf_check_arch(hdr)	((hdr)->e_machine == EM_HEXAGON)
+
+/*
+ * These are used to set parameters in the core dumps.
+ */
+#define ELF_CLASS	ELFCLASS32
+#define ELF_DATA	ELFDATA2LSB
+#define ELF_ARCH	EM_HEXAGON
+
+#ifdef CONFIG_HEXAGON_ARCH_V2
+#define ELF_CORE_EFLAGS 0x1
+#endif
+
+#ifdef CONFIG_HEXAGON_ARCH_V3
+#define ELF_CORE_EFLAGS 0x2
+#endif
+
+#ifdef CONFIG_HEXAGON_ARCH_V4
+#define ELF_CORE_EFLAGS 0x3
+#endif
+
+/*
+ * Some architectures have ld.so set up a pointer to a function
+ * to be registered using atexit, to facilitate cleanup.  So that
+ * static executables will be well-behaved, we would null the register
+ * in question here, in the pt_regs structure passed.  For now,
+ * leave it a null macro.
+ */
+#define ELF_PLAT_INIT(regs, load_addr) do { } while (0)
+
+#define USE_ELF_CORE_DUMP
+#define ELF_FDPIC_CORE_EFLAGS	0x0
+
+/* Hrm is this going to cause problems for changing PAGE_SIZE?  */
+#define ELF_EXEC_PAGESIZE	4096
+
+/*
+ * This is the location that an ET_DYN program is loaded if exec'ed.  Typical
+ * use of this is to invoke "./ld.so someprog" to test out a new version of
+ * the loader.  We need to make sure that it is out of the way of the program
+ * that it will "exec", and that there is sufficient room for the brk.
+ */
+#define ELF_ET_DYN_BASE         0x08000000UL
+
+/*
+ * This yields a mask that user programs can use to figure out what
+ * instruction set this cpu supports.
+ */
+#define ELF_HWCAP	(0)
+
+/*
+ * This yields a string that ld.so will use to load implementation
+ * specific libraries for optimization.  This is more specific in
+ * intent than poking at uname or /proc/cpuinfo.
+ */
+#define ELF_PLATFORM  (NULL)
+
+#ifdef __KERNEL__
+#define SET_PERSONALITY(ex) set_personality(PER_LINUX)
+#endif
+
+#define ARCH_HAS_SETUP_ADDITIONAL_PAGES 1
+struct linux_binprm;
+extern int arch_setup_additional_pages(struct linux_binprm *bprm,
+				       int uses_interp);
+
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/fpu.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/fpu.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,4 @@
+/*
+ * If the FPU is used inside the kernel,
+ * kernel_fpu_end() will be defined here.
+ */
Index: linux-hexagon-kernel/arch/hexagon/include/asm/futex.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/futex.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,137 @@
+#ifndef _ASM_HEXAGON_FUTEX_H
+#define _ASM_HEXAGON_FUTEX_H
+
+#ifdef __KERNEL__
+
+#include <linux/futex.h>
+#include <linux/uaccess.h>
+#include <asm/errno.h>
+
+/* XXX TODO-- need to add sync barriers! */
+
+#define __futex_atomic_op(insn, ret, oldval, uaddr, oparg) \
+	__asm__ __volatile( \
+	"1: %0 = memw_locked(%3);\n" \
+	    /* For example: %1 = %4 */ \
+	    insn \
+	"2: memw_locked(%3,p2) = %1;\n" \
+	"   if !p2 jump 1b;\n" \
+	"   %1 = #0;\n" \
+	"3:\n" \
+	".section .fixup,\"ax\"\n" \
+	"4: %1 = #%5;\n" \
+	"   jump 3b\n" \
+	".previous\n" \
+	".section __ex_table,\"a\"\n" \
+	".long 1b,4b,2b,4b\n" \
+	".previous\n" \
+	: "=&r" (oldval), "=&r" (ret), "+m" (*uaddr) \
+	: "r" (uaddr), "r" (oparg), "i" (-EFAULT) \
+	: "p2", "memory")
+
+
+static inline int
+futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
+{
+	int op = (encoded_op >> 28) & 7;
+	int cmp = (encoded_op >> 24) & 15;
+	int oparg = (encoded_op << 8) >> 20;
+	int cmparg = (encoded_op << 20) >> 20;
+	int oldval = 0, ret;
+	if (encoded_op & (FUTEX_OP_OPARG_SHIFT << 28))
+		oparg = 1 << oparg;
+
+	if (!access_ok(VERIFY_WRITE, uaddr, sizeof(int)))
+		return -EFAULT;
+
+	pagefault_disable();
+
+	switch (op) {
+	case FUTEX_OP_SET:
+		__futex_atomic_op("%1 = %4\n", ret, oldval, uaddr, oparg);
+		break;
+	case FUTEX_OP_ADD:
+		__futex_atomic_op("%1 = add(%0,%4)\n", ret, oldval, uaddr,
+				  oparg);
+		break;
+	case FUTEX_OP_OR:
+		__futex_atomic_op("%1 = or(%0,%4)\n", ret, oldval, uaddr,
+				  oparg);
+		break;
+	case FUTEX_OP_ANDN:
+		__futex_atomic_op("%1 = not(%4); %1 = and(%0,%1)\n", ret,
+				  oldval, uaddr, oparg);
+		break;
+	case FUTEX_OP_XOR:
+		__futex_atomic_op("%1 = xor(%0,%4)\n", ret, oldval, uaddr,
+				  oparg);
+		break;
+	default:
+		ret = -ENOSYS;
+	}
+
+	pagefault_enable();
+
+	if (!ret) {
+		switch (cmp) {
+		case FUTEX_OP_CMP_EQ:
+			ret = (oldval == cmparg);
+			break;
+		case FUTEX_OP_CMP_NE:
+			ret = (oldval != cmparg);
+			break;
+		case FUTEX_OP_CMP_LT:
+			ret = (oldval < cmparg);
+			break;
+		case FUTEX_OP_CMP_GE:
+			ret = (oldval >= cmparg);
+			break;
+		case FUTEX_OP_CMP_LE:
+			ret = (oldval <= cmparg);
+			break;
+		case FUTEX_OP_CMP_GT:
+			ret = (oldval > cmparg);
+			break;
+		default:
+			ret = -ENOSYS;
+		}
+	}
+	return ret;
+}
+
+static inline int
+futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr, u32 oldval,
+			      u32 newval)
+{
+	int prev;
+	int ret;
+
+	if (!access_ok(VERIFY_WRITE, uaddr, sizeof(u32)))
+		return -EFAULT;
+
+	__asm__ __volatile__ (
+	"1: %1 = memw_locked(%3)\n"
+	"   {\n"
+	"      p2 = cmp.eq(%1,%4)\n"
+	"      if !p2.new jump:NT 3f\n"
+	"   }\n"
+	"2: memw_locked(%3,p2) = %5\n"
+	"   if !p2 jump 1b\n"
+	"3:\n"
+	".section .fixup,\"ax\"\n"
+	"4: %0 = #%6\n"
+	"   jump 3b\n"
+	".previous\n"
+	".section __ex_table,\"a\"\n"
+	".long 1b,4b,2b,4b\n"
+	".previous\n"
+	: "+r" (ret), "=&r" (prev), "+m" (*uaddr)
+	: "r" (uaddr), "r" (oldval), "r" (newval), "i"(-EFAULT)
+	: "p2", "memory");
+
+	*uval = prev;
+	return ret;
+}
+
+#endif /* __KERNEL__ */
+#endif /* _ASM_HEXAGON_FUTEX_H */
Index: linux-hexagon-kernel/arch/hexagon/include/asm/linkage.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/linkage.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,25 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef __ASM_LINKAGE_H
+#define __ASM_LINKAGE_H
+
+#define __ALIGN		.align 4
+#define __ALIGN_STR	".align 4"
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/module.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/module.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_MODULE_H
+#define _ASM_MODULE_H
+
+#include <asm-generic/module.h>
+
+#define MODULE_ARCH_VERMAGIC __stringify(PROCESSOR_MODEL_NAME) " "
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/param.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/param.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_PARAM_H
+#define _ASM_PARAM_H
+
+#define EXEC_PAGESIZE	16384
+
+#include <asm-generic/param.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/string.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/string.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_STRING_H_
+#define _ASM_STRING_H_
+
+#ifdef __KERNEL__
+#define __HAVE_ARCH_MEMCPY
+extern void *memcpy(void *__to, __const__ void *__from, size_t __n);
+
+/*  ToDo:  use dczeroa, accelerate the compiler-constant zero case  */
+#define __HAVE_ARCH_MEMSET
+extern void *memset(void *__to, int c, size_t __n);
+#endif
+
+
+#endif /* _ASM_STRING_H_ */
Index: linux-hexagon-kernel/arch/hexagon/include/asm/suspend.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/suspend.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SUSPEND_H
+#define _ASM_SUSPEND_H
+
+static inline int arch_prepare_suspend(void)
+{
+	return 0;
+}
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/swab.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/swab.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,24 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SWAB_H
+#define _ASM_SWAB_H
+
+#define __SWAB_64_THRU_32__
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/timex.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/timex.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,37 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_TIMEX_H
+#define _ASM_TIMEX_H
+
+#include <asm-generic/timex.h>
+#include <asm/timer-regs.h>
+
+/* Using TCX0 as our clock.  CLOCK_TICK_RATE scheduled to be removed. */
+#define CLOCK_TICK_RATE              TCX0_CLK_RATE
+
+#define ARCH_HAS_READ_CURRENT_TIMER
+
+static inline int read_current_timer(unsigned long *timer_val)
+{
+	*timer_val = (unsigned long) __vmgettime();
+	return 0;
+}
+
+#endif
+
Index: linux-hexagon-kernel/arch/hexagon/include/asm/perf_event.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/perf_event.h	2011-09-08 19:22:01.000238581 -0500
@@ -0,0 +1,24 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_PERF_EVENT_H
+#define _ASM_PERF_EVENT_H
+
+#define PERF_EVENT_INDEX_OFFSET	0
+
+#endif /* _ASM_PERF_EVENT_H */

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 03/36] Hexagon: Add bitops support

[Richard Kuo]

[-- Attachment #1: bitops.diff --]
[-- Type: text/plain, Size: 7870 bytes --]

Merged sub-files into bitops.h; fixed return types and volatiles.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/bitops.h |  301 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 301 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/bitops.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/bitops.h	2011-09-03 20:14:59.504980775 -0500
@@ -0,0 +1,301 @@
+/*
+ * Bit operations for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_BITOPS_H
+#define _ASM_BITOPS_H
+
+#include <linux/compiler.h>
+#include <asm/byteorder.h>
+#include <asm/system.h>
+#include <asm/atomic.h>
+
+#ifdef __KERNEL__
+
+#define smp_mb__before_clear_bit()	barrier()
+#define smp_mb__after_clear_bit()	barrier()
+
+/*
+ * The offset calculations for these are based on BITS_PER_LONG == 32
+ * (i.e. I get to shift by #5-2 (32 bits per long, 4 bytes per access),
+ * mask by 0x0000001F)
+ *
+ * Typically, R10 is clobbered for address, R11 bit nr, and R12 is temp
+ */
+
+/**
+ * test_and_clear_bit - clear a bit and return its old value
+ * @nr:  bit number to clear
+ * @addr:  pointer to memory
+ */
+static inline int test_and_clear_bit(int nr, volatile void *addr)
+{
+	int oldval;
+
+	__asm__ __volatile__ (
+	"	{R10 = %1; R11 = asr(%2,#5); }\n"
+	"	{R10 += asl(R11,#2); R11 = and(%2,#0x1f)}\n"
+	"1:	R12 = memw_locked(R10);\n"
+	"	{ P0 = tstbit(R12,R11); R12 = clrbit(R12,R11); }\n"
+	"	memw_locked(R10,P1) = R12;\n"
+	"	{if !P1 jump 1b; %0 = mux(P0,#1,#0);}\n"
+	: "=&r" (oldval)
+	: "r" (addr), "r" (nr)
+	: "r10", "r11", "r12", "p0", "p1", "memory"
+	);
+
+	return oldval;
+}
+
+/**
+ * test_and_set_bit - set a bit and return its old value
+ * @nr:  bit number to set
+ * @addr:  pointer to memory
+ */
+static inline int test_and_set_bit(int nr, volatile void *addr)
+{
+	int oldval;
+
+	__asm__ __volatile__ (
+	"	{R10 = %1; R11 = asr(%2,#5); }\n"
+	"	{R10 += asl(R11,#2); R11 = and(%2,#0x1f)}\n"
+	"1:	R12 = memw_locked(R10);\n"
+	"	{ P0 = tstbit(R12,R11); R12 = setbit(R12,R11); }\n"
+	"	memw_locked(R10,P1) = R12;\n"
+	"	{if !P1 jump 1b; %0 = mux(P0,#1,#0);}\n"
+	: "=&r" (oldval)
+	: "r" (addr), "r" (nr)
+	: "r10", "r11", "r12", "p0", "p1", "memory"
+	);
+
+
+	return oldval;
+
+}
+
+/**
+ * test_and_change_bit - toggle a bit and return its old value
+ * @nr:  bit number to set
+ * @addr:  pointer to memory
+ */
+static inline int test_and_change_bit(int nr, volatile void *addr)
+{
+	int oldval;
+
+	__asm__ __volatile__ (
+	"	{R10 = %1; R11 = asr(%2,#5); }\n"
+	"	{R10 += asl(R11,#2); R11 = and(%2,#0x1f)}\n"
+	"1:	R12 = memw_locked(R10);\n"
+	"	{ P0 = tstbit(R12,R11); R12 = togglebit(R12,R11); }\n"
+	"	memw_locked(R10,P1) = R12;\n"
+	"	{if !P1 jump 1b; %0 = mux(P0,#1,#0);}\n"
+	: "=&r" (oldval)
+	: "r" (addr), "r" (nr)
+	: "r10", "r11", "r12", "p0", "p1", "memory"
+	);
+
+	return oldval;
+
+}
+
+/*
+ * Atomic, but doesn't care about the return value.
+ * Rewrite later to save a cycle or two.
+ */
+
+static inline void clear_bit(int nr, volatile void *addr)
+{
+	test_and_clear_bit(nr, addr);
+}
+
+static inline void set_bit(int nr, volatile void *addr)
+{
+	test_and_set_bit(nr, addr);
+}
+
+static inline void change_bit(int nr, volatile void *addr)
+{
+	test_and_change_bit(nr, addr);
+}
+
+
+/*
+ * These are allowed to be non-atomic.  In fact the generic flavors are
+ * in non-atomic.h.  Would it be better to use intrinsics for this?
+ *
+ * OK, writes in our architecture do not invalidate LL/SC, so this has to
+ * be atomic, particularly for things like slab_lock and slab_unlock.
+ *
+ */
+static inline void __clear_bit(int nr, volatile unsigned long *addr)
+{
+	test_and_clear_bit(nr, addr);
+}
+
+static inline void __set_bit(int nr, volatile unsigned long *addr)
+{
+	test_and_set_bit(nr, addr);
+}
+
+static inline void __change_bit(int nr, volatile unsigned long *addr)
+{
+	test_and_change_bit(nr, addr);
+}
+
+/*  Apparently, at least some of these are allowed to be non-atomic  */
+static inline int __test_and_clear_bit(int nr, volatile unsigned long *addr)
+{
+	return test_and_clear_bit(nr, addr);
+}
+
+static inline int __test_and_set_bit(int nr, volatile unsigned long *addr)
+{
+	return test_and_set_bit(nr, addr);
+}
+
+static inline int __test_and_change_bit(int nr, volatile unsigned long *addr)
+{
+	return test_and_change_bit(nr, addr);
+}
+
+static inline int __test_bit(int nr, const volatile unsigned long *addr)
+{
+	int retval;
+
+	asm volatile(
+	"{P0 = tstbit(%1,%2); if (P0.new) %0 = #1; if (!P0.new) %0 = #0;}\n"
+	: "=&r" (retval)
+	: "r" (addr[BIT_WORD(nr)]), "r" (nr % BITS_PER_LONG)
+	: "p0"
+	);
+
+	return retval;
+}
+
+#define test_bit(nr, addr) __test_bit(nr, addr)
+
+/*
+ * ffz - find first zero in word.
+ * @word: The word to search
+ *
+ * Undefined if no zero exists, so code should check against ~0UL first.
+ */
+static inline long ffz(int x)
+{
+	int r;
+
+	asm("%0 = ct1(%1);\n"
+		: "=&r" (r)
+		: "r" (x));
+	return r;
+}
+
+/*
+ * fls - find last (most-significant) bit set
+ * @x: the word to search
+ *
+ * This is defined the same way as ffs.
+ * Note fls(0) = 0, fls(1) = 1, fls(0x80000000) = 32.
+ */
+static inline long fls(int x)
+{
+	int r;
+
+	asm("{ %0 = cl0(%1);}\n"
+		"%0 = sub(#32,%0);\n"
+		: "=&r" (r)
+		: "r" (x)
+		: "p0");
+
+	return r;
+}
+
+/*
+ * ffs - find first bit set
+ * @x: the word to search
+ *
+ * This is defined the same way as
+ * the libc and compiler builtin ffs routines, therefore
+ * differs in spirit from the above ffz (man ffs).
+ */
+static inline long ffs(int x)
+{
+	int r;
+
+	asm("{ P0 = cmp.eq(%1,#0); %0 = ct0(%1);}\n"
+		"{ if P0 %0 = #0; if !P0 %0 = add(%0,#1);}\n"
+		: "=&r" (r)
+		: "r" (x)
+		: "p0");
+
+	return r;
+}
+
+/*
+ * __ffs - find first bit in word.
+ * @word: The word to search
+ *
+ * Undefined if no bit exists, so code should check against 0 first.
+ *
+ * bits_per_long assumed to be 32
+ * numbering starts at 0 I think (instead of 1 like ffs)
+ */
+static inline unsigned long __ffs(unsigned long word)
+{
+	int num;
+
+	asm("%0 = ct0(%1);\n"
+		: "=&r" (num)
+		: "r" (word));
+
+	return num;
+}
+
+/*
+ * __fls - find last (most-significant) set bit in a long word
+ * @word: the word to search
+ *
+ * Undefined if no set bit exists, so code should check against 0 first.
+ * bits_per_long assumed to be 32
+ */
+static inline unsigned long __fls(unsigned long word)
+{
+	int num;
+
+	asm("%0 = cl0(%1);\n"
+		"%0 = sub(#31,%0);\n"
+		: "=&r" (num)
+		: "r" (word));
+
+	return num;
+}
+
+#include <asm-generic/bitops/lock.h>
+#include <asm-generic/bitops/find.h>
+
+#include <asm-generic/bitops/fls64.h>
+#include <asm-generic/bitops/sched.h>
+#include <asm-generic/bitops/hweight.h>
+
+#include <asm-generic/bitops/le.h>
+#include <asm-generic/bitops/ext2-atomic.h>
+
+#endif /* __KERNEL__ */
+#endif

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 04/36] Hexagon: Add atomic ops support

[Richard Kuo]

[-- Attachment #1: atomic.diff --]
[-- Type: text/plain, Size: 5105 bytes --]

Our architecture uses load locked/store conditional type semantics for atomic
ops.

Added inline assembly version of atomic_add_unless.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/atomic.h |  164 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 164 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/atomic.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/atomic.h	2011-09-07 13:00:30.890646728 -0500
@@ -0,0 +1,164 @@
+/*
+ * Atomic operations for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_ATOMIC_H
+#define _ASM_ATOMIC_H
+
+#include <linux/types.h>
+
+#define ATOMIC_INIT(i)		{ (i) }
+#define atomic_set(v, i)	((v)->counter = (i))
+
+/**
+ * atomic_read - reads a word, atomically
+ * @v: pointer to atomic value
+ *
+ * Assumes all word reads on our architecture are atomic.
+ */
+#define atomic_read(v)		((v)->counter)
+
+/**
+ * atomic_xchg - atomic
+ * @v: pointer to memory to change
+ * @new: new value (technically passed in a register -- see xchg)
+ */
+#define atomic_xchg(v, new)	(xchg(&((v)->counter), (new)))
+
+
+/**
+ * atomic_cmpxchg - atomic compare-and-exchange values
+ * @v: pointer to value to change
+ * @old:  desired old value to match
+ * @new:  new value to put in
+ *
+ * Parameters are then pointer, value-in-register, value-in-register,
+ * and the output is the old value.
+ *
+ * Apparently this is complicated for archs that don't support
+ * the memw_locked like we do (or it's broken or whatever).
+ *
+ * Kind of the lynchpin of the rest of the generically defined routines.
+ * Remember V2 had that bug with dotnew predicate set by memw_locked.
+ *
+ * "old" is "expected" old val, __oldval is actual old value
+ */
+static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
+{
+	int __oldval;
+
+	asm volatile(
+		"1:	%0 = memw_locked(%1);\n"
+		"	{ P0 = cmp.eq(%0,%2);\n"
+		"	  if (!P0.new) jump:nt 2f; }\n"
+		"	memw_locked(%1,P0) = %3;\n"
+		"	if (!P0) jump 1b;\n"
+		"2:\n"
+		: "=&r" (__oldval)
+		: "r" (&v->counter), "r" (old), "r" (new)
+		: "memory", "p0"
+	);
+
+	return __oldval;
+}
+
+static inline int atomic_add_return(int i, atomic_t *v)
+{
+	int output;
+
+	__asm__ __volatile__ (
+		"1:	%0 = memw_locked(%1);\n"
+		"	%0 = add(%0,%2);\n"
+		"	memw_locked(%1,P3)=%0;\n"
+		"	if !P3 jump 1b;\n"
+		: "=&r" (output)
+		: "r" (&v->counter), "r" (i)
+		: "memory", "p3"
+	);
+	return output;
+
+}
+
+#define atomic_add(i, v) atomic_add_return(i, (v))
+
+static inline int atomic_sub_return(int i, atomic_t *v)
+{
+	int output;
+	__asm__ __volatile__ (
+		"1:	%0 = memw_locked(%1);\n"
+		"	%0 = sub(%0,%2);\n"
+		"	memw_locked(%1,P3)=%0\n"
+		"	if !P3 jump 1b;\n"
+		: "=&r" (output)
+		: "r" (&v->counter), "r" (i)
+		: "memory", "p3"
+	);
+	return output;
+}
+
+#define atomic_sub(i, v) atomic_sub_return(i, (v))
+
+/**
+ * atomic_add_unless - add unless the number is a given value
+ * @v: pointer to value
+ * @a: amount to add
+ * @u: unless value is equal to u
+ *
+ * Returns 1 if the add happened, 0 if it didn't.
+ */
+static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+{
+	int output;
+	asm volatile(
+		"1:	R15 = memw_locked(%1);"
+		"	{"
+		"		p3 = cmp.eq(R15, %3);"
+		"		if (p3.new) jump:nt 2f;"
+		"		R15 = add(R15,%2);"
+		"		%0 = #0;"
+		"	}"
+		"	memw_locked(%1,p3) = r15;"
+		"	{"
+		"		if !p3 jump 1b;"
+		"		%0 = #1;"
+		"	}"
+		"2:"
+		: "=&r" (output)
+		: "r" (v), "r" (a), "r" (u)
+		: "memory", "p3", "r15"
+	);
+	return output;
+}
+
+#define atomic_inc_not_zero(v) atomic_add_unless((v), 1, 0)
+
+#define atomic_inc(v) atomic_add(1, (v))
+#define atomic_dec(v) atomic_sub(1, (v))
+
+#define atomic_inc_and_test(v) (atomic_add_return(1, (v)) == 0)
+#define atomic_dec_and_test(v) (atomic_sub_return(1, (v)) == 0)
+#define atomic_sub_and_test(i, v) (atomic_sub_return(i, (v)) == 0)
+#define atomic_add_negative(i, v) (atomic_add_return(i, (v)) < 0)
+
+
+#define atomic_inc_return(v) (atomic_add_return(1, v))
+#define atomic_dec_return(v) (atomic_sub_return(1, v))
+
+#endif

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 05/36] Hexagon: Add syscalls

[Richard Kuo]

[-- Attachment #1: syscalls.diff --]
[-- Type: text/plain, Size: 8080 bytes --]

Removed sys_mmap* and replaced with sys_mmap_pgoff()


Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

---
 arch/hexagon/include/asm/syscall.h |   40 ++++++++++++++++
 arch/hexagon/include/asm/unistd.h  |   36 ++++++++++++++
 arch/hexagon/kernel/syscall.c      |   90 +++++++++++++++++++++++++++++++++++++
 arch/hexagon/kernel/syscalltab.c   |   32 +++++++++++++
 4 files changed, 198 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/syscall.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/syscall.h	2011-09-07 13:00:30.940646803 -0500
@@ -0,0 +1,40 @@
+/*
+ * Syscall support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_HEXAGON_SYSCALL_H
+#define _ASM_HEXAGON_SYSCALL_H
+
+typedef long (*syscall_fn)(unsigned long, unsigned long,
+	unsigned long, unsigned long,
+	unsigned long, unsigned long);
+
+asmlinkage int sys_execve(char __user *ufilename, char __user * __user *argv,
+			  char __user * __user *envp);
+asmlinkage int sys_clone(unsigned long clone_flags, unsigned long newsp,
+			 unsigned long parent_tidp, unsigned long child_tidp);
+
+#define sys_execve	sys_execve
+#define sys_clone	sys_clone
+
+#include <asm-generic/syscalls.h>
+
+extern void *sys_call_table[];
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/unistd.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/unistd.h	2011-09-07 13:00:31.010646910 -0500
@@ -0,0 +1,36 @@
+/*
+ * Syscall support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#if !defined(_ASM_HEXAGON_UNISTD_H) || defined(__SYSCALL)
+#define _ASM_HEXAGON_UNISTD_H
+
+/*
+ *  The kernel pulls this unistd.h in three different ways:
+ *  1.  the "normal" way which gets all the __NR defines
+ *  2.  with __SYSCALL defined to produce function declarations
+ *  3.  with __SYSCALL defined to produce syscall table initialization
+ *  See also:  syscalltab.c
+ */
+
+#define sys_mmap2 sys_mmap_pgoff
+
+#include <asm-generic/unistd.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/syscall.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/syscall.c	2011-09-07 13:00:31.150647121 -0500
@@ -0,0 +1,90 @@
+/*
+ * Hexagon system calls
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/linkage.h>
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/syscalls.h>
+#include <linux/unistd.h>
+#include <asm/mman.h>
+#include <asm/registers.h>
+
+/*
+ * System calls with architecture-specific wrappers.
+ * See signal.c for signal-related system call wrappers.
+ */
+
+asmlinkage int sys_execve(char __user *ufilename,
+			  const char __user *const __user *argv,
+			  const char __user *const __user *envp)
+{
+	struct pt_regs *pregs = current_thread_info()->regs;
+	char *filename;
+	int retval;
+
+	filename = getname(ufilename);
+	retval = PTR_ERR(filename);
+	if (IS_ERR(filename))
+		return retval;
+
+	retval = do_execve(filename, argv, envp, pregs);
+	putname(filename);
+
+	return retval;
+}
+
+asmlinkage int sys_clone(unsigned long clone_flags, unsigned long newsp,
+			 unsigned long parent_tidp, unsigned long child_tidp)
+{
+	struct pt_regs *pregs = current_thread_info()->regs;
+
+	if (!newsp)
+		newsp = pregs->SP;
+	return do_fork(clone_flags, newsp, pregs, 0, (int __user *)parent_tidp,
+		       (int __user *)child_tidp);
+}
+
+/*
+ * Do a system call from the kernel, so as to have a proper pt_regs
+ * and recycle the sys_execvpe infrustructure.
+ */
+int kernel_execve(const char *filename,
+		  const char *const argv[], const char *const envp[])
+{
+	register unsigned long __a0 asm("r0") = (unsigned long) filename;
+	register unsigned long __a1 asm("r1") = (unsigned long) argv;
+	register unsigned long __a2 asm("r2") = (unsigned long) envp;
+	int retval;
+
+	__asm__ volatile(
+		"	R6 = #%4;\n"
+		"	trap0(#1);\n"
+		"	%0 = R0;\n"
+		: "=r" (retval)
+		: "r" (__a0), "r" (__a1), "r" (__a2), "i" (__NR_execve)
+	);
+
+	return retval;
+}
+EXPORT_SYMBOL(kernel_execve);
Index: linux-hexagon-kernel/arch/hexagon/kernel/syscalltab.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/syscalltab.c	2011-09-03 20:14:59.524980831 -0500
@@ -0,0 +1,32 @@
+/*
+ * System call table for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/syscalls.h>
+#include <linux/signal.h>
+#include <linux/unistd.h>
+
+#include <asm/syscall.h>
+
+#undef __SYSCALL
+#define __SYSCALL(nr, call) [nr] = (call),
+
+void *sys_call_table[__NR_syscalls] = {
+#include <asm/unistd.h>
+};

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 05/36] Hexagon: Add syscalls

[Arnd Bergmann]

On Thursday 08 September 2011 20:08:52 Richard Kuo wrote:
> Removed sys_mmap* and replaced with sys_mmap_pgoff()
> 
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

Acked-by: Arnd Bergmann <arnd@arndb.de>

[patch v3 06/36] Hexagon: Add processor and system headers

[Richard Kuo]

[-- Attachment #1: processor.diff --]
[-- Type: text/plain, Size: 16852 bytes --]

Adds Hexagon information for processor/system type stuff
like context switching, register access, and irqflags.

Changed cpu_relax() to call __vmyield().

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/asm-offsets.h |    1 
 arch/hexagon/include/asm/irqflags.h    |   62 ++++++++
 arch/hexagon/include/asm/processor.h   |  123 +++++++++++++++++
 arch/hexagon/include/asm/registers.h   |  236 +++++++++++++++++++++++++++++++++
 arch/hexagon/include/asm/system.h      |  126 +++++++++++++++++
 5 files changed, 548 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/irqflags.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/irqflags.h	2011-09-03 20:14:59.534980859 -0500
@@ -0,0 +1,62 @@
+/*
+ * IRQ support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_IRQFLAGS_H
+#define _ASM_IRQFLAGS_H
+
+#include <asm/hexagon_vm.h>
+#include <linux/types.h>
+
+static inline unsigned long arch_local_save_flags(void)
+{
+	return __vmgetie();
+}
+
+static inline unsigned long arch_local_irq_save(void)
+{
+	return __vmsetie(VM_INT_DISABLE);
+}
+
+static inline bool arch_irqs_disabled_flags(unsigned long flags)
+{
+	return !flags;
+}
+
+static inline bool arch_irqs_disabled(void)
+{
+	return !__vmgetie();
+}
+
+static inline void arch_local_irq_enable(void)
+{
+	__vmsetie(VM_INT_ENABLE);
+}
+
+static inline void arch_local_irq_disable(void)
+{
+	__vmsetie(VM_INT_DISABLE);
+}
+
+static inline void arch_local_irq_restore(unsigned long flags)
+{
+	__vmsetie(flags);
+}
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/processor.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/processor.h	2011-09-07 13:00:30.940646803 -0500
@@ -0,0 +1,123 @@
+/*
+ * Process/processor support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_PROCESSOR_H
+#define _ASM_PROCESSOR_H
+
+#ifndef __ASSEMBLY__
+
+#include <asm/mem-layout.h>
+#include <asm/registers.h>
+#include <asm/hexagon_vm.h>
+
+/*  must be a macro  */
+#define current_text_addr() ({ __label__ _l; _l: &&_l; })
+
+/*  task_struct, defined elsewhere, is the "process descriptor" */
+struct task_struct;
+
+/*  this is defined in arch/process.c  */
+extern pid_t kernel_thread(int (*fn)(void *), void * arg, unsigned long flags);
+extern unsigned long thread_saved_pc(struct task_struct *tsk);
+
+extern void start_thread(struct pt_regs *, unsigned long, unsigned long);
+
+/*
+ * thread_struct is supposed to be for context switch data.
+ * Specifically, to hold the state necessary to perform switch_to...
+ */
+struct thread_struct {
+	void *switch_sp;
+};
+
+/*
+ * initializes thread_struct
+ * The only thing we have in there is switch_sp
+ * which doesn't really need to be initialized.
+ */
+
+#define INIT_THREAD { \
+}
+
+#define cpu_relax() __vmyield()
+
+/*
+ * "Unlazying all lazy status" occurs here.
+ */
+static inline void prepare_to_copy(struct task_struct *tsk)
+{
+}
+
+/*
+ * Decides where the kernel will search for a free chunk of vm space during
+ * mmaps.
+ * See also arch_get_unmapped_area.
+ * Doesn't affect if you have MAX_FIXED in the page flags set though...
+ *
+ * Apparently the convention is that ld.so will ask for "unmapped" private
+ * memory to be allocated SOMEWHERE, but it also asks for memory explicitly
+ * via MAP_FIXED at the lower * addresses starting at VA=0x0.
+ *
+ * If the two requests collide, you get authentic segfaulting action, so
+ * you have to kick the "unmapped" base requests higher up.
+ */
+#define TASK_UNMAPPED_BASE	(PAGE_ALIGN(TASK_SIZE/3))
+
+
+#define task_pt_regs(task) \
+	((struct pt_regs *)(task_stack_page(task) + THREAD_SIZE) - 1)
+
+#define KSTK_EIP(tsk) (pt_elr(task_pt_regs(tsk)))
+#define KSTK_ESP(tsk) (pt_psp(task_pt_regs(tsk)))
+
+/*  Free all resources held by a thread; defined in process.c  */
+extern void release_thread(struct task_struct *dead_task);
+
+/* Get wait channel for task P.  */
+extern unsigned long get_wchan(struct task_struct *p);
+
+/*  The following stuff is pretty HEXAGON specific.  */
+
+/*  This is really just here for __switch_to.
+    Offsets are pulled via asm-offsets.c  */
+
+/*
+ * No real reason why VM and native switch stacks should be different.
+ * Ultimately this should merge.  Note that Rev C. ABI called out only
+ * R24-27 as callee saved GPRs needing explicit attention (R29-31 being
+ * dealt with automagically by allocframe), but the current ABI has
+ * more, R16-R27.  By saving more, the worst case is that we waste some
+ * cycles if building with the old compilers.
+ */
+
+struct hexagon_switch_stack {
+	unsigned long long	r1716;
+	unsigned long long	r1918;
+	unsigned long long	r2120;
+	unsigned long long	r2322;
+	unsigned long long	r2524;
+	unsigned long long	r2726;
+	unsigned long		fp;
+	unsigned long		lr;
+};
+
+#endif /* !__ASSEMBLY__ */
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/registers.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/registers.h	2011-09-03 20:14:59.544980887 -0500
@@ -0,0 +1,236 @@
+/*
+ * Register definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_REGISTERS_H
+#define _ASM_REGISTERS_H
+
+#define SP r29
+
+#ifndef __ASSEMBLY__
+
+/*  See kernel/entry.S for further documentation.  */
+
+/*
+ * Entry code copies the event record out of guest registers into
+ * this structure (which is on the stack).
+ */
+
+struct hvm_event_record {
+	unsigned long vmel;     /* Event Linkage (return address) */
+	unsigned long vmest;    /* Event context - pre-event SSR values */
+	unsigned long vmpsp;    /* Previous stack pointer */
+	unsigned long vmbadva;  /* Bad virtual address for addressing events */
+};
+
+struct pt_regs {
+	long restart_r0;        /* R0 checkpoint for syscall restart */
+	long syscall_nr;        /* Only used in system calls */
+	union {
+		struct {
+			unsigned long usr;
+			unsigned long preds;
+		};
+		long long int predsusr;
+	};
+	union {
+		struct {
+			unsigned long m0;
+			unsigned long m1;
+		};
+		long long int m1m0;
+	};
+	union {
+		struct {
+			unsigned long sa1;
+			unsigned long lc1;
+		};
+		long long int lc1sa1;
+	};
+	union {
+		struct {
+			unsigned long sa0;
+			unsigned long lc0;
+		};
+		long long int lc0sa0;
+	};
+	union {
+		struct {
+			unsigned long gp;
+			unsigned long ugp;
+		};
+		long long int ugpgp;
+	};
+	/*
+	* Be extremely careful with rearranging these, if at all.  Some code
+	* assumes the 32 registers exist exactly like this in memory;
+	* e.g. kernel/ptrace.c
+	* e.g. kernel/signal.c (restore_sigcontext)
+	*/
+	union {
+		struct {
+			unsigned long r00;
+			unsigned long r01;
+		};
+		long long int r0100;
+	};
+	union {
+		struct {
+			unsigned long r02;
+			unsigned long r03;
+		};
+		long long int r0302;
+	};
+	union {
+		struct {
+			unsigned long r04;
+			unsigned long r05;
+		};
+		long long int r0504;
+	};
+	union {
+		struct {
+			unsigned long r06;
+			unsigned long r07;
+		};
+		long long int r0706;
+	};
+	union {
+		struct {
+			unsigned long r08;
+			unsigned long r09;
+		};
+		long long int r0908;
+	};
+	union {
+	       struct {
+			unsigned long r10;
+			unsigned long r11;
+	       };
+	       long long int r1110;
+	};
+	union {
+	       struct {
+			unsigned long r12;
+			unsigned long r13;
+	       };
+	       long long int r1312;
+	};
+	union {
+	       struct {
+			unsigned long r14;
+			unsigned long r15;
+	       };
+	       long long int r1514;
+	};
+	union {
+		struct {
+			unsigned long r16;
+			unsigned long r17;
+		};
+		long long int r1716;
+	};
+	union {
+		struct {
+			unsigned long r18;
+			unsigned long r19;
+		};
+		long long int r1918;
+	};
+	union {
+		struct {
+			unsigned long r20;
+			unsigned long r21;
+		};
+		long long int r2120;
+	};
+	union {
+		struct {
+			unsigned long r22;
+			unsigned long r23;
+		};
+		long long int r2322;
+	};
+	union {
+		struct {
+			unsigned long r24;
+			unsigned long r25;
+		};
+		long long int r2524;
+	};
+	union {
+		struct {
+			unsigned long r26;
+			unsigned long r27;
+		};
+		long long int r2726;
+	};
+	union {
+		struct {
+			unsigned long r28;
+			unsigned long r29;
+	       };
+	       long long int r2928;
+	};
+	union {
+		struct {
+			unsigned long r30;
+			unsigned long r31;
+		};
+		long long int r3130;
+	};
+	/* VM dispatch pushes event record onto stack - we can build on it */
+	struct hvm_event_record hvmer;
+};
+
+/* Defines to conveniently access the values  */
+
+/*
+ * As of the VM spec 0.5, these registers are now set/retrieved via a
+ * VM call.  On the in-bound side, we just fetch the values
+ * at the entry points and stuff them into the old record in pt_regs.
+ * However, on the outbound side, probably at VM rte, we set the
+ * registers back.
+ */
+
+#define pt_elr(regs) ((regs)->hvmer.vmel)
+#define pt_set_elr(regs, val) ((regs)->hvmer.vmel = (val))
+#define pt_cause(regs) ((regs)->hvmer.vmest & (HVM_VMEST_CAUSE_MSK))
+#define user_mode(regs) \
+	(((regs)->hvmer.vmest & (HVM_VMEST_UM_MSK << HVM_VMEST_UM_SFT)) != 0)
+#define ints_enabled(regs) \
+	(((regs)->hvmer.vmest & (HVM_VMEST_IE_MSK << HVM_VMEST_IE_SFT)) != 0)
+#define pt_psp(regs) ((regs)->hvmer.vmpsp)
+#define pt_badva(regs) ((regs)->hvmer.vmbadva)
+
+#define pt_set_rte_sp(regs, sp) do {\
+	pt_psp(regs) = (sp);\
+	(regs)->SP = (unsigned long) &((regs)->hvmer);\
+	} while (0)
+
+#define pt_set_kmode(regs) \
+	(regs)->hvmer.vmest = (HVM_VMEST_IE_MSK << HVM_VMEST_IE_SFT)
+
+#define pt_set_usermode(regs) \
+	(regs)->hvmer.vmest = (HVM_VMEST_UM_MSK << HVM_VMEST_UM_SFT) \
+			    | (HVM_VMEST_IE_MSK << HVM_VMEST_IE_SFT)
+
+#endif  /*  ifndef __ASSEMBLY  */
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/system.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/system.h	2011-09-03 20:14:59.544980887 -0500
@@ -0,0 +1,126 @@
+/*
+ * System level definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SYSTEM_H
+#define _ASM_SYSTEM_H
+
+#include <linux/linkage.h>
+#include <linux/irqflags.h>
+#include <asm/atomic.h>
+#include <asm/hexagon_vm.h>
+
+struct thread_struct;
+
+extern struct task_struct *__switch_to(struct task_struct *,
+	struct task_struct *,
+	struct task_struct *);
+
+#define switch_to(p, n, r) do {\
+	r = __switch_to((p), (n), (r));\
+} while (0)
+
+
+#define rmb()				barrier()
+#define read_barrier_depends()		barrier()
+#define wmb()				barrier()
+#define mb()				barrier()
+#define smp_rmb()			barrier()
+#define smp_read_barrier_depends()	barrier()
+#define smp_wmb()			barrier()
+#define smp_mb()			barrier()
+#define smp_mb__before_atomic_dec()	barrier()
+#define smp_mb__after_atomic_dec()	barrier()
+#define smp_mb__before_atomic_inc()	barrier()
+#define smp_mb__after_atomic_inc()	barrier()
+
+/*
+ * __xchg - atomically exchange a register and a memory location
+ * @x: value to swap
+ * @ptr: pointer to memory
+ * @size:  size of the value
+ *
+ * Only 4 bytes supported currently.
+ *
+ * Note:  there was an errata for V2 about .new's and memw_locked.
+ *
+ */
+static inline unsigned long __xchg(unsigned long x, volatile void *ptr,
+				   int size)
+{
+	unsigned long retval;
+
+	/*  Can't seem to use printk or panic here, so just stop  */
+	if (size != 4) do { asm volatile("brkpt;\n"); } while (1);
+
+	__asm__ __volatile__ (
+	"1:	%0 = memw_locked(%1);\n"    /*  load into retval */
+	"	memw_locked(%1,P0) = %2;\n" /*  store into memory */
+	"	if !P0 jump 1b;\n"
+	: "=&r" (retval)
+	: "r" (ptr), "r" (x)
+	: "memory", "p0"
+	);
+	return retval;
+}
+
+/*
+ * Atomically swap the contents of a register with memory.  Should be atomic
+ * between multiple CPU's and within interrupts on the same CPU.
+ */
+#define xchg(ptr, v) ((__typeof__(*(ptr)))__xchg((unsigned long)(v), (ptr), \
+	sizeof(*(ptr))))
+
+/*  Set a value and use a memory barrier.  Used by the scheduler somewhere.  */
+#define set_mb(var, value) \
+	do { var = value; mb(); } while (0)
+
+/*
+ *  see rt-mutex-design.txt; cmpxchg supposedly checks if *ptr == A and swaps.
+ *  looks just like atomic_cmpxchg on our arch currently with a bunch of
+ *  variable casting.
+ */
+#define __HAVE_ARCH_CMPXCHG 1
+
+#define cmpxchg(ptr, old, new)					\
+({								\
+	__typeof__(ptr) __ptr = (ptr);				\
+	__typeof__(*(ptr)) __old = (old);			\
+	__typeof__(*(ptr)) __new = (new);			\
+	__typeof__(*(ptr)) __oldval = 0;			\
+								\
+	asm volatile(						\
+		"1:	%0 = memw_locked(%1);\n"		\
+		"	{ P0 = cmp.eq(%0,%2);\n"		\
+		"	  if (!P0.new) jump:nt 2f; }\n"		\
+		"	memw_locked(%1,p0) = %3;\n"		\
+		"	if (!P0) jump 1b;\n"			\
+		"2:\n"						\
+		: "=&r" (__oldval)				\
+		: "r" (__ptr), "r" (__old), "r" (__new)		\
+		: "memory", "p0"				\
+	);							\
+	__oldval;						\
+})
+
+/*  Should probably shoot for an 8-byte aligned stack pointer  */
+#define STACK_MASK (~7)
+#define arch_align_stack(x) (x & STACK_MASK)
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/asm-offsets.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/asm-offsets.h	2011-09-03 20:14:59.544980887 -0500
@@ -0,0 +1 @@
+#include <generated/asm-offsets.h>

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 07/36] Hexagon: Add threadinfo

[Richard Kuo]

[-- Attachment #1: thread.diff --]
[-- Type: text/plain, Size: 6009 bytes --]

Removed some cruft and changed to allow kernel stack to be smaller than a page
size.

I left the thread_info alignment alone for the time being; may revisit that in
the future.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/thread_info.h |  154 +++++++++++++++++++++++++++++++++
 1 file changed, 154 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/thread_info.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/thread_info.h	2011-09-07 13:00:30.960646833 -0500
@@ -0,0 +1,154 @@
+/*
+ * Thread support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_THREAD_INFO_H
+#define _ASM_THREAD_INFO_H
+
+#ifdef __KERNEL__
+
+#ifndef __ASSEMBLY__
+#include <asm/processor.h>
+#include <asm/registers.h>
+#include <asm/page.h>
+#endif
+
+#define THREAD_SHIFT		12
+#define THREAD_SIZE		(1<<THREAD_SHIFT)
+
+#if THREAD_SHIFT >= PAGE_SHIFT
+#define THREAD_SIZE_ORDER	(THREAD_SHIFT - PAGE_SHIFT)
+#else  /*  don't use standard allocator  */
+#define __HAVE_ARCH_THREAD_INFO_ALLOCATOR
+extern struct thread_info *alloc_thread_info_node(struct task_struct *tsk, int node);
+extern void free_thread_info(struct thread_info *ti);
+#endif
+
+
+#ifndef __ASSEMBLY__
+
+typedef struct {
+	unsigned long seg;
+} mm_segment_t;
+
+/*
+ * This is union'd with the "bottom" of the kernel stack.
+ * It keeps track of thread info which is handy for routines
+ * to access quickly.
+ */
+
+struct thread_info {
+	struct task_struct	*task;		/* main task structure */
+	struct exec_domain      *exec_domain;   /* execution domain */
+	unsigned long		flags;          /* low level flags */
+	__u32                   cpu;            /* current cpu */
+	int                     preempt_count;  /* 0=>preemptible,<0=>BUG */
+	mm_segment_t            addr_limit;     /* segmentation sux */
+	/*
+	 * used for syscalls somehow;
+	 * seems to have a function pointer and four arguments
+	 */
+	struct restart_block    restart_block;
+	/* Points to the current pt_regs frame  */
+	struct pt_regs		*regs;
+	/*
+	 * saved kernel sp at switch_to time;
+	 * not sure if this is used (it's not in the VM model it seems;
+	 * see thread_struct)
+	 */
+	unsigned long		sp;
+};
+
+#else /* !__ASSEMBLY__ */
+
+#include <asm/asm-offsets.h>
+
+#endif  /* __ASSEMBLY__  */
+
+/*  looks like "linux/hardirq.h" uses this.  */
+
+#define PREEMPT_ACTIVE		0x10000000
+
+#ifndef __ASSEMBLY__
+
+#define INIT_THREAD_INFO(tsk)                   \
+{                                               \
+	.task           = &tsk,                 \
+	.exec_domain    = &default_exec_domain, \
+	.flags          = 0,                    \
+	.cpu            = 0,                    \
+	.preempt_count  = 1,                    \
+	.addr_limit     = KERNEL_DS,            \
+	.restart_block = {                      \
+		.fn = do_no_restart_syscall,    \
+	},                                      \
+	.sp = 0,				\
+	.regs = NULL,			\
+}
+
+#define init_thread_info        (init_thread_union.thread_info)
+#define init_stack              (init_thread_union.stack)
+
+/* Tacky preprocessor trickery */
+#define	qqstr(s) qstr(s)
+#define qstr(s) #s
+#define QUOTED_THREADINFO_REG qqstr(THREADINFO_REG)
+
+register struct thread_info *__current_thread_info asm(QUOTED_THREADINFO_REG);
+#define current_thread_info()  __current_thread_info
+
+#endif /* __ASSEMBLY__ */
+
+/*
+ * thread information flags
+ * - these are process state flags that various assembly files
+ *   may need to access
+ * - pending work-to-be-done flags are in LSW
+ * - other flags in MSW
+ */
+
+#define TIF_SYSCALL_TRACE       0       /* syscall trace active */
+#define TIF_NOTIFY_RESUME       1       /* resumption notification requested */
+#define TIF_SIGPENDING          2       /* signal pending */
+#define TIF_NEED_RESCHED        3       /* rescheduling necessary */
+#define TIF_SINGLESTEP          4       /* restore ss @ return to usr mode */
+#define TIF_IRET                5       /* return with iret */
+#define TIF_RESTORE_SIGMASK     6       /* restore sig mask in do_signal() */
+/* true if poll_idle() is polling TIF_NEED_RESCHED */
+#define TIF_POLLING_NRFLAG      16
+#define TIF_MEMDIE              17      /* OOM killer killed process */
+
+#define _TIF_SYSCALL_TRACE      (1 << TIF_SYSCALL_TRACE)
+#define _TIF_NOTIFY_RESUME      (1 << TIF_NOTIFY_RESUME)
+#define _TIF_SIGPENDING         (1 << TIF_SIGPENDING)
+#define _TIF_NEED_RESCHED       (1 << TIF_NEED_RESCHED)
+#define _TIF_SINGLESTEP         (1 << TIF_SINGLESTEP)
+#define _TIF_IRET               (1 << TIF_IRET)
+#define _TIF_RESTORE_SIGMASK    (1 << TIF_RESTORE_SIGMASK)
+#define _TIF_POLLING_NRFLAG     (1 << TIF_POLLING_NRFLAG)
+
+/* work to do on interrupt/exception return - All but TIF_SYSCALL_TRACE */
+#define _TIF_WORK_MASK          (0x0000FFFF & ~_TIF_SYSCALL_TRACE)
+
+/* work to do on any return to u-space */
+#define _TIF_ALLWORK_MASK       0x0000FFFF
+
+#endif /* __KERNEL__ */
+
+#endif

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 08/36] Hexagon: Add delay functions

[Richard Kuo]

[-- Attachment #1: delay.diff --]
[-- Type: text/plain, Size: 1708 bytes --]

Added read_current_timer elsewhere.

udelay is based off of __vmgettime() which returns processor ticks.

Removed __delay and CONFIG_GENERIC_CALIBRATE_DELAY.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

---
 arch/hexagon/include/asm/delay.h |   28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/delay.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/delay.h	2011-09-07 13:00:30.900646743 -0500
@@ -0,0 +1,28 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_DELAY_H
+#define _ASM_DELAY_H
+
+#include <asm/param.h>
+
+extern void __udelay(unsigned long usecs);
+
+#define udelay(usecs) __udelay((usecs))
+
+#endif /* _ASM_DELAY_H */

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 08/36] Hexagon: Add delay functions

[Arnd Bergmann]

On Thursday 08 September 2011 20:08:55 Richard Kuo wrote:
> Added read_current_timer elsewhere.
> 
> udelay is based off of __vmgettime() which returns processor ticks.
> 
> Removed __delay and CONFIG_GENERIC_CALIBRATE_DELAY.
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

Acked-by: Arnd Bergmann <arnd@arndb.de>

[patch v3 09/36] Hexagon: Add checksum functions

[Richard Kuo]

[-- Attachment #1: checksum.diff --]
[-- Type: text/plain, Size: 10138 bytes --]

Removed most non-assembly functions and use the generic ones instead.

Made a few more minor changes to use the generic header file.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/checksum.h   |   50 ++++++++
 arch/hexagon/include/asm/intrinsics.h |   26 ++++
 arch/hexagon/lib/checksum.c           |  203 ++++++++++++++++++++++++++++++++++
 3 files changed, 279 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/checksum.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/checksum.h	2011-09-03 20:14:59.564980943 -0500
@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_CHECKSUM_H
+#define _ASM_CHECKSUM_H
+
+#define do_csum	do_csum
+unsigned int do_csum(const void *voidptr, int len);
+
+/*
+ * the same as csum_partial, but copies from src while it
+ * checksums
+ *
+ * here even more important to align src and dst on a 32-bit (or even
+ * better 64-bit) boundary
+ */
+#define csum_partial_copy_nocheck csum_partial_copy_nocheck
+__wsum csum_partial_copy_nocheck(const void *src, void *dst,
+					int len, __wsum sum);
+
+/*
+ * computes the checksum of the TCP/UDP pseudo-header
+ * returns a 16-bit checksum, already complemented
+ */
+#define csum_tcpudp_nofold csum_tcpudp_nofold
+__wsum csum_tcpudp_nofold(unsigned long saddr, unsigned long daddr,
+	unsigned short len, unsigned short proto, __wsum sum);
+
+#define csum_tcpudp_magic csum_tcpudp_magic
+__sum16 csum_tcpudp_magic(unsigned long saddr, unsigned long daddr,
+	unsigned short len, unsigned short proto, __wsum sum);
+
+#include <asm-generic/checksum.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/intrinsics.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/intrinsics.h	2011-09-03 20:14:59.574980971 -0500
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_HEXAGON_INTRINSICS_H
+#define _ASM_HEXAGON_INTRINSICS_H
+
+#define HEXAGON_P_vrmpyhacc_PP	__builtin_HEXAGON_M2_vrmac_s0
+#define HEXAGON_P_vrmpyh_PP	__builtin_HEXAGON_M2_vrmpy_s0
+#define HEXAGON_R_cl0_R		__builtin_HEXAGON_S2_cl0
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/lib/checksum.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/lib/checksum.c	2011-09-03 20:14:59.574980971 -0500
@@ -0,0 +1,203 @@
+/*
+ * Checksum functions for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*  This was derived from arch/alpha/lib/checksum.c  */
+
+
+#include <linux/module.h>
+#include <linux/string.h>
+
+#include <asm/byteorder.h>
+#include <net/checksum.h>
+#include <linux/uaccess.h>
+#include <asm/intrinsics.h>
+
+
+/*  Vector value operations  */
+#define SIGN(x, y)	((0x8000ULL*x)<<y)
+#define CARRY(x, y)	((0x0002ULL*x)<<y)
+#define SELECT(x, y)	((0x0001ULL*x)<<y)
+
+#define VR_NEGATE(a, b, c, d)	(SIGN(a, 48) + SIGN(b, 32) + SIGN(c, 16) \
+	+ SIGN(d, 0))
+#define VR_CARRY(a, b, c, d)	(CARRY(a, 48) + CARRY(b, 32) + CARRY(c, 16) \
+	+ CARRY(d, 0))
+#define VR_SELECT(a, b, c, d)	(SELECT(a, 48) + SELECT(b, 32) + SELECT(c, 16) \
+	+ SELECT(d, 0))
+
+
+/* optimized HEXAGON V3 intrinsic version */
+static inline unsigned short from64to16(u64 x)
+{
+	u64 sum;
+
+	sum = HEXAGON_P_vrmpyh_PP(x^VR_NEGATE(1, 1, 1, 1),
+			     VR_SELECT(1, 1, 1, 1));
+	sum += VR_CARRY(0, 0, 1, 0);
+	sum = HEXAGON_P_vrmpyh_PP(sum, VR_SELECT(0, 0, 1, 1));
+
+	return 0xFFFF & sum;
+}
+
+/*
+ * computes the checksum of the TCP/UDP pseudo-header
+ * returns a 16-bit checksum, already complemented.
+ */
+__sum16 csum_tcpudp_magic(unsigned long saddr, unsigned long daddr,
+			  unsigned short len, unsigned short proto,
+			  __wsum sum)
+{
+	return (__force __sum16)~from64to16(
+		(__force u64)saddr + (__force u64)daddr +
+		(__force u64)sum + ((len + proto) << 8));
+}
+
+__wsum csum_tcpudp_nofold(unsigned long saddr, unsigned long daddr,
+			  unsigned short len, unsigned short proto,
+			  __wsum sum)
+{
+	u64 result;
+
+	result = (__force u64)saddr + (__force u64)daddr +
+		 (__force u64)sum + ((len + proto) << 8);
+
+	/* Fold down to 32-bits so we don't lose in the typedef-less
+	   network stack.  */
+	/* 64 to 33 */
+	result = (result & 0xffffffffUL) + (result >> 32);
+	/* 33 to 32 */
+	result = (result & 0xffffffffUL) + (result >> 32);
+	return (__force __wsum)result;
+}
+EXPORT_SYMBOL(csum_tcpudp_nofold);
+
+/*
+ * Do a 64-bit checksum on an arbitrary memory area..
+ *
+ * This isn't a great routine, but it's not _horrible_ either. The
+ * inner loop could be unrolled a bit further, and there are better
+ * ways to do the carry, but this is reasonable.
+ */
+
+/* optimized HEXAGON intrinsic version, with over read fixed */
+unsigned int do_csum(const void *voidptr, int len)
+{
+	u64 sum0, sum1, x0, x1, *ptr8_o, *ptr8_e, *ptr8;
+	int i, start, mid, end, mask;
+	const char *ptr = voidptr;
+	unsigned short *ptr2;
+	unsigned int *ptr4;
+
+	if (len <= 0)
+		return 0;
+
+	start = 0xF & (16-(((int) ptr) & 0xF)) ;
+	mask  = 0x7fffffffUL >> HEXAGON_R_cl0_R(len);
+	start = start & mask ;
+
+	mid = len - start;
+	end = mid & 0xF;
+	mid = mid>>4;
+	sum0 = mid << 18;
+	sum1 = 0;
+
+	if (start & 1)
+		sum0 += (u64) (ptr[0] << 8);
+	ptr2 = (unsigned short *) &ptr[start & 1];
+	if (start & 2)
+		sum1 += (u64) ptr2[0];
+	ptr4 = (unsigned int *) &ptr[start & 3];
+	if (start & 4) {
+		sum0 = HEXAGON_P_vrmpyhacc_PP(sum0,
+			VR_NEGATE(0, 0, 1, 1)^((u64)ptr4[0]),
+			VR_SELECT(0, 0, 1, 1));
+		sum0 += VR_SELECT(0, 0, 1, 0);
+	}
+	ptr8 = (u64 *) &ptr[start & 7];
+	if (start & 8) {
+		sum1 = HEXAGON_P_vrmpyhacc_PP(sum1,
+			VR_NEGATE(1, 1, 1, 1)^(ptr8[0]),
+			VR_SELECT(1, 1, 1, 1));
+		sum1 += VR_CARRY(0, 0, 1, 0);
+	}
+	ptr8_o = (u64 *) (ptr + start);
+	ptr8_e = (u64 *) (ptr + start + 8);
+
+	if (mid) {
+		x0 = *ptr8_e; ptr8_e += 2;
+		x1 = *ptr8_o; ptr8_o += 2;
+		if (mid > 1)
+			for (i = 0; i < mid-1; i++) {
+				sum0 = HEXAGON_P_vrmpyhacc_PP(sum0,
+					x0^VR_NEGATE(1, 1, 1, 1),
+					VR_SELECT(1, 1, 1, 1));
+				sum1 = HEXAGON_P_vrmpyhacc_PP(sum1,
+					x1^VR_NEGATE(1, 1, 1, 1),
+					VR_SELECT(1, 1, 1, 1));
+				x0 = *ptr8_e; ptr8_e += 2;
+				x1 = *ptr8_o; ptr8_o += 2;
+			}
+		sum0 = HEXAGON_P_vrmpyhacc_PP(sum0, x0^VR_NEGATE(1, 1, 1, 1),
+			VR_SELECT(1, 1, 1, 1));
+		sum1 = HEXAGON_P_vrmpyhacc_PP(sum1, x1^VR_NEGATE(1, 1, 1, 1),
+			VR_SELECT(1, 1, 1, 1));
+	}
+
+	ptr4 = (unsigned int *) &ptr[start + (mid * 16) + (end & 8)];
+	if (end & 4) {
+		sum1 = HEXAGON_P_vrmpyhacc_PP(sum1,
+			VR_NEGATE(0, 0, 1, 1)^((u64)ptr4[0]),
+			VR_SELECT(0, 0, 1, 1));
+		sum1 += VR_SELECT(0, 0, 1, 0);
+	}
+	ptr2 = (unsigned short *) &ptr[start + (mid * 16) + (end & 12)];
+	if (end & 2)
+		sum0 += (u64) ptr2[0];
+
+	if (end & 1)
+		sum1 += (u64) ptr[start + (mid * 16) + (end & 14)];
+
+	ptr8 = (u64 *) &ptr[start + (mid * 16)];
+	if (end & 8) {
+		sum0 = HEXAGON_P_vrmpyhacc_PP(sum0,
+			VR_NEGATE(1, 1, 1, 1)^(ptr8[0]),
+			VR_SELECT(1, 1, 1, 1));
+		sum0 += VR_CARRY(0, 0, 1, 0);
+	}
+	sum0 = HEXAGON_P_vrmpyh_PP((sum0+sum1)^VR_NEGATE(0, 0, 0, 1),
+		VR_SELECT(0, 0, 1, 1));
+	sum0 += VR_NEGATE(0, 0, 0, 1);
+	sum0 = HEXAGON_P_vrmpyh_PP(sum0, VR_SELECT(0, 0, 1, 1));
+
+	if (start & 1)
+		sum0 = (sum0 << 8) | (0xFF & (sum0 >> 8));
+
+	return 0xFFFF & sum0;
+}
+
+/*
+ * copy from ds while checksumming, otherwise like csum_partial
+ */
+__wsum
+csum_partial_copy_nocheck(const void *src, void *dst, int len, __wsum sum)
+{
+	memcpy(dst, src, len);
+	return csum_partial(dst, len, sum);
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 10/36] Hexagon: Add memcpy and memset accelerated functions

[Richard Kuo]

[-- Attachment #1: memcpy.diff --]
[-- Type: text/plain, Size: 23124 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/lib/memcpy.S |  543 ++++++++++++++++++++++++++++++++++++++++++++++
 arch/hexagon/lib/memset.S |  316 ++++++++++++++++++++++++++
 2 files changed, 859 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/lib/memcpy.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/lib/memcpy.S	2011-09-03 20:14:59.584980999 -0500
@@ -0,0 +1,543 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * Description
+ *
+ *   library function for memcpy where length bytes are copied from
+ *   ptr_in to ptr_out. ptr_out is returned unchanged.
+ *   Allows any combination of alignment on input and output pointers
+ *   and length from 0 to 2^32-1
+ *
+ * Restrictions
+ *   The arrays should not overlap, the program will produce undefined output
+ *   if they do.
+ *   For blocks less than 16 bytes a byte by byte copy is performed. For
+ *   8byte alignments, and length multiples, a dword copy is performed up to
+ *   96bytes
+ * History
+ *
+ *   DJH  5/15/09 Initial version 1.0
+ *   DJH  6/ 1/09 Version 1.1 modified ABI to inlcude R16-R19
+ *   DJH  7/12/09 Version 1.2 optimized codesize down to 760 was 840
+ *   DJH 10/14/09 Version 1.3 added special loop for aligned case, was
+ *                            overreading bloated codesize back up to 892
+ *   DJH  4/20/10 Version 1.4 fixed Ldword_loop_epilog loop to prevent loads
+ *                            occuring if only 1 left outstanding, fixes bug
+ *                            # 3888, corrected for all alignments. Peeled off
+ *                            1 32byte chunk from kernel loop and extended 8byte
+ *                            loop at end to solve all combinations and prevent
+ *                            over read.  Fixed Ldword_loop_prolog to prevent
+ *                            overread for blocks less than 48bytes. Reduced
+ *                            codesize to 752 bytes
+ *   DJH  4/21/10 version 1.5 1.4 fix broke code for input block ends not
+ *                            aligned to dword boundaries,underwriting by 1
+ *                            byte, added detection for this and fixed. A
+ *                            little bloat.
+ *   DJH  4/23/10 version 1.6 corrected stack error, R20 was not being restored
+ *                            always, fixed the error of R20 being modified
+ *                            before it was being saved
+ * Natural c model
+ * ===============
+ * void * memcpy(char * ptr_out, char * ptr_in, int length) {
+ *   int i;
+ *   if(length) for(i=0; i < length; i++) { ptr_out[i] = ptr_in[i]; }
+ *   return(ptr_out);
+ * }
+ *
+ * Optimized memcpy function
+ * =========================
+ * void * memcpy(char * ptr_out, char * ptr_in, int len) {
+ *   int i, prolog, kernel, epilog, mask;
+ *   u8 offset;
+ *   s64 data0, dataF8, data70;
+ *
+ *   s64 * ptr8_in;
+ *   s64 * ptr8_out;
+ *   s32 * ptr4;
+ *   s16 * ptr2;
+ *
+ *   offset = ((int) ptr_in) & 7;
+ *   ptr8_in = (s64 *) &ptr_in[-offset];   //read in the aligned pointers
+ *
+ *   data70 = *ptr8_in++;
+ *   dataF8 = *ptr8_in++;
+ *
+ *   data0 = HEXAGON_P_valignb_PPp(dataF8, data70, offset);
+ *
+ *   prolog = 32 - ((int) ptr_out);
+ *   mask  = 0x7fffffff >> HEXAGON_R_cl0_R(len);
+ *   prolog = prolog & mask;
+ *   kernel = len - prolog;
+ *   epilog = kernel & 0x1F;
+ *   kernel = kernel>>5;
+ *
+ *   if (prolog & 1) { ptr_out[0] = (u8) data0; data0 >>= 8; ptr_out += 1;}
+ *   ptr2 = (s16 *) &ptr_out[0];
+ *   if (prolog & 2) { ptr2[0] = (u16) data0;  data0 >>= 16; ptr_out += 2;}
+ *   ptr4 = (s32 *) &ptr_out[0];
+ *   if (prolog & 4) { ptr4[0] = (u32) data0;  data0 >>= 32; ptr_out += 4;}
+ *
+ *   offset = offset + (prolog & 7);
+ *   if (offset >= 8) {
+ *     data70 = dataF8;
+ *     dataF8 = *ptr8_in++;
+ *   }
+ *   offset = offset & 0x7;
+ *
+ *   prolog = prolog >> 3;
+ *   if (prolog) for (i=0; i < prolog; i++) {
+ *       data0 = HEXAGON_P_valignb_PPp(dataF8, data70, offset);
+ *       ptr8_out = (s64 *) &ptr_out[0]; *ptr8_out = data0; ptr_out += 8;
+ *       data70 = dataF8;
+ *       dataF8 = *ptr8_in++;
+ *   }
+ *   if(kernel) { kernel -= 1; epilog += 32; }
+ *   if(kernel) for(i=0; i < kernel; i++) {
+ *       data0 = HEXAGON_P_valignb_PPp(dataF8, data70, offset);
+ *       ptr8_out = (s64 *) &ptr_out[0]; *ptr8_out = data0; ptr_out += 8;
+ *       data70 = *ptr8_in++;
+ *
+ *       data0 = HEXAGON_P_valignb_PPp(data70, dataF8, offset);
+ *       ptr8_out = (s64 *) &ptr_out[0]; *ptr8_out = data0; ptr_out += 8;
+ *       dataF8 = *ptr8_in++;
+ *
+ *       data0 = HEXAGON_P_valignb_PPp(dataF8, data70, offset);
+ *       ptr8_out = (s64 *) &ptr_out[0]; *ptr8_out = data0; ptr_out += 8;
+ *       data70 = *ptr8_in++;
+ *
+ *       data0 = HEXAGON_P_valignb_PPp(data70, dataF8, offset);
+ *       ptr8_out = (s64 *) &ptr_out[0]; *ptr8_out = data0; ptr_out += 8;
+ *       dataF8 = *ptr8_in++;
+ *   }
+ *   epilogdws = epilog >> 3;
+ *   if (epilogdws) for (i=0; i < epilogdws; i++) {
+ *       data0 = HEXAGON_P_valignb_PPp(dataF8, data70, offset);
+ *       ptr8_out = (s64 *) &ptr_out[0]; *ptr8_out = data0; ptr_out += 8;
+ *       data70 = dataF8;
+ *       dataF8 = *ptr8_in++;
+ *   }
+ *   data0 = HEXAGON_P_valignb_PPp(dataF8, data70, offset);
+ *
+ *   ptr4 = (s32 *) &ptr_out[0];
+ *   if (epilog & 4) { ptr4[0] = (u32) data0; data0 >>= 32; ptr_out += 4;}
+ *   ptr2 = (s16 *) &ptr_out[0];
+ *   if (epilog & 2) { ptr2[0] = (u16) data0; data0 >>= 16; ptr_out += 2;}
+ *   if (epilog & 1) { *ptr_out++ = (u8) data0; }
+ *
+ *   return(ptr_out - length);
+ * }
+ *
+ * Codesize : 784 bytes
+ */
+
+
+#define ptr_out		R0	/*  destination  pounter  */
+#define ptr_in		R1	/*  source pointer  */
+#define len		R2	/*  length of copy in bytes  */
+
+#define data70		R13:12	/*  lo 8 bytes of non-aligned transfer  */
+#define dataF8		R11:10	/*  hi 8 bytes of non-aligned transfer  */
+#define ldata0		R7:6	/*  even 8 bytes chunks  */
+#define ldata1		R25:24	/*  odd 8 bytes chunks  */
+#define data1		R7	/*  lower 8 bytes of ldata1  */
+#define data0		R6	/*  lower 8 bytes of ldata0  */
+
+#define ifbyte		p0	/*  if transfer has bytes in epilog/prolog  */
+#define ifhword		p0	/*  if transfer has shorts in epilog/prolog  */
+#define ifword		p0	/*  if transfer has words in epilog/prolog  */
+#define noprolog	p0	/*  no prolog, xfer starts at 32byte  */
+#define nokernel	p1	/*  no 32byte multiple block in the transfer  */
+#define noepilog	p0	/*  no epilog, xfer ends on 32byte boundary  */
+#define align		p2	/*  alignment of input rel to 8byte boundary  */
+#define kernel1		p0	/*  kernel count == 1  */
+
+#define dalign		R25	/*  rel alignment of input to output data  */
+#define star3		R16	/*  number bytes in prolog - dwords  */
+#define rest		R8	/*  length - prolog bytes  */
+#define back		R7	/*  nr bytes > dword boundary in src block  */
+#define epilog		R3	/*  bytes in epilog  */
+#define inc		R15:14	/*  inc kernel by -1 and defetch ptr by 32  */
+#define kernel		R4	/*  number of 32byte chunks in kernel  */
+#define ptr_in_p_128	R5	/*  pointer for prefetch of input data  */
+#define mask		R8	/*  mask used to determine prolog size  */
+#define shift		R8	/*  used to work a shifter to extract bytes  */
+#define shift2		R5	/*  in epilog to workshifter to extract bytes */
+#define prolog		R15	/*  bytes in  prolog  */
+#define epilogdws	R15	/*  number dwords in epilog  */
+#define shiftb		R14	/*  used to extract bytes  */
+#define offset		R9	/*  same as align in reg  */
+#define ptr_out_p_32	R17	/*  pointer to output dczero  */
+#define align888	R14	/*  if simple dword loop can be used  */
+#define len8		R9	/*  number of dwords in length  */
+#define over		R20	/*  nr of bytes > last inp buf dword boundary */
+
+#define ptr_in_p_128kernel	R5:4	/*  packed fetch pointer & kernel cnt */
+
+	.section .text
+	.p2align 4
+        .global memcpy
+        .type memcpy, @function
+memcpy:
+{
+	p2 = cmp.eq(len, #0);		/*  =0 */
+	align888 = or(ptr_in, ptr_out);	/*  %8 < 97 */
+	p0 = cmp.gtu(len, #23);		/*  %1, <24 */
+	p1 = cmp.eq(ptr_in, ptr_out);	/*  attempt to overwrite self */
+}
+{
+	p1 = or(p2, p1);
+	p3 = cmp.gtu(len, #95);		/*  %8 < 97 */
+	align888 = or(align888, len);	/*  %8 < 97 */
+	len8 = lsr(len, #3);		/*  %8 < 97 */
+}
+{
+	dcfetch(ptr_in);		/*  zero/ptrin=ptrout causes fetch */
+	p2 = bitsclr(align888, #7);	/*  %8 < 97  */
+	if(p1) jumpr r31;		/*  =0  */
+}
+{
+	p2 = and(p2,!p3);			/*  %8 < 97  */
+	if (p2.new) len = add(len, #-8);	/*  %8 < 97  */
+	if (p2.new) jump:NT .Ldwordaligned; 	/*  %8 < 97  */
+}
+{
+	if(!p0) jump .Lbytes23orless;	/*  %1, <24  */
+	mask.l = #LO(0x7fffffff);
+	/*  all bytes before line multiples of data  */
+	prolog = sub(#0, ptr_out);
+}
+{
+	/*  save r31 on stack, decrement sp by 16  */
+	allocframe(#24);
+	mask.h = #HI(0x7fffffff);
+	ptr_in_p_128 = add(ptr_in, #32);
+	back = cl0(len);
+}
+{
+	memd(sp+#0) = R17:16;		/*  save r16,r17 on stack6  */
+	r31.l = #LO(.Lmemcpy_return);	/*  set up final return pointer  */
+	prolog &= lsr(mask, back);
+	offset = and(ptr_in, #7);
+}
+{
+	memd(sp+#8) = R25:24;		/*  save r25,r24 on stack  */
+	dalign = sub(ptr_out, ptr_in);
+	r31.h = #HI(.Lmemcpy_return);	/*  set up final return pointer  */
+}
+{
+	/*  see if there if input buffer end if aligned  */
+	over = add(len, ptr_in);
+	back = add(len, offset);
+	memd(sp+#16) = R21:20;		/*  save r20,r21 on stack  */
+}
+{
+	noprolog = bitsclr(prolog, #7);
+	prolog = and(prolog, #31);
+	dcfetch(ptr_in_p_128);
+	ptr_in_p_128 = add(ptr_in_p_128, #32);
+}
+{
+	kernel = sub(len, prolog);
+	shift = asl(prolog, #3);
+	star3 = and(prolog, #7);
+	ptr_in = and(ptr_in, #-8);
+}
+{
+	prolog = lsr(prolog, #3);
+	epilog = and(kernel, #31);
+	ptr_out_p_32 = add(ptr_out, prolog);
+	over = and(over, #7);
+}
+{
+	p3 = cmp.gtu(back, #8);
+	kernel = lsr(kernel, #5);
+	dcfetch(ptr_in_p_128);
+	ptr_in_p_128 = add(ptr_in_p_128, #32);
+}
+{
+	p1 = cmp.eq(prolog, #0);
+	if(!p1.new) prolog = add(prolog, #1);
+	dcfetch(ptr_in_p_128);	/*  reserve the line 64bytes on  */
+	ptr_in_p_128 = add(ptr_in_p_128, #32);
+}
+{
+	nokernel = cmp.eq(kernel,#0);
+	dcfetch(ptr_in_p_128);	/* reserve the line 64bytes on  */
+	ptr_in_p_128 = add(ptr_in_p_128, #32);
+	shiftb = and(shift, #8);
+}
+{
+	dcfetch(ptr_in_p_128);		/*  reserve the line 64bytes on  */
+	ptr_in_p_128 = add(ptr_in_p_128, #32);
+	if(nokernel) jump .Lskip64;
+	p2 = cmp.eq(kernel, #1);	/*  skip ovr if kernel == 0  */
+}
+{
+	dczeroa(ptr_out_p_32);
+	/*  don't advance pointer  */
+	if(!p2) ptr_out_p_32 = add(ptr_out_p_32, #32);
+}
+{
+	dalign = and(dalign, #31);
+	dczeroa(ptr_out_p_32);
+}
+.Lskip64:
+{
+	data70 = memd(ptr_in++#16);
+	if(p3) dataF8 = memd(ptr_in+#8);
+	if(noprolog) jump .Lnoprolog32;
+	align = offset;
+}
+/*  upto initial 7 bytes  */
+{
+	ldata0 = valignb(dataF8, data70, align);
+	ifbyte = tstbit(shift,#3);
+	offset = add(offset, star3);
+}
+{
+	if(ifbyte) memb(ptr_out++#1) = data0;
+	ldata0 = lsr(ldata0, shiftb);
+	shiftb = and(shift, #16);
+	ifhword = tstbit(shift,#4);
+}
+{
+	if(ifhword) memh(ptr_out++#2) = data0;
+	ldata0 = lsr(ldata0, shiftb);
+	ifword = tstbit(shift,#5);
+	p2 = cmp.gtu(offset, #7);
+}
+{
+	if(ifword) memw(ptr_out++#4) = data0;
+	if(p2) data70 = dataF8;
+	if(p2) dataF8 = memd(ptr_in++#8);	/*  another 8 bytes  */
+	align = offset;
+}
+.Lnoprolog32:
+{
+	p3 = sp1loop0(.Ldword_loop_prolog, prolog)
+	rest = sub(len, star3);	/*  whats left after the loop  */
+	p0 = cmp.gt(over, #0);
+}
+	if(p0) rest = add(rest, #16);
+.Ldword_loop_prolog:
+{
+	if(p3) memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(dataF8, data70, align);
+	p0 = cmp.gt(rest, #16);
+}
+{
+	data70 = dataF8;
+	if(p0) dataF8 = memd(ptr_in++#8);
+	rest = add(rest, #-8);
+}:endloop0
+.Lkernel:
+{
+	/*  kernel is at least 32bytes  */
+	p3 = cmp.gtu(kernel, #0);
+	/*  last itn. remove edge effects  */
+	if(p3.new) kernel = add(kernel, #-1);
+	/*  dealt with in last dword loop  */
+	if(p3.new) epilog = add(epilog, #32);
+}
+{
+	nokernel = cmp.eq(kernel, #0);		/*  after adjustment, recheck */
+	if(nokernel.new) jump:NT .Lepilog;	/*  likely not taken  */
+	inc = combine(#32, #-1);
+	p3 = cmp.gtu(dalign, #24);
+}
+{
+	if(p3) jump .Lodd_alignment;
+}
+{
+	loop0(.Loword_loop_25to31, kernel);
+	kernel1 = cmp.gtu(kernel, #1);
+	rest = kernel;
+}
+	.falign
+.Loword_loop_25to31:
+{
+	dcfetch(ptr_in_p_128);	/*  prefetch 4 lines ahead  */
+	if(kernel1) ptr_out_p_32 = add(ptr_out_p_32, #32);
+}
+{
+	dczeroa(ptr_out_p_32);	/*  reserve the next 32bytes in cache  */
+	p3 = cmp.eq(kernel, rest);
+}
+{
+	/*  kernel -= 1  */
+	ptr_in_p_128kernel = vaddw(ptr_in_p_128kernel, inc);
+	/*  kill write on first iteration  */
+	if(!p3) memd(ptr_out++#8) = ldata1;
+	ldata1 = valignb(dataF8, data70, align);
+	data70 = memd(ptr_in++#8);
+}
+{
+	memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(data70, dataF8, align);
+	dataF8 = memd(ptr_in++#8);
+}
+{
+	memd(ptr_out++#8) = ldata1;
+	ldata1 = valignb(dataF8, data70, align);
+	data70 = memd(ptr_in++#8);
+}
+{
+	memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(data70, dataF8, align);
+	dataF8 = memd(ptr_in++#8);
+	kernel1 = cmp.gtu(kernel, #1);
+}:endloop0
+{
+	memd(ptr_out++#8) = ldata1;
+	jump .Lepilog;
+}
+.Lodd_alignment:
+{
+	loop0(.Loword_loop_00to24, kernel);
+	kernel1 = cmp.gtu(kernel, #1);
+	rest = add(kernel, #-1);
+}
+	.falign
+.Loword_loop_00to24:
+{
+	dcfetch(ptr_in_p_128);	/*  prefetch 4 lines ahead  */
+	ptr_in_p_128kernel = vaddw(ptr_in_p_128kernel, inc);
+	if(kernel1) ptr_out_p_32 = add(ptr_out_p_32, #32);
+}
+{
+	dczeroa(ptr_out_p_32);	/*  reserve the next 32bytes in cache  */
+}
+{
+	memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(dataF8, data70, align);
+	data70 = memd(ptr_in++#8);
+}
+{
+	memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(data70, dataF8, align);
+	dataF8 = memd(ptr_in++#8);
+}
+{
+	memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(dataF8, data70, align);
+	data70 = memd(ptr_in++#8);
+}
+{
+	memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(data70, dataF8, align);
+	dataF8 = memd(ptr_in++#8);
+	kernel1 = cmp.gtu(kernel, #1);
+}:endloop0
+.Lepilog:
+{
+	noepilog = cmp.eq(epilog,#0);
+	epilogdws = lsr(epilog, #3);
+	kernel = and(epilog, #7);
+}
+{
+	if(noepilog) jumpr r31;
+	if(noepilog) ptr_out = sub(ptr_out, len);
+	p3 = cmp.eq(epilogdws, #0);
+	shift2 = asl(epilog, #3);
+}
+{
+	shiftb = and(shift2, #32);
+	ifword = tstbit(epilog,#2);
+	if(p3) jump .Lepilog60;
+	if(!p3) epilog = add(epilog, #-16);
+}
+{
+	loop0(.Ldword_loop_epilog, epilogdws);
+	/*  stop criteria is lsbs unless = 0 then its 8  */
+	p3 = cmp.eq(kernel, #0);
+	if(p3.new) kernel= #8;
+	p1 = cmp.gt(over, #0);
+}
+	/*  if not aligned to end of buffer execute 1 more iteration  */
+	if(p1) kernel= #0;
+.Ldword_loop_epilog:
+{
+	memd(ptr_out++#8) = ldata0;
+	ldata0 = valignb(dataF8, data70, align);
+	p3 = cmp.gt(epilog, kernel);
+}
+{
+	data70 = dataF8;
+	if(p3) dataF8 = memd(ptr_in++#8);
+	epilog = add(epilog, #-8);
+}:endloop0
+/* copy last 7 bytes */
+.Lepilog60:
+{
+	if(ifword) memw(ptr_out++#4) = data0;
+	ldata0 = lsr(ldata0, shiftb);
+	ifhword = tstbit(epilog,#1);
+	shiftb = and(shift2, #16);
+}
+{
+	if(ifhword) memh(ptr_out++#2) = data0;
+	ldata0 = lsr(ldata0, shiftb);
+	ifbyte = tstbit(epilog,#0);
+	if(ifbyte.new) len = add(len, #-1);
+}
+{
+	if(ifbyte) memb(ptr_out) = data0;
+	ptr_out = sub(ptr_out, len);	/*  return dest pointer  */
+        jumpr r31;
+}
+/*  do byte copy for small n  */
+.Lbytes23orless:
+{
+	p3 = sp1loop0(.Lbyte_copy, len);
+	len = add(len, #-1);
+}
+.Lbyte_copy:
+{
+	data0 = memb(ptr_in++#1);
+	if(p3) memb(ptr_out++#1) = data0;
+}:endloop0
+{
+	memb(ptr_out) = data0;
+	ptr_out = sub(ptr_out, len);
+	jumpr r31;
+}
+/*  do dword copies for aligned in, out and length  */
+.Ldwordaligned:
+{
+	p3 = sp1loop0(.Ldword_copy, len8);
+}
+.Ldword_copy:
+{
+	if(p3) memd(ptr_out++#8) = ldata0;
+	ldata0 = memd(ptr_in++#8);
+}:endloop0
+{
+	memd(ptr_out) = ldata0;
+	ptr_out = sub(ptr_out, len);
+	jumpr r31;	/*  return to function caller  */
+}
+.Lmemcpy_return:
+	r21:20 = memd(sp+#16);	/*  restore r20+r21  */
+{
+	r25:24 = memd(sp+#8);	/*  restore r24+r25  */
+	r17:16 = memd(sp+#0);	/*  restore r16+r17  */
+}
+	deallocframe;	/*  restore r31 and incrment stack by 16  */
+	jumpr r31
Index: linux-hexagon-kernel/arch/hexagon/lib/memset.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/lib/memset.S	2011-09-03 20:14:59.584980999 -0500
@@ -0,0 +1,316 @@
+/*
+ * Copyright (c) 2011 Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+
+/* HEXAGON assembly optimized memset */
+/* Replaces the standard library function memset */
+
+
+        .macro HEXAGON_OPT_FUNC_BEGIN name
+	.text
+	.p2align 4
+	.globl \name
+	.type  \name, @function
+\name:
+	.endm
+
+	.macro HEXAGON_OPT_FUNC_FINISH name
+	.size  \name, . - \name
+	.endm
+
+/* FUNCTION: memset (v2 version) */
+#if __HEXAGON_ARCH__ < 3
+HEXAGON_OPT_FUNC_BEGIN memset
+	{
+		r6 = #8
+		r7 = extractu(r0, #3 , #0)
+		p0 = cmp.eq(r2, #0)
+		p1 = cmp.gtu(r2, #7)
+	}
+	{
+		r4 = vsplatb(r1)
+		r8 = r0           /* leave r0 intact for return val  */
+		r9 = sub(r6, r7)  /* bytes until double alignment  */
+		if p0 jumpr r31   /* count == 0, so return  */
+	}
+	{
+		r3 = #0
+		r7 = #0
+		p0 = tstbit(r9, #0)
+		if p1 jump 2f /* skip byte loop */
+	}
+
+/* less than 8 bytes to set, so just set a byte at a time and return  */
+
+		loop0(1f, r2) /* byte loop */
+	.falign
+1: /* byte loop */
+	{
+		memb(r8++#1) = r4
+	}:endloop0
+		jumpr r31
+	.falign
+2: /* skip byte loop */
+	{
+		r6 = #1
+		p0 = tstbit(r9, #1)
+		p1 = cmp.eq(r2, #1)
+		if !p0 jump 3f /* skip initial byte store */
+	}
+	{
+		memb(r8++#1) = r4
+		r3:2 = sub(r3:2, r7:6)
+		if p1 jumpr r31
+	}
+	.falign
+3: /* skip initial byte store */
+	{
+		r6 = #2
+		p0 = tstbit(r9, #2)
+		p1 = cmp.eq(r2, #2)
+		if !p0 jump 4f /* skip initial half store */
+	}
+	{
+		memh(r8++#2) = r4
+		r3:2 = sub(r3:2, r7:6)
+		if p1 jumpr r31
+	}
+	.falign
+4: /* skip initial half store */
+	{
+		r6 = #4
+		p0 = cmp.gtu(r2, #7)
+		p1 = cmp.eq(r2, #4)
+		if !p0 jump 5f /* skip initial word store */
+	}
+	{
+		memw(r8++#4) = r4
+		r3:2 = sub(r3:2, r7:6)
+		p0 = cmp.gtu(r2, #11)
+		if p1 jumpr r31
+	}
+	.falign
+5: /* skip initial word store */
+	{
+		r10 = lsr(r2, #3)
+		p1 = cmp.eq(r3, #1)
+		if !p0 jump 7f /* skip double loop */
+	}
+	{
+		r5 = r4
+		r6 = #8
+		loop0(6f, r10) /* double loop */
+	}
+
+/* set bytes a double word at a time  */
+
+	.falign
+6: /* double loop */
+	{
+		memd(r8++#8) = r5:4
+		r3:2 = sub(r3:2, r7:6)
+		p1 = cmp.eq(r2, #8)
+	}:endloop0
+	.falign
+7: /* skip double loop */
+	{
+		p0 = tstbit(r2, #2)
+		if p1 jumpr r31
+	}
+	{
+		r6 = #4
+		p0 = tstbit(r2, #1)
+		p1 = cmp.eq(r2, #4)
+		if !p0 jump 8f /* skip final word store */
+	}
+	{
+		memw(r8++#4) = r4
+		r3:2 = sub(r3:2, r7:6)
+		if p1 jumpr r31
+	}
+	.falign
+8: /* skip final word store */
+	{
+		p1 = cmp.eq(r2, #2)
+		if !p0 jump 9f /* skip final half store */
+	}
+	{
+		memh(r8++#2) = r4
+		if p1 jumpr r31
+	}
+	.falign
+9: /* skip final half store */
+	{
+		memb(r8++#1) = r4
+		jumpr r31
+	}
+HEXAGON_OPT_FUNC_FINISH memset
+#endif
+
+
+/*  FUNCTION: memset (v3 and higher version)  */
+#if __HEXAGON_ARCH__ >= 3
+HEXAGON_OPT_FUNC_BEGIN memset
+	{
+		r7=vsplatb(r1)
+		r6 = r0
+		if (r2==#0) jump:nt .L1
+	}
+	{
+		r5:4=combine(r7,r7)
+		p0 = cmp.gtu(r2,#8)
+		if (p0.new) jump:nt .L3
+	}
+	{
+		r3 = r0
+		loop0(.L47,r2)
+	}
+	.falign
+.L47:
+	{
+		memb(r3++#1) = r1
+	}:endloop0 /* start=.L47 */
+		jumpr r31
+.L3:
+	{
+		p0 = tstbit(r0,#0)
+		if (!p0.new) jump:nt .L8
+		p1 = cmp.eq(r2, #1)
+	}
+	{
+		r6 = add(r0, #1)
+		r2 = add(r2,#-1)
+		memb(r0) = r1
+		if (p1) jump .L1
+	}
+.L8:
+	{
+		p0 = tstbit(r6,#1)
+		if (!p0.new) jump:nt .L10
+	}
+	{
+		r2 = add(r2,#-2)
+		memh(r6++#2) = r7
+		p0 = cmp.eq(r2, #2)
+		if (p0.new) jump:nt .L1
+	}
+.L10:
+	{
+		p0 = tstbit(r6,#2)
+		if (!p0.new) jump:nt .L12
+	}
+	{
+		r2 = add(r2,#-4)
+		memw(r6++#4) = r7
+		p0 = cmp.eq(r2, #4)
+		if (p0.new) jump:nt .L1
+	}
+.L12:
+	{
+		p0 = cmp.gtu(r2,#127)
+		if (!p0.new) jump:nt .L14
+	}
+		r3 = and(r6,#31)
+		if (r3==#0) jump:nt .L17
+	{
+		memd(r6++#8) = r5:4
+		r2 = add(r2,#-8)
+	}
+		r3 = and(r6,#31)
+		if (r3==#0) jump:nt .L17
+	{
+		memd(r6++#8) = r5:4
+		r2 = add(r2,#-8)
+	}
+		r3 = and(r6,#31)
+		if (r3==#0) jump:nt .L17
+	{
+		memd(r6++#8) = r5:4
+		r2 = add(r2,#-8)
+	}
+.L17:
+	{
+		r3 = lsr(r2,#5)
+		if (r1!=#0) jump:nt .L18
+	}
+	{
+		r8 = r3
+		r3 = r6
+		loop0(.L46,r3)
+	}
+	.falign
+.L46:
+	{
+		dczeroa(r6)
+		r6 = add(r6,#32)
+		r2 = add(r2,#-32)
+	}:endloop0 /* start=.L46 */
+.L14:
+	{
+		p0 = cmp.gtu(r2,#7)
+		if (!p0.new) jump:nt .L28
+		r8 = lsr(r2,#3)
+	}
+		loop0(.L44,r8)
+	.falign
+.L44:
+	{
+		memd(r6++#8) = r5:4
+		r2 = add(r2,#-8)
+	}:endloop0 /* start=.L44 */
+.L28:
+	{
+		p0 = tstbit(r2,#2)
+		if (!p0.new) jump:nt .L33
+	}
+	{
+		r2 = add(r2,#-4)
+		memw(r6++#4) = r7
+	}
+.L33:
+	{
+		p0 = tstbit(r2,#1)
+		if (!p0.new) jump:nt .L35
+	}
+	{
+		r2 = add(r2,#-2)
+		memh(r6++#2) = r7
+	}
+.L35:
+		p0 = cmp.eq(r2,#1)
+		if (p0) memb(r6) = r1
+.L1:
+		jumpr r31
+.L18:
+		loop0(.L45,r3)
+	.falign
+.L45:
+		dczeroa(r6)
+	{
+		memd(r6++#8) = r5:4
+		r2 = add(r2,#-32)
+	}
+		memd(r6++#8) = r5:4
+		memd(r6++#8) = r5:4
+	{
+		memd(r6++#8) = r5:4
+	}:endloop0 /* start=.L45  */
+		jump .L14
+HEXAGON_OPT_FUNC_FINISH memset
+#endif
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 11/36] Hexagon: Add hypervisor interface

[Richard Kuo]

[-- Attachment #1: hypervisor.diff --]
[-- Type: text/plain, Size: 44612 bytes --]

The port of Linux for Hexagon runs on top of a hypervisor layer,
which presents an interface for memory management, events, hardware
control, etc.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/hexagon_vm.h  |  281 ++++++++++++++++++++
 arch/hexagon/include/asm/vm_mmu.h      |  111 ++++++++
 arch/hexagon/kernel/vm_entry.S         |  269 ++++++++++++++++++++
 arch/hexagon/kernel/vm_events.c        |  101 +++++++
 arch/hexagon/kernel/vm_init_segtable.S |  442 +++++++++++++++++++++++++++++++++
 arch/hexagon/kernel/vm_ops.S           |  102 +++++++
 arch/hexagon/kernel/vm_switch.S        |   95 +++++++
 arch/hexagon/kernel/vm_vectors.S       |   50 +++
 8 files changed, 1451 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/hexagon_vm.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/hexagon_vm.h	2011-09-08 19:25:43.090921680 -0500
@@ -0,0 +1,281 @@
+/*
+ * Declarations for to Hexagon Virtal Machine.
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef ASM_HEXAGON_VM_H
+#define ASM_HEXAGON_VM_H
+
+/*
+ * In principle, a Linux kernel for the VM could
+ * selectively define the virtual instructions
+ * as inline assembler macros, but for a first pass,
+ * we'll use subroutines for both the VM and the native
+ * kernels.  It's costing a subroutine call/return,
+ * but it makes for a single set of entry points
+ * for tracing/debugging.
+ */
+
+/*
+ * Lets make this stuff visible only if configured,
+ * so we can unconditionally include the file.
+ */
+
+#ifndef __ASSEMBLY__
+
+enum VM_CACHE_OPS {
+	ickill,
+	dckill,
+	l2kill,
+	dccleaninva,
+	icinva,
+	idsync,
+	fetch_cfg
+};
+
+enum VM_INT_OPS {
+	nop,
+	globen,
+	globdis,
+	locen,
+	locdis,
+	affinity,
+	get,
+	peek,
+	status,
+	post,
+	clear
+};
+
+extern void _K_VM_event_vector(void);
+
+void __vmrte(void);
+long __vmsetvec(void *);
+long __vmsetie(long);
+long __vmgetie(void);
+long __vmintop(enum VM_INT_OPS, long, long, long, long);
+long __vmclrmap(void *, unsigned long);
+long __vmnewmap(void *);
+long __vmcache(enum VM_CACHE_OPS op, unsigned long addr, unsigned long len);
+unsigned long long __vmgettime(void);
+long __vmsettime(unsigned long long);
+long __vmstart(void *, void *);
+void __vmstop(void);
+long __vmwait(void);
+void __vmyield(void);
+long __vmvpid(void);
+
+static inline long __vmcache_ickill(void)
+{
+	return __vmcache(ickill, 0, 0);
+}
+
+static inline long __vmcache_dckill(void)
+{
+	return __vmcache(dckill, 0, 0);
+}
+
+static inline long __vmcache_l2kill(void)
+{
+	return __vmcache(l2kill, 0, 0);
+}
+
+static inline long __vmcache_dccleaninva(unsigned long addr, unsigned long len)
+{
+	return __vmcache(dccleaninva, addr, len);
+}
+
+static inline long __vmcache_icinva(unsigned long addr, unsigned long len)
+{
+	return __vmcache(icinva, addr, len);
+}
+
+static inline long __vmcache_idsync(unsigned long addr,
+					   unsigned long len)
+{
+	return __vmcache(idsync, addr, len);
+}
+
+static inline long __vmcache_fetch_cfg(unsigned long val)
+{
+	return __vmcache(fetch_cfg, val, 0);
+}
+
+/* interrupt operations  */
+
+static inline long __vmintop_nop(void)
+{
+	return __vmintop(nop, 0, 0, 0, 0);
+}
+
+static inline long __vmintop_globen(long i)
+{
+	return __vmintop(globen, i, 0, 0, 0);
+}
+
+static inline long __vmintop_globdis(long i)
+{
+	return __vmintop(globdis, i, 0, 0, 0);
+}
+
+static inline long __vmintop_locen(long i)
+{
+	return __vmintop(locen, i, 0, 0, 0);
+}
+
+static inline long __vmintop_locdis(long i)
+{
+	return __vmintop(locdis, i, 0, 0, 0);
+}
+
+static inline long __vmintop_affinity(long i, long cpu)
+{
+	return __vmintop(locdis, i, cpu, 0, 0);
+}
+
+static inline long __vmintop_get(void)
+{
+	return __vmintop(get, 0, 0, 0, 0);
+}
+
+static inline long __vmintop_peek(void)
+{
+	return __vmintop(peek, 0, 0, 0, 0);
+}
+
+static inline long __vmintop_status(long i)
+{
+	return __vmintop(status, i, 0, 0, 0);
+}
+
+static inline long __vmintop_post(long i)
+{
+	return __vmintop(post, i, 0, 0, 0);
+}
+
+static inline long __vmintop_clear(long i)
+{
+	return __vmintop(clear, i, 0, 0, 0);
+}
+
+#else /* Only assembly code should reference these */
+
+#define HVM_TRAP1_VMRTE			1
+#define HVM_TRAP1_VMSETVEC		2
+#define HVM_TRAP1_VMSETIE		3
+#define HVM_TRAP1_VMGETIE		4
+#define HVM_TRAP1_VMINTOP		5
+#define HVM_TRAP1_VMCLRMAP		10
+#define HVM_TRAP1_VMNEWMAP		11
+#define HVM_TRAP1_FORMERLY_VMWIRE	12
+#define HVM_TRAP1_VMCACHE		13
+#define HVM_TRAP1_VMGETTIME		14
+#define HVM_TRAP1_VMSETTIME		15
+#define HVM_TRAP1_VMWAIT		16
+#define HVM_TRAP1_VMYIELD		17
+#define HVM_TRAP1_VMSTART		18
+#define HVM_TRAP1_VMSTOP		19
+#define HVM_TRAP1_VMVPID		20
+#define HVM_TRAP1_VMSETREGS		21
+#define HVM_TRAP1_VMGETREGS		22
+
+#endif /* __ASSEMBLY__ */
+
+/*
+ * Constants for virtual instruction parameters and return values
+ */
+
+/* vmsetie arguments */
+
+#define VM_INT_DISABLE	0
+#define VM_INT_ENABLE	1
+
+/* vmsetimask arguments */
+
+#define VM_INT_UNMASK	0
+#define VM_INT_MASK	1
+
+#define VM_NEWMAP_TYPE_LINEAR	0
+#define VM_NEWMAP_TYPE_PGTABLES	1
+
+
+/*
+ * Event Record definitions useful to both C and Assembler
+ */
+
+/* VMEST Layout */
+
+#define HVM_VMEST_UM_SFT	31
+#define HVM_VMEST_UM_MSK	1
+#define HVM_VMEST_IE_SFT	30
+#define HVM_VMEST_IE_MSK	1
+#define HVM_VMEST_EVENTNUM_SFT	16
+#define HVM_VMEST_EVENTNUM_MSK	0xff
+#define HVM_VMEST_CAUSE_SFT	0
+#define HVM_VMEST_CAUSE_MSK	0xffff
+
+/*
+ * The initial program gets to find a system environment descriptor
+ * on its stack when it begins exection. The first word is a version
+ * code to indicate what is there.  Zero means nothing more.
+ */
+
+#define HEXAGON_VM_SED_NULL	0
+
+/*
+ * Event numbers for vector binding
+ */
+
+#define HVM_EV_RESET		0
+#define HVM_EV_MACHCHECK	1
+#define HVM_EV_GENEX		2
+#define HVM_EV_TRAP		8
+#define HVM_EV_INTR		15
+/* These shoud be nuked as soon as we know the VM is up to spec v0.1.1 */
+#define HVM_EV_INTR_0		16
+#define HVM_MAX_INTR		240
+
+/*
+ * Cause values for General Exception
+ */
+
+#define HVM_GE_C_BUS	0x01
+#define HVM_GE_C_XPROT	0x11
+#define HVM_GE_C_XUSER	0x14
+#define HVM_GE_C_INVI	0x15
+#define HVM_GE_C_PRIVI	0x1B
+#define HVM_GE_C_XMAL	0x1C
+#define HVM_GE_C_RMAL	0x20
+#define HVM_GE_C_WMAL	0x21
+#define HVM_GE_C_RPROT	0x22
+#define HVM_GE_C_WPROT	0x23
+#define HVM_GE_C_RUSER	0x24
+#define HVM_GE_C_WUSER	0x25
+#define HVM_GE_C_CACHE	0x28
+
+/*
+ * Cause codes for Machine Check
+ */
+
+#define	HVM_MCHK_C_DOWN		0x00
+#define	HVM_MCHK_C_BADSP	0x01
+#define	HVM_MCHK_C_BADEX	0x02
+#define	HVM_MCHK_C_BADPT	0x03
+#define	HVM_MCHK_C_REGWR	0x29
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/vm_mmu.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/vm_mmu.h	2011-09-08 19:22:01.120238955 -0500
@@ -0,0 +1,111 @@
+/*
+ * Hexagon VM page table entry definitions
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_VM_MMU_H
+#define _ASM_VM_MMU_H
+
+/*
+ * Shift, mask, and other constants for the Hexagon Virtual Machine
+ * page tables.
+ *
+ * Virtual machine MMU allows first-level entries to either be
+ * single-level lookup PTEs for very large pages, or PDEs pointing
+ * to second-level PTEs for smaller pages. If PTE is single-level,
+ * the least significant bits cannot be used as software bits to encode
+ * virtual memory subsystem information about the page, and that state
+ * must be maintained in some parallel data structure.
+ */
+
+/* S or Page Size field in PDE */
+#define	__HVM_PDE_S		(0x7 << 0)
+#define __HVM_PDE_S_4KB		0
+#define __HVM_PDE_S_16KB	1
+#define __HVM_PDE_S_64KB	2
+#define __HVM_PDE_S_256KB	3
+#define __HVM_PDE_S_1MB		4
+#define __HVM_PDE_S_4MB		5
+#define __HVM_PDE_S_16MB	6
+#define __HVM_PDE_S_INVALID	7
+
+/* Masks for L2 page table pointer, as function of page size */
+#define __HVM_PDE_PTMASK_4KB	0xfffff000
+#define __HVM_PDE_PTMASK_16KB	0xfffffc00
+#define __HVM_PDE_PTMASK_64KB	0xffffff00
+#define __HVM_PDE_PTMASK_256KB	0xffffffc0
+#define __HVM_PDE_PTMASK_1MB	0xfffffff0
+
+/*
+ * Virtual Machine PTE Bits/Fields
+ */
+#define __HVM_PTE_T		(1<<4)
+#define __HVM_PTE_U		(1<<5)
+#define	__HVM_PTE_C		(0x7<<6)
+#define __HVM_PTE_CVAL(pte)	(((pte) & __HVM_PTE_C) >> 6)
+#define __HVM_PTE_R		(1<<9)
+#define __HVM_PTE_W		(1<<10)
+#define __HVM_PTE_X		(1<<11)
+
+/*
+ * Cache Attributes, to be shifted as necessary for virtual/physical PTEs
+ */
+
+#define __HEXAGON_C_WB		0x0	/* Write-back, no L2 */
+#define	__HEXAGON_C_WT		0x1	/* Write-through, no L2 */
+#define	__HEXAGON_C_DEV		0x4	/* Device register space */
+#define	__HEXAGON_C_WT_L2	0x5	/* Write-through, with L2 */
+/* this really should be #if CONFIG_HEXAGON_ARCH = 2 but that's not defined */
+#if defined(CONFIG_HEXAGON_COMET) || defined(CONFIG_QDSP6_ST1)
+#define __HEXAGON_C_UNC		__HEXAGON_C_DEV
+#else
+#define	__HEXAGON_C_UNC		0x6	/* Uncached memory */
+#endif
+#define	__HEXAGON_C_WB_L2	0x7	/* Write-back, with L2 */
+
+/*
+ * This can be overriden, but we're defaulting to the most aggressive
+ * cache policy, the better to find bugs sooner.
+ */
+
+#define	CACHE_DEFAULT	__HEXAGON_C_WB_L2
+
+/* Masks for physical page address, as a function of page size */
+
+#define __HVM_PTE_PGMASK_4KB	0xfffff000
+#define __HVM_PTE_PGMASK_16KB	0xffffc000
+#define __HVM_PTE_PGMASK_64KB	0xffff0000
+#define __HVM_PTE_PGMASK_256KB	0xfffc0000
+#define __HVM_PTE_PGMASK_1MB	0xfff00000
+
+/* Masks for single-level large page lookups */
+
+#define __HVM_PTE_PGMASK_4MB	0xffc00000
+#define __HVM_PTE_PGMASK_16MB	0xff000000
+
+/*
+ * "Big kernel page mappings" (see vm_init_segtable.S)
+ * are currently 16MB
+ */
+
+#define BIG_KERNEL_PAGE_SHIFT 24
+#define BIG_KERNEL_PAGE_SIZE (1 << BIG_KERNEL_PAGE_SHIFT)
+
+
+
+#endif /* _ASM_VM_MMU_H */
Index: linux-hexagon-kernel/arch/hexagon/kernel/vm_entry.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vm_entry.S	2011-09-08 19:22:01.120238955 -0500
@@ -0,0 +1,269 @@
+/*
+ * Event entry/exit for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <asm/asm-offsets.h>  /*  assembly-safer versions of C defines */
+#include <asm/mem-layout.h>   /*  sigh, except for page_offset  */
+#include <asm/hexagon_vm.h>
+#include <asm/thread_info.h>
+
+/*
+ * Entry into guest-mode Linux under Hexagon Virtual Machine.
+ * Stack pointer points to event record - build pt_regs on top of it,
+ * set up a plausible C stack frame, and dispatch to the C handler.
+ * On return, do vmrte virtual instruction with SP where we started.
+ *
+ * VM Spec 0.5 uses a trap to fetch HVM record now.
+ */
+
+/*
+ * Save full register state, while setting up thread_info struct
+ * pointer derived from kernel stack pointer in THREADINFO_REG
+ * register, putting prior thread_info.regs pointer in a callee-save
+ * register (R24, which had better not ever be assigned to THREADINFO_REG),
+ * and updating thread_info.regs to point to current stack frame,
+ * so as to support nested events in kernel mode.
+ *
+ * As this is common code, we set the pt_regs system call number
+ * to -1 for all events.  It will be replaced with the system call
+ * number in the case where we decode a system call (trap0(#1)).
+ */
+
+#define save_pt_regs()\
+	memd(R0 + #_PT_R3130) = R31:30; \
+	{ memw(R0 + #_PT_R2928) = R28; \
+	  R31 = memw(R0 + #_PT_ER_VMPSP); }\
+	{ memw(R0 + #(_PT_R2928 + 4)) = R31; \
+	  R31 = ugp; } \
+	{ memd(R0 + #_PT_R2726) = R27:26; \
+	  R30 = gp ; } \
+	memd(R0 + #_PT_R2524) = R25:24; \
+	memd(R0 + #_PT_R2322) = R23:22; \
+	memd(R0 + #_PT_R2120) = R21:20; \
+	memd(R0 + #_PT_R1918) = R19:18; \
+	memd(R0 + #_PT_R1716) = R17:16; \
+	memd(R0 + #_PT_R1514) = R15:14; \
+	memd(R0 + #_PT_R1312) = R13:12; \
+	{ memd(R0 + #_PT_R1110) = R11:10; \
+	  R15 = lc0; } \
+	{ memd(R0 + #_PT_R0908) = R9:8; \
+	  R14 = sa0; } \
+	{ memd(R0 + #_PT_R0706) = R7:6; \
+	  R13 = lc1; } \
+	{ memd(R0 + #_PT_R0504) = R5:4; \
+	  R12 = sa1; } \
+	{ memd(R0 + #_PT_UGPGP) = R31:30; \
+	  R11 = m1; \
+	  R2.H = #HI(_THREAD_SIZE); } \
+	{ memd(R0 + #_PT_LC0SA0) = R15:14; \
+	  R10 = m0; \
+	  R2.L = #LO(_THREAD_SIZE); } \
+	{ memd(R0 + #_PT_LC1SA1) = R13:12; \
+	  R15 = p3:0; \
+	  R2 = neg(R2); } \
+	{ memd(R0 + #_PT_M1M0) = R11:10; \
+	  R14  = usr; \
+	  R2 = and(R0,R2); } \
+	{ memd(R0 + #_PT_PREDSUSR) =  R15:14; \
+	  THREADINFO_REG = R2; } \
+	{ r24 = memw(THREADINFO_REG + #_THREAD_INFO_PT_REGS); \
+	  memw(THREADINFO_REG + #_THREAD_INFO_PT_REGS) = R0; \
+	  R2 = #-1; } \
+	{ memw(R0 + #_PT_SYSCALL_NR) = R2; \
+	  R30 = #0; }
+
+/*
+ * Restore registers and thread_info.regs state. THREADINFO_REG
+ * is assumed to still be sane, and R24 to have been correctly
+ * preserved. Don't restore R29 (SP) until later.
+ */
+
+#define restore_pt_regs() \
+	{ memw(THREADINFO_REG + #_THREAD_INFO_PT_REGS) = R24; \
+	  R15:14 = memd(R0 + #_PT_PREDSUSR); } \
+	{ R11:10 = memd(R0 + #_PT_M1M0); \
+	  p3:0 = R15; } \
+	{ R13:12 = memd(R0 + #_PT_LC1SA1); \
+	  usr = R14; } \
+	{ R15:14 = memd(R0 + #_PT_LC0SA0); \
+	  m1 = R11; } \
+	{ R3:2 = memd(R0 + #_PT_R0302); \
+	  m0 = R10; } \
+	{ R5:4 = memd(R0 + #_PT_R0504); \
+	  lc1 = R13; } \
+	{ R7:6 = memd(R0 + #_PT_R0706); \
+	  sa1 = R12; } \
+	{ R9:8 = memd(R0 + #_PT_R0908); \
+	  lc0 = R15; } \
+	{ R11:10 = memd(R0 + #_PT_R1110); \
+	  sa0 = R14; } \
+	{ R13:12 = memd(R0 + #_PT_R1312); \
+	  R15:14 = memd(R0 + #_PT_R1514); } \
+	{ R17:16 = memd(R0 + #_PT_R1716); \
+	  R19:18 = memd(R0 + #_PT_R1918); } \
+	{ R21:20 = memd(R0 + #_PT_R2120); \
+	  R23:22 = memd(R0 + #_PT_R2322); } \
+	{ R25:24 = memd(R0 + #_PT_R2524); \
+	  R27:26 = memd(R0 + #_PT_R2726); } \
+	R31:30 = memd(R0 + #_PT_UGPGP); \
+	{ R28 = memw(R0 + #_PT_R2928); \
+	  ugp = R31; } \
+	{ R31:30 = memd(R0 + #_PT_R3130); \
+	  gp = R30; }
+
+	/*
+	 * Clears off enough space for the rest of pt_regs; evrec is a part
+	 * of pt_regs in HVM mode.  Save R0/R1, set handler's address in R1.
+	 * R0 is the address of pt_regs and is the parameter to save_pt_regs.
+	 */
+
+/*
+ * Since the HVM isn't automagically pushing the EVREC onto the stack anymore,
+ * we'll subract the entire size out and then fill it in ourselves.
+ * Need to save off R0, R1, R2, R3 immediately.
+ */
+
+#define	vm_event_entry(CHandler) \
+	{ \
+		R29 = add(R29, #-(_PT_REGS_SIZE)); \
+		memd(R29 + #(_PT_R0100 + -_PT_REGS_SIZE)) = R1:0; \
+	} \
+	{ \
+		memd(R29 +#_PT_R0302) = R3:2; \
+	} \
+	trap1(#HVM_TRAP1_VMGETREGS); \
+	{ \
+		memd(R29 + #_PT_ER_VMEL) = R1:0; \
+		R0 = R29; \
+		R1.L = #LO(CHandler); \
+	} \
+	{ \
+		memd(R29 + #_PT_ER_VMPSP) = R3:2; \
+		R1.H = #HI(CHandler); \
+		jump event_dispatch; \
+	}
+
+.text
+	/*
+	 * Do bulk save/restore in one place.
+	 * Adds a jump to dispatch latency, but
+	 * saves hundreds of bytes.
+	 */
+
+event_dispatch:
+	save_pt_regs()
+	callr	r1
+
+	/*
+	 * If we were in kernel mode, we don't need to check scheduler
+	 * or signals if CONFIG_PREEMPT is not set.  If set, then it has
+	 * to jump to a need_resched kind of block.
+	 * BTW, CONFIG_PREEMPT is not supported yet.
+	 */
+
+#ifdef CONFIG_PREEMPT
+	R0 = #VM_INT_DISABLE
+	trap1(#HVM_TRAP1_VMSETIE)
+#endif
+
+	/*  "Nested control path" -- if the previous mode was kernel  */
+	R0 = memw(R29 + #_PT_ER_VMEST);
+	P0 = tstbit(R0, #HVM_VMEST_UM_SFT);
+	if !P0 jump restore_all;
+	/*
+	 * Returning from system call, normally coming back from user mode
+	 */
+return_from_syscall:
+	/*  Disable interrupts while checking TIF  */
+	R0 = #VM_INT_DISABLE
+	trap1(#HVM_TRAP1_VMSETIE)
+
+	/*
+	 * Coming back from the C-world, our thread info pointer
+	 * should be in the designated register (usually R19)
+	 */
+	R1.L = #LO(_TIF_ALLWORK_MASK)
+	{
+		R1.H = #HI(_TIF_ALLWORK_MASK);
+		R0 = memw(THREADINFO_REG + #_THREAD_INFO_FLAGS);
+	}
+
+	/*
+	 * Compare against the "return to userspace" _TIF_WORK_MASK
+	 */
+	R1 = and(R1,R0);
+	{ P0 = cmp.eq(R1,#0); if (!P0.new) jump:t work_pending;}
+	jump restore_all;  /*  we're outta here!  */
+
+work_pending:
+	{
+		P0 = tstbit(R1, #TIF_NEED_RESCHED);
+		if (!P0.new) jump:nt work_notifysig;
+	}
+	call schedule
+	jump return_from_syscall;  /*  check for more work  */
+
+work_notifysig:
+	/*  this is the part that's kind of fuzzy.  */
+	R1 = and(R0, #(_TIF_SIGPENDING | _TIF_NOTIFY_RESUME));
+	P0 = cmp.eq(R1, #0);
+	if P0 jump restore_all
+	R1 = R0; 	/* unsigned long thread_info_flags */
+	R0 = R29;	/* regs should still be at top of stack  */
+	call do_notify_resume
+
+restore_all:
+	/* Disable interrupts, if they weren't already, before reg restore.  */
+	R0 = #VM_INT_DISABLE
+	trap1(#HVM_TRAP1_VMSETIE)
+
+	/*  do the setregs here for VM 0.5  */
+	/*  R29 here should already be pointing at pt_regs  */
+	R1:0 = memd(R29 + #_PT_ER_VMEL);
+	R3:2 = memd(R29 + #_PT_ER_VMPSP);
+	trap1(#HVM_TRAP1_VMSETREGS);
+
+	R0 = R29
+	restore_pt_regs()
+	R1:0 = memd(R29 + #_PT_R0100);
+	R29 = add(R29, #_PT_REGS_SIZE);
+	trap1(#HVM_TRAP1_VMRTE)
+	/* Notreached */
+
+	.globl _K_enter_genex
+_K_enter_genex:
+	vm_event_entry(do_genex)
+
+	.globl _K_enter_interrupt
+_K_enter_interrupt:
+	vm_event_entry(arch_do_IRQ)
+
+	.globl _K_enter_trap0
+_K_enter_trap0:
+	vm_event_entry(do_trap0)
+
+	.globl _K_enter_machcheck
+_K_enter_machcheck:
+	vm_event_entry(do_machcheck)
+
+
+	.globl ret_from_fork
+ret_from_fork:
+	call schedule_tail
+	jump return_from_syscall
Index: linux-hexagon-kernel/arch/hexagon/kernel/vm_events.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vm_events.c	2011-09-08 19:22:01.120238955 -0500
@@ -0,0 +1,101 @@
+/*
+ * Mostly IRQ support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/kernel.h>
+#include <asm/registers.h>
+#include <linux/irq.h>
+#include <linux/hardirq.h>
+#include <asm/system.h>
+
+/*
+ * show_regs - print pt_regs structure
+ * @regs: pointer to pt_regs
+ *
+ * To-do:  add all the accessor definitions to registers.h
+ *
+ * Will make this routine a lot easier to write.
+ */
+void show_regs(struct pt_regs *regs)
+{
+	printk(KERN_EMERG "restart_r0: \t0x%08lx   syscall_nr: %ld\n",
+	       regs->restart_r0, regs->syscall_nr);
+	printk(KERN_EMERG "preds: \t\t0x%08lx\n", regs->preds);
+	printk(KERN_EMERG "lc0: \t0x%08lx   sa0: 0x%08lx   m0:  0x%08lx\n",
+	       regs->lc0, regs->sa0, regs->m0);
+	printk(KERN_EMERG "lc1: \t0x%08lx   sa1: 0x%08lx   m1:  0x%08lx\n",
+	       regs->lc1, regs->sa1, regs->m1);
+	printk(KERN_EMERG "gp: \t0x%08lx   ugp: 0x%08lx   usr: 0x%08lx\n",
+	       regs->gp, regs->ugp, regs->usr);
+	printk(KERN_EMERG "r0: \t0x%08lx %08lx %08lx %08lx\n", regs->r00,
+		regs->r01,
+		regs->r02,
+		regs->r03);
+	printk(KERN_EMERG "r4:  \t0x%08lx %08lx %08lx %08lx\n", regs->r04,
+		regs->r05,
+		regs->r06,
+		regs->r07);
+	printk(KERN_EMERG "r8:  \t0x%08lx %08lx %08lx %08lx\n", regs->r08,
+		regs->r09,
+		regs->r10,
+		regs->r11);
+	printk(KERN_EMERG "r12: \t0x%08lx %08lx %08lx %08lx\n", regs->r12,
+		regs->r13,
+		regs->r14,
+		regs->r15);
+	printk(KERN_EMERG "r16: \t0x%08lx %08lx %08lx %08lx\n", regs->r16,
+		regs->r17,
+		regs->r18,
+		regs->r19);
+	printk(KERN_EMERG "r20: \t0x%08lx %08lx %08lx %08lx\n", regs->r20,
+		regs->r21,
+		regs->r22,
+		regs->r23);
+	printk(KERN_EMERG "r24: \t0x%08lx %08lx %08lx %08lx\n", regs->r24,
+		regs->r25,
+		regs->r26,
+		regs->r27);
+	printk(KERN_EMERG "r28: \t0x%08lx %08lx %08lx %08lx\n", regs->r28,
+		regs->r29,
+		regs->r30,
+		regs->r31);
+
+	printk(KERN_EMERG "elr: \t0x%08lx   cause: 0x%08lx   user_mode: %d\n",
+		pt_elr(regs), pt_cause(regs), user_mode(regs));
+	printk(KERN_EMERG "psp: \t0x%08lx   badva: 0x%08lx   int_enabled: %d\n",
+		pt_psp(regs), pt_badva(regs), ints_enabled(regs));
+}
+
+void dummy_handler(struct pt_regs *regs)
+{
+	unsigned int elr = pt_elr(regs);
+	printk(KERN_ERR "Unimplemented handler; ELR=0x%08x\n", elr);
+}
+
+
+void arch_do_IRQ(struct pt_regs *regs)
+{
+	int irq = pt_cause(regs);
+	struct pt_regs *old_regs = set_irq_regs(regs);
+
+	irq_enter();
+	generic_handle_irq(irq);
+	irq_exit();
+	set_irq_regs(old_regs);
+}
Index: linux-hexagon-kernel/arch/hexagon/kernel/vm_init_segtable.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vm_init_segtable.S	2011-09-08 19:22:01.120238955 -0500
@@ -0,0 +1,442 @@
+/*
+ * Initial page table for Linux kernel under Hexagon VM,
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * These tables are pre-computed and linked into kernel.
+ */
+
+#include <asm/vm_mmu.h>
+/*  #include <asm/iomap.h>  */
+
+/*
+ * Start with mapping PA=0 to both VA=0x0 and VA=0xc000000 as 16MB large pages.
+ * No user mode access, RWX, write-back cache.  The entry needs
+ * to be replicated for all 4 virtual segments mapping to the page.
+ */
+
+/* "Big Kernel Page"  */
+#define BKP(pa) (((pa) & __HVM_PTE_PGMASK_4MB)		\
+		| __HVM_PTE_R | __HVM_PTE_W | __HVM_PTE_X	\
+		| __HEXAGON_C_WB_L2 << 6			\
+		| __HVM_PDE_S_16MB)
+
+/*  No cache version  */
+
+#define BKPG_IO(pa) (((pa) & __HVM_PTE_PGMASK_16MB) \
+			| __HVM_PTE_R | __HVM_PTE_W | __HVM_PTE_X \
+			| __HVM_PDE_S_16MB | __HEXAGON_C_DEV << 6 )
+
+#define FOURK_IO(pa) (((pa) & __HVM_PTE_PGMASK_4KB) \
+			| __HVM_PTE_R | __HVM_PTE_W | __HVM_PTE_X \
+			| __HEXAGON_C_DEV << 6 )
+
+#define L2_PTR(pa) (((pa) & __HVM_PTE_PGMASK_4KB) \
+			| __HVM_PDE_S_4KB  )
+
+#define X __HVM_PDE_S_INVALID
+
+	.p2align 12
+	.globl swapper_pg_dir
+	.globl _K_init_segtable
+swapper_pg_dir:
+/* VA 0x00000000 */
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+/* VA 0x40000000 */
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+/* VA 0x80000000 */
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+/*0xa8*/.word X,X,X,X
+#ifdef CONFIG_COMET_EARLY_UART_DEBUG
+UART_PTE_ENTRY:
+/*0xa9*/.word BKPG_IO(0xa9000000),BKPG_IO(0xa9000000),BKPG_IO(0xa9000000),BKPG_IO(0xa9000000)
+#else
+/*0xa9*/.word X,X,X,X
+#endif
+/*0xaa*/.word X,X,X,X
+/*0xab*/.word X,X,X,X
+/*0xac*/.word X,X,X,X
+/*0xad*/.word X,X,X,X
+/*0xae*/.word X,X,X,X
+/*0xaf*/.word X,X,X,X
+/*0xb0*/.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+	.word X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X,X
+_K_init_segtable:
+/* VA 0xC0000000 */
+	.word BKP(0x00000000), BKP(0x00400000), BKP(0x00800000), BKP(0x00c00000)
+	.word BKP(0x01000000), BKP(0x01400000), BKP(0x01800000), BKP(0x01c00000)
+	.word BKP(0x02000000), BKP(0x02400000), BKP(0x02800000), BKP(0x02c00000)
+	.word BKP(0x03000000), BKP(0x03400000), BKP(0x03800000), BKP(0x03c00000)
+	.word BKP(0x04000000), BKP(0x04400000), BKP(0x04800000), BKP(0x04c00000)
+	.word BKP(0x05000000), BKP(0x05400000), BKP(0x05800000), BKP(0x05c00000)
+	.word BKP(0x06000000), BKP(0x06400000), BKP(0x06800000), BKP(0x06c00000)
+	.word BKP(0x07000000), BKP(0x07400000), BKP(0x07800000), BKP(0x07c00000)
+
+	.word BKP(0x08000000), BKP(0x08400000), BKP(0x08800000), BKP(0x08c00000)
+	.word BKP(0x09000000), BKP(0x09400000), BKP(0x09800000), BKP(0x09c00000)
+	.word BKP(0x0a000000), BKP(0x0a400000), BKP(0x0a800000), BKP(0x0ac00000)
+	.word BKP(0x0b000000), BKP(0x0b400000), BKP(0x0b800000), BKP(0x0bc00000)
+	.word BKP(0x0c000000), BKP(0x0c400000), BKP(0x0c800000), BKP(0x0cc00000)
+	.word BKP(0x0d000000), BKP(0x0d400000), BKP(0x0d800000), BKP(0x0dc00000)
+	.word BKP(0x0e000000), BKP(0x0e400000), BKP(0x0e800000), BKP(0x0ec00000)
+	.word BKP(0x0f000000), BKP(0x0f400000), BKP(0x0f800000), BKP(0x0fc00000)
+
+	.word BKP(0x10000000), BKP(0x10400000), BKP(0x10800000), BKP(0x10c00000)
+	.word BKP(0x11000000), BKP(0x11400000), BKP(0x11800000), BKP(0x11c00000)
+	.word BKP(0x12000000), BKP(0x12400000), BKP(0x12800000), BKP(0x12c00000)
+	.word BKP(0x13000000), BKP(0x13400000), BKP(0x13800000), BKP(0x13c00000)
+	.word BKP(0x14000000), BKP(0x14400000), BKP(0x14800000), BKP(0x14c00000)
+	.word BKP(0x15000000), BKP(0x15400000), BKP(0x15800000), BKP(0x15c00000)
+	.word BKP(0x16000000), BKP(0x16400000), BKP(0x16800000), BKP(0x16c00000)
+	.word BKP(0x17000000), BKP(0x17400000), BKP(0x17800000), BKP(0x17c00000)
+
+	.word BKP(0x18000000), BKP(0x18400000), BKP(0x18800000), BKP(0x18c00000)
+	.word BKP(0x19000000), BKP(0x19400000), BKP(0x19800000), BKP(0x19c00000)
+	.word BKP(0x1a000000), BKP(0x1a400000), BKP(0x1a800000), BKP(0x1ac00000)
+	.word BKP(0x1b000000), BKP(0x1b400000), BKP(0x1b800000), BKP(0x1bc00000)
+	.word BKP(0x1c000000), BKP(0x1c400000), BKP(0x1c800000), BKP(0x1cc00000)
+	.word BKP(0x1d000000), BKP(0x1d400000), BKP(0x1d800000), BKP(0x1dc00000)
+	.word BKP(0x1e000000), BKP(0x1e400000), BKP(0x1e800000), BKP(0x1ec00000)
+	.word BKP(0x1f000000), BKP(0x1f400000), BKP(0x1f800000), BKP(0x1fc00000)
+
+	.word BKP(0x20000000), BKP(0x20400000), BKP(0x20800000), BKP(0x20c00000)
+	.word BKP(0x21000000), BKP(0x21400000), BKP(0x21800000), BKP(0x21c00000)
+	.word BKP(0x22000000), BKP(0x22400000), BKP(0x22800000), BKP(0x22c00000)
+	.word BKP(0x23000000), BKP(0x23400000), BKP(0x23800000), BKP(0x23c00000)
+	.word BKP(0x24000000), BKP(0x24400000), BKP(0x24800000), BKP(0x24c00000)
+	.word BKP(0x25000000), BKP(0x25400000), BKP(0x25800000), BKP(0x25c00000)
+	.word BKP(0x26000000), BKP(0x26400000), BKP(0x26800000), BKP(0x26c00000)
+	.word BKP(0x27000000), BKP(0x27400000), BKP(0x27800000), BKP(0x27c00000)
+
+	.word BKP(0x28000000), BKP(0x28400000), BKP(0x28800000), BKP(0x28c00000)
+	.word BKP(0x29000000), BKP(0x29400000), BKP(0x29800000), BKP(0x29c00000)
+	.word BKP(0x2a000000), BKP(0x2a400000), BKP(0x2a800000), BKP(0x2ac00000)
+	.word BKP(0x2b000000), BKP(0x2b400000), BKP(0x2b800000), BKP(0x2bc00000)
+	.word BKP(0x2c000000), BKP(0x2c400000), BKP(0x2c800000), BKP(0x2cc00000)
+	.word BKP(0x2d000000), BKP(0x2d400000), BKP(0x2d800000), BKP(0x2dc00000)
+	.word BKP(0x2e000000), BKP(0x2e400000), BKP(0x2e800000), BKP(0x2ec00000)
+	.word BKP(0x2f000000), BKP(0x2f400000), BKP(0x2f800000), BKP(0x2fc00000)
+
+	.word BKP(0x30000000), BKP(0x30400000), BKP(0x30800000), BKP(0x30c00000)
+	.word BKP(0x31000000), BKP(0x31400000), BKP(0x31800000), BKP(0x31c00000)
+	.word BKP(0x32000000), BKP(0x32400000), BKP(0x32800000), BKP(0x32c00000)
+	.word BKP(0x33000000), BKP(0x33400000), BKP(0x33800000), BKP(0x33c00000)
+	.word BKP(0x34000000), BKP(0x34400000), BKP(0x34800000), BKP(0x34c00000)
+	.word BKP(0x35000000), BKP(0x35400000), BKP(0x35800000), BKP(0x35c00000)
+	.word BKP(0x36000000), BKP(0x36400000), BKP(0x36800000), BKP(0x36c00000)
+	.word BKP(0x37000000), BKP(0x37400000), BKP(0x37800000), BKP(0x37c00000)
+
+	.word BKP(0x38000000), BKP(0x38400000), BKP(0x38800000), BKP(0x38c00000)
+	.word BKP(0x39000000), BKP(0x39400000), BKP(0x39800000), BKP(0x39c00000)
+	.word BKP(0x3a000000), BKP(0x3a400000), BKP(0x3a800000), BKP(0x3ac00000)
+	.word BKP(0x3b000000), BKP(0x3b400000), BKP(0x3b800000), BKP(0x3bc00000)
+	.word BKP(0x3c000000), BKP(0x3c400000), BKP(0x3c800000), BKP(0x3cc00000)
+	.word BKP(0x3d000000), BKP(0x3d400000), BKP(0x3d800000), BKP(0x3dc00000)
+_K_io_map:
+	.word X,X,X,X /* 0x3e000000 - device IO early remap */
+	.word X,X,X,X /* 0x3f000000 - hypervisor space*/
+
+#if 0
+/*
+ * This is in here as an example for devices which need to be mapped really
+ * early.
+ */
+	.p2align 12
+	.globl _K_io_kmap
+	.globl _K_init_devicetable
+_K_init_devicetable:  /*  Should be 4MB worth of entries  */
+	.word FOURK_IO(MSM_GPIO1_PHYS),FOURK_IO(MSM_GPIO2_PHYS),FOURK_IO(MSM_SIRC_PHYS),X
+	.word FOURK_IO(TLMM_GPIO1_PHYS),X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+	.word X,X,X,X
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/vm_ops.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vm_ops.S	2011-09-08 19:22:01.120238955 -0500
@@ -0,0 +1,102 @@
+/*
+ * Hexagon VM instruction support
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/linkage.h>
+#include <asm/hexagon_vm.h>
+
+/*
+ * C wrappers for virtual machine "instructions".  These
+ * could be, and perhaps some day will be, handled as in-line
+ * macros, but for tracing/debugging it's handy to have
+ * a single point of invocation for each of them.
+ * Conveniently, they take paramters and return values
+ * consistent with the ABI calling convention.
+ */
+
+ENTRY(__vmrte)
+	trap1(#HVM_TRAP1_VMRTE);
+	jumpr	R31;
+
+ENTRY(__vmsetvec)
+	trap1(#HVM_TRAP1_VMSETVEC);
+	jumpr	R31;
+
+ENTRY(__vmsetie)
+	trap1(#HVM_TRAP1_VMSETIE);
+	jumpr	R31;
+
+ENTRY(__vmgetie)
+	trap1(#HVM_TRAP1_VMGETIE);
+	jumpr	R31;
+
+ENTRY(__vmintop)
+	trap1(#HVM_TRAP1_VMINTOP);
+	jumpr	R31;
+
+ENTRY(__vmclrmap)
+	trap1(#HVM_TRAP1_VMCLRMAP);
+	jumpr	R31;
+
+ENTRY(__vmnewmap)
+	r1 = #VM_NEWMAP_TYPE_PGTABLES;
+	trap1(#HVM_TRAP1_VMNEWMAP);
+	jumpr	R31;
+
+ENTRY(__vmcache)
+	trap1(#HVM_TRAP1_VMCACHE);
+	jumpr	R31;
+
+ENTRY(__vmgettime)
+	trap1(#HVM_TRAP1_VMGETTIME);
+	jumpr	R31;
+
+ENTRY(__vmsettime)
+	trap1(#HVM_TRAP1_VMSETTIME);
+	jumpr	R31;
+
+ENTRY(__vmwait)
+	trap1(#HVM_TRAP1_VMWAIT);
+	jumpr	R31;
+
+ENTRY(__vmyield)
+	trap1(#HVM_TRAP1_VMYIELD);
+	jumpr	R31;
+
+ENTRY(__vmstart)
+	trap1(#HVM_TRAP1_VMSTART);
+	jumpr	R31;
+
+ENTRY(__vmstop)
+	trap1(#HVM_TRAP1_VMSTOP);
+	jumpr	R31;
+
+ENTRY(__vmvpid)
+	trap1(#HVM_TRAP1_VMVPID);
+	jumpr	R31;
+
+/*  Probably not actually going to use these; see vm_entry.S  */
+
+ENTRY(__vmsetregs)
+	trap1(#HVM_TRAP1_VMSETREGS);
+	jumpr	R31;
+
+ENTRY(__vmgetregs)
+	trap1(#HVM_TRAP1_VMGETREGS);
+	jumpr	R31;
Index: linux-hexagon-kernel/arch/hexagon/kernel/vm_switch.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vm_switch.S	2011-09-08 19:22:01.120238955 -0500
@@ -0,0 +1,95 @@
+/*
+ * Context switch support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <asm/asm-offsets.h>
+
+.text
+
+/*
+ * The register used as a fast-path thread information pointer
+ * is determined as a kernel configuration option.  If it happens
+ * to be a callee-save register, we're going to be saving and
+ * restoring it twice here.
+ *
+ * This code anticipates a revised ABI where R20-23 are added
+ * to the set of callee-save registers, but this should be
+ * backward compatible to legacy tools.
+ */
+
+
+/*
+ *	void switch_to(struct task_struct *prev,
+ *		struct task_struct *next, struct task_struct *last);
+ */
+	.p2align 2
+	.globl __switch_to
+	.type	__switch_to, @function
+
+/*
+ * When we exit the wormhole, we need to store the previous task
+ * in the new R0's pointer.  Technically it should be R2, but they should
+ * be the same; seems like a legacy thing.  In short, don't butcher
+ * R0, let it go back out unmolested.
+ */
+
+__switch_to:
+	/*
+	 * Push callee-saves onto "prev" stack.
+	 * Here, we're sneaky because the LR and FP
+	 * storage of the thread_stack structure
+	 * is automagically allocated by allocframe,
+	 * so we pass struct size less 8.
+	 */
+	allocframe(#(_SWITCH_STACK_SIZE - 8));
+	memd(R29+#(_SWITCH_R2726))=R27:26;
+	memd(R29+#(_SWITCH_R2524))=R25:24;
+	memd(R29+#(_SWITCH_R2322))=R23:22;
+	memd(R29+#(_SWITCH_R2120))=R21:20;
+	memd(R29+#(_SWITCH_R1918))=R19:18;
+	memd(R29+#(_SWITCH_R1716))=R17:16;
+	/* Stash thread_info pointer in task_struct */
+	memw(R0+#_TASK_THREAD_INFO) = THREADINFO_REG;
+	memw(R0 +#(_TASK_STRUCT_THREAD + _THREAD_STRUCT_SWITCH_SP)) = R29;
+	/* Switch to "next" stack and restore callee saves from there */
+	R29 = memw(R1 + #(_TASK_STRUCT_THREAD + _THREAD_STRUCT_SWITCH_SP));
+	{
+	    R27:26 = memd(R29+#(_SWITCH_R2726));
+	    R25:24 = memd(R29+#(_SWITCH_R2524));
+	}
+	{
+	    R23:22 = memd(R29+#(_SWITCH_R2322));
+	    R21:20 = memd(R29+#(_SWITCH_R2120));
+	}
+	{
+	    R19:18 = memd(R29+#(_SWITCH_R1918));
+	    R17:16 = memd(R29+#(_SWITCH_R1716));
+	}
+	{
+	    /* THREADINFO_REG is currently one of the callee-saved regs
+	     * above, and so be sure to re-load it last.
+	     */
+	    THREADINFO_REG = memw(R1 + #_TASK_THREAD_INFO);
+	    R31:30 = memd(R29+#_SWITCH_FP);
+	}
+	{
+	    R29 = add(R29,#_SWITCH_STACK_SIZE);
+	    jumpr R31;
+	}
+	.size	__switch_to, .-__switch_to
Index: linux-hexagon-kernel/arch/hexagon/kernel/vm_vectors.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vm_vectors.S	2011-09-08 19:22:01.120238955 -0500
@@ -0,0 +1,50 @@
+/*
+ * Event jump tables
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <asm/hexagon_vm.h>
+
+.text
+
+/*  This is registered early on to allow angel  */
+.global _K_provisional_vec
+_K_provisional_vec:
+	jump 1f;
+	jump 1f;
+	jump 1f;
+	jump 1f;
+	jump 1f;
+	trap1(#HVM_TRAP1_VMRTE)
+	jump 1f;
+	jump 1f;
+
+
+.global _K_VM_event_vector
+_K_VM_event_vector:
+1:
+	jump 1b;  /*  Reset  */
+	jump _K_enter_machcheck;
+	jump _K_enter_genex;
+	jump 1b;  /*  3 Rsvd  */
+	jump 1b;  /*  4 Rsvd  */
+	jump _K_enter_trap0;
+	jump 1b;  /*  6 Rsvd  */
+	jump _K_enter_interrupt;
+
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 12/36] Hexagon: Export ksyms defined in assembly files.

[Richard Kuo]

[-- Attachment #1: ksyms.diff --]
[-- Type: text/plain, Size: 2101 bytes --]

Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/kernel/hexagon_ksyms.c |   42 ++++++++++++++++++++++++++++++++++++
 1 file changed, 42 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/kernel/hexagon_ksyms.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/hexagon_ksyms.c	2011-09-07 13:00:31.040646956 -0500
@@ -0,0 +1,42 @@
+/*
+ * Export of symbols defined in assembly files and/or libgcc.
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <asm/hexagon_vm.h>
+#include <asm/uaccess.h>
+
+EXPORT_SYMBOL(__copy_from_user_hexagon);
+EXPORT_SYMBOL(__copy_to_user_hexagon);
+EXPORT_SYMBOL(__vmgetie);
+EXPORT_SYMBOL(__vmsetie);
+EXPORT_SYMBOL(memcpy);
+EXPORT_SYMBOL(memset);
+
+#define DECLARE_EXPORT(name)     \
+	extern void name(void); EXPORT_SYMBOL(name)
+
+/* Symbols found in libgcc that assorted kernel modules need */
+DECLARE_EXPORT(__hexagon_memcpy_likely_aligned_min32bytes_mult8bytes);
+
+DECLARE_EXPORT(__hexagon_divsi3);
+DECLARE_EXPORT(__hexagon_modsi3);
+DECLARE_EXPORT(__hexagon_udivsi3);
+DECLARE_EXPORT(__hexagon_umodsi3);
+
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 13/36] Hexagon: Support dynamic module loading.

[Richard Kuo]

[-- Attachment #1: module.diff --]
[-- Type: text/plain, Size: 5678 bytes --]

Modules should be compiled as ordinary .o's; shared objects are not
supported.

Deleted module functions that already have generic equivalents.

Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/kernel/module.c |  162 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 162 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/kernel/module.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/module.c	2011-09-07 13:00:31.090647031 -0500
@@ -0,0 +1,162 @@
+/*
+ * Kernel module loader for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <asm/module.h>
+#include <linux/elf.h>
+#include <linux/module.h>
+#include <linux/moduleloader.h>
+#include <linux/vmalloc.h>
+
+#if 0
+#define DEBUGP printk
+#else
+#define DEBUGP(fmt , ...)
+#endif
+
+/*
+ * module_frob_arch_sections - tweak got/plt sections.
+ * @hdr - pointer to elf header
+ * @sechdrs - pointer to elf load section headers
+ * @secstrings - symbol names
+ * @mod - pointer to module
+ */
+int module_frob_arch_sections(Elf_Ehdr *hdr, Elf_Shdr *sechdrs,
+				char *secstrings,
+				struct module *mod)
+{
+	unsigned int i;
+	int found = 0;
+
+	/* Look for .plt and/or .got.plt and/or .init.plt sections */
+	for (i = 0; i < hdr->e_shnum; i++) {
+		DEBUGP("Section %d is %s\n", i,
+		       secstrings + sechdrs[i].sh_name);
+		if (strcmp(secstrings + sechdrs[i].sh_name, ".plt") == 0)
+			found = i+1;
+		if (strcmp(secstrings + sechdrs[i].sh_name, ".got.plt") == 0)
+			found = i+1;
+		if (strcmp(secstrings + sechdrs[i].sh_name, ".rela.plt") == 0)
+			found = i+1;
+	}
+
+	/* At this time, we don't support modules comiled with -shared */
+	if (found) {
+		printk(KERN_WARNING
+			"Module '%s' contains unexpected .plt/.got sections.\n",
+			mod->name);
+		/*  return -ENOEXEC;  */
+	}
+
+	return 0;
+}
+
+/*
+ * apply_relocate_add - perform rela relocations.
+ * @sechdrs - pointer to section headers
+ * @strtab - some sort of start address?
+ * @symindex - symbol index offset or something?
+ * @relsec - address to relocate to?
+ * @module - pointer to module
+ *
+ * Perform rela relocations.
+ */
+int apply_relocate_add(Elf_Shdr *sechdrs, const char *strtab,
+			unsigned int symindex, unsigned int relsec,
+			struct module *module)
+{
+	unsigned int i;
+	Elf32_Sym *sym;
+	uint32_t *location;
+	uint32_t value;
+	unsigned int nrelocs = sechdrs[relsec].sh_size / sizeof(Elf32_Rela);
+	Elf32_Rela *rela = (void *)sechdrs[relsec].sh_addr;
+	Elf32_Word sym_info = sechdrs[relsec].sh_info;
+	Elf32_Sym *sym_base = (Elf32_Sym *) sechdrs[symindex].sh_addr;
+	void *loc_base = (void *) sechdrs[sym_info].sh_addr;
+
+	DEBUGP("Applying relocations in section %u to section %u base=%p\n",
+	       relsec, sym_info, loc_base);
+
+	for (i = 0; i < nrelocs; i++) {
+
+		/* Symbol to relocate */
+		sym = sym_base + ELF32_R_SYM(rela[i].r_info);
+
+		/* Where to make the change */
+		location = loc_base + rela[i].r_offset;
+
+		/* `Everything is relative'. */
+		value = sym->st_value + rela[i].r_addend;
+
+		DEBUGP("%d: value=%08x loc=%p reloc=%d symbol=%s\n",
+		       i, value, location, ELF32_R_TYPE(rela[i].r_info),
+		       sym->st_name ?
+		       &strtab[sym->st_name] : "(anonymous)");
+
+		switch (ELF32_R_TYPE(rela[i].r_info)) {
+		case R_HEXAGON_B22_PCREL: {
+			int dist = (int)(value - (uint32_t)location);
+			if ((dist < -0x00800000) ||
+			    (dist >= 0x00800000)) {
+				printk(KERN_ERR
+				       "%s: %s: %08x=%08x-%08x %s\n",
+				       module->name,
+				       "R_HEXAGON_B22_PCREL reloc out of range",
+				       dist, value, (uint32_t)location,
+				       sym->st_name ?
+				       &strtab[sym->st_name] : "(anonymous)");
+				return -ENOEXEC;
+			}
+			DEBUGP("B22_PCREL contents: %08X.\n", *location);
+			*location &= ~0x01ff3fff;
+			*location |= 0x00003fff & dist;
+			*location |= 0x01ff0000 & (dist<<2);
+			DEBUGP("Contents after reloc: %08x\n", *location);
+			break;
+		}
+		case R_HEXAGON_HI16:
+			value = (value>>16) & 0xffff;
+			/* fallthrough */
+		case R_HEXAGON_LO16:
+			*location &= ~0x00c03fff;
+			*location |= value & 0x3fff;
+			*location |= (value & 0xc000) << 8;
+			break;
+		case R_HEXAGON_32:
+			*location = value;
+			break;
+		case R_HEXAGON_32_PCREL:
+			*location = value - (uint32_t)location;
+			break;
+		case R_HEXAGON_PLT_B22_PCREL:
+		case R_HEXAGON_GOTOFF_LO16:
+		case R_HEXAGON_GOTOFF_HI16:
+			printk(KERN_ERR "%s: GOT/PLT relocations unsupported\n",
+			       module->name);
+			return -ENOEXEC;
+		default:
+			printk(KERN_ERR "%s: unknown relocation: %u\n",
+			       module->name,
+			       ELF32_R_TYPE(rela[i].r_info));
+			return -ENOEXEC;
+		}
+	}
+	return 0;
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 14/36] Hexagon: Add signal functions

[Richard Kuo]

[-- Attachment #1: signals.diff --]
[-- Type: text/plain, Size: 17922 bytes --]

Added vDSO support; signal trampoline is no longer on the stack
so it no longer needs to be executable.


Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>

---
 arch/hexagon/include/asm/sigcontext.h |   32 +++
 arch/hexagon/include/asm/signal.h     |   26 ++
 arch/hexagon/include/asm/vdso.h       |   31 +++
 arch/hexagon/kernel/signal.c          |  320 ++++++++++++++++++++++++++++++++++
 arch/hexagon/kernel/trampoline.S      |   35 +++
 arch/hexagon/kernel/vdso.c            |  101 ++++++++++
 6 files changed, 545 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/sigcontext.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/sigcontext.h	2011-09-08 19:22:01.150239048 -0500
@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SIGCONTEXT_H
+#define _ASM_SIGCONTEXT_H
+
+#include <asm/registers.h>
+
+/*
+ * Signal context structure - contains all info to do with the state
+ * before the signal handler was invoked.
+ */
+struct sigcontext {
+	struct pt_regs sc_regs;
+};
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/signal.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/signal.h	2011-09-08 19:22:01.150239048 -0500
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SIGNAL_H
+#define _ASM_SIGNAL_H
+
+extern unsigned long __rt_sigtramp_template[2];
+
+#include <asm-generic/signal.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/signal.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/signal.c	2011-09-08 19:22:01.150239048 -0500
@@ -0,0 +1,320 @@
+/*
+ * Signal support for Hexagon processor
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/linkage.h>
+#include <linux/syscalls.h>
+#include <linux/freezer.h>
+#include <asm/registers.h>
+#include <asm/thread_info.h>
+#include <asm/unistd.h>
+#include <asm/uaccess.h>
+#include <asm/ucontext.h>
+#include <asm/cacheflush.h>
+#include <asm/signal.h>
+#include <asm/vdso.h>
+
+#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
+
+struct sigframe {
+	unsigned long pad[2]; /* Used to be signal tramp */
+	struct sigcontext sc;
+	sigset_t mask;
+};
+
+struct rt_sigframe {
+	unsigned long pad[2]; /* Used to be signal tramp */
+	struct siginfo info;
+	struct ucontext uc;
+};
+
+static void __user *get_sigframe(struct k_sigaction *ka, struct pt_regs *regs,
+			  size_t frame_size)
+{
+	unsigned long sp = regs->r29;
+
+	/* Switch to signal stack if appropriate */
+	if ((ka->sa.sa_flags & SA_ONSTACK) && (sas_ss_flags(sp) == 0))
+		sp = current->sas_ss_sp + current->sas_ss_size;
+
+	return (void __user *)((sp - frame_size) & ~(sizeof(long long) - 1));
+}
+
+static int setup_sigcontext(struct pt_regs *regs, struct sigcontext __user *sc)
+{
+	int err = 0;
+
+	/*
+	 * Need to revisit if we're leaking anything inappropriate to user
+	 * stack.
+	 */
+
+	err |= copy_to_user(&sc->sc_regs, regs, sizeof(struct pt_regs));
+	return err;
+}
+
+static int restore_sigcontext(struct pt_regs *regs,
+			      struct sigcontext __user *sc)
+{
+	int err = 0;
+
+	/*
+	 * Should only restore vmel in case user stack is corrupted,
+	 * bogus status shouldn't be used in the rte.
+	 */
+
+	err |= copy_from_user(regs, &sc->sc_regs,
+			      offsetof(struct pt_regs, hvmer.vmel)
+			      + sizeof(unsigned long));
+	return err;
+}
+
+/*
+ * Setup signal stack frame with siginfo structure
+ */
+static int setup_rt_frame(int signr, struct k_sigaction *ka, siginfo_t *info,
+			  sigset_t *set,  struct pt_regs *regs)
+{
+	int err = 0;
+	struct rt_sigframe __user *frame;
+	struct hexagon_vdso *vdso = current->mm->context.vdso;
+
+	frame = get_sigframe(ka, regs, sizeof(struct rt_sigframe));
+
+	if (!access_ok(VERIFY_WRITE, frame, sizeof(struct rt_sigframe)))
+		goto	sigsegv;
+
+	if (copy_siginfo_to_user(&frame->info, info))
+		goto	sigsegv;
+
+	mb();
+	err |= setup_sigcontext(regs, &frame->uc.uc_mcontext);
+	err |= __copy_to_user(&frame->uc.uc_sigmask, set, sizeof(*set));
+	if (err)
+		goto sigsegv;
+
+	/* Load r0/r1 pair with signumber/siginfo pointer... */
+	regs->r0100 = ((unsigned long long)((unsigned long)&frame->info) << 32)
+		| (unsigned long long)signr;
+	regs->r02 = (unsigned long) &frame->uc;
+	regs->r31 = (unsigned long) vdso->rt_signal_trampoline;
+	pt_psp(regs) = (unsigned long) frame;
+	pt_set_elr(regs, (unsigned long)ka->sa.sa_handler);
+
+	return 0;
+
+sigsegv:
+	force_sigsegv(signr, current);
+	return -EFAULT;
+}
+
+/*
+ * Setup invocation of signal handler
+ */
+static int handle_signal(int sig, siginfo_t *info, struct k_sigaction *ka,
+			 sigset_t *oldset, struct pt_regs *regs)
+{
+	int rc;
+
+	/*
+	 * If we're handling a signal that aborted a system call,
+	 * set up the error return value before adding the signal
+	 * frame to the stack.
+	 */
+
+	if (regs->syscall_nr >= 0) {
+		switch (regs->r00) {
+		case -ERESTART_RESTARTBLOCK:
+		case -ERESTARTNOHAND:
+			regs->r00 = -EINTR;
+			break;
+		case -ERESTARTSYS:
+			if (!(ka->sa.sa_flags & SA_RESTART)) {
+				regs->r00 = -EINTR;
+				break;
+			}
+			/* Fall through */
+		case -ERESTARTNOINTR:
+			regs->r06 = regs->syscall_nr;
+			pt_set_elr(regs, pt_elr(regs) - 4);
+			regs->r00 = regs->restart_r0;
+			break;
+		default:
+			break;
+		}
+	}
+
+	/*
+	 * Set up the stack frame; not doing the SA_SIGINFO thing.  We
+	 * only set up the rt_frame flavor.
+	 */
+	rc = setup_rt_frame(sig, ka, info, oldset, regs);
+
+	/* If there was an error on setup, no signal was delivered. */
+	if (rc)
+		return rc;
+
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
+		sigaddset(&current->blocked, sig);
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
+
+	return 0;
+}
+
+/*
+ * Called from return-from-event code.
+ */
+static void do_signal(struct pt_regs *regs)
+{
+	struct k_sigaction sigact;
+	siginfo_t info;
+	int signo;
+
+	if (!user_mode(regs))
+		return;
+
+	if (try_to_freeze())
+		goto no_signal;
+
+	signo = get_signal_to_deliver(&info, &sigact, regs, NULL);
+
+	if (signo > 0) {
+		sigset_t *oldset;
+
+		if (test_thread_flag(TIF_RESTORE_SIGMASK))
+			oldset = &current->saved_sigmask;
+		else
+			oldset = &current->blocked;
+
+		if (handle_signal(signo, &info, &sigact, oldset, regs) == 0) {
+			/*
+			 * Successful delivery case.  The saved sigmask is
+			 * stored in the signal frame, and will be restored
+			 * by sigreturn.  We can clear the TIF flag.
+			 */
+			clear_thread_flag(TIF_RESTORE_SIGMASK);
+		}
+		return;
+	}
+
+no_signal:
+	/*
+	 * If we came from a system call, handle the restart.
+	 */
+	if (regs->syscall_nr >= 0) {
+		switch (regs->r00) {
+		case -ERESTARTNOHAND:
+		case -ERESTARTSYS:
+		case -ERESTARTNOINTR:
+			regs->r06 = regs->syscall_nr;
+			break;
+		case -ERESTART_RESTARTBLOCK:
+			regs->r06 = __NR_restart_syscall;
+			break;
+		default:
+			goto no_restart;
+		}
+		pt_set_elr(regs, pt_elr(regs) - 4);
+		regs->r00 = regs->restart_r0;
+	}
+
+no_restart:
+	/* If there's no signal to deliver, put the saved sigmask back */
+	if (test_thread_flag(TIF_RESTORE_SIGMASK)) {
+		clear_thread_flag(TIF_RESTORE_SIGMASK);
+		sigprocmask(SIG_SETMASK, &current->saved_sigmask, NULL);
+	}
+}
+
+void do_notify_resume(struct pt_regs *regs, unsigned long thread_info_flags)
+{
+	if (thread_info_flags & _TIF_SIGPENDING)
+		do_signal(regs);
+
+	if (thread_info_flags & _TIF_NOTIFY_RESUME) {
+		clear_thread_flag(TIF_NOTIFY_RESUME);
+		if (current->replacement_session_keyring)
+			key_replace_session_keyring();
+	}
+}
+
+/*
+ * Architecture-specific wrappers for signal-related system calls
+ */
+asmlinkage int sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss)
+{
+	struct pt_regs *regs = current_thread_info()->regs;
+
+	return do_sigaltstack(uss, uoss, regs->r29);
+}
+
+asmlinkage int sys_rt_sigreturn(void)
+{
+	struct pt_regs *regs = current_thread_info()->regs;
+	struct rt_sigframe __user *frame;
+	sigset_t blocked;
+
+	frame = (struct rt_sigframe __user *)pt_psp(regs);
+	if (!access_ok(VERIFY_READ, frame, sizeof(*frame)))
+		goto badframe;
+	if (__copy_from_user(&blocked, &frame->uc.uc_sigmask, sizeof(blocked)))
+		goto badframe;
+
+	sigdelsetmask(&blocked, ~_BLOCKABLE);
+	spin_lock_irq(&current->sighand->siglock);
+	current->blocked = blocked;
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
+
+	if (restore_sigcontext(regs, &frame->uc.uc_mcontext))
+		goto badframe;
+
+	/* Restore the user's stack as well */
+	pt_psp(regs) = regs->r29;
+
+	/*
+	 * Leave a trace in the stack frame that this was a sigreturn.
+	 * If the system call is to replay, we've already restored the
+	 * number in the GPR slot and it will be regenerated on the
+	 * new system call trap entry. Note that if restore_sigcontext()
+	 * did something other than a bulk copy of the pt_regs struct,
+	 * we could avoid this assignment by simply not overwriting
+	 * regs->syscall_nr.
+	 */
+	regs->syscall_nr = __NR_rt_sigreturn;
+
+	/*
+	 * If we were meticulous, we'd only call this if we knew that
+	 * we were actually going to use an alternate stack, and we'd
+	 * consider any error to be fatal.  What we do here, in common
+	 * with many other architectures, is call it blindly and only
+	 * consider the -EFAULT return case to be proof of a problem.
+	 */
+	if (do_sigaltstack(&frame->uc.uc_stack, NULL, pt_psp(regs)) == -EFAULT)
+		goto badframe;
+
+	return 0;
+
+badframe:
+	force_sig(SIGSEGV, current);
+	return 0;
+}
Index: linux-hexagon-kernel/arch/hexagon/kernel/trampoline.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/trampoline.S	2011-09-08 19:22:01.150239048 -0500
@@ -0,0 +1,35 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * Trampoline sequences to be copied onto user stack.
+ * This consumes a little more space than hand-assembling
+ * immediate constants for use in C, but is more portable
+ * to future tweaks to the Hexagon instruction set.
+ */
+
+#include <asm/unistd.h>
+
+/*  Sig trampolines - call sys_sigreturn or sys_rt_sigreturn as appropriate */
+
+/*  plain sigreturn is gone.  */
+
+	.globl __rt_sigtramp_template
+__rt_sigtramp_template:
+	r6 = #__NR_rt_sigreturn;
+	trap0(#1);
Index: linux-hexagon-kernel/arch/hexagon/include/asm/vdso.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/vdso.h	2011-09-08 19:36:15.052799920 -0500
@@ -0,0 +1,31 @@
+/*
+ * vDSO implementation for Hexagon
+ *
+ * Copyright (c) 2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef __ASM_VDSO_H
+#define __ASM_VDSO_H
+
+#include <linux/types.h>
+
+struct hexagon_vdso {
+	u32 rt_signal_trampoline[2];
+};
+
+#endif /* __ASM_VDSO_H */
+
Index: linux-hexagon-kernel/arch/hexagon/kernel/vdso.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vdso.c	2011-09-08 19:22:01.150239048 -0500
@@ -0,0 +1,101 @@
+/*
+ * vDSO implementation for Hexagon
+ *
+ * Copyright (c) 2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/err.h>
+#include <linux/mm.h>
+#include <linux/vmalloc.h>
+
+#include <asm/vdso.h>
+
+static struct page *vdso_page;
+
+/* Create a vDSO page holding the signal trampoline.
+ * We want this for a non-executable stack.
+ */
+static int __init vdso_init(void)
+{
+	struct hexagon_vdso *vdso;
+
+	vdso_page = alloc_page(GFP_KERNEL);
+	if (!vdso_page)
+		panic("Cannot allocate vdso");
+
+	vdso = vmap(&vdso_page, 1, 0, PAGE_KERNEL);
+	if (!vdso)
+		panic("Cannot map vdso");
+	clear_page(vdso);
+
+	/* Install the signal trampoline; currently looks like this:
+	 *	r6 = #__NR_rt_sigreturn;
+	 *	trap0(#1);
+	 */
+	vdso->rt_signal_trampoline[0] = __rt_sigtramp_template[0];
+	vdso->rt_signal_trampoline[1] = __rt_sigtramp_template[1];
+
+	vunmap(vdso);
+
+	return 0;
+}
+arch_initcall(vdso_init);
+
+/*
+ * Called from binfmt_elf.  Create a VMA for the vDSO page.
+ */
+int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
+{
+	int ret;
+	unsigned long vdso_base;
+	struct mm_struct *mm = current->mm;
+
+	down_write(&mm->mmap_sem);
+
+	/* Try to get it loaded right near ld.so/glibc. */
+	vdso_base = STACK_TOP;
+
+	vdso_base = get_unmapped_area(NULL, vdso_base, PAGE_SIZE, 0, 0);
+	if (IS_ERR_VALUE(vdso_base)) {
+		ret = vdso_base;
+		goto up_fail;
+	}
+
+	/* MAYWRITE to allow gdb to COW and set breakpoints. */
+	ret = install_special_mapping(mm, vdso_base, PAGE_SIZE,
+				      VM_READ|VM_EXEC|
+				      VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC|
+				      VM_ALWAYSDUMP,
+				      &vdso_page);
+
+	if (ret)
+		goto up_fail;
+
+	mm->context.vdso = (void *)vdso_base;
+
+up_fail:
+	up_write(&mm->mmap_sem);
+	return ret;
+}
+
+const char *arch_vma_name(struct vm_area_struct *vma)
+{
+	if (vma->vm_mm && vma->vm_start == (long)vma->vm_mm->context.vdso)
+		return "[vdso]";
+	return NULL;
+}
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 14/36] Hexagon: Add signal functions

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:01 Richard Kuo wrote:
> +/*
> + * Signal context structure - contains all info to do with the state
> + * before the signal handler was invoked.
> + */
> +struct sigcontext {
> +       struct pt_regs sc_regs;
> +};

I've recently discussed this with Martin Schwidefsky for an unrelated reason,
but I think now that you should define the sigcontext independent of pt_regs.
The pt_regs should really be internal 

> +struct sigframe {
> +       unsigned long pad[2]; /* Used to be signal tramp */
> +       struct sigcontext sc;
> +       sigset_t mask;
> +};
> +
> +struct rt_sigframe {
> +       unsigned long pad[2]; /* Used to be signal tramp */
> +       struct siginfo info;
> +       struct ucontext uc;
> +};

I hadn't noticed this earlier, but I think you can remove the struct sigframe
definition. There should be absolutely no need to have both sigframe and
rt_sigframe.

	Arnd

Re: [patch v3 14/36] Hexagon: Add signal functions

[Benjamin Herrenschmidt]

On Thu, 2011-09-08 at 20:09 -0500, Richard Kuo wrote:

> +#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
> +
> +struct sigframe {
> +	unsigned long pad[2]; /* Used to be signal tramp */
> +	struct sigcontext sc;
> +	sigset_t mask;
> +};
> +
> +struct rt_sigframe {
> +	unsigned long pad[2]; /* Used to be signal tramp */
> +	struct siginfo info;
> +	struct ucontext uc;
> +};

Why did you make two types of signal frames ? It's an unnecessary
complication... Dunno if you can still fix that (if you can still
control your ABI that is), but you should look how ppc64 uses the same
frame for RT and non-RT, avoids duplication of bugs :-)

Cheers,
Ben.

[patch v3 15/36] Hexagon: Add init_task and process functions

[Richard Kuo]

[-- Attachment #1: init-task.diff --]
[-- Type: text/plain, Size: 10593 bytes --]

Fixed potential race-to-sleep condition in cpu_idle() and calling schedule
with irq's disabled.  I still have to have that extra enable/disbale in the
while() loop else the interrupt never gets serviced, and thus need_resched()
will always be false.

If called with interrupts disabled, our __vmwait() actually just returns
without servicing it.


Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/kernel/init_task.c |   54 +++++++
 arch/hexagon/kernel/process.c   |  279 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 333 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/kernel/init_task.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/init_task.c	2011-09-03 20:14:59.634981139 -0500
@@ -0,0 +1,54 @@
+/*
+ * Init task definition
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/init_task.h>
+#include <linux/fs.h>
+#include <linux/mqueue.h>
+#include <asm/thread_info.h>
+#include <asm/uaccess.h>
+#include <asm/pgtable.h>
+
+static struct signal_struct init_signals = INIT_SIGNALS(init_signals);
+static struct sighand_struct init_sighand = INIT_SIGHAND(init_sighand);
+
+/*
+ * Initial thread structure.
+ *
+ * We need to make sure that this is 8192-byte aligned due to the
+ * way process stacks are handled. This is done by making sure
+ * the linker maps this in the .text segment right after head.S,
+ * and making head.S ensure the proper alignment.
+ */
+union thread_union init_thread_union
+	__attribute__((__section__(".data.init_task"),
+		__aligned__(THREAD_SIZE))) = {
+			INIT_THREAD_INFO(init_task)
+		};
+
+/*
+ * Initial task structure.
+ *
+ * All other task structs will be allocated on slabs in fork.c
+ */
+struct task_struct init_task = INIT_TASK(init_task);
+EXPORT_SYMBOL(init_task);
Index: linux-hexagon-kernel/arch/hexagon/kernel/process.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/process.c	2011-09-03 20:14:59.634981139 -0500
@@ -0,0 +1,279 @@
+/*
+ * Process creation support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/sched.h>
+#include <linux/types.h>
+#include <linux/module.h>
+#include <linux/tick.h>
+#include <linux/uaccess.h>
+#include <linux/slab.h>
+
+/*
+ * Kernel thread creation.  The desired kernel function is "wrapped"
+ * in the kernel_thread_helper function, which does cleanup
+ * afterwards.
+ */
+static void __noreturn kernel_thread_helper(void *arg, int (*fn)(void *))
+{
+	do_exit(fn(arg));
+}
+
+int kernel_thread(int (*fn)(void *), void *arg, unsigned long flags)
+{
+	struct pt_regs regs;
+
+	memset(&regs, 0, sizeof(regs));
+	/*
+	 * Yes, we're exploting illicit knowledge of the ABI here.
+	 */
+	regs.r00 = (unsigned long) arg;
+	regs.r01 = (unsigned long) fn;
+	pt_set_elr(&regs, (unsigned long)kernel_thread_helper);
+	pt_set_kmode(&regs);
+
+	return do_fork(flags|CLONE_VM|CLONE_UNTRACED, 0, &regs, 0, NULL, NULL);
+}
+EXPORT_SYMBOL(kernel_thread);
+
+/*
+ * Program thread launch.  Often defined as a macro in processor.h,
+ * but we're shooting for a small footprint and it's not an inner-loop
+ * performance-critical operation.
+ *
+ * The Hexagon ABI specifies that R28 is zero'ed before program launch,
+ * so that gets automatically done here.  If we ever stop doing that here,
+ * we'll probably want to define the ELF_PLAT_INIT macro.
+ */
+void start_thread(struct pt_regs *regs, unsigned long pc, unsigned long sp)
+{
+	/* Set to run with user-mode data segmentation */
+	set_fs(USER_DS);
+	/* We want to zero all data-containing registers. Is this overkill? */
+	memset(regs, 0, sizeof(*regs));
+	/* We might want to also zero all Processor registers here */
+	pt_set_usermode(regs);
+	pt_set_elr(regs, pc);
+	pt_set_rte_sp(regs, sp);
+}
+
+/*
+ *  Spin, or better still, do a hardware or VM wait instruction
+ *  If hardware or VM offer wait termination even though interrupts
+ *  are disabled.
+ */
+static void default_idle(void)
+{
+	__vmwait();
+}
+
+void (*idle_sleep)(void) = default_idle;
+
+void cpu_idle(void)
+{
+	while (1) {
+		tick_nohz_stop_sched_tick(1);
+		local_irq_disable();
+		while (!need_resched()) {
+			idle_sleep();
+			/*  interrupts wake us up, but aren't serviced  */
+			local_irq_enable();	/* service interrupt   */
+			local_irq_disable();
+		}
+		local_irq_enable();
+		tick_nohz_restart_sched_tick();
+		schedule();
+	}
+}
+
+/*
+ *  Return saved PC of a blocked thread
+ */
+unsigned long thread_saved_pc(struct task_struct *tsk)
+{
+	return 0;
+}
+
+/*
+ * Copy architecture-specific thread state
+ */
+int copy_thread(unsigned long clone_flags, unsigned long usp,
+		unsigned long unused, struct task_struct *p,
+		struct pt_regs *regs)
+{
+	struct thread_info *ti = task_thread_info(p);
+	struct hexagon_switch_stack *ss;
+	struct pt_regs *childregs;
+	asmlinkage void ret_from_fork(void);
+
+	childregs = (struct pt_regs *) (((unsigned long) ti + THREAD_SIZE) -
+					sizeof(*childregs));
+
+	memcpy(childregs, regs, sizeof(*childregs));
+	ti->regs = childregs;
+
+	/*
+	 * Establish kernel stack pointer and initial PC for new thread
+	 */
+	ss = (struct hexagon_switch_stack *) ((unsigned long) childregs -
+						    sizeof(*ss));
+	ss->lr = (unsigned long)ret_from_fork;
+	p->thread.switch_sp = ss;
+
+	/* If User mode thread, set pt_reg stack pointer as per parameter */
+	if (user_mode(childregs)) {
+		pt_set_rte_sp(childregs, usp);
+
+		/* Child sees zero return value */
+		childregs->r00 = 0;
+
+		/*
+		 * The clone syscall has the C signature:
+		 * int [r0] clone(int flags [r0],
+		 *           void *child_frame [r1],
+		 *           void *parent_tid [r2],
+		 *           void *child_tid [r3],
+		 *           void *thread_control_block [r4]);
+		 * ugp is used to provide TLS support.
+		 */
+		if (clone_flags & CLONE_SETTLS)
+			childregs->ugp = childregs->r04;
+
+		/*
+		 * Parent sees new pid -- not necessary, not even possible at
+		 * this point in the fork process
+		 * Might also want to set things like ti->addr_limit
+		 */
+	} else {
+		/*
+		 * If kernel thread, resume stack is kernel stack base.
+		 * Note that this is pointer arithmetic on pt_regs *
+		 */
+		pt_set_rte_sp(childregs, (unsigned long)(childregs + 1));
+		/*
+		 * We need the current thread_info fast path pointer
+		 * set up in pt_regs.  The register to be used is
+		 * parametric for assembler code, but the mechanism
+		 * doesn't drop neatly into C.  Needs to be fixed.
+		 */
+		childregs->THREADINFO_REG = (unsigned long) ti;
+	}
+
+	/*
+	 * thread_info pointer is pulled out of task_struct "stack"
+	 * field on switch_to.
+	 */
+	p->stack = (void *)ti;
+
+	return 0;
+}
+
+/*
+ * Release any architecture-specific resources locked by thread
+ */
+void release_thread(struct task_struct *dead_task)
+{
+}
+
+/*
+ * Free any architecture-specific thread data structures, etc.
+ */
+void exit_thread(void)
+{
+}
+
+/*
+ * Some archs flush debug and FPU info here
+ */
+void flush_thread(void)
+{
+}
+
+/*
+ * The "wait channel" terminology is archaic, but what we want
+ * is an identification of the point at which the scheduler
+ * was invoked by a blocked thread.
+ */
+unsigned long get_wchan(struct task_struct *p)
+{
+	unsigned long fp, pc;
+	unsigned long stack_page;
+	int count = 0;
+	if (!p || p == current || p->state == TASK_RUNNING)
+		return 0;
+
+	stack_page = (unsigned long)task_stack_page(p);
+	fp = ((struct hexagon_switch_stack *)p->thread.switch_sp)->fp;
+	do {
+		if (fp < (stack_page + sizeof(struct thread_info)) ||
+			fp >= (THREAD_SIZE - 8 + stack_page))
+			return 0;
+		pc = ((unsigned long *)fp)[1];
+		if (!in_sched_functions(pc))
+			return pc;
+		fp = *(unsigned long *) fp;
+	} while (count++ < 16);
+
+	return 0;
+}
+
+/*
+ * Borrowed from PowerPC -- basically allow smaller kernel stacks if we
+ * go crazy with the page sizes.
+ */
+#if THREAD_SHIFT < PAGE_SHIFT
+
+static struct kmem_cache *thread_info_cache;
+
+struct thread_info *alloc_thread_info_node(struct task_struct *tsk, int node)
+{
+	struct thread_info *ti;
+
+	ti = kmem_cache_alloc_node(thread_info_cache, GFP_KERNEL, node);
+	if (unlikely(ti == NULL))
+		return NULL;
+#ifdef CONFIG_DEBUG_STACK_USAGE
+	memset(ti, 0, THREAD_SIZE);
+#endif
+	return ti;
+}
+
+void free_thread_info(struct thread_info *ti)
+{
+	kmem_cache_free(thread_info_cache, ti);
+}
+
+/*  Weak symbol; called by init/main.c  */
+
+void thread_info_cache_init(void)
+{
+	thread_info_cache = kmem_cache_create("thread_info", THREAD_SIZE,
+					      THREAD_SIZE, 0, NULL);
+	BUG_ON(thread_info_cache == NULL);
+}
+
+#endif /* THREAD_SHIFT < PAGE_SHIFT */
+
+/*
+ * Required placeholder.
+ */
+int dump_fpu(struct pt_regs *regs, elf_fpregset_t *fpu)
+{
+	return 0;
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 16/36] Hexagon: Add startup code

[Richard Kuo]

[-- Attachment #1: startup.diff --]
[-- Type: text/plain, Size: 10634 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/kernel/head.S  |  162 ++++++++++++++++++++++++++++++++++++++++++++
 arch/hexagon/kernel/reset.c |   38 ++++++++++
 arch/hexagon/kernel/setup.c |  145 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 345 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/kernel/head.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/head.S	2011-09-07 13:00:31.040646956 -0500
@@ -0,0 +1,162 @@
+/*
+ * Early kernel startup code for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/linkage.h>
+#include <linux/init.h>
+#include <asm/asm-offsets.h>
+#include <asm/mem-layout.h>
+#include <asm/vm_mmu.h>
+#include <asm/page.h>
+
+       	__INIT
+ENTRY(stext)
+	/*
+	 * VMM will already have set up true vector page, MMU, etc.
+	 * To set up initial kernel identity map, we have to pass
+	 * the VMM a pointer to some canonical page tables. In
+	 * this implementation, we're assuming that we've got
+	 * them precompiled. Generate value in R24, as we'll need
+	 * it again shortly.
+	 */
+	r24.L = #LO(swapper_pg_dir)
+	r24.H = #HI(swapper_pg_dir)
+
+	/*
+	 * Symbol is kernel segment address, but we need
+	 * the logical/physical address.
+	 */
+	r24 = asl(r24, #2)
+	r24 = lsr(r24, #2)
+
+	r0 = r24
+
+	/*
+	 * Initialize a 16MB PTE to make the virtual and physical
+	 * addresses where the kernel was loaded be identical.
+	 */
+#define PTE_BITS ( __HVM_PTE_R | __HVM_PTE_W | __HVM_PTE_X	\
+		  | __HEXAGON_C_WB_L2 << 6			\
+		  | __HVM_PDE_S_4MB)
+
+	r1 = pc
+	r2.H = #0xffc0
+	r2.L = #0x0000
+	r1 = and(r1,r2)		/* round PC to 4MB boundary	*/
+	r2 = lsr(r1, #22)	/* 4MB page number		*/
+	r2 = asl(r2, #2)	/* times sizeof(PTE) (4bytes)	*/
+	r0 = add(r0,r2)		/* r0 = address of correct PTE	*/
+	r2 = #PTE_BITS
+	r1 = add(r1,r2)		/* r1 = 4MB PTE for the first entry	*/
+	r2.h = #0x0040
+	r2.l = #0x0000		/* 4MB	*/
+	memw(r0 ++ #4) = r1
+	r1 = add(r1, r2)
+	memw(r0 ++ #4) = r1
+
+	r0 = r24
+
+	/*
+	 * The subroutine wrapper around the virtual instruction touches
+	 * no memory, so we should be able to use it even here.
+	 */
+	call	__vmnewmap;
+
+	/*  Jump into virtual address range.  */
+
+	r31.h = #hi(__head_s_vaddr_target)
+	r31.l = #lo(__head_s_vaddr_target)
+	jumpr r31
+
+	/*  Insert trippy space effects.  */
+
+__head_s_vaddr_target:
+	/*
+	 * Tear down VA=PA translation now that we are running
+	 * in the desgnated kernel segments.
+	 */
+	r0 = #__HVM_PDE_S_INVALID
+	r1 = r24
+	loop0(1f,#0x100)
+1:
+	{
+		memw(R1 ++ #4) = R0
+	}:endloop0
+
+	r0 = r24
+	call __vmnewmap
+
+	/*  Go ahead and install the trap0 return so angel calls work  */
+	r0.h = #hi(_K_provisional_vec)
+	r0.l = #lo(_K_provisional_vec)
+	call __vmsetvec
+
+	/*
+	 * OK, at this point we should start to be much more careful,
+	 * we're going to enter C code and start touching memory
+	 * in all sorts of places.
+	 * This means:
+	 *      SGP needs to be OK
+	 *	Need to lock shared resources
+	 *	A bunch of other things that will cause
+	 * 	all kinds of painful bugs
+	 */
+
+	/*
+	 * Stack pointer should be pointed at the init task's
+	 * thread stack, which should have been declared in arch/init_task.c.
+	 * So uhhhhh...
+	 * It's accessible via the init_thread_union, which is a union
+	 * of a thread_info struct and a stack; of course, the top
+	 * of the stack is not for you.  The end of the stack
+	 * is simply init_thread_union + THREAD_SIZE.
+	 */
+
+	{r29.H = #HI(init_thread_union); r0.H = #HI(_THREAD_SIZE); }
+	{r29.L = #LO(init_thread_union); r0.L = #LO(_THREAD_SIZE); }
+
+	/*  initialize the register used to point to current_thread_info */
+	/*  Fixme:  THREADINFO_REG can't be R2 because of that memset thing. */
+	{r29 = add(r29,r0); THREADINFO_REG = r29; }
+
+	/*  Hack:  zero bss; */
+	{ r0.L = #LO(__bss_start);  r1 = #0; r2.l = #LO(__bss_stop); }
+	{ r0.H = #HI(__bss_start);           r2.h = #HI(__bss_stop); }
+
+	r2 = sub(r2,r0);
+	call memset;
+
+	/* Time to make the doughnuts.   */
+	call start_kernel
+
+	/*
+	 * Should not reach here.
+	 */
+1:
+	jump 1b
+
+.p2align PAGE_SHIFT
+ENTRY(external_cmdline_buffer)
+        .fill _PAGE_SIZE,1,0
+
+.data
+.p2align PAGE_SHIFT
+ENTRY(empty_zero_page)
+        .fill _PAGE_SIZE,1,0
Index: linux-hexagon-kernel/arch/hexagon/kernel/reset.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/reset.c	2011-09-03 20:14:59.644981167 -0500
@@ -0,0 +1,38 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/smp.h>
+#include <asm/hexagon_vm.h>
+
+void machine_power_off(void)
+{
+	smp_send_stop();
+	__vmstop();
+}
+
+void machine_halt(void)
+{
+}
+
+void machine_restart(char *cmd)
+{
+}
+
+void pm_power_off(void)
+{
+}
Index: linux-hexagon-kernel/arch/hexagon/kernel/setup.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/setup.c	2011-09-07 13:00:31.120647076 -0500
@@ -0,0 +1,145 @@
+/*
+ * Arch related setup for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/init.h>
+#include <linux/bootmem.h>
+#include <linux/mmzone.h>
+#include <linux/mm.h>
+#include <linux/seq_file.h>
+#include <linux/console.h>
+#include <linux/of_fdt.h>
+#include <asm/io.h>
+#include <asm/sections.h>
+#include <asm/setup.h>
+#include <asm/processor.h>
+#include <asm/hexagon_vm.h>
+#include <asm/vm_mmu.h>
+#include <asm/time.h>
+#ifdef CONFIG_OF
+#include <asm/prom.h>
+#endif
+
+char cmd_line[COMMAND_LINE_SIZE];
+static char default_command_line[COMMAND_LINE_SIZE] __initdata = CONFIG_CMDLINE;
+
+int on_simulator;
+
+void __cpuinit calibrate_delay(void)
+{
+	loops_per_jiffy = thread_freq_mhz * 1000000 / HZ;
+}
+
+/*
+ * setup_arch -  high level architectural setup routine
+ * @cmdline_p: pointer to pointer to command-line arguments
+ */
+
+void __init setup_arch(char **cmdline_p)
+{
+	char *p = &external_cmdline_buffer;
+
+	/*
+	 * These will eventually be pulled in via either some hypervisor
+	 * or devicetree description.  Hardwiring for now.
+	 */
+	pcycle_freq_mhz = 600;
+	thread_freq_mhz = 100;
+	sleep_clk_freq = 32000;
+
+	/*
+	 * Set up event bindings to handle exceptions and interrupts.
+	 */
+	__vmsetvec(_K_VM_event_vector);
+
+	/*
+	 * Simulator has a few differences from the hardware.
+	 * For now, check uninitialized-but-mapped memory
+	 * prior to invoking setup_arch_memory().
+	 */
+	if (*(int *)((unsigned long)_end + 8) == 0x1f1f1f1f)
+		on_simulator = 1;
+	else
+		on_simulator = 0;
+
+	if (p[0] != '\0')
+		strlcpy(boot_command_line, p, COMMAND_LINE_SIZE);
+	else
+		strlcpy(boot_command_line, default_command_line,
+			COMMAND_LINE_SIZE);
+
+	/*
+	 * boot_command_line and the value set up by setup_arch
+	 * are both picked up by the init code. If no reason to
+	 * make them different, pass the same pointer back.
+	 */
+	strlcpy(cmd_line, boot_command_line, COMMAND_LINE_SIZE);
+	*cmdline_p = cmd_line;
+
+	parse_early_param();
+
+	setup_arch_memory();
+
+#ifdef CONFIG_SMP
+	smp_start_cpus();
+#endif
+}
+
+/*
+ * Functions for dumping CPU info via /proc
+ * Probably should move to kernel/proc.c or something.
+ */
+static void *c_start(struct seq_file *m, loff_t *pos)
+{
+	return *pos < nr_cpu_ids ? (void *)((unsigned long) *pos + 1) : NULL;
+}
+
+static void *c_next(struct seq_file *m, void *v, loff_t *pos)
+{
+	++*pos;
+	return c_start(m, pos);
+}
+
+static void c_stop(struct seq_file *m, void *v)
+{
+}
+
+/*
+ * Eventually this will dump information about
+ * CPU properties like ISA level, TLB size, etc.
+ */
+static int show_cpuinfo(struct seq_file *m, void *v)
+{
+	int cpu = (unsigned long) v - 1;
+
+	seq_printf(m, "processor\t: %d\n", cpu);
+	seq_printf(m, "model name\t: Hexagon Virtual Machine\n");
+	seq_printf(m, "BogoMips\t: %lu.%02lu\n",
+		(loops_per_jiffy * HZ) / 500000,
+		((loops_per_jiffy * HZ) / 5000) % 100);
+	seq_printf(m, "\n");
+	return 0;
+}
+
+const struct seq_operations cpuinfo_op = {
+	.start  = &c_start,
+	.next   = &c_next,
+	.stop   = &c_stop,
+	.show   = &show_cpuinfo,
+};

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 17/36] Hexagon: Add interrupts

[Richard Kuo]

[-- Attachment #1: interrupts.diff --]
[-- Type: text/plain, Size: 7370 bytes --]

Removed init_IRQ wrapper and fixed up some comments.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/irq.h |   36 ++++++++++++++++
 arch/hexagon/kernel/irq.c      |   67 ++++++++++++++++++++++++++++++
 arch/hexagon/kernel/irq_cpu.c  |   90 +++++++++++++++++++++++++++++++++++++++++
 3 files changed, 193 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/irq.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/irq.h	2011-09-07 13:00:30.930646788 -0500
@@ -0,0 +1,36 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_IRQ_H_
+#define _ASM_IRQ_H_
+
+/* Number of first-level interrupts associated with the CPU core. */
+#define HEXAGON_CPUINTS 32
+
+/*
+ * Must define NR_IRQS before including <asm-generic/irq.h>
+ * 64 == the two SIRC's, 176 == the two gpio's
+ *
+ * IRQ configuration is still in flux; defining this to a comfortably
+ * large number.
+ */
+#define NR_IRQS 512
+
+#include <asm-generic/irq.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/irq.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/irq.c	2011-09-07 13:00:31.060646986 -0500
@@ -0,0 +1,67 @@
+/*
+ * Interrupt support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/kernel_stat.h>
+#include <linux/interrupt.h>
+#include <linux/seq_file.h>
+#include <linux/version.h>
+
+int show_interrupts(struct seq_file *p, void *v)
+{
+	int i = *(loff_t *)v, cpu;
+	struct irqaction *action;
+	unsigned long flags;
+
+	if (i == 0) {
+		seq_puts(p, "           ");
+		for_each_online_cpu(cpu)
+			seq_printf(p, "CPU%d       ", cpu);
+		seq_putc(p, '\n');
+	}
+
+	if (i < NR_IRQS) {
+		struct irq_desc *desc = irq_to_desc(i);
+
+		raw_spin_lock_irqsave(&desc->lock, flags);
+
+		action = irq_desc[i].action;
+		if (action) {
+			seq_printf(p, "%3d: ", i);
+			for_each_online_cpu(cpu) {
+				seq_printf(p, "%10u ", kstat_irqs_cpu(i, cpu));
+			}
+			seq_printf(p, " %8s",
+				   irq_desc_get_chip(&irq_desc[i])->name ?
+				   : "-");
+			seq_printf(p, "  %s", action->name);
+			for (action = action->next; action;
+			     action = action->next) {
+				seq_printf(p, ", %s", action->name);
+			}
+
+			seq_putc(p, '\n');
+		}
+		raw_spin_unlock_irqrestore(&desc->lock, flags);
+	}
+
+	return 0;
+}
Index: linux-hexagon-kernel/arch/hexagon/kernel/irq_cpu.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/irq_cpu.c	2011-09-07 13:00:31.080647016 -0500
@@ -0,0 +1,90 @@
+/*
+ * First-level interrupt controller model for Hexagon.
+ *
+ * Copyright (c) 2010-2011 Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/interrupt.h>
+#include <asm/irq.h>
+#include <asm/hexagon_vm.h>
+
+static void mask_irq(struct irq_data *data)
+{
+	__vmintop_locdis((long) data->irq);
+}
+
+static void mask_irq_num(unsigned int irq)
+{
+	__vmintop_locdis((long) irq);
+}
+
+static void unmask_irq(struct irq_data *data)
+{
+	__vmintop_locen((long) data->irq);
+}
+
+/*  This is actually all we need for handle_fasteoi_irq  */
+static void eoi_irq(struct irq_data *data)
+{
+	__vmintop_globen((long) data->irq);
+}
+
+/* Power mamangement wake call. We don't need this, however,
+ * if this is absent, then an -ENXIO error is returned to the
+ * msm_serial driver, and it fails to correctly initialize.
+ * This is a bug in the msm_serial driver, but, for now, we
+ * work around it here, by providing this bogus handler.
+ * XXX FIXME!!! remove this when msm_serial is fixed.
+ */
+static int set_wake(struct irq_data *data, unsigned int on)
+{
+	return 0;
+}
+
+static struct irq_chip hexagon_irq_chip = {
+	.name		= "HEXAGON",
+	.irq_mask	= mask_irq,
+	.irq_unmask	= unmask_irq,
+	.irq_set_wake	= set_wake,
+	.irq_eoi	= eoi_irq
+};
+
+/**
+ * The hexagon core comes with a first-level interrupt controller
+ * with 32 total possible interrupts.  When the core is embedded
+ * into different systems/platforms, it is typically wrapped by
+ * macro cells that provide one or more second-level interrupt
+ * controllers that are cascaded into one or more of the first-level
+ * interrupts handled here. The precise wiring of these other
+ * irqs varies from platform to platform, and are set up & configured
+ * in the platform-specific files.
+ *
+ * The first-level interrupt controller is wrapped by the VM, which
+ * virtualizes the interrupt controller for us.  It provides a very
+ * simple, fast & efficient API, and so the fasteoi handler is
+ * appropriate for this case.
+ */
+void __init init_IRQ(void)
+{
+	int irq;
+
+	for (irq = 0; irq < HEXAGON_CPUINTS; irq++) {
+		mask_irq_num(irq);
+		irq_set_chip_and_handler(irq, &hexagon_irq_chip,
+						 handle_fasteoi_irq);
+	}
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 17/36] Hexagon: Add interrupts

[Thomas Gleixner]

On Thu, 8 Sep 2011, Richard Kuo wrote:

> Removed init_IRQ wrapper and fixed up some comments.

> +int show_interrupts(struct seq_file *p, void *v)

Please use the generic version of show_interrupts.

Otherwise please add:

Reviewed-by: Thomas Gleixner <tglx@linutronix.de>

Re: [patch v3 17/36] Hexagon: Add interrupts

[Linas Vepstas (Code Aurora)]

On Thu, Sep 08, 2011 at 08:09:04PM -0500, Richard Kuo wrote:
> Removed init_IRQ wrapper and fixed up some comments.
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
> Acked-by: Arnd Bergmann <arnd@arndb.de>

This should have a 
Signed-off-by: Linas Vepstas <linas@codeaurora.org>

since I think I wrote about 100% of some of these files... 

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 18/36] Hexagon: Add time and timer functions

[Richard Kuo]

[-- Attachment #1: timer.diff --]
[-- Type: text/plain, Size: 11383 bytes --]

Cleaned up iomem and register declarations, unnecessary
SMP ifdefs, cpu_relax() in udelay.

Platform stuff is in development; platform defines will be cleaned up.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

---
 arch/hexagon/include/asm/time.h       |   29 +++
 arch/hexagon/include/asm/timer-regs.h |   39 ++++
 arch/hexagon/kernel/time.c            |  275 ++++++++++++++++++++++++++++++++++
 3 files changed, 343 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/time.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/time.h	2011-09-07 13:00:30.970646848 -0500
@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef ASM_TIME_H
+#define ASM_TIME_H
+
+extern cycles_t        pcycle_freq_mhz;
+extern cycles_t        thread_freq_mhz;
+extern cycles_t        sleep_clk_freq;
+
+void setup_percpu_clockdev(void);
+void ipi_timer(void);
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/timer-regs.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/timer-regs.h	2011-09-07 13:00:30.970646848 -0500
@@ -0,0 +1,39 @@
+/*
+ * Timer support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_TIMER_REGS_H
+#define _ASM_TIMER_REGS_H
+
+/*  This stuff should go into a platform specific file  */
+#define TCX0_CLK_RATE		19200
+#define TIMER_ENABLE		0
+#define TIMER_CLR_ON_MATCH	1
+
+/*
+ * 8x50 HDD Specs 5-8.  Simulator co-sim not fixed until
+ * release 1.1, and then it's "adjustable" and probably not defaulted.
+ */
+#define RTOS_TIMER_INT		3
+#ifdef CONFIG_HEXAGON_COMET
+#define RTOS_TIMER_REGS_ADDR	0xAB000000UL
+#endif
+#define SLEEP_CLK_RATE		32000
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/time.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/time.c	2011-09-07 13:00:31.160647137 -0500
@@ -0,0 +1,275 @@
+/*
+ * Time related functions for Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/init.h>
+#include <linux/clockchips.h>
+#include <linux/clocksource.h>
+#include <linux/interrupt.h>
+#include <linux/err.h>
+#include <linux/platform_device.h>
+#include <linux/ioport.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+
+#include <asm/timer-regs.h>
+#include <asm/hexagon_vm.h>
+
+/*
+ * For the clocksource we need:
+ *	pcycle frequency (600MHz)
+ * For the loops_per_jiffy we need:
+ *	thread/cpu frequency (100MHz)
+ * And for the timer, we need:
+ *	sleep clock rate
+ */
+
+cycles_t	pcycle_freq_mhz;
+cycles_t	thread_freq_mhz;
+cycles_t	sleep_clk_freq;
+
+static struct resource rtos_timer_resources[] = {
+	{
+		.start	= RTOS_TIMER_REGS_ADDR,
+		.end	= RTOS_TIMER_REGS_ADDR+PAGE_SIZE-1,
+		.flags	= IORESOURCE_MEM,
+	},
+};
+
+static struct platform_device rtos_timer_device = {
+	.name		= "rtos_timer",
+	.id		= -1,
+	.num_resources	= ARRAY_SIZE(rtos_timer_resources),
+	.resource	= rtos_timer_resources,
+};
+
+/*  A lot of this stuff should move into a platform specific section.  */
+struct adsp_hw_timer_struct {
+	u32 match;   /*  Match value  */
+	u32 count;
+	u32 enable;  /*  [1] - CLR_ON_MATCH_EN, [0] - EN  */
+	u32 clear;   /*  one-shot register that clears the count  */
+};
+
+/*  Look for "TCX0" for related constants.  */
+static __iomem struct adsp_hw_timer_struct *rtos_timer;
+
+void debug_adsp_timers(void)
+{
+	printk(KERN_INFO "rtos_timer->match=0x%08x\n", rtos_timer->match);
+	printk(KERN_INFO "rtos_timer->count=%d\n", rtos_timer->count);
+	printk(KERN_INFO "rtos_timer->enable=%d\n", rtos_timer->enable);
+	printk(KERN_INFO "rtos_timer->clear=%d\n", rtos_timer->clear);
+}
+
+static cycle_t timer_get_cycles(struct clocksource *cs)
+{
+	return (cycle_t) __vmgettime();
+}
+
+static struct clocksource hexagon_clocksource = {
+	.name		= "pcycles",
+	.rating		= 250,
+	.shift		= 16,
+	.read		= timer_get_cycles,
+	.mask		= CLOCKSOURCE_MASK(64),
+	.flags		= CLOCK_SOURCE_IS_CONTINUOUS,
+};
+
+static int set_next_event(unsigned long delta, struct clock_event_device *evt)
+{
+	/*  Assuming the timer will be disabled when we enter here.  */
+	rtos_timer->clear = 1;
+	rtos_timer->clear = 0;
+
+	/*  Does this need massaging?  */
+	rtos_timer->match = delta;
+	rtos_timer->enable = 1 << TIMER_ENABLE;
+	return 0;
+}
+
+/*
+ * Sets the mode (periodic, shutdown, oneshot, etc) of a timer.
+ */
+static void set_mode(enum clock_event_mode mode,
+	struct clock_event_device *evt)
+{
+	switch (mode) {
+	case CLOCK_EVT_MODE_SHUTDOWN:
+		/* XXX implement me */
+	default:
+		break;
+	}
+}
+
+#ifdef CONFIG_SMP
+/*  Broadcast mechanism  */
+static void broadcast(const struct cpumask *mask)
+{
+	send_ipi(mask, IPI_TIMER);
+}
+#endif
+
+static struct clock_event_device hexagon_clockevent_dev = {
+	.name		= "clockevent",
+	.features	= CLOCK_EVT_FEAT_ONESHOT,
+	.rating		= 400,
+	.shift		= 32,
+	.irq		= RTOS_TIMER_INT,
+	.set_next_event = set_next_event,
+	.set_mode	= set_mode,
+#ifdef CONFIG_SMP
+	.broadcast	= broadcast,
+#endif
+};
+
+#ifdef CONFIG_SMP
+static DEFINE_PER_CPU(struct clock_event_device, clock_events);
+
+void setup_percpu_clockdev(void)
+{
+	int cpu = smp_processor_id();
+	struct clock_event_device *ce_dev = &hexagon_clockevent_dev;
+	struct clock_event_device *dummy_clock_dev =
+		&per_cpu(clock_events, cpu);
+
+	memcpy(dummy_clock_dev, ce_dev, sizeof(*dummy_clock_dev));
+	INIT_LIST_HEAD(&dummy_clock_dev->list);
+
+	dummy_clock_dev->features = CLOCK_EVT_FEAT_DUMMY;
+	dummy_clock_dev->cpumask = cpumask_of(cpu);
+	dummy_clock_dev->mode = CLOCK_EVT_MODE_UNUSED;
+
+	clockevents_register_device(dummy_clock_dev);
+}
+
+/*  Called from smp.c for each CPU's timer ipi call  */
+void ipi_timer(void)
+{
+	int cpu = smp_processor_id();
+	struct clock_event_device *ce_dev = &hexagon_clockevent_dev;
+
+	ce_dev = &per_cpu(clock_events, cpu);
+	ce_dev->event_handler(ce_dev);
+}
+#endif /* CONFIG_SMP */
+
+static irqreturn_t timer_interrupt(int irq, void *devid)
+{
+	struct clock_event_device *ce_dev = &hexagon_clockevent_dev;
+
+	rtos_timer->enable = 0;
+	ce_dev->event_handler(ce_dev);
+
+	return IRQ_HANDLED;
+}
+
+/*  This should also be pulled from devtree  */
+static struct irqaction rtos_timer_intdesc = {
+	.handler = timer_interrupt,
+	.flags = IRQF_DISABLED | IRQF_TIMER | IRQF_TRIGGER_RISING,
+	.name = "rtos_timer"
+};
+
+void clocksource_debug(struct clocksource *cs)
+{
+	printk(KERN_DEBUG "cs->mult=0x%08x\n", cs->mult);
+	printk(KERN_DEBUG "cs->shift=%d\n", cs->shift);
+}
+
+void clockevent_debug(struct clock_event_device *ce)
+{
+	printk(KERN_DEBUG "ce->mult=0x%08x\n", ce->mult);
+	printk(KERN_DEBUG "ce->shift=%d\n", ce->shift);
+}
+
+/*
+ * time_init_deferred - called by start_kernel to set up timer/clock source
+ *
+ * Install the IRQ handler for the clock, setup timers.
+ * This is done late, as that way, we can use ioremap().
+ *
+ * This runs just before the delay loop is calibrated, and
+ * is used for delay calibration.
+ */
+void __init time_init_deferred(void)
+{
+	struct resource *resource = NULL;
+	struct clock_event_device *ce_dev = &hexagon_clockevent_dev;
+	struct device_node *dn;
+	struct resource r;
+	int err;
+
+	ce_dev->cpumask = cpu_all_mask;
+
+	if (!resource)
+		resource = rtos_timer_device.resource;
+
+	/*  ioremap here means this has to run later, after paging init  */
+	rtos_timer = ioremap(resource->start, resource->end
+		- resource->start + 1);
+
+	if (!rtos_timer) {
+		release_mem_region(resource->start, resource->end
+			- resource->start + 1);
+	}
+	/*  Change to clocksource_register_khz in newer kernels  */
+	clocksource_register_khz(&hexagon_clocksource, pcycle_freq_mhz * 1000);
+	clocksource_debug(&hexagon_clocksource);
+
+	/*  Note: the sim generic RTOS clock is apparently really 18750Hz  */
+
+	/*
+	 * Last arg is some guaranteed seconds for which the conversion will
+	 * work without overflow.
+	 */
+	clockevents_calc_mult_shift(ce_dev, sleep_clk_freq, 4);
+
+	ce_dev->max_delta_ns = clockevent_delta2ns(0x7fffffff, ce_dev);
+	ce_dev->min_delta_ns = clockevent_delta2ns(0xf, ce_dev);
+
+#ifdef CONFIG_SMP
+	setup_percpu_clockdev();
+#endif
+
+	clockevents_register_device(ce_dev);
+	setup_irq(ce_dev->irq, &rtos_timer_intdesc);
+}
+
+void __init time_init(void)
+{
+	late_time_init = time_init_deferred;
+}
+
+/*
+ * This could become parametric or perhaps even computed at run-time,
+ * but for now we take the observed simulator jitter.
+ */
+static long long fudgefactor = 350;  /* Maybe lower if kernel optimized. */
+
+void __udelay(unsigned long usecs)
+{
+	unsigned long long start = __vmgettime();
+	unsigned long long finish = (pcycle_freq_mhz * usecs) - fudgefactor;
+
+	while ((__vmgettime() - start) < finish)
+		cpu_relax(); /*  not sure how this improves readability  */
+}
+EXPORT_SYMBOL(__udelay);

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 18/36] Hexagon: Add time and timer functions

[Arnd Bergmann]

> + * 8x50 HDD Specs 5-8.  Simulator co-sim not fixed until
> + * release 1.1, and then it's "adjustable" and probably not defaulted.
> + */
> +#define RTOS_TIMER_INT         3
> +#ifdef CONFIG_HEXAGON_COMET
> +#define RTOS_TIMER_REGS_ADDR   0xAB000000UL
> +#endif
> +#define SLEEP_CLK_RATE         32000

I think it would be better to pull RTOS_TIMER_REGS_ADDR from the device
tree, but this is probably good enough for 3.2.

Acked-by: Arnd Bergmann <arnd@arndb.de>

Re: [patch v3 18/36] Hexagon: Add time and timer functions

[Thomas Gleixner]

On Thu, 8 Sep 2011, Richard Kuo wrote:
> +static cycle_t timer_get_cycles(struct clocksource *cs)
> +{
> +	return (cycle_t) __vmgettime();
> +}
> +
> +static struct clocksource hexagon_clocksource = {
> +	.name		= "pcycles",
> +	.rating		= 250,
> +	.shift		= 16,

clocksource_register_khz() calcs mult and shift.

> +	.read		= timer_get_cycles,
> +	.mask		= CLOCKSOURCE_MASK(64),
> +	.flags		= CLOCK_SOURCE_IS_CONTINUOUS,
> +};
> +

> +static struct clock_event_device hexagon_clockevent_dev = {
> +	.name		= "clockevent",
> +	.features	= CLOCK_EVT_FEAT_ONESHOT,
> +	.rating		= 400,
> +	.shift		= 32,

You recalc this shift as well

> +	.irq		= RTOS_TIMER_INT,
> +	.set_next_event = set_next_event,
> +	.set_mode	= set_mode,
> +#ifdef CONFIG_SMP
> +	.broadcast	= broadcast,
> +#endif
> +};
> +

> +/*  Called from smp.c for each CPU's timer ipi call  */
> +void ipi_timer(void)
> +{
> +	int cpu = smp_processor_id();
> +	struct clock_event_device *ce_dev = &hexagon_clockevent_dev;

This initialization is pointless

> +	ce_dev = &per_cpu(clock_events, cpu);
> +	ce_dev->event_handler(ce_dev);
> +}
> +#endif /* CONFIG_SMP */
> +
> +static irqreturn_t timer_interrupt(int irq, void *devid)
> +{
> +	struct clock_event_device *ce_dev = &hexagon_clockevent_dev;
> +
> +	rtos_timer->enable = 0;
> +	ce_dev->event_handler(ce_dev);
> +
> +	return IRQ_HANDLED;
> +}
> +
> +/*  This should also be pulled from devtree  */
> +static struct irqaction rtos_timer_intdesc = {
> +	.handler = timer_interrupt,
> +	.flags = IRQF_DISABLED | IRQF_TIMER | IRQF_TRIGGER_RISING,

  IRQF_DISABLED is about to be gone and IRQF_TIMER has it already.

> +	.name = "rtos_timer"
> +};
> +
> +void clocksource_debug(struct clocksource *cs)
> +{
> +	printk(KERN_DEBUG "cs->mult=0x%08x\n", cs->mult);
> +	printk(KERN_DEBUG "cs->shift=%d\n", cs->shift);
> +}
> +
> +void clockevent_debug(struct clock_event_device *ce)
> +{
> +	printk(KERN_DEBUG "ce->mult=0x%08x\n", ce->mult);
> +	printk(KERN_DEBUG "ce->shift=%d\n", ce->shift);
> +}

Important debug leftovers ?

Thanks,

	tglx

[patch v3 19/36] Hexagon: Add ptrace support

[Richard Kuo]

[-- Attachment #1: ptrace.diff --]
[-- Type: text/plain, Size: 11547 bytes --]

Will look at the more generic ptrace support in the future.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>

---
 arch/hexagon/include/asm/ptrace.h |   46 +++++
 arch/hexagon/kernel/ptrace.c      |  333 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 379 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/ptrace.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/ptrace.h	2011-09-03 20:14:59.684981279 -0500
@@ -0,0 +1,46 @@
+/*
+ * Ptrace definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_PTRACE_H
+#define _ASM_PTRACE_H
+
+#include <asm/registers.h>
+
+/*  arch_ptrace and ptrace_disable use task_struct in arg lists  */
+struct task_struct;
+
+/*
+ * ptrace commands not part of Linux base set (but common to essentially
+ * all supported architectures).
+ */
+
+#define PTRACE_GETREGS	12
+#define PTRACE_SETREGS	13
+
+#define N_PTRACE_REGS	44
+
+#define profile_pc(regs) instruction_pointer(regs)
+#define instruction_pointer(regs) pt_elr(regs)
+
+/* kprobe-based event tracer support */
+extern int regs_query_register_offset(const char *name);
+extern const char *regs_query_register_name(unsigned int offset);
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/ptrace.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/ptrace.c	2011-09-07 13:00:31.110647061 -0500
@@ -0,0 +1,333 @@
+/*
+ * Ptrace support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <generated/compile.h>
+
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/errno.h>
+#include <linux/ptrace.h>
+#include <linux/regset.h>
+#include <linux/user.h>
+
+#include <asm/system.h>
+#include <asm/user.h>
+
+struct pt_regs_offset {
+	const char *name;
+	int offset;
+};
+
+#define REG_OFFSET_NAME(r) {.name = #r, .offset = offsetof(struct pt_regs, r)}
+#define REG_OFFSET_HVM_NAME(r, s) {.name = #r, \
+				  .offset = offsetof(struct pt_regs, s)}
+#define REG_OFFSET_END {.name = NULL, .offset = 0}
+
+static const struct pt_regs_offset regoffset_table[] = {
+	REG_OFFSET_NAME(r00),
+	REG_OFFSET_NAME(r01),
+	REG_OFFSET_NAME(r02),
+	REG_OFFSET_NAME(r03),
+	REG_OFFSET_NAME(r04),
+	REG_OFFSET_NAME(r05),
+	REG_OFFSET_NAME(r06),
+	REG_OFFSET_NAME(r07),
+	REG_OFFSET_NAME(r08),
+	REG_OFFSET_NAME(r09),
+
+	REG_OFFSET_NAME(r10),
+	REG_OFFSET_NAME(r11),
+	REG_OFFSET_NAME(r12),
+	REG_OFFSET_NAME(r13),
+	REG_OFFSET_NAME(r14),
+	REG_OFFSET_NAME(r15),
+	REG_OFFSET_NAME(r16),
+	REG_OFFSET_NAME(r17),
+	REG_OFFSET_NAME(r18),
+	REG_OFFSET_NAME(r19),
+
+	REG_OFFSET_NAME(r20),
+	REG_OFFSET_NAME(r21),
+	REG_OFFSET_NAME(r22),
+	REG_OFFSET_NAME(r23),
+	REG_OFFSET_NAME(r24),
+	REG_OFFSET_NAME(r25),
+	REG_OFFSET_NAME(r26),
+	REG_OFFSET_NAME(r27),
+	REG_OFFSET_NAME(r28),
+	REG_OFFSET_NAME(r29),
+
+	REG_OFFSET_NAME(r30),
+	REG_OFFSET_NAME(r31),
+
+	REG_OFFSET_NAME(gp),
+	REG_OFFSET_NAME(ugp),
+	REG_OFFSET_NAME(sa0),
+	REG_OFFSET_NAME(lc0),
+	REG_OFFSET_NAME(sa1),
+	REG_OFFSET_NAME(lc1),
+	REG_OFFSET_NAME(m0),
+	REG_OFFSET_NAME(m1),
+	REG_OFFSET_NAME(preds),
+	REG_OFFSET_HVM_NAME(pc, hvmer.vmel),
+	REG_OFFSET_HVM_NAME(cause, hvmer.vmest),
+	REG_OFFSET_HVM_NAME(badva, hvmer.vmbadva),
+};
+
+/**
+ * regs_query_register_offset() - query register offset from its name
+ * @name:      the name of a register
+ *
+ * regs_query_register_offset() returns the offset of a register in struct
+ * pt_regs from its name. If the name is invalid, this returns -EINVAL;
+ */
+int regs_query_register_offset(const char *name)
+{
+	const struct pt_regs_offset *roff;
+	for (roff = regoffset_table; roff->name != NULL; roff++)
+		if (!strcmp(roff->name, name))
+			return roff->offset;
+	return -EINVAL;
+}
+
+/**
+ * regs_query_register_name() - query register name from its offset
+ * @offset:    the offset of a register in struct pt_regs.
+ *
+ * regs_query_register_name() returns the name of a register from its
+ * offset in struct pt_regs. If the @offset is invalid, this returns NULL;
+ */
+const char *regs_query_register_name(unsigned int offset)
+{
+	const struct pt_regs_offset *roff;
+	for (roff = regoffset_table; roff->name != NULL; roff++)
+		if (roff->offset == offset)
+			return roff->name;
+	return NULL;
+}
+
+static int gpr_get(struct task_struct *target,
+		   const struct user_regset *regset,
+		   unsigned int pos, unsigned int count,
+		   void *kbuf, void __user *ubuf)
+{
+	int ret;
+	unsigned int dummy;
+	struct pt_regs *regs = task_pt_regs(target);
+
+
+	if (!regs)
+		return -EIO;
+
+	/* The general idea here is that the copyout must happen in
+	 * exactly the same order in which the userspace expects these
+	 * regs. Now, the sequence in userspace does not match the
+	 * sequence in the kernel, so everything past the 32 gprs
+	 * happens one at a time.
+	 */
+	ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf,
+				  &regs->r00, 0, 32*sizeof(unsigned long));
+
+#define ONEXT(KPT_REG, USR_REG) \
+	if (!ret) \
+		ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, \
+			KPT_REG, offsetof(struct user_regs_struct, USR_REG), \
+			offsetof(struct user_regs_struct, USR_REG) + \
+				 sizeof(unsigned long));
+
+	/* Must be exactly same sequence as struct user_regs_struct */
+	ONEXT(&regs->gp, gp);
+	ONEXT(&regs->ugp, ugp);
+	ONEXT(&regs->sa0, sa0);
+	ONEXT(&regs->lc0, lc0);
+	ONEXT(&regs->sa1, sa1);
+	ONEXT(&regs->lc1, lc1);
+	ONEXT(&regs->m0, m0);
+	ONEXT(&regs->m1, m1);
+	ONEXT(&regs->preds, p3_0);
+	ONEXT(&pt_elr(regs), pc);
+	dummy = pt_cause(regs);
+	ONEXT(&dummy, cause);
+	ONEXT(&pt_badva(regs), badva);
+
+	/* Pad the rest with zeros, if needed */
+	if (!ret)
+		ret = user_regset_copyout_zero(&pos, &count, &kbuf, &ubuf,
+					sizeof(struct user_regs_struct), -1);
+	return ret;
+}
+
+static int gpr_set(struct task_struct *target,
+		   const struct user_regset *regset,
+		   unsigned int pos, unsigned int count,
+		   const void *kbuf, const void __user *ubuf)
+{
+	int ret;
+	unsigned long bucket;
+	struct pt_regs *regs = task_pt_regs(target);
+
+	if (!regs)
+		return -EIO;
+
+	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf,
+				 &regs->r00, 0, 32*sizeof(unsigned long));
+
+#define INEXT(KPT_REG, USR_REG) \
+	if (!ret) \
+		ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, \
+			KPT_REG, offsetof(struct user_regs_struct, USR_REG), \
+			offsetof(struct user_regs_struct, USR_REG) + \
+				sizeof(unsigned long));
+
+	/* Must be exactly same sequence as struct user_regs_struct */
+	INEXT(&regs->gp, gp);
+	INEXT(&regs->ugp, ugp);
+	INEXT(&regs->sa0, sa0);
+	INEXT(&regs->lc0, lc0);
+	INEXT(&regs->sa1, sa1);
+	INEXT(&regs->lc1, lc1);
+	INEXT(&regs->m0, m0);
+	INEXT(&regs->m1, m1);
+	INEXT(&regs->preds, p3_0);
+	INEXT(&pt_elr(regs), pc);
+
+	/* CAUSE and BADVA aren't writeable. */
+	INEXT(&bucket, cause);
+	INEXT(&bucket, badva);
+
+	/* Ignore the rest, if needed */
+	if (!ret)
+		ret = user_regset_copyin_ignore(&pos, &count, &kbuf, &ubuf,
+					sizeof(struct user_regs_struct), -1);
+
+	if (ret)
+		return ret;
+
+	/*
+	 * This is special; SP is actually restored by the VM via the
+	 * special event record which is set by the special trap.
+	 */
+	regs->hvmer.vmpsp = regs->r29;
+	return 0;
+}
+
+enum hexagon_regset {
+	REGSET_GPR,
+};
+
+static const struct user_regset hexagon_regsets[] = {
+	[REGSET_GPR] = {
+		.core_note_type = NT_PRSTATUS,
+		.n = N_PTRACE_REGS,
+		.size = sizeof(unsigned long),
+		.align = sizeof(unsigned long),
+		.get = gpr_get,
+		.set = gpr_set
+	},
+};
+
+static const struct user_regset_view hexagon_user_view = {
+	.name = UTS_MACHINE,
+	.e_machine = ELF_ARCH,
+	.ei_osabi = ELF_OSABI,
+	.regsets = hexagon_regsets,
+	.n = ARRAY_SIZE(hexagon_regsets)
+};
+
+static int ptrace_pokeusr(struct task_struct *target, int regnum,
+			  unsigned long data)
+{
+	struct pt_regs *regs = task_pt_regs(target);
+
+	if (!regs || (regnum < 0))
+		return -EIO;
+
+	if (regnum < 32) {
+		unsigned long *p = &regs->r00;
+		*(p+regnum) = data;
+
+		/* Return happens via HVM */
+		if (regnum == 29)
+			regs->hvmer.vmpsp = regs->r29;
+		return 0;
+	}
+
+/*
+ * Todo:  sync up with gdb on register passing numbering.
+ */
+
+#define SETREG(KPT_REG, USR_REG) \
+	case offsetof(struct user_regs_struct, USR_REG)>>2: \
+		KPT_REG = data; \
+		return 0;
+
+	/* Other regs are scattered about */
+	switch (regnum) {
+	SETREG(regs->gp, gp);
+	SETREG(regs->ugp, ugp);
+	SETREG(regs->sa0, sa0);
+	SETREG(regs->lc0, lc0);
+	SETREG(regs->sa1, sa1);
+	SETREG(regs->lc1, lc1);
+	SETREG(regs->m0, m0);
+	SETREG(regs->m1, m1);
+	SETREG(regs->preds, p3_0);
+	SETREG(pt_elr(regs), pc);
+	}
+
+	/* CAUSE and BADVA aren't writeable. */
+	return -EINVAL;
+}
+
+void ptrace_disable(struct task_struct *child)
+{
+	/* Boilerplate - resolves to null inline if no HW single-step */
+	user_disable_single_step(child);
+}
+
+long arch_ptrace(struct task_struct *child, long request,
+		 unsigned long addr, unsigned long data)
+{
+	void __user *udata = (void __user *) data;
+
+	switch (request) {
+	case PTRACE_POKETEXT:	/* write the word at location addr. */
+	case PTRACE_POKEDATA:
+		return generic_ptrace_pokedata(child, addr, data);
+	case PTRACE_POKEUSR:	/* write register specified by addr. */
+		return ptrace_pokeusr(child, addr, data);
+	case PTRACE_GETREGS:
+		return copy_regset_to_user(child, &hexagon_user_view,
+					   REGSET_GPR, 0,
+					   sizeof(struct user_regs_struct),
+					   udata);
+	case PTRACE_SETREGS:
+		return copy_regset_from_user(child, &hexagon_user_view,
+					     REGSET_GPR, 0,
+					     sizeof(struct user_regs_struct),
+					     udata);
+	default:
+		return ptrace_request(child, request, addr, data);
+	}
+
+	return 0;
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:06 Richard Kuo wrote:
> Will look at the more generic ptrace support in the future.
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
> Signed-off-by: Linas Vepstas <linas@codeaurora.org>

Note that this absolutely has to be completed before the code is merged upstream,
because it is relevant to the ABI.

	Arnd

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Jonas Bonn]

On 9 September 2011 03:09, Richard Kuo <rkuo@codeaurora.org> wrote:
> +long arch_ptrace(struct task_struct *child, long request,
> +                unsigned long addr, unsigned long data)
> +{
> +       void __user *udata = (void __user *) data;
> +
> +       switch (request) {
> +       case PTRACE_POKETEXT:   /* write the word at location addr. */
> +       case PTRACE_POKEDATA:
> +               return generic_ptrace_pokedata(child, addr, data);
> +       case PTRACE_POKEUSR:    /* write register specified by addr. */
> +               return ptrace_pokeusr(child, addr, data);
> +       case PTRACE_GETREGS:
> +               return copy_regset_to_user(child, &hexagon_user_view,
> +                                          REGSET_GPR, 0,
> +                                          sizeof(struct user_regs_struct),
> +                                          udata);
> +       case PTRACE_SETREGS:
> +               return copy_regset_from_user(child, &hexagon_user_view,
> +                                            REGSET_GPR, 0,
> +                                            sizeof(struct user_regs_struct),
> +                                            udata);
> +       default:
> +               return ptrace_request(child, request, addr, data);
> +       }
> +
> +       return 0;
> +}
>

I think that... other than for compatibility with legacy userspace
code, you shouldn't need any of these cases other than the 'default'.

PTRACE_GETREGS is replaced by PTRACE_GETREGSET and this is handled in
generic code.  POKETEXT, POKEDATA are handled in the generic code.
POKEUSR can also be done via the SETREGSET mechanism.

That said, I don't think gdbserver has been updated to use
GETREGSET/SETREGSET.  This is a bit like the uClibc/glibc issue,
though... somebody just needs to do the work to make userspace work
with the new, "clean" kernel interface.

Given all that, you shouldn't need to export pt_regs to userspace
either.  You already have a user_regs_struct which is userspace's view
of the process registers.  For OpenRISC we made pt_regs a kernel
private structure, giving us the freedom to change the stack layout of
the registers in the future while providing a stable view for
userspace.  I would suggest doing the same here.

If you do decide to make pt_regs private, you'll need to change your
struct sigcontext to use user_regs_struct, as well.

/Jonas
--
To unsubscribe from this list: send the line "unsubscribe linux-hexagon" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Jonas Bonn]

On 9 September 2011 03:09, Richard Kuo <rkuo@codeaurora.org> wrote:
> +long arch_ptrace(struct task_struct *child, long request,
> +                unsigned long addr, unsigned long data)
> +{
> +       void __user *udata = (void __user *) data;
> +
> +       switch (request) {
> +       case PTRACE_POKETEXT:   /* write the word at location addr. */
> +       case PTRACE_POKEDATA:
> +               return generic_ptrace_pokedata(child, addr, data);
> +       case PTRACE_POKEUSR:    /* write register specified by addr. */
> +               return ptrace_pokeusr(child, addr, data);
> +       case PTRACE_GETREGS:
> +               return copy_regset_to_user(child, &hexagon_user_view,
> +                                          REGSET_GPR, 0,
> +                                          sizeof(struct user_regs_struct),
> +                                          udata);
> +       case PTRACE_SETREGS:
> +               return copy_regset_from_user(child, &hexagon_user_view,
> +                                            REGSET_GPR, 0,
> +                                            sizeof(struct user_regs_struct),
> +                                            udata);
> +       default:
> +               return ptrace_request(child, request, addr, data);
> +       }
> +
> +       return 0;
> +}
>

I think that... other than for compatibility with legacy userspace
code, you shouldn't need any of these cases other than the 'default'.

PTRACE_GETREGS is replaced by PTRACE_GETREGSET and this is handled in
generic code.  POKETEXT, POKEDATA are handled in the generic code.
POKEUSR can also be done via the SETREGSET mechanism.

That said, I don't think gdbserver has been updated to use
GETREGSET/SETREGSET.  This is a bit like the uClibc/glibc issue,
though... somebody just needs to do the work to make userspace work
with the new, "clean" kernel interface.

Given all that, you shouldn't need to export pt_regs to userspace
either.  You already have a user_regs_struct which is userspace's view
of the process registers.  For OpenRISC we made pt_regs a kernel
private structure, giving us the freedom to change the stack layout of
the registers in the future while providing a stable view for
userspace.  I would suggest doing the same here.

If you do decide to make pt_regs private, you'll need to change your
struct sigcontext to use user_regs_struct, as well.

/Jonas

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Linas Vepstas (Code Aurora)]

On Fri, Sep 09, 2011 at 10:15:20PM +0200, Jonas Bonn wrote:
> 
> That said, I don't think gdbserver has been updated to use
> GETREGSET/SETREGSET.  This is a bit like the uClibc/glibc issue,

Some dumb questions then: I notice that a few arches declare
struct user_regs_struct in glibc, most of the others in the
kernel.  If I were to make pt_regs completely kernel-private,
then I really do need to have struct user_regs_struct declared
the kernel headers, right? 

Putting a printk into case PTRACE_POKEUSR seemed to show that
gdb was using this (and was somehow mis-numbering the registers
... sigh.) Are you saying that the default should handle this?

--linas

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Jonas Bonn]

On 9 September 2011 23:18, Linas Vepstas (Code Aurora)
<linas@codeaurora.org> wrote:
> On Fri, Sep 09, 2011 at 10:15:20PM +0200, Jonas Bonn wrote:
>>
>> That said, I don't think gdbserver has been updated to use
>> GETREGSET/SETREGSET.  This is a bit like the uClibc/glibc issue,
>
> Some dumb questions then: I notice that a few arches declare
> struct user_regs_struct in glibc, most of the others in the
> kernel.  If I were to make pt_regs completely kernel-private,
> then I really do need to have struct user_regs_struct declared
> the kernel headers, right?

Yes, on OpenRISC we declare it in ptrace.h (see
arch/openrisc/include/asm/ptrace.h).

>
> Putting a printk into case PTRACE_POKEUSR seemed to show that
> gdb was using this (and was somehow mis-numbering the registers
> ... sigh.) Are you saying that the default should handle this?

I'm really not an expert on GDB, but I believe it can use either the
PEEK/POKE functions to access regs, or it can use GET/SETREGS; it's an
architecture-dependent configuration.  I think the second variant can
be easily modified to become GET/SETREGSET instead, and the PEEK/POKE
variant wouldn't be needed at all.

/Jonas
--
To unsubscribe from this list: send the line "unsubscribe linux-hexagon" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Jonas Bonn]

On 9 September 2011 23:18, Linas Vepstas (Code Aurora)
<linas@codeaurora.org> wrote:
> On Fri, Sep 09, 2011 at 10:15:20PM +0200, Jonas Bonn wrote:
>>
>> That said, I don't think gdbserver has been updated to use
>> GETREGSET/SETREGSET.  This is a bit like the uClibc/glibc issue,
>
> Some dumb questions then: I notice that a few arches declare
> struct user_regs_struct in glibc, most of the others in the
> kernel.  If I were to make pt_regs completely kernel-private,
> then I really do need to have struct user_regs_struct declared
> the kernel headers, right?

Yes, on OpenRISC we declare it in ptrace.h (see
arch/openrisc/include/asm/ptrace.h).

>
> Putting a printk into case PTRACE_POKEUSR seemed to show that
> gdb was using this (and was somehow mis-numbering the registers
> ... sigh.) Are you saying that the default should handle this?

I'm really not an expert on GDB, but I believe it can use either the
PEEK/POKE functions to access regs, or it can use GET/SETREGS; it's an
architecture-dependent configuration.  I think the second variant can
be easily modified to become GET/SETREGSET instead, and the PEEK/POKE
variant wouldn't be needed at all.

/Jonas

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Arnd Bergmann]

On Saturday 10 September 2011 08:42:26 Jonas Bonn wrote:
> > Putting a printk into case PTRACE_POKEUSR seemed to show that
> > gdb was using this (and was somehow mis-numbering the registers
> > ... sigh.) Are you saying that the default should handle this?
> 
> I'm really not an expert on GDB, but I believe it can use either the
> PEEK/POKE functions to access regs, or it can use GET/SETREGS; it's an
> architecture-dependent configuration.  I think the second variant can
> be easily modified to become GET/SETREGSET instead, and the PEEK/POKE
> variant wouldn't be needed at all.

Yes, I think that is the way to go for new architectures, although
practically no architecture in the upstream kernel does it this
way.

The PEEKUSR/POKEUSR stuff is a relic from ancient Unix time where you
would directly access a kernel data structure with it, we should get
over it.

	Arnd

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Pedro Alves]

On Saturday 10 September 2011 07:42:26, Jonas Bonn wrote:

> I'm really not an expert on GDB, but I believe it can use either the
> PEEK/POKE functions to access regs, or it can use GET/SETREGS; it's an
> architecture-dependent configuration.  I think the second variant can
> be easily modified to become GET/SETREGSET instead, and the PEEK/POKE
> variant wouldn't be needed at all.

Yes, the x86/x86-64 ports or gdb and gdbserver are already
using GET/SETREGSET even.

-- 
Pedro Alves

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Linas Vepstas (Code Aurora)]

On Sat, Sep 10, 2011 at 12:29:57PM +0100, Pedro Alves wrote:
> On Saturday 10 September 2011 07:42:26, Jonas Bonn wrote:
> 
> > I'm really not an expert on GDB, but I believe it can use either the
> > PEEK/POKE functions to access regs, or it can use GET/SETREGS; it's an
> > architecture-dependent configuration.  I think the second variant can
> > be easily modified to become GET/SETREGSET instead, and the PEEK/POKE
> > variant wouldn't be needed at all.
> 
> Yes, the x86/x86-64 ports or gdb and gdbserver are already
> using GET/SETREGSET even.

Reading the code, it seemed to only fetch "extended state" in
this way, and not all of the registers.  Am I mistaken?

--linas

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Pedro Alves]

On Monday 19 September 2011 16:25:05, Linas Vepstas (Code Aurora) wrote:
> On Sat, Sep 10, 2011 at 12:29:57PM +0100, Pedro Alves wrote:
> > On Saturday 10 September 2011 07:42:26, Jonas Bonn wrote:
> > 
> > > I'm really not an expert on GDB, but I believe it can use either the
> > > PEEK/POKE functions to access regs, or it can use GET/SETREGS; it's an
> > > architecture-dependent configuration.  I think the second variant can
> > > be easily modified to become GET/SETREGSET instead, and the PEEK/POKE
> > > variant wouldn't be needed at all.
> > 
> > Yes, the x86/x86-64 ports or gdb and gdbserver are already
> > using GET/SETREGSET even.
> 
> Reading the code, it seemed to only fetch "extended state" in
> this way, and not all of the registers.  Am I mistaken?

True.  gdb/gdbserver need to keep working with older x86 kernels, so
we avoid having two ways of doing things with the other registers.
No reason to support legacy interfaces with new ports though.

I wish there was an equivalent of PTRACE_GETREGSET for siginfo...

-- 
Pedro Alves

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Linas Vepstas (Code Aurora)]

On Wed, Sep 21, 2011 at 05:15:20PM +0100, Pedro Alves wrote:
> > On Sat, Sep 10, 2011 at 12:29:57PM +0100, Pedro Alves wrote:
> > > Yes, the x86/x86-64 ports or gdb and gdbserver are already
> > > using GET/SETREGSET even.

Thanks; after a bit of initial confusion, I've gotten gdb to
work with this, so I'm happy. 

> I wish there was an equivalent of PTRACE_GETREGSET for siginfo...

How? Why?

-- Linas

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 19/36] Hexagon: Add ptrace support

[Pedro Alves]

On Wednesday 21 September 2011 18:50:09, Linas Vepstas (Code Aurora) wrote:
> On Wed, Sep 21, 2011 at 05:15:20PM +0100, Pedro Alves wrote:
> > > On Sat, Sep 10, 2011 at 12:29:57PM +0100, Pedro Alves wrote:
> > > > Yes, the x86/x86-64 ports or gdb and gdbserver are already
> > > > using GET/SETREGSET even.
> 
> Thanks; after a bit of initial confusion, I've gotten gdb to
> work with this, so I'm happy. 
> 
> > I wish there was an equivalent of PTRACE_GETREGSET for siginfo...
> 
> How? Why?

One of the the cool features of PTRACE_GETREGSET is that it returns
the regset in the natural layout of the tracee.  So when debugging a 
32-bit process on a 64-bit kernel, PTRACE_GETREGSET returns the
regset in 32-bit layout.  Recent enough GDBs can print the siginfo of a
caught signal (before its delivered with (PTRACE_CONT, sig).
Try "(gdb) p $_siginfo" on x86.  Trouble is 32-bit processes on 64-bit
kernels or other biarch setups.  PTRACE_GETSIGINFO returns the siginfo in
64-bit layout, so GDB gets to try to immitate what the kernel's 32-bit
compat layer does.  Both for reading and writting siginfo.
See compat_siginfo_from_siginfo and siginfo_from_compat_siginfo
at [1].
The kernel already knows to convert siginfo to the native layout
of the process -- needed to pass to the user space signal handler
at least.  But AFAIK, there's no way for the debugger to get at it
before the signal is actually delivered.

[1]
http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/amd64-linux-nat.c?rev=1.35&content-type=text/x-cvsweb-markup&cvsroot=src

-- 
Pedro Alves

[patch v3 20/36] Hexagon: Provide basic debugging and system trap support.

[Richard Kuo]

[-- Attachment #1: traps.diff --]
[-- Type: text/plain, Size: 10904 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/traps.h |   29 +++
 arch/hexagon/kernel/traps.c      |  356 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 385 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/kernel/traps.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/traps.c	2011-09-03 20:14:59.694981307 -0500
@@ -0,0 +1,356 @@
+/*
+ * Kernel traps/events for Hexagon processor
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/module.h>
+#include <linux/kallsyms.h>
+#include <linux/kdebug.h>
+#include <linux/syscalls.h>
+#include <linux/signal.h>
+#include <linux/tracehook.h>
+#include <asm/traps.h>
+#include <asm/vm_fault.h>
+#include <asm/syscall.h>
+#include <asm/registers.h>
+#include <asm/unistd.h>
+#ifdef CONFIG_KGDB
+# include <linux/kgdb.h>
+#endif
+
+#define TRAP_SYSCALL	1
+#define TRAP_DEBUG	0xdb
+
+void __init trap_init(void)
+{
+}
+
+#ifdef CONFIG_GENERIC_BUG
+/* Maybe should resemble arch/sh/kernel/traps.c ?? */
+int is_valid_bugaddr(unsigned long addr)
+{
+	return 1;
+}
+#endif /* CONFIG_GENERIC_BUG */
+
+void show_stack(struct task_struct *task, unsigned long *fp)
+{
+	int kstack_depth_to_print = 20;
+	unsigned long offset, size;
+	const char *name = NULL;
+	unsigned long *newfp;
+	unsigned long ip;
+	char tmpstr[128];
+	char *modname;
+	int i;
+
+	if (task == NULL)
+		task = current;
+
+	if (fp == NULL) {
+		if (task == current) {
+			asm("%0 = r30" : "=r" (fp));
+		} else {
+			fp = (unsigned long *)
+			     ((struct hexagon_switch_stack *)
+			     task->thread.switch_sp)->fp;
+		}
+	}
+
+	printk(KERN_INFO "CPU#%d, %s/%d, Call Trace:\n",
+	       raw_smp_processor_id(), current->comm,
+	       task_pid_nr(current));
+	for (i = 0; i < kstack_depth_to_print; i++) {
+		if (0 /* Check irq stack at some point */)
+			printk(KERN_INFO " <EOI> ");
+		else if (((long) fp & (THREAD_SIZE-1)) == 0)
+			break;
+
+		newfp = (unsigned long *) *fp;
+
+		ip = *(unsigned long *)((char *)fp + sizeof(unsigned long));
+		name = kallsyms_lookup(ip, &size, &offset, &modname, tmpstr);
+
+		printk(KERN_INFO "[%p] 0x%lx: %s + 0x%lx", fp, ip, name,
+			offset);
+		if (modname)
+			printk(KERN_CONT " [%s] ", modname);
+		if (i == 0)
+			printk(KERN_CONT " (unreliable)");
+		printk(KERN_CONT "\n");
+
+		/* If newfp isn't larger, we're tracing garbage */
+		if (newfp > fp)
+			fp = newfp;
+		else
+			break;
+	}
+}
+
+void dump_state(struct pt_regs *regs)
+{
+	show_regs(regs);
+	show_stack(current, &regs->r30);
+}
+
+void dump_stack(void)
+{
+	show_stack(current, NULL);
+}
+EXPORT_SYMBOL(dump_stack);
+
+int die(const char *str, struct pt_regs *regs, long err)
+{
+	static struct {
+		spinlock_t lock;
+		int counter;
+	} die = {
+		.lock = __SPIN_LOCK_UNLOCKED(die.lock),
+		.counter = 0
+	};
+
+	spin_lock_irq(&die.lock);
+	bust_spinlocks(1);
+	printk(KERN_INFO "%s[#%d]:\n", str, ++die.counter);
+
+	if (notify_die(DIE_OOPS, str, regs, err, pt_cause(regs), SIGSEGV) ==
+	    NOTIFY_STOP)
+		return 1;
+
+	print_modules();
+	dump_state(regs);
+
+	do_exit(err);
+	return 0;
+}
+
+int die_if_kernel(char *str, struct pt_regs *regs, long err)
+{
+	if (!user_mode(regs))
+		return die(str, regs, err);
+	else
+		return 0;
+}
+
+/*
+ * It's not clear that misaligned fetches are ever recoverable.
+ */
+static void misaligned_instruction(struct pt_regs *regs)
+{
+	die_if_kernel("Misaligned Instruction", regs, 0);
+	force_sig(SIGBUS, current);
+}
+
+/*
+ * Misaligned loads and stores, on the other hand, can be
+ * emulated, and probably should be, some day.  But for now
+ * they will be considered fatal.
+ */
+static void misaligned_data_load(struct pt_regs *regs)
+{
+	die_if_kernel("Misaligned Data Load", regs, 0);
+	force_sig(SIGBUS, current);
+}
+
+static void misaligned_data_store(struct pt_regs *regs)
+{
+	die_if_kernel("Misaligned Data Store", regs, 0);
+	force_sig(SIGBUS, current);
+}
+
+static void illegal_instruction(struct pt_regs *regs)
+{
+	die_if_kernel("Illegal Instruction", regs, 0);
+	force_sig(SIGILL, current);
+}
+
+/*
+ * Precise bus errors may be recoverable with a a retry,
+ * but for now, treat them as irrecoverable.
+ */
+static void precise_bus_error(struct pt_regs *regs)
+{
+	die_if_kernel("Precise Bus Error", regs, 0);
+	force_sig(SIGBUS, current);
+}
+
+/*
+ * If anything is to be done here other than panic,
+ * it will probably be complex and migrate to another
+ * source module.  For now, just die.
+ */
+static void cache_error(struct pt_regs *regs)
+{
+	die("Cache Error", regs, 0);
+}
+
+/*
+ * General exception handler
+ */
+void do_genex(struct pt_regs *regs)
+{
+
+	/*
+	 * Decode Cause and Dispatch
+	 */
+	switch (pt_cause(regs)) {
+	case HVM_GE_C_XPROT:
+	case HVM_GE_C_XUSER:
+		execute_protection_fault(regs);
+		break;
+	case HVM_GE_C_RPROT:
+	case HVM_GE_C_RUSER:
+		read_protection_fault(regs);
+		break;
+	case HVM_GE_C_WPROT:
+	case HVM_GE_C_WUSER:
+		write_protection_fault(regs);
+		break;
+	case HVM_GE_C_XMAL:
+		misaligned_instruction(regs);
+		break;
+	case HVM_GE_C_RMAL:
+		misaligned_data_load(regs);
+		break;
+	case HVM_GE_C_WMAL:
+		misaligned_data_store(regs);
+		break;
+	case HVM_GE_C_INVI:
+	case HVM_GE_C_PRIVI:
+		illegal_instruction(regs);
+		break;
+	case HVM_GE_C_BUS:
+		precise_bus_error(regs);
+		break;
+	case HVM_GE_C_CACHE:
+		cache_error(regs);
+		break;
+	default:
+		/* Halt and catch fire */
+		panic("Unrecognized exception 0x%lx\n", pt_cause(regs));
+		break;
+	}
+}
+
+/* Indirect system call dispatch */
+long sys_syscall(void)
+{
+	printk(KERN_ERR "sys_syscall invoked!\n");
+	return -ENOSYS;
+}
+
+void do_trap0(struct pt_regs *regs)
+{
+	unsigned long syscallret = 0;
+	syscall_fn syscall;
+
+	switch (pt_cause(regs)) {
+	case TRAP_SYSCALL:
+		/* System call is trap0 #1 */
+
+		/* allow strace to catch syscall args  */
+		if (unlikely(test_thread_flag(TIF_SYSCALL_TRACE) &&
+			tracehook_report_syscall_entry(regs)))
+			return;  /*  return -ENOSYS somewhere?  */
+
+		/* Interrupts should be re-enabled for syscall processing */
+		__vmsetie(VM_INT_ENABLE);
+
+		/*
+		 * System call number is in r6, arguments in r0..r5.
+		 * Fortunately, no Linux syscall has more than 6 arguments,
+		 * and Hexagon ABI passes first 6 arguments in registers.
+		 * 64-bit arguments are passed in odd/even register pairs.
+		 * Fortunately, we have no system calls that take more
+		 * than three arguments with more than one 64-bit value.
+		 * Should that change, we'd need to redesign to copy
+		 * between user and kernel stacks.
+		 */
+		regs->syscall_nr = regs->r06;
+
+		/*
+		 * GPR R0 carries the first parameter, and is also used
+		 * to report the return value.  We need a backup of
+		 * the user's value in case we need to do a late restart
+		 * of the system call.
+		 */
+		regs->restart_r0 = regs->r00;
+
+		if ((unsigned long) regs->syscall_nr >= __NR_syscalls) {
+			regs->r00 = -1;
+		} else {
+			syscall = (syscall_fn)
+				  (sys_call_table[regs->syscall_nr]);
+			syscallret = syscall(regs->r00, regs->r01,
+				   regs->r02, regs->r03,
+				   regs->r04, regs->r05);
+		}
+
+		/*
+		 * If it was a sigreturn system call, don't overwrite
+		 * r0 value in stack frame with return value.
+		 *
+		 * __NR_sigreturn doesn't seem to exist in new unistd.h
+		 */
+
+		if (regs->syscall_nr != __NR_rt_sigreturn)
+			regs->r00 = syscallret;
+
+		/* allow strace to get the syscall return state  */
+		if (unlikely(test_thread_flag(TIF_SYSCALL_TRACE)))
+			tracehook_report_syscall_exit(regs, 0);
+
+		break;
+	case TRAP_DEBUG:
+		/* Trap0 0xdb is debug breakpoint */
+		if (user_mode(regs)) {
+			struct siginfo info;
+
+			info.si_signo = SIGTRAP;
+			info.si_errno = 0;
+			/*
+			 * Some architecures add some per-thread state
+			 * to distinguish between breakpoint traps and
+			 * trace traps.  We may want to do that, and
+			 * set the si_code value appropriately, or we
+			 * may want to use a different trap0 flavor.
+			 */
+			info.si_code = TRAP_BRKPT;
+			info.si_addr = (void __user *) pt_elr(regs);
+			send_sig_info(SIGTRAP, &info, current);
+		} else {
+#ifdef CONFIG_KGDB
+			kgdb_handle_exception(pt_cause(regs), SIGTRAP,
+					      TRAP_BRKPT, regs);
+#endif
+		}
+		break;
+	}
+	/* Ignore other trap0 codes for now, especially 0 (Angel calls) */
+}
+
+/*
+ * Machine check exception handler
+ */
+void do_machcheck(struct pt_regs *regs)
+{
+	/* Halt and catch fire */
+	__vmstop();
+}
Index: linux-hexagon-kernel/arch/hexagon/include/asm/traps.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/traps.h	2011-09-03 20:14:59.694981307 -0500
@@ -0,0 +1,29 @@
+/*
+ * Trap support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_HEXAGON_TRAPS_H
+#define _ASM_HEXAGON_TRAPS_H
+
+#include <asm/registers.h>
+
+extern int die(const char *str, struct pt_regs *regs, long err);
+extern int die_if_kernel(char *str, struct pt_regs *regs, long err);
+
+#endif /* _ASM_HEXAGON_TRAPS_H */

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 21/36] Hexagon: Add SMP support

[Richard Kuo]

[-- Attachment #1: smp.diff --]
[-- Type: text/plain, Size: 10516 bytes --]

Changed smp.c to use generic IPI handler functions.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>

---
 arch/hexagon/include/asm/smp.h |   44 ++++++
 arch/hexagon/kernel/smp.c      |  282 +++++++++++++++++++++++++++++++++++++++++
 arch/hexagon/kernel/topology.c |   52 +++++++
 3 files changed, 378 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/smp.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/smp.h	2011-09-07 13:01:00.480691689 -0500
@@ -0,0 +1,44 @@
+/*
+ * SMP definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef __ASM_SMP_H
+#define __ASM_SMP_H
+
+#include <linux/cpumask.h>
+
+#define raw_smp_processor_id() (current_thread_info()->cpu)
+
+enum ipi_message_type {
+	IPI_NOP = 0,
+	IPI_RESCHEDULE = 1,
+	IPI_CALL_FUNC,
+	IPI_CALL_FUNC_SINGLE,
+	IPI_CPU_STOP,
+	IPI_TIMER,
+};
+
+extern void send_ipi(const struct cpumask *cpumask, enum ipi_message_type msg);
+extern void smp_start_cpus(void);
+extern void arch_send_call_function_single_ipi(int cpu);
+extern void arch_send_call_function_ipi_mask(const struct cpumask *mask);
+
+extern void smp_vm_unmask_irq(void *info);
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/smp.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/smp.c	2011-09-07 13:01:00.500691719 -0500
@@ -0,0 +1,282 @@
+/*
+ * SMP support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/err.h>
+#include <linux/errno.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <linux/percpu.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/spinlock.h>
+
+#include <asm/system.h>  /*  xchg  */
+#include <asm/time.h>    /*  timer_interrupt  */
+#include <asm/hexagon_vm.h>
+
+#define BASE_IPI_IRQ 26
+
+/*
+ * cpu_possible_map needs to be filled out prior to setup_per_cpu_areas
+ * (which is prior to any of our smp_prepare_cpu crap), in order to set
+ * up the...  per_cpu areas.
+ */
+
+struct ipi_data {
+	spinlock_t lock;
+	unsigned long ipi_count;
+	unsigned long bits;
+};
+
+static DEFINE_PER_CPU(struct ipi_data, ipi_data) = {
+	.lock = __SPIN_LOCK_UNLOCKED(ipi_data.lock)
+};
+
+static inline void __handle_ipi(unsigned long *ops, struct ipi_data *ipi,
+				int cpu)
+{
+	do {
+		unsigned long msg = ffz(~*ops);
+
+		*ops &= ~(1 << msg);
+
+		switch (msg) {
+		case IPI_TIMER:
+			ipi_timer();
+			break;
+
+		case IPI_CALL_FUNC:
+			generic_smp_call_function_interrupt();
+			break;
+
+		case IPI_CALL_FUNC_SINGLE:
+			generic_smp_call_function_single_interrupt();
+			break;
+
+		case IPI_CPU_STOP:
+			/*
+			 * call vmstop()
+			 */
+			__vmstop();
+			break;
+
+		case IPI_RESCHEDULE:
+			/*
+			 * need to revisit our exit path; see comments for
+			 * scheduler_ipi()
+			 */
+			scheduler_ipi();
+			break;
+		}
+	} while (*ops);
+}
+
+/*  Used for IPI call from other CPU's to unmask int  */
+void smp_vm_unmask_irq(void *info)
+{
+	__vmintop_locen((long) info);
+}
+
+
+/*
+ * This is based on Alpha's IPI stuff.
+ * Supposed to take (int, void*) as args now.
+ * Specifically, first arg is irq, second is the irq_desc.
+ */
+
+irqreturn_t handle_ipi(int irq, void *desc)
+{
+	int cpu = smp_processor_id();
+	struct ipi_data *ipi = &per_cpu(ipi_data, cpu);
+	unsigned long ops;
+
+	while ((ops = xchg(&ipi->bits, 0)) != 0)
+		__handle_ipi(&ops, ipi, cpu);
+	return IRQ_HANDLED;
+}
+
+void send_ipi(const struct cpumask *cpumask, enum ipi_message_type msg)
+{
+	unsigned long flags;
+	unsigned long cpu;
+	unsigned long retval;
+
+	local_irq_save(flags);
+
+	for_each_cpu(cpu, cpumask) {
+		struct ipi_data *ipi = &per_cpu(ipi_data, cpu);
+
+		spin_lock(&ipi->lock);
+		ipi->bits |= 1 << msg;  /*  one bit per message  */
+		spin_unlock(&ipi->lock);
+
+		retval = __vmintop_post(BASE_IPI_IRQ+cpu);
+		if (retval != 0) {
+			printk(KERN_ERR "interrupt %ld not configured?\n",
+				BASE_IPI_IRQ+cpu);
+		}
+	}
+
+	local_irq_restore(flags);
+}
+
+static struct irqaction ipi_intdesc = {
+	.handler = handle_ipi,
+	.flags = IRQF_DISABLED | IRQF_TRIGGER_RISING,
+	.name = "ipi_handler"
+};
+
+void __init smp_prepare_boot_cpu(void)
+{
+}
+
+/*
+ * interrupts should already be disabled from the VM
+ * SP should already be correct; need to set THREADINFO_REG
+ * to point to current thread info
+ */
+
+void __cpuinit start_secondary(void)
+{
+	unsigned int cpu;
+	unsigned long thread_ptr;
+
+	/*  Calculate thread_info pointer from stack pointer  */
+	__asm__ __volatile__(
+		"%0 = SP;\n"
+		: "=r" (thread_ptr)
+	);
+
+	thread_ptr = thread_ptr & ~(THREAD_SIZE-1);
+
+	__asm__ __volatile__(
+		QUOTED_THREADINFO_REG " = %0;\n"
+		:
+		: "r" (thread_ptr)
+	);
+
+	/*  Set the memory struct  */
+	atomic_inc(&init_mm.mm_count);
+	current->active_mm = &init_mm;
+
+	cpu = smp_processor_id();
+
+	setup_irq(BASE_IPI_IRQ + cpu, &ipi_intdesc);
+
+	/*  Register the clock_event dummy  */
+	setup_percpu_clockdev();
+
+	printk(KERN_INFO "%s cpu %d\n", __func__, current_thread_info()->cpu);
+
+	local_irq_enable();
+	set_cpu_online(cpu, true);
+	cpu_idle();
+}
+
+
+/*
+ * called once for each present cpu
+ * apparently starts up the CPU and then
+ * maintains control until "cpu_online(cpu)" is set.
+ */
+
+int __cpuinit __cpu_up(unsigned int cpu)
+{
+	struct task_struct *idle;
+	struct thread_info *thread;
+	void *stack_start;
+
+	/*  Create new init task for the CPU  */
+	idle = fork_idle(cpu);
+	if (IS_ERR(idle))
+		panic(KERN_ERR "fork_idle failed\n");
+
+	thread = (struct thread_info *)idle->stack;
+	thread->cpu = cpu;
+
+	/*  Boot to the head.  */
+	stack_start =  ((void *) thread) + THREAD_SIZE;
+	__vmstart(start_secondary, stack_start);
+
+	while (!cpu_isset(cpu, cpu_online_map))
+		barrier();
+
+	return 0;
+}
+
+void __init smp_cpus_done(unsigned int max_cpus)
+{
+}
+
+void __init smp_prepare_cpus(unsigned int max_cpus)
+{
+	int i;
+
+	/*
+	 * should eventually have some sort of machine
+	 * descriptor that has this stuff
+	 */
+
+	/*  Right now, let's just fake it. */
+	for (i = 0; i < max_cpus; i++)
+		cpu_set(i, cpu_present_map);
+
+	/*  Also need to register the interrupts for IPI  */
+	if (max_cpus > 1)
+		setup_irq(BASE_IPI_IRQ, &ipi_intdesc);
+}
+
+void smp_send_reschedule(int cpu)
+{
+	send_ipi(cpumask_of(cpu), IPI_RESCHEDULE);
+}
+
+void smp_send_stop(void)
+{
+	struct cpumask targets;
+	cpumask_copy(&targets, cpu_online_mask);
+	cpumask_clear_cpu(smp_processor_id(), &targets);
+	send_ipi(&targets, IPI_CPU_STOP);
+}
+
+void arch_send_call_function_single_ipi(int cpu)
+{
+	send_ipi(cpumask_of(cpu), IPI_CALL_FUNC_SINGLE);
+}
+
+void arch_send_call_function_ipi_mask(const struct cpumask *mask)
+{
+	send_ipi(mask, IPI_CALL_FUNC);
+}
+
+int setup_profiling_timer(unsigned int multiplier)
+{
+	return -EINVAL;
+}
+
+void smp_start_cpus(void)
+{
+	int i;
+
+	for (i = 0; i < NR_CPUS; i++)
+		cpu_set(i, cpu_possible_map);
+}
Index: linux-hexagon-kernel/arch/hexagon/kernel/topology.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/topology.c	2011-09-03 20:14:59.704981335 -0500
@@ -0,0 +1,52 @@
+/*
+ * CPU topology for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/cpu.h>
+#include <linux/cpumask.h>
+#include <linux/init.h>
+#include <linux/node.h>
+#include <linux/nodemask.h>
+#include <linux/percpu.h>
+
+/*  Swiped from MIPS.  */
+
+static DEFINE_PER_CPU(struct cpu, cpu_devices);
+
+static int __init topology_init(void)
+{
+	int i, ret;
+
+	for_each_present_cpu(i) {
+
+		/*
+		 * register_cpu takes a per_cpu pointer and
+		 * just points it at another per_cpu struct...
+		 */
+
+		ret = register_cpu(&per_cpu(cpu_devices, i), i);
+		if (ret)
+			printk(KERN_WARNING "topology_init: register_cpu %d "
+			       "failed (%d)\n", i, ret);
+	}
+
+	return 0;
+}
+
+subsys_initcall(topology_init);

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 21/36] Hexagon: Add SMP support

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:08 Richard Kuo wrote:
> Changed smp.c to use generic IPI handler functions.
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
> Signed-off-by: Linas Vepstas <linas@codeaurora.org>

Acked-by: Arnd Bergmann <arnd@arndb.de>

Re: [patch v3 21/36] Hexagon: Add SMP support

[Thomas Gleixner]

On Thu, 8 Sep 2011, Richard Kuo wrote:

> +	/*  Set the memory struct  */
> +	atomic_inc(&init_mm.mm_count);
> +	current->active_mm = &init_mm;
> +
> +	cpu = smp_processor_id();
> +
> +	setup_irq(BASE_IPI_IRQ + cpu, &ipi_intdesc);
> +
> +	/*  Register the clock_event dummy  */
> +	setup_percpu_clockdev();
> +
> +	printk(KERN_INFO "%s cpu %d\n", __func__, current_thread_info()->cpu);
> +
> +	local_irq_enable();
> +	set_cpu_online(cpu, true);

This wants to be:

     	set_cpu_online(cpu, true);
        while (!cpumask_test_cpu(cpu, cpu_active_mask))
                cpu_relax();
	local_irq_enable();

See x86/kernel/smpboot.c and 
http://lists.arm.linux.org.uk/lurker/message/20110908.215756.a2ffbe5f.en.html

Thanks,

	tglx

Re: [patch v3 21/36] Hexagon: Add SMP support

[Benjamin Herrenschmidt]

On Thu, 2011-09-08 at 20:09 -0500, Richard Kuo wrote:

> +irqreturn_t handle_ipi(int irq, void *desc)
> +{
> +	int cpu = smp_processor_id();
> +	struct ipi_data *ipi = &per_cpu(ipi_data, cpu);
> +	unsigned long ops;
> +
> +	while ((ops = xchg(&ipi->bits, 0)) != 0)
> +		__handle_ipi(&ops, ipi, cpu);
> +	return IRQ_HANDLED;
> +}

So on the consumer side you are using xchg() ...

> +void send_ipi(const struct cpumask *cpumask, enum ipi_message_type msg)
> +{
> +	unsigned long flags;
> +	unsigned long cpu;
> +	unsigned long retval;
> +
> +	local_irq_save(flags);
> +
> +	for_each_cpu(cpu, cpumask) {
> +		struct ipi_data *ipi = &per_cpu(ipi_data, cpu);
> +
> +		spin_lock(&ipi->lock);
> +		ipi->bits |= 1 << msg;  /*  one bit per message  */
> +		spin_unlock(&ipi->lock);
> +
> +		retval = __vmintop_post(BASE_IPI_IRQ+cpu);
> +		if (retval != 0) {
> +			printk(KERN_ERR "interrupt %ld not configured?\n",
> +				BASE_IPI_IRQ+cpu);
> +		}
> +	}
> +
> +	local_irq_restore(flags);
> +}

And on the producer side, a spinlock for one or... that looks like wheel
re-inventing to me :-)

In fact your bit iteration loop also re-invents find_*_bit interfaces.

Any reason why you don't simply use bitops or atomics here ?

Also what is your memory model ? ordered or weakly ordered ? Do you need
some kind of memory barrier between setting ipi->bits and
__vmintop_post() ?

Cheers,
Ben.

Re: [patch v3 21/36] Hexagon: Add SMP support

[Richard Kuo]

On Sun, Sep 11, 2011 at 11:51:52AM -0300, Benjamin Herrenschmidt wrote:
> And on the producer side, a spinlock for one or... that looks like wheel
> re-inventing to me :-)
> 
> In fact your bit iteration loop also re-invents find_*_bit interfaces.
> 
> Any reason why you don't simply use bitops or atomics here ?

No particular reason; just the way it turned out when I was getting things
working.  I'll get them cleaned up.


> Also what is your memory model ? ordered or weakly ordered ? Do you need
> some kind of memory barrier between setting ipi->bits and
> __vmintop_post() ?

Our model is ordered, so there's no need for a barrier there yet.


Thanks,
Richard Kuo


-- 

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 22/36] Hexagon: Add locking types and functions

[Richard Kuo]

[-- Attachment #1: rwsem-spin.diff --]
[-- Type: text/plain, Size: 10247 bytes --]

Moved rwsem to asm-generic.  Will submit a subsequent patch for PPC
use it from the new location.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

---
 arch/hexagon/include/asm/spinlock.h       |  186 ++++++++++++++++++++++++++++++
 arch/hexagon/include/asm/spinlock_types.h |   42 ++++++
 include/asm-generic/rwsem.h               |  132 +++++++++++++++++++++
 3 files changed, 360 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/spinlock.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/spinlock.h	2011-09-03 20:14:59.714981363 -0500
@@ -0,0 +1,186 @@
+/*
+ * Spinlock support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SPINLOCK_H
+#define _ASM_SPINLOCK_H
+
+#include <asm/irqflags.h>
+
+/*
+ * This file is pulled in for SMP builds.
+ * Really need to check all the barrier stuff for "true" SMP
+ */
+
+/*
+ * Read locks:
+ * - load the lock value
+ * - increment it
+ * - if the lock value is still negative, go back and try again.
+ * - unsuccessful store is unsuccessful.  Go back and try again.  Loser.
+ * - successful store new lock value if positive -> lock acquired
+ */
+static inline void arch_read_lock(arch_rwlock_t *lock)
+{
+	__asm__ __volatile__(
+		"1:	R6 = memw_locked(%0);\n"
+		"	{ P3 = cmp.ge(R6,#0); R6 = add(R6,#1);}\n"
+		"	{ if !P3 jump 1b; }\n"
+		"	memw_locked(%0,P3) = R6;\n"
+		"	{ if !P3 jump 1b; }\n"
+		:
+		: "r" (&lock->lock)
+		: "memory", "r6", "p3"
+	);
+
+}
+
+static inline void arch_read_unlock(arch_rwlock_t *lock)
+{
+	__asm__ __volatile__(
+		"1:	R6 = memw_locked(%0);\n"
+		"	R6 = add(R6,#-1);\n"
+		"	memw_locked(%0,P3) = R6\n"
+		"	if !P3 jump 1b;\n"
+		:
+		: "r" (&lock->lock)
+		: "memory", "r6", "p3"
+	);
+
+}
+
+/*  I think this returns 0 on fail, 1 on success.  */
+static inline int arch_read_trylock(arch_rwlock_t *lock)
+{
+	int temp;
+	__asm__ __volatile__(
+		"	R6 = memw_locked(%1);\n"
+		"	{ %0 = #0; P3 = cmp.ge(R6,#0); R6 = add(R6,#1);}\n"
+		"	{ if !P3 jump 1f; }\n"
+		"	memw_locked(%1,P3) = R6;\n"
+		"	{ %0 = P3 }\n"
+		"1:\n"
+		: "=&r" (temp)
+		: "r" (&lock->lock)
+		: "memory", "r6", "p3"
+	);
+	return temp;
+}
+
+static inline int arch_read_can_lock(arch_rwlock_t *rwlock)
+{
+	return rwlock->lock == 0;
+}
+
+static inline int arch_write_can_lock(arch_rwlock_t *rwlock)
+{
+	return rwlock->lock == 0;
+}
+
+/*  Stuffs a -1 in the lock value?  */
+static inline void arch_write_lock(arch_rwlock_t *lock)
+{
+	__asm__ __volatile__(
+		"1:	R6 = memw_locked(%0)\n"
+		"	{ P3 = cmp.eq(R6,#0);  R6 = #-1;}\n"
+		"	{ if !P3 jump 1b; }\n"
+		"	memw_locked(%0,P3) = R6;\n"
+		"	{ if !P3 jump 1b; }\n"
+		:
+		: "r" (&lock->lock)
+		: "memory", "r6", "p3"
+	);
+}
+
+
+static inline int arch_write_trylock(arch_rwlock_t *lock)
+{
+	int temp;
+	__asm__ __volatile__(
+		"	R6 = memw_locked(%1)\n"
+		"	{ %0 = #0; P3 = cmp.eq(R6,#0);  R6 = #-1;}\n"
+		"	{ if !P3 jump 1f; }\n"
+		"	memw_locked(%1,P3) = R6;\n"
+		"	%0 = P3;\n"
+		"1:\n"
+		: "=&r" (temp)
+		: "r" (&lock->lock)
+		: "memory", "r6", "p3"
+	);
+	return temp;
+
+}
+
+static inline void arch_write_unlock(arch_rwlock_t *lock)
+{
+	smp_mb();
+	lock->lock = 0;
+}
+
+static inline void arch_spin_lock(arch_spinlock_t *lock)
+{
+	__asm__ __volatile__(
+		"1:	R6 = memw_locked(%0);\n"
+		"	P3 = cmp.eq(R6,#0);\n"
+		"	{ if !P3 jump 1b; R6 = #1; }\n"
+		"	memw_locked(%0,P3) = R6;\n"
+		"	{ if !P3 jump 1b; }\n"
+		:
+		: "r" (&lock->lock)
+		: "memory", "r6", "p3"
+	);
+
+}
+
+static inline void arch_spin_unlock(arch_spinlock_t *lock)
+{
+	smp_mb();
+	lock->lock = 0;
+}
+
+static inline unsigned int arch_spin_trylock(arch_spinlock_t *lock)
+{
+	int temp;
+	__asm__ __volatile__(
+		"	R6 = memw_locked(%1);\n"
+		"	P3 = cmp.eq(R6,#0);\n"
+		"	{ if !P3 jump 1f; R6 = #1; %0 = #0; }\n"
+		"	memw_locked(%1,P3) = R6;\n"
+		"	%0 = P3;\n"
+		"1:\n"
+		: "=&r" (temp)
+		: "r" (&lock->lock)
+		: "memory", "r6", "p3"
+	);
+	return temp;
+}
+
+/*
+ * SMP spinlocks are intended to allow only a single CPU at the lock
+ */
+#define arch_spin_lock_flags(lock, flags) arch_spin_lock(lock)
+#define arch_spin_unlock_wait(lock) \
+	do {while (arch_spin_is_locked(lock)) cpu_relax(); } while (0)
+#define arch_spin_is_locked(x) ((x)->lock != 0)
+
+#define arch_read_lock_flags(lock, flags) arch_read_lock(lock)
+#define arch_write_lock_flags(lock, flags) arch_write_lock(lock)
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/spinlock_types.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/spinlock_types.h	2011-09-03 20:14:59.714981363 -0500
@@ -0,0 +1,42 @@
+/*
+ * Spinlock support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_SPINLOCK_TYPES_H
+#define _ASM_SPINLOCK_TYPES_H
+
+#include <linux/version.h>
+
+#ifndef __LINUX_SPINLOCK_TYPES_H
+# error "please don't include this file directly"
+#endif
+
+typedef struct {
+	volatile unsigned int lock;
+} arch_spinlock_t;
+
+#define __ARCH_SPIN_LOCK_UNLOCKED	{ 0 }
+
+typedef struct {
+	volatile unsigned int lock;
+} arch_rwlock_t;
+
+#define __ARCH_RW_LOCK_UNLOCKED		{ 0 }
+
+#endif
Index: linux-hexagon-kernel/include/asm-generic/rwsem.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/include/asm-generic/rwsem.h	2011-09-07 13:00:37.190656296 -0500
@@ -0,0 +1,132 @@
+#ifndef _ASM_POWERPC_RWSEM_H
+#define _ASM_POWERPC_RWSEM_H
+
+#ifndef _LINUX_RWSEM_H
+#error "Please don't include <asm/rwsem.h> directly, use <linux/rwsem.h> instead."
+#endif
+
+#ifdef __KERNEL__
+
+/*
+ * R/W semaphores for PPC using the stuff in lib/rwsem.c.
+ * Adapted largely from include/asm-i386/rwsem.h
+ * by Paul Mackerras <paulus@samba.org>.
+ */
+
+/*
+ * the semaphore definition
+ */
+#ifdef CONFIG_PPC64
+# define RWSEM_ACTIVE_MASK		0xffffffffL
+#else
+# define RWSEM_ACTIVE_MASK		0x0000ffffL
+#endif
+
+#define RWSEM_UNLOCKED_VALUE		0x00000000L
+#define RWSEM_ACTIVE_BIAS		0x00000001L
+#define RWSEM_WAITING_BIAS		(-RWSEM_ACTIVE_MASK-1)
+#define RWSEM_ACTIVE_READ_BIAS		RWSEM_ACTIVE_BIAS
+#define RWSEM_ACTIVE_WRITE_BIAS		(RWSEM_WAITING_BIAS + RWSEM_ACTIVE_BIAS)
+
+/*
+ * lock for reading
+ */
+static inline void __down_read(struct rw_semaphore *sem)
+{
+	if (unlikely(atomic_long_inc_return((atomic_long_t *)&sem->count) <= 0))
+		rwsem_down_read_failed(sem);
+}
+
+static inline int __down_read_trylock(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	while ((tmp = sem->count) >= 0) {
+		if (tmp == cmpxchg(&sem->count, tmp,
+				   tmp + RWSEM_ACTIVE_READ_BIAS)) {
+			return 1;
+		}
+	}
+	return 0;
+}
+
+/*
+ * lock for writing
+ */
+static inline void __down_write_nested(struct rw_semaphore *sem, int subclass)
+{
+	long tmp;
+
+	tmp = atomic_long_add_return(RWSEM_ACTIVE_WRITE_BIAS,
+				     (atomic_long_t *)&sem->count);
+	if (unlikely(tmp != RWSEM_ACTIVE_WRITE_BIAS))
+		rwsem_down_write_failed(sem);
+}
+
+static inline void __down_write(struct rw_semaphore *sem)
+{
+	__down_write_nested(sem, 0);
+}
+
+static inline int __down_write_trylock(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	tmp = cmpxchg(&sem->count, RWSEM_UNLOCKED_VALUE,
+		      RWSEM_ACTIVE_WRITE_BIAS);
+	return tmp == RWSEM_UNLOCKED_VALUE;
+}
+
+/*
+ * unlock after reading
+ */
+static inline void __up_read(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	tmp = atomic_long_dec_return((atomic_long_t *)&sem->count);
+	if (unlikely(tmp < -1 && (tmp & RWSEM_ACTIVE_MASK) == 0))
+		rwsem_wake(sem);
+}
+
+/*
+ * unlock after writing
+ */
+static inline void __up_write(struct rw_semaphore *sem)
+{
+	if (unlikely(atomic_long_sub_return(RWSEM_ACTIVE_WRITE_BIAS,
+				 (atomic_long_t *)&sem->count) < 0))
+		rwsem_wake(sem);
+}
+
+/*
+ * implement atomic add functionality
+ */
+static inline void rwsem_atomic_add(long delta, struct rw_semaphore *sem)
+{
+	atomic_long_add(delta, (atomic_long_t *)&sem->count);
+}
+
+/*
+ * downgrade write lock to read lock
+ */
+static inline void __downgrade_write(struct rw_semaphore *sem)
+{
+	long tmp;
+
+	tmp = atomic_long_add_return(-RWSEM_WAITING_BIAS,
+				     (atomic_long_t *)&sem->count);
+	if (tmp < 0)
+		rwsem_downgrade_wake(sem);
+}
+
+/*
+ * implement exchange and add functionality
+ */
+static inline long rwsem_atomic_update(long delta, struct rw_semaphore *sem)
+{
+	return atomic_long_add_return(delta, (atomic_long_t *)&sem->count);
+}
+
+#endif	/* __KERNEL__ */
+#endif	/* _ASM_POWERPC_RWSEM_H */

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 22/36] Hexagon: Add locking types and functions

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:09 Richard Kuo wrote:
> Moved rwsem to asm-generic.  Will submit a subsequent patch for PPC
> use it from the new location.

Excellent, thanks!
 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

Acked-by: Arnd Bergmann <arnd@arndb.de>

[patch v3 23/36] Hexagon: Add user access functions

[Richard Kuo]

[-- Attachment #1: userspace.diff --]
[-- Type: text/plain, Size: 19345 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/uaccess.h   |  116 +++++++++++++++++++++
 arch/hexagon/mm/copy_from_user.S     |  114 +++++++++++++++++++++
 arch/hexagon/mm/copy_to_user.S       |   92 +++++++++++++++++
 arch/hexagon/mm/copy_user_template.S |  185 +++++++++++++++++++++++++++++++++++
 arch/hexagon/mm/strnlen_user.S       |  140 ++++++++++++++++++++++++++
 arch/hexagon/mm/uaccess.c            |   59 +++++++++++
 6 files changed, 706 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/uaccess.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/uaccess.h	2011-09-03 20:14:59.724981391 -0500
@@ -0,0 +1,116 @@
+/*
+ * User memory access support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_UACCESS_H
+#define _ASM_UACCESS_H
+/*
+ * User space memory access functions
+ */
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <asm/segment.h>
+#include <asm/sections.h>
+
+/*
+ * access_ok: - Checks if a user space pointer is valid
+ * @type: Type of access: %VERIFY_READ or %VERIFY_WRITE.  Note that
+ *        %VERIFY_WRITE is a superset of %VERIFY_READ - if it is safe
+ *        to write to a block, it is always safe to read from it.
+ * @addr: User space pointer to start of block to check
+ * @size: Size of block to check
+ *
+ * Context: User context only.  This function may sleep.
+ *
+ * Checks if a pointer to a block of memory in user space is valid.
+ *
+ * Returns true (nonzero) if the memory block *may* be valid, false (zero)
+ * if it is definitely invalid.
+ *
+ * User address space in Hexagon, like x86, goes to 0xbfffffff, so the
+ * simple MSB-based tests used by MIPS won't work.  Some further
+ * optimization is probably possible here, but for now, keep it
+ * reasonably simple and not *too* slow.  After all, we've got the
+ * MMU for backup.
+ */
+#define VERIFY_READ     0
+#define VERIFY_WRITE    1
+
+#define __access_ok(addr, size) \
+	((get_fs().seg == KERNEL_DS.seg) || \
+	(((unsigned long)addr < get_fs().seg) && \
+	  (unsigned long)size < (get_fs().seg - (unsigned long)addr)))
+
+/*
+ * When a kernel-mode page fault is taken, the faulting instruction
+ * address is checked against a table of exception_table_entries.
+ * Each entry is a tuple of the address of an instruction that may
+ * be authorized to fault, and the address at which execution should
+ * be resumed instead of the faulting instruction, so as to effect
+ * a workaround.
+ */
+
+/*  Assembly somewhat optimized copy routines  */
+unsigned long __copy_from_user_hexagon(void *to, const void __user *from,
+				     unsigned long n);
+unsigned long __copy_to_user_hexagon(void __user *to, const void *from,
+				   unsigned long n);
+
+#define __copy_from_user(to, from, n) __copy_from_user_hexagon(to, from, n)
+#define __copy_to_user(to, from, n) __copy_to_user_hexagon(to, from, n)
+
+/*
+ * XXX todo: some additonal performance gain is possible by
+ * implementing __copy_to/from_user_inatomic, which is much
+ * like __copy_to/from_user, but performs slightly less checking.
+ */
+
+__kernel_size_t __clear_user_hexagon(void __user *dest, unsigned long count);
+#define __clear_user(a, s) __clear_user_hexagon((a), (s))
+
+#define __strncpy_from_user(dst, src, n) hexagon_strncpy_from_user(dst, src, n)
+
+/*  get around the ifndef in asm-generic/uaccess.h  */
+#define __strnlen_user __strnlen_user
+
+extern long __strnlen_user(const char __user *src, long n);
+
+static inline long hexagon_strncpy_from_user(char *dst, const char __user *src,
+					     long n);
+
+#include <asm-generic/uaccess.h>
+
+/*  Todo:  an actual accelerated version of this.  */
+static inline long hexagon_strncpy_from_user(char *dst, const char __user *src,
+					     long n)
+{
+	long res = __strnlen_user(src, n);
+
+	/* return from strnlen can't be zero -- that would be rubbish. */
+
+	if (res > n) {
+		copy_from_user(dst, src, n);
+		return n;
+	} else {
+		copy_from_user(dst, src, res);
+		return res-1;
+	}
+}
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/mm/copy_from_user.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/copy_from_user.S	2011-09-03 20:14:59.724981391 -0500
@@ -0,0 +1,114 @@
+/*
+ * User memory copy functions for kernel
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * The right way to do this involves valignb
+ * The easy way to do this is only speed up src/dest similar alignment.
+ */
+
+/*
+ * Copy to/from user are the same, except that for packets with a load and
+ * a store, I don't know how to tell which kind of exception we got.
+ * Therefore, we duplicate the function, and handle faulting addresses
+ * differently for each function
+ */
+
+/*
+ * copy from user: loads can fault
+ */
+#define src_sav r13
+#define dst_sav r12
+#define src_dst_sav r13:12
+#define d_dbuf r15:14
+#define w_dbuf r15
+
+#define dst r0
+#define src r1
+#define bytes r2
+#define loopcount r5
+
+#define FUNCNAME __copy_from_user_hexagon
+#include "copy_user_template.S"
+
+	/* LOAD FAULTS from COPY_FROM_USER */
+
+	/* Alignment loop.  r2 has been updated. Return it. */
+	.falign
+1009:
+2009:
+4009:
+	{
+		r0 = r2
+		jumpr r31
+	}
+	/* Normal copy loops. Do epilog. Use src-src_sav to compute distance */
+	/* X - (A - B) == X + B - A */
+	.falign
+8089:
+	{
+		memd(dst) = d_dbuf
+		r2 += sub(src_sav,src)
+	}
+	{
+		r0 = r2
+		jumpr r31
+	}
+	.falign
+4089:
+	{
+		memw(dst) = w_dbuf
+		r2 += sub(src_sav,src)
+	}
+	{
+		r0 = r2
+		jumpr r31
+	}
+	.falign
+2089:
+	{
+		memh(dst) = w_dbuf
+		r2 += sub(src_sav,src)
+	}
+	{
+		r0 = r2
+		jumpr r31
+	}
+	.falign
+1089:
+	{
+		memb(dst) = w_dbuf
+		r2 += sub(src_sav,src)
+	}
+	{
+		r0 = r2
+		jumpr r31
+	}
+
+	/* COPY FROM USER: only loads can fail */
+
+	.section __ex_table,"a"
+	.long 1000b,1009b
+	.long 2000b,2009b
+	.long 4000b,4009b
+	.long 8080b,8089b
+	.long 4080b,4089b
+	.long 2080b,2089b
+	.long 1080b,1089b
+	.previous
Index: linux-hexagon-kernel/arch/hexagon/mm/copy_to_user.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/copy_to_user.S	2011-09-03 20:14:59.724981391 -0500
@@ -0,0 +1,92 @@
+/*
+ * User memory copying routines for the Hexagon Kernel
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/* The right way to do this involves valignb
+ * The easy way to do this is only speed up src/dest similar alignment.
+ */
+
+/*
+ * Copy to/from user are the same, except that for packets with a load and
+ * a store, I don't know how to tell which kind of exception we got.
+ * Therefore, we duplicate the function, and handle faulting addresses
+ * differently for each function
+ */
+
+/*
+ * copy to user: stores can fault
+ */
+#define src_sav r13
+#define dst_sav r12
+#define src_dst_sav r13:12
+#define d_dbuf r15:14
+#define w_dbuf r15
+
+#define dst r0
+#define src r1
+#define bytes r2
+#define loopcount r5
+
+#define FUNCNAME __copy_to_user_hexagon
+#include "copy_user_template.S"
+
+	/* STORE FAULTS from COPY_TO_USER */
+	.falign
+1109:
+2109:
+4109:
+	/* Alignment loop.  r2 has been updated.  Return it. */
+	{
+		r0 = r2
+		jumpr r31
+	}
+	/* Normal copy loops.  Use dst-dst_sav to compute distance */
+	/* dst holds best write, no need to unwind any loops */
+	/* X - (A - B) == X + B - A */
+	.falign
+8189:
+8199:
+4189:
+4199:
+2189:
+2199:
+1189:
+1199:
+	{
+		r2 += sub(dst_sav,dst)
+	}
+	{
+		r0 = r2
+		jumpr r31
+	}
+
+	/* COPY TO USER: only stores can fail */
+	.section __ex_table,"a"
+	.long 1100b,1109b
+	.long 2100b,2109b
+	.long 4100b,4109b
+	.long 8180b,8189b
+	.long 8190b,8199b
+	.long 4180b,4189b
+	.long 4190b,4199b
+	.long 2180b,2189b
+	.long 2190b,2199b
+	.long 1180b,1189b
+	.long 1190b,1199b
+	.previous
Index: linux-hexagon-kernel/arch/hexagon/mm/copy_user_template.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/copy_user_template.S	2011-09-03 20:14:59.724981391 -0500
@@ -0,0 +1,185 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/* Numerology:
+ * WXYZ
+ * W: width in bytes
+ * X: Load=0, Store=1
+ * Y: Location 0=preamble,8=loop,9=epilog
+ * Z: Location=0,handler=9
+ */
+	.text
+	.global FUNCNAME
+	.type FUNCNAME, @function
+	.p2align 5
+FUNCNAME:
+	{
+		p0 = cmp.gtu(bytes,#0)
+		if (!p0.new) jump:nt .Ldone
+		r3 = or(dst,src)
+		r4 = xor(dst,src)
+	}
+	{
+		p1 = cmp.gtu(bytes,#15)
+		p0 = bitsclr(r3,#7)
+		if (!p0.new) jump:nt .Loop_not_aligned_8
+		src_dst_sav = combine(src,dst)
+	}
+
+	{
+		loopcount = lsr(bytes,#3)
+		if (!p1) jump .Lsmall
+	}
+	p3=sp1loop0(.Loop8,loopcount)
+.Loop8:
+8080:
+8180:
+	{
+		if (p3) memd(dst++#8) = d_dbuf
+		d_dbuf = memd(src++#8)
+	}:endloop0
+8190:
+	{
+		memd(dst++#8) = d_dbuf
+		bytes -= asl(loopcount,#3)
+		jump .Lsmall
+	}
+
+.Loop_not_aligned_8:
+	{
+		p0 = bitsclr(r4,#7)
+		if (p0.new) jump:nt .Lalign
+	}
+	{
+		p0 = bitsclr(r3,#3)
+		if (!p0.new) jump:nt .Loop_not_aligned_4
+		p1 = cmp.gtu(bytes,#7)
+	}
+
+	{
+		if (!p1) jump .Lsmall
+		loopcount = lsr(bytes,#2)
+	}
+	p3=sp1loop0(.Loop4,loopcount)
+.Loop4:
+4080:
+4180:
+	{
+		if (p3) memw(dst++#4) = w_dbuf
+		w_dbuf = memw(src++#4)
+	}:endloop0
+4190:
+	{
+		memw(dst++#4) = w_dbuf
+		bytes -= asl(loopcount,#2)
+		jump .Lsmall
+	}
+
+.Loop_not_aligned_4:
+	{
+		p0 = bitsclr(r3,#1)
+		if (!p0.new) jump:nt .Loop_not_aligned
+		p1 = cmp.gtu(bytes,#3)
+	}
+
+	{
+		if (!p1) jump .Lsmall
+		loopcount = lsr(bytes,#1)
+	}
+	p3=sp1loop0(.Loop2,loopcount)
+.Loop2:
+2080:
+2180:
+	{
+		if (p3) memh(dst++#2) = w_dbuf
+		w_dbuf = memuh(src++#2)
+	}:endloop0
+2190:
+	{
+		memh(dst++#2) = w_dbuf
+		bytes -= asl(loopcount,#1)
+		jump .Lsmall
+	}
+
+.Loop_not_aligned: /* Works for as small as one byte */
+	p3=sp1loop0(.Loop1,bytes)
+.Loop1:
+1080:
+1180:
+	{
+		if (p3) memb(dst++#1) = w_dbuf
+		w_dbuf = memub(src++#1)
+	}:endloop0
+	/* Done */
+1190:
+	{
+		memb(dst) = w_dbuf
+		jumpr r31
+		r0 = #0
+	}
+
+.Lsmall:
+	{
+		p0 = cmp.gtu(bytes,#0)
+		if (p0.new) jump:nt .Loop_not_aligned
+	}
+.Ldone:
+	{
+		r0 = #0
+		jumpr r31
+	}
+	.falign
+.Lalign:
+1000:
+	{
+		if (p0.new) w_dbuf = memub(src)
+		p0 = tstbit(src,#0)
+		if (!p1) jump .Lsmall
+	}
+1100:
+	{
+		if (p0) memb(dst++#1) = w_dbuf
+		if (p0) bytes = add(bytes,#-1)
+		if (p0) src = add(src,#1)
+	}
+2000:
+	{
+		if (p0.new) w_dbuf = memuh(src)
+		p0 = tstbit(src,#1)
+		if (!p1) jump .Lsmall
+	}
+2100:
+	{
+		if (p0) memh(dst++#2) = w_dbuf
+		if (p0) bytes = add(bytes,#-2)
+		if (p0) src = add(src,#2)
+	}
+4000:
+	{
+		if (p0.new) w_dbuf = memw(src)
+		p0 = tstbit(src,#2)
+		if (!p1) jump .Lsmall
+	}
+4100:
+	{
+		if (p0) memw(dst++#4) = w_dbuf
+		if (p0) bytes = add(bytes,#-4)
+		if (p0) src = add(src,#4)
+		jump FUNCNAME
+	}
+	.size FUNCNAME,.-FUNCNAME
Index: linux-hexagon-kernel/arch/hexagon/mm/uaccess.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/uaccess.c	2011-09-03 20:14:59.724981391 -0500
@@ -0,0 +1,59 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * Support for user memory access from kernel.  This will
+ * probably be inlined for performance at some point, but
+ * for ease of debug, and to a lesser degree for code size,
+ * we implement here as subroutines.
+ */
+#include <linux/types.h>
+#include <asm/uaccess.h>
+#include <asm/pgtable.h>
+
+/*
+ * For clear_user(), exploit previously defined copy_to_user function
+ * and the fact that we've got a handy zero page defined in kernel/head.S
+ *
+ * dczero here would be even faster.
+ */
+__kernel_size_t __clear_user_hexagon(void __user *dest, unsigned long count)
+{
+	long uncleared;
+
+	while (count > PAGE_SIZE) {
+		uncleared = __copy_to_user_hexagon(dest, &empty_zero_page,
+						PAGE_SIZE);
+		if (uncleared)
+			return count - (PAGE_SIZE - uncleared);
+		count -= PAGE_SIZE;
+		dest += PAGE_SIZE;
+	}
+	if (count)
+		count = __copy_to_user_hexagon(dest, &empty_zero_page, count);
+
+	return count;
+}
+
+unsigned long clear_user_hexagon(void __user *dest, unsigned long count)
+{
+	if (!access_ok(VERIFY_WRITE, dest, count))
+		return count;
+	else
+		return __clear_user_hexagon(dest, count);
+}
Index: linux-hexagon-kernel/arch/hexagon/mm/strnlen_user.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/strnlen_user.S	2011-09-03 20:14:59.724981391 -0500
@@ -0,0 +1,140 @@
+/*
+ * User string length functions for kernel
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#define isrc	r0
+#define max	r1	/*  Do not change!  */
+
+#define end	r2
+#define tmp1	r3
+
+#define obo	r6	/*  off-by-one  */
+#define start	r7
+#define mod8	r8
+#define dbuf    r15:14
+#define dcmp	r13:12
+
+/*
+ * The vector mask version of this turned out *really* badly.
+ * The hardware loop version also turned out *really* badly.
+ * Seems straight pointer arithmetic basically wins here.
+ */
+
+#define fname __strnlen_user
+
+	.text
+	.global fname
+	.type fname, @function
+	.p2align 5  /*  why?  */
+fname:
+	{
+		mod8 = and(isrc,#7);
+		end = add(isrc,max);
+		start = isrc;
+	}
+	{
+		P0 = cmp.eq(mod8,#0);
+		mod8 = and(end,#7);
+		dcmp = #0;
+		if (P0.new) jump:t dw_loop;	/*  fire up the oven  */
+	}
+
+alignment_loop:
+fail_1:	{
+		tmp1 = memb(start++#1);
+	}
+	{
+		P0 = cmp.eq(tmp1,#0);
+		if (P0.new) jump:nt exit_found;
+		P1 = cmp.gtu(end,start);
+		mod8 = and(start,#7);
+	}
+	{
+		if (!P1) jump exit_error;  /*  hit the end  */
+		P0 = cmp.eq(mod8,#0);
+	}
+	{
+		if (!P0) jump alignment_loop;
+	}
+
+
+
+dw_loop:
+fail_2:	{
+		dbuf = memd(start);
+		obo = add(start,#1);
+	}
+	{
+		P0 = vcmpb.eq(dbuf,dcmp);
+	}
+	{
+		tmp1 = P0;
+		P0 = cmp.gtu(end,start);
+	}
+	{
+		tmp1 = ct0(tmp1);
+		mod8 = and(end,#7);
+		if (!P0) jump end_check;
+	}
+	{
+		P0 = cmp.eq(tmp1,#32);
+		if (!P0.new) jump:nt exit_found;
+		if (!P0.new) start = add(obo,tmp1);
+	}
+	{
+		start = add(start,#8);
+		jump dw_loop;
+	}	/*  might be nice to combine these jumps...   */
+
+
+end_check:
+	{
+		P0 = cmp.gt(tmp1,mod8);
+		if (P0.new) jump:nt exit_error;	/*  neverfound!  */
+		start = add(obo,tmp1);
+	}
+
+exit_found:
+	{
+		R0 = sub(start,isrc);
+		jumpr R31;
+	}
+
+exit_error:
+	{
+		R0 = add(max,#1);
+		jumpr R31;
+	}
+
+	/*  Uh, what does the "fixup" return here?  */
+	.falign
+fix_1:
+	{
+		R0 = #0;
+		jumpr R31;
+	}
+
+	.size fname,.-fname
+
+
+.section __ex_table,"a"
+.long fail_1,fix_1
+.long fail_2,fix_1
+.previous
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Richard Kuo]

[-- Attachment #1: io.diff --]
[-- Type: text/plain, Size: 10790 bytes --]

Filled in the ins/outs; leaving them in for now.

Change _IO_BASE and IO_SPACE_LIMIT for safety.

I have compiled this with sparse and it builds; I've also picked over
various files related to the HAS_IOPORT_MAP from arm-soc.git and
tried them out, and they work, so I guess we'll just wait until
they're merged.


Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>

---
 arch/hexagon/include/asm/io.h |  323 ++++++++++++++++++++++++++++++++++++++++++
 arch/hexagon/lib/io.c         |   91 +++++++++++
 2 files changed, 414 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/io.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/io.h	2011-09-07 13:00:30.910646758 -0500
@@ -0,0 +1,323 @@
+/*
+ * IO definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_IO_H
+#define _ASM_IO_H
+
+#ifdef __KERNEL__
+
+#include <linux/types.h>
+#include <linux/delay.h>
+#include <linux/vmalloc.h>
+#include <asm/string.h>
+#include <asm/mem-layout.h>
+#include <asm/iomap.h>
+#include <asm/page.h>
+#include <asm/cacheflush.h>
+#include <asm/tlbflush.h>
+
+/*
+ * We don't have PCI yet.
+ * _IO_BASE is pointing at what should be unused virtual space.
+ */
+#define IO_SPACE_LIMIT 0xffff
+#define _IO_BASE ((void __iomem *)0xfe000000)
+
+extern int remap_area_pages(unsigned long start, unsigned long phys_addr,
+				unsigned long end, unsigned long flags);
+
+extern void __iounmap(const volatile void __iomem *addr);
+
+/* Defined in lib/io.c, needed for smc91x driver. */
+extern void __raw_readsw(const void __iomem *addr, void *data, int wordlen);
+extern void __raw_writesw(void __iomem *addr, const void *data, int wordlen);
+
+extern void __raw_readsl(const void __iomem *addr, void *data, int wordlen);
+extern void __raw_writesl(void __iomem *addr, const void *data, int wordlen);
+
+#define readsw(p, d, l)	__raw_readsw(p, d, l)
+#define writesw(p, d, l) __raw_writesw(p, d, l)
+
+#define readsl(p, d, l)   __raw_readsl(p, d, l)
+#define writesl(p, d, l)  __raw_writesl(p, d, l)
+
+/*
+ * virt_to_phys - map virtual address to physical
+ * @address:  address to map
+ */
+static inline unsigned long virt_to_phys(volatile void *address)
+{
+	return __pa(address);
+}
+
+/*
+ * phys_to_virt - map physical address to virtual
+ * @address: address to map
+ */
+static inline void *phys_to_virt(unsigned long address)
+{
+	return __va(address);
+}
+
+/*
+ * convert a physical pointer to a virtual kernel pointer for
+ * /dev/mem access.
+ */
+#define xlate_dev_kmem_ptr(p)    __va(p)
+#define xlate_dev_mem_ptr(p)    __va(p)
+
+/*
+ * IO port access primitives.  Hexagon doesn't have special IO access
+ * instructions; all I/O is memory mapped.
+ *
+ * in/out are used for "ports", but we don't have "port instructions",
+ * so these are really just memory mapped too.
+ */
+
+/*
+ * readb - read byte from memory mapped device
+ * @addr:  pointer to memory
+ *
+ * Operates on "I/O bus memory space"
+ */
+static inline u8 readb(const volatile void __iomem *addr)
+{
+	u8 val;
+	asm volatile(
+		"%0 = memb(%1);"
+		: "=&r" (val)
+		: "r" (addr)
+	);
+	return val;
+}
+
+static inline u16 readw(const volatile void __iomem *addr)
+{
+	u16 val;
+	asm volatile(
+		"%0 = memh(%1);"
+		: "=&r" (val)
+		: "r" (addr)
+	);
+	return val;
+}
+
+static inline u32 readl(const volatile void __iomem *addr)
+{
+	u32 val;
+	asm volatile(
+		"%0 = memw(%1);"
+		: "=&r" (val)
+		: "r" (addr)
+	);
+	return val;
+}
+
+/*
+ * writeb - write a byte to a memory location
+ * @data: data to write to
+ * @addr:  pointer to memory
+ *
+ */
+static inline void writeb(u8 data, volatile void __iomem *addr)
+{
+	asm volatile(
+		"memb(%0) = %1;"
+		:
+		: "r" (addr), "r" (data)
+	);
+}
+
+static inline void writew(u16 data, volatile void __iomem *addr)
+{
+	asm volatile(
+		"memh(%0) = %1;"
+		:
+		: "r" (addr), "r" (data)
+	);
+
+}
+
+static inline void writel(u32 data, volatile void __iomem *addr)
+{
+	asm volatile(
+		"memw(%0) = %1;"
+		:
+		: "r" (addr), "r" (data)
+	);
+}
+
+#define __raw_writeb writeb
+#define __raw_writew writew
+#define __raw_writel writel
+
+#define __raw_readb readb
+#define __raw_readw readw
+#define __raw_readl readl
+
+/*
+ * Need an mtype somewhere in here, for cache type deals?
+ * This is probably too long for an inline.
+ */
+void __iomem *ioremap_nocache(unsigned long phys_addr, unsigned long size);
+
+static inline void __iomem *ioremap(unsigned long phys_addr, unsigned long size)
+{
+	return ioremap_nocache(phys_addr, size);
+}
+
+static inline void iounmap(volatile void __iomem *addr)
+{
+	__iounmap(addr);
+}
+
+#define __raw_writel writel
+
+static inline void memcpy_fromio(void *dst, const volatile void __iomem *src,
+	int count)
+{
+	memcpy(dst, (void *) src, count);
+}
+
+static inline void memcpy_toio(volatile void __iomem *dst, const void *src,
+	int count)
+{
+	memcpy((void *) dst, src, count);
+}
+
+#define PCI_IO_ADDR	(volatile void __iomem *)
+
+/*
+ * inb - read byte from I/O port or something
+ * @port:  address in I/O space
+ *
+ * Operates on "I/O bus I/O space"
+ */
+static inline u8 inb(unsigned long port)
+{
+	return readb(_IO_BASE + (port & IO_SPACE_LIMIT));
+}
+
+static inline u16 inw(unsigned long port)
+{
+	return readw(_IO_BASE + (port & IO_SPACE_LIMIT));
+}
+
+static inline u32 inl(unsigned long port)
+{
+	return readl(_IO_BASE + (port & IO_SPACE_LIMIT));
+}
+
+/*
+ * outb - write a byte to a memory location
+ * @data: data to write to
+ * @addr:  address in I/O space
+ */
+static inline void outb(u8 data, unsigned long port)
+{
+	writeb(data, _IO_BASE + (port & IO_SPACE_LIMIT));
+}
+
+static inline void outw(u16 data, unsigned long port)
+{
+	writew(data, _IO_BASE + (port & IO_SPACE_LIMIT));
+}
+
+static inline void outl(u32 data, unsigned long port)
+{
+	writel(data, _IO_BASE + (port & IO_SPACE_LIMIT));
+}
+
+#define outb_p outb
+#define outw_p outw
+#define outl_p outl
+
+#define inb_p inb
+#define inw_p inw
+#define inl_p inl
+
+static inline void insb(unsigned long port, void *buffer, int count)
+{
+	if (count) {
+		u8 *buf = buffer;
+		do {
+			u8 x = inb(port);
+			*buf++ = x;
+		} while (--count);
+	}
+}
+
+static inline void insw(unsigned long port, void *buffer, int count)
+{
+	if (count) {
+		u16 *buf = buffer;
+		do {
+			u16 x = inw(port);
+			*buf++ = x;
+		} while (--count);
+	}
+}
+
+static inline void insl(unsigned long port, void *buffer, int count)
+{
+	if (count) {
+		u32 *buf = buffer;
+		do {
+			u32 x = inw(port);
+			*buf++ = x;
+		} while (--count);
+	}
+}
+
+static inline void outsb(unsigned long port, const void *buffer, int count)
+{
+	if (count) {
+		const u8 *buf = buffer;
+		do {
+			outb(*buf++, port);
+		} while (--count);
+	}
+}
+
+static inline void outsw(unsigned long port, const void *buffer, int count)
+{
+	if (count) {
+		const u16 *buf = buffer;
+		do {
+			outw(*buf++, port);
+		} while (--count);
+	}
+}
+
+static inline void outsl(unsigned long port, const void *buffer, int count)
+{
+	if (count) {
+		const u32 *buf = buffer;
+		do {
+			outl(*buf++, port);
+		} while (--count);
+	}
+}
+
+#define flush_write_buffers() do { } while (0)
+
+#endif /* __KERNEL__ */
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/lib/io.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/lib/io.c	2011-09-03 20:14:59.734981419 -0500
@@ -0,0 +1,91 @@
+/*
+ * I/O access functions for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <asm/io.h>
+
+/*  These are all FIFO routines!  */
+
+/*
+ * __raw_readsw - read words a short at a time
+ * @addr:  source address
+ * @data:  data address
+ * @len: number of shorts to read
+ */
+void __raw_readsw(const void __iomem *addr, void *data, int len)
+{
+	const volatile short int *src = (short int *) addr;
+	short int *dst = (short int *) data;
+
+	if ((u32)data & 0x1)
+		panic("unaligned pointer to readsw");
+
+	while (len-- > 0)
+		*dst++ = *src;
+
+}
+
+/*
+ * __raw_writesw - read words a short at a time
+ * @addr:  source address
+ * @data:  data address
+ * @len: number of shorts to read
+ */
+void __raw_writesw(void __iomem *addr, const void *data, int len)
+{
+	const short int *src = (short int *)data;
+	volatile short int *dst = (short int *)addr;
+
+	if ((u32)data & 0x1)
+		panic("unaligned pointer to writesw");
+
+	while (len-- > 0)
+		*dst = *src++;
+
+
+}
+
+/*  Pretty sure len is pre-adjusted for the length of the access already */
+void __raw_readsl(const void __iomem *addr, void *data, int len)
+{
+	const volatile long *src = (long *) addr;
+	long *dst = (long *) data;
+
+	if ((u32)data & 0x3)
+		panic("unaligned pointer to readsl");
+
+	while (len-- > 0)
+		*dst++ = *src;
+
+
+}
+
+void __raw_writesl(void __iomem *addr, const void *data, int len)
+{
+	const long *src = (long *)data;
+	volatile long *dst = (long *)addr;
+
+	if ((u32)data & 0x3)
+		panic("unaligned pointer to writesl");
+
+	while (len-- > 0)
+		*dst = *src++;
+
+
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:11 Richard Kuo wrote:
> Filled in the ins/outs; leaving them in for now.
> 
> Change _IO_BASE and IO_SPACE_LIMIT for safety.
> 
> I have compiled this with sparse and it builds; I've also picked over
> various files related to the HAS_IOPORT_MAP from arm-soc.git and
> tried them out, and they work, so I guess we'll just wait until
> they're merged.
> 
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
> Signed-off-by: Linas Vepstas <linas@codeaurora.org>

Acked-by: Arnd Bergmann <arnd@arndb.de>

Re: [patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Linas Vepstas (Code Aurora)]

On Thu, Sep 08, 2011 at 08:09:11PM -0500, Richard Kuo wrote:
> +	asm volatile(
> +		"memb(%0) = %1;"
> +		:
> +		: "r" (addr), "r" (data)
> +	);

I'm kind-of vague on this, but I think that this also needs a
: "memory" at the end (to indicate that the asm snippet clobbers
memory), as otherwise the compiler may be tempted to reorder
these around other memory accesses. That is, the "volatile"
keyword is not enough (well, I'm not sure...), but I think
"volatile" only prevents the compiler from optimizing these
away completely, but is not enough to prevent unwanted
re-ordering.

--linas

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Arnd Bergmann]

On Friday 09 September 2011 14:14:13 Linas Vepstas wrote:
> On Thu, Sep 08, 2011 at 08:09:11PM -0500, Richard Kuo wrote:
> > +     asm volatile(
> > +             "memb(%0) = %1;"
> > +             :
> > +             : "r" (addr), "r" (data)
> > +     );
> 
> I'm kind-of vague on this, but I think that this also needs a
> : "memory" at the end (to indicate that the asm snippet clobbers
> memory), as otherwise the compiler may be tempted to reorder
> these around other memory accesses. That is, the "volatile"
> keyword is not enough (well, I'm not sure...), but I think
> "volatile" only prevents the compiler from optimizing these
> away completely, but is not enough to prevent unwanted
> re-ordering.

Yes, absolutely correct. Specifically, this is required for the
case where you first store values into regular memory and then
use the writel() to trigger a DMA operation. Without the memory
clobber, the dma may read from memory before the CPU has accessed
it in case the store gets scheduled after the writel by the compiler.

	Arnd

RE: [patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Taylor Simpson]

Using "memory" is a pretty big hammer.  Wouldn't it be better to use a
memory operand?
     asm volatile(
             "memb(%0) = %1;"
             : "m" (*(u8*)addr)
             : "r" (data)
     );

This may be a dumb question, but why do we need inline asm?  Why can't we
write this with plain C code?
/*
 * writeb - write a byte to a memory location
 * @data: data to write to
 * @addr:  pointer to memory
 *
 */
static inline void writeb(u8 data, volatile void __iomem *addr) {
	volatile u8 *ptr = (volatile u8*)addr;
	*ptr = data;
}


Taylor



-----Original Message-----
From: linux-hexagon-owner@vger.kernel.org
[mailto:linux-hexagon-owner@vger.kernel.org] On Behalf Of Arnd Bergmann
Sent: Friday, September 09, 2011 4:13 PM
To: Linas Vepstas (Code Aurora)
Cc: Richard Kuo; linux-arch@vger.kernel.org; linux-hexagon@vger.kernel.org;
linux-kernel@vger.kernel.org
Subject: Re: [patch v3 24/36] Hexagon: Provide basic implementation and/or
stubs for I/O routines.

On Friday 09 September 2011 14:14:13 Linas Vepstas wrote:
> On Thu, Sep 08, 2011 at 08:09:11PM -0500, Richard Kuo wrote:
> > +     asm volatile(
> > +             "memb(%0) = %1;"
> > +             :
> > +             : "r" (addr), "r" (data)
> > +     );
> 
> I'm kind-of vague on this, but I think that this also needs a
> : "memory" at the end (to indicate that the asm snippet clobbers
> memory), as otherwise the compiler may be tempted to reorder
> these around other memory accesses. That is, the "volatile"
> keyword is not enough (well, I'm not sure...), but I think
> "volatile" only prevents the compiler from optimizing these
> away completely, but is not enough to prevent unwanted
> re-ordering.

Yes, absolutely correct. Specifically, this is required for the
case where you first store values into regular memory and then
use the writel() to trigger a DMA operation. Without the memory
clobber, the dma may read from memory before the CPU has accessed
it in case the store gets scheduled after the writel by the compiler.

	Arnd

RE: [patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Taylor Simpson]

Using "memory" is a pretty big hammer.  Wouldn't it be better to use a
memory operand?
     asm volatile(
             "memb(%0) = %1;"
             : "m" (*(u8*)addr)
             : "r" (data)
     );

This may be a dumb question, but why do we need inline asm?  Why can't we
write this with plain C code?
/*
 * writeb - write a byte to a memory location
 * @data: data to write to
 * @addr:  pointer to memory
 *
 */
static inline void writeb(u8 data, volatile void __iomem *addr) {
	volatile u8 *ptr = (volatile u8*)addr;
	*ptr = data;
}


Taylor



-----Original Message-----
From: linux-hexagon-owner@vger.kernel.org
[mailto:linux-hexagon-owner@vger.kernel.org] On Behalf Of Arnd Bergmann
Sent: Friday, September 09, 2011 4:13 PM
To: Linas Vepstas (Code Aurora)
Cc: Richard Kuo; linux-arch@vger.kernel.org; linux-hexagon@vger.kernel.org;
linux-kernel@vger.kernel.org
Subject: Re: [patch v3 24/36] Hexagon: Provide basic implementation and/or
stubs for I/O routines.

On Friday 09 September 2011 14:14:13 Linas Vepstas wrote:
> On Thu, Sep 08, 2011 at 08:09:11PM -0500, Richard Kuo wrote:
> > +     asm volatile(
> > +             "memb(%0) = %1;"
> > +             :
> > +             : "r" (addr), "r" (data)
> > +     );
> 
> I'm kind-of vague on this, but I think that this also needs a
> : "memory" at the end (to indicate that the asm snippet clobbers
> memory), as otherwise the compiler may be tempted to reorder
> these around other memory accesses. That is, the "volatile"
> keyword is not enough (well, I'm not sure...), but I think
> "volatile" only prevents the compiler from optimizing these
> away completely, but is not enough to prevent unwanted
> re-ordering.

Yes, absolutely correct. Specifically, this is required for the
case where you first store values into regular memory and then
use the writel() to trigger a DMA operation. Without the memory
clobber, the dma may read from memory before the CPU has accessed
it in case the store gets scheduled after the writel by the compiler.

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe linux-hexagon" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Taylor Simpson]

Using "memory" is a pretty big hammer.  Wouldn't it be better to use a
memory operand?
     asm volatile(
             "memb(%0) = %1;"
             : "m" (*(u8*)addr)
             : "r" (data)
     );

This may be a dumb question, but why do we need inline asm?  Why can't we
write this with plain C code?
/*
 * writeb - write a byte to a memory location
 * @data: data to write to
 * @addr:  pointer to memory
 *
 */
static inline void writeb(u8 data, volatile void __iomem *addr) {
	volatile u8 *ptr = (volatile u8*)addr;
	*ptr = data;
}


Taylor



-----Original Message-----
From: linux-hexagon-owner@vger.kernel.org
[mailto:linux-hexagon-owner@vger.kernel.org] On Behalf Of Arnd Bergmann
Sent: Friday, September 09, 2011 4:13 PM
To: Linas Vepstas (Code Aurora)
Cc: Richard Kuo; linux-arch@vger.kernel.org; linux-hexagon@vger.kernel.org;
linux-kernel@vger.kernel.org
Subject: Re: [patch v3 24/36] Hexagon: Provide basic implementation and/or
stubs for I/O routines.

On Friday 09 September 2011 14:14:13 Linas Vepstas wrote:
> On Thu, Sep 08, 2011 at 08:09:11PM -0500, Richard Kuo wrote:
> > +     asm volatile(
> > +             "memb(%0) = %1;"
> > +             :
> > +             : "r" (addr), "r" (data)
> > +     );
> 
> I'm kind-of vague on this, but I think that this also needs a
> : "memory" at the end (to indicate that the asm snippet clobbers
> memory), as otherwise the compiler may be tempted to reorder
> these around other memory accesses. That is, the "volatile"
> keyword is not enough (well, I'm not sure...), but I think
> "volatile" only prevents the compiler from optimizing these
> away completely, but is not enough to prevent unwanted
> re-ordering.

Yes, absolutely correct. Specifically, this is required for the
case where you first store values into regular memory and then
use the writel() to trigger a DMA operation. Without the memory
clobber, the dma may read from memory before the CPU has accessed
it in case the store gets scheduled after the writel by the compiler.

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe linux-hexagon" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: [patch v3 24/36] Hexagon: Provide basic implementation and/or stubs for I/O routines.

[Benjamin Herrenschmidt]

On Sat, 2011-09-10 at 15:02 -0500, Taylor Simpson wrote:
> Using "memory" is a pretty big hammer.  Wouldn't it be better to use a
> memory operand?
>      asm volatile(
>              "memb(%0) = %1;"
>              : "m" (*(u8*)addr)
>              : "r" (data)
>      );

No you probably really want a full "memory" clobber, because typically
your IO ops should also act as barriers. The compiler has no way to know
for example that you want to order previous stores to memory (some DMA
descriptor you are populating) vs. an MMIO access that will trigger the
DMA from that location.

Linux defines the readX/writeX accessors as being fully ordered in that
case.

In fact, I'm not familiar with Hexagon, but if it's a weakly ordered
memory model you may also need some kind of memory barrier in those
accessors.

Cheers,
Ben.

> This may be a dumb question, but why do we need inline asm?  Why can't we
> write this with plain C code?
> /*
>  * writeb - write a byte to a memory location
>  * @data: data to write to
>  * @addr:  pointer to memory
>  *
>  */
> static inline void writeb(u8 data, volatile void __iomem *addr) {
> 	volatile u8 *ptr = (volatile u8*)addr;
> 	*ptr = data;
> }
> 
> 
> Taylor
> 
> 
> 
> -----Original Message-----
> From: linux-hexagon-owner@vger.kernel.org
> [mailto:linux-hexagon-owner@vger.kernel.org] On Behalf Of Arnd Bergmann
> Sent: Friday, September 09, 2011 4:13 PM
> To: Linas Vepstas (Code Aurora)
> Cc: Richard Kuo; linux-arch@vger.kernel.org; linux-hexagon@vger.kernel.org;
> linux-kernel@vger.kernel.org
> Subject: Re: [patch v3 24/36] Hexagon: Provide basic implementation and/or
> stubs for I/O routines.
> 
> On Friday 09 September 2011 14:14:13 Linas Vepstas wrote:
> > On Thu, Sep 08, 2011 at 08:09:11PM -0500, Richard Kuo wrote:
> > > +     asm volatile(
> > > +             "memb(%0) = %1;"
> > > +             :
> > > +             : "r" (addr), "r" (data)
> > > +     );
> > 
> > I'm kind-of vague on this, but I think that this also needs a
> > : "memory" at the end (to indicate that the asm snippet clobbers
> > memory), as otherwise the compiler may be tempted to reorder
> > these around other memory accesses. That is, the "volatile"
> > keyword is not enough (well, I'm not sure...), but I think
> > "volatile" only prevents the compiler from optimizing these
> > away completely, but is not enough to prevent unwanted
> > re-ordering.
> 
> Yes, absolutely correct. Specifically, this is required for the
> case where you first store values into regular memory and then
> use the writel() to trigger a DMA operation. Without the memory
> clobber, the dma may read from memory before the CPU has accessed
> it in case the store gets scheduled after the writel by the compiler.
> 
> 	Arnd
> --
> To unsubscribe from this list: send the line "unsubscribe linux-hexagon" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

[patch v3 25/36] Hexagon: Implement basic cache-flush support

[Richard Kuo]

[-- Attachment #1: cache.diff --]
[-- Type: text/plain, Size: 8942 bytes --]

We have separate I/D caches.  Data caches are physically indexed.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/cache.h      |   34 +++++++++
 arch/hexagon/include/asm/cacheflush.h |   99 ++++++++++++++++++++++++++
 arch/hexagon/mm/cache.c               |  128 ++++++++++++++++++++++++++++++++++
 3 files changed, 261 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/cacheflush.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/cacheflush.h	2011-09-08 19:22:01.300239513 -0500
@@ -0,0 +1,99 @@
+/*
+ * Cache flush operations for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_CACHEFLUSH_H
+#define _ASM_CACHEFLUSH_H
+
+#include <linux/cache.h>
+#include <linux/mm.h>
+#include <asm/string.h>
+#include <asm-generic/cacheflush.h>
+
+/* Cache flushing:
+ *
+ *  - flush_cache_all() flushes entire cache
+ *  - flush_cache_mm(mm) flushes the specified mm context's cache lines
+ *  - flush_cache_page(mm, vmaddr, pfn) flushes a single page
+ *  - flush_cache_range(vma, start, end) flushes a range of pages
+ *  - flush_icache_range(start, end) flush a range of instructions
+ *  - flush_dcache_page(pg) flushes(wback&invalidates) a page for dcache
+ *  - flush_icache_page(vma, pg) flushes(invalidates) a page for icache
+ *
+ *  Need to doublecheck which one is really needed for ptrace stuff to work.
+ */
+#define LINESIZE	32
+#define LINEBITS	5
+
+/*
+ * Flush Dcache range through current map.
+ */
+extern void flush_dcache_range(unsigned long start, unsigned long end);
+
+/*
+ * Flush Icache range through current map.
+ */
+#undef flush_icache_range
+extern void flush_icache_range(unsigned long start, unsigned long end);
+
+/*
+ * Memory-management related flushes are there to ensure in non-physically
+ * indexed cache schemes that stale lines belonging to a given ASID aren't
+ * in the cache to confuse things.  The prototype Hexagon Virtual Machine
+ * only uses a single ASID for all user-mode maps, which should
+ * mean that they aren't necessary.  A brute-force, flush-everything
+ * implementation, with the name xxxxx_hexagon() is present in
+ * arch/hexagon/mm/cache.c, but let's not wire it up until we know
+ * it is needed.
+ */
+extern void flush_cache_all_hexagon(void);
+
+/*
+ * This may or may not ever have to be non-null, depending on the
+ * virtual machine MMU.  For a native kernel, it's definitiely  a no-op
+ *
+ * This is also the place where deferred cache coherency stuff seems
+ * to happen, classically...  but instead we do it like ia64 and
+ * clean the cache when the PTE is set.
+ *
+ */
+static inline void update_mmu_cache(struct vm_area_struct *vma,
+					unsigned long address, pte_t *ptep)
+{
+	/*  generic_ptrace_pokedata doesn't wind up here, does it?  */
+}
+
+#undef copy_to_user_page
+static inline void copy_to_user_page(struct vm_area_struct *vma,
+					     struct page *page,
+					     unsigned long vaddr,
+					     void *dst, void *src, int len)
+{
+	memcpy(dst, src, len);
+	if (vma->vm_flags & VM_EXEC) {
+		flush_icache_range((unsigned long) dst,
+		(unsigned long) dst + len);
+	}
+}
+
+
+extern void hexagon_inv_dcache_range(unsigned long start, unsigned long end);
+extern void hexagon_clean_dcache_range(unsigned long start, unsigned long end);
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/cache.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/cache.h	2011-09-08 19:24:27.480690545 -0500
@@ -0,0 +1,34 @@
+/*
+ * Cache definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef __ASM_CACHE_H
+#define __ASM_CACHE_H
+
+/* Bytes per L1 cache line */
+#define L1_CACHE_SHIFT		(5)
+#define L1_CACHE_BYTES		(1 << L1_CACHE_SHIFT)
+
+#define __cacheline_aligned	__aligned(L1_CACHE_BYTES)
+#define ____cacheline_aligned	__aligned(L1_CACHE_BYTES)
+
+/* See http://kerneltrap.org/node/15100  */
+#define __read_mostly
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/mm/cache.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/cache.c	2011-09-08 19:22:01.300239513 -0500
@@ -0,0 +1,128 @@
+/*
+ * Cache management functions for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/mm.h>
+#include <asm/cacheflush.h>
+#include <asm/hexagon_vm.h>
+
+#define spanlines(start, end) \
+	(((end - (start & ~(LINESIZE - 1))) >> LINEBITS) + 1)
+
+void flush_dcache_range(unsigned long start, unsigned long end)
+{
+	unsigned long lines = spanlines(start, end-1);
+	unsigned long i, flags;
+
+	start &= ~(LINESIZE - 1);
+
+	local_irq_save(flags);
+
+	for (i = 0; i < lines; i++) {
+		__asm__ __volatile__ (
+		"	dccleaninva(%0);	"
+		:
+		: "r" (start)
+		);
+		start += LINESIZE;
+	}
+	local_irq_restore(flags);
+}
+
+void flush_icache_range(unsigned long start, unsigned long end)
+{
+	unsigned long lines = spanlines(start, end-1);
+	unsigned long i, flags;
+
+	start &= ~(LINESIZE - 1);
+
+	local_irq_save(flags);
+
+	for (i = 0; i < lines; i++) {
+		__asm__ __volatile__ (
+			"	dccleana(%0); "
+			"	icinva(%0);	"
+			:
+			: "r" (start)
+		);
+		start += LINESIZE;
+	}
+	__asm__ __volatile__ (
+		"isync"
+	);
+	local_irq_restore(flags);
+}
+
+void hexagon_clean_dcache_range(unsigned long start, unsigned long end)
+{
+	unsigned long lines = spanlines(start, end-1);
+	unsigned long i, flags;
+
+	start &= ~(LINESIZE - 1);
+
+	local_irq_save(flags);
+
+	for (i = 0; i < lines; i++) {
+		__asm__ __volatile__ (
+		"	dccleana(%0);	"
+		:
+		: "r" (start)
+		);
+		start += LINESIZE;
+	}
+	local_irq_restore(flags);
+}
+
+void hexagon_inv_dcache_range(unsigned long start, unsigned long end)
+{
+	unsigned long lines = spanlines(start, end-1);
+	unsigned long i, flags;
+
+	start &= ~(LINESIZE - 1);
+
+	local_irq_save(flags);
+
+	for (i = 0; i < lines; i++) {
+		__asm__ __volatile__ (
+		"	dcinva(%0);	"
+		:
+		: "r" (start)
+		);
+		start += LINESIZE;
+	}
+	local_irq_restore(flags);
+}
+
+
+
+
+/*
+ * This is just really brutal and shouldn't be used anyways,
+ * especially on V2.  Left here just in case.
+ */
+void flush_cache_all_hexagon(void)
+{
+	unsigned long flags;
+	local_irq_save(flags);
+	__vmcache_ickill();
+	__vmcache_dckill();
+	__vmcache_l2kill();
+	local_irq_restore(flags);
+	mb();
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 26/36] Hexagon: Implement basic TLB management routines for Hexagon.

[Richard Kuo]

[-- Attachment #1: tlb.diff --]
[-- Type: text/plain, Size: 7777 bytes --]

Mostly all stubs, as the TLB is managed by the hypervisor.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/tlb.h      |   39 +++++++++++++++
 arch/hexagon/include/asm/tlbflush.h |   58 ++++++++++++++++++++++
 arch/hexagon/mm/vm_tlb.c            |   93 ++++++++++++++++++++++++++++++++++++
 3 files changed, 190 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/tlbflush.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/tlbflush.h	2011-09-03 20:14:59.754981475 -0500
@@ -0,0 +1,58 @@
+/*
+ * TLB flush support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_TLBFLUSH_H
+#define _ASM_TLBFLUSH_H
+
+#include <linux/mm.h>
+#include <asm/processor.h>
+
+/*
+ * TLB flushing -- in "SMP", these routines get defined to be the
+ * ones from smp.c, else they are some local flavors.
+ */
+
+/*
+ * These functions are commonly macros, but in the interests of
+ * VM vs. native implementation and code size, we simply declare
+ * the function prototypes here.
+ */
+extern void tlb_flush_all(void);
+extern void flush_tlb_mm(struct mm_struct *mm);
+extern void flush_tlb_page(struct vm_area_struct *vma, unsigned long addr);
+extern void flush_tlb_range(struct vm_area_struct *vma,
+				unsigned long start, unsigned long end);
+extern void flush_tlb_kernel_range(unsigned long start, unsigned long end);
+extern void flush_tlb_one(unsigned long);
+
+/*
+ * "This is called in munmap when we have freed up some page-table pages.
+ * We don't need to do anything here..."
+ *
+ * The VM kernel doesn't walk page tables, and they are passed to the VMM
+ * by logical address. There doesn't seem to be any possibility that they
+ * could be referenced by the VM kernel based on a stale mapping, since
+ * they would only be located by consulting the mm structure, and they
+ * will have been purged from that structure by the munmap.  Seems like
+ * a noop on HVM as well.
+ */
+#define flush_tlb_pgtables(mm, start, end)
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/tlb.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/tlb.h	2011-09-03 20:14:59.754981475 -0500
@@ -0,0 +1,39 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_TLB_H
+#define _ASM_TLB_H
+
+#include <linux/pagemap.h>
+#include <asm/tlbflush.h>
+
+/*
+ * We don't need any special per-pte or per-vma handling...
+ */
+#define tlb_start_vma(tlb, vma)				do { } while (0)
+#define tlb_end_vma(tlb, vma)				do { } while (0)
+#define __tlb_remove_tlb_entry(tlb, ptep, address)	do { } while (0)
+
+/*
+ * .. because we flush the whole mm when it fills up
+ */
+#define tlb_flush(tlb)		flush_tlb_mm((tlb)->mm)
+
+#include <asm-generic/tlb.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/mm/vm_tlb.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/vm_tlb.c	2011-09-03 20:14:59.754981475 -0500
@@ -0,0 +1,93 @@
+/*
+ * Hexagon Virtual Machine TLB functions
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * The Hexagon Virtual Machine conceals the real workings of
+ * the TLB, but there are one or two functions that need to
+ * be instantiated for it, differently from a native build.
+ */
+#include <linux/mm.h>
+#include <asm/page.h>
+#include <asm/hexagon_vm.h>
+
+/*
+ * Initial VM implementation has only one map active at a time, with
+ * TLB purgings on changes.  So either we're nuking the current map,
+ * or it's a no-op.  This operation is messy on true SMPs where other
+ * processors must be induced to flush the copies in their local TLBs,
+ * but Hexagon thread-based virtual processors share the same MMU.
+ */
+void flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
+			unsigned long end)
+{
+	struct mm_struct *mm = vma->vm_mm;
+
+	if (mm->context.ptbase == current->active_mm->context.ptbase)
+		__vmclrmap((void *)start, end - start);
+}
+
+/*
+ * Flush a page from the kernel virtual map - used by highmem
+ */
+void flush_tlb_one(unsigned long vaddr)
+{
+	__vmclrmap((void *)vaddr, PAGE_SIZE);
+}
+
+/*
+ * Flush all TLBs across all CPUs, virtual or real.
+ * A single Hexagon core has 6 thread contexts but
+ * only one TLB.
+ */
+void tlb_flush_all(void)
+{
+	/*  should probably use that fixaddr end or whateve label  */
+	__vmclrmap(0, 0xffff0000);
+}
+
+/*
+ * Flush TLB entries associated with a given mm_struct mapping.
+ */
+void flush_tlb_mm(struct mm_struct *mm)
+{
+	/* Current Virtual Machine has only one map active at a time */
+	if (current->active_mm->context.ptbase == mm->context.ptbase)
+		tlb_flush_all();
+}
+
+/*
+ * Flush TLB state associated with a page of a vma.
+ */
+void flush_tlb_page(struct vm_area_struct *vma, unsigned long vaddr)
+{
+	struct mm_struct *mm = vma->vm_mm;
+
+	if (mm->context.ptbase  == current->active_mm->context.ptbase)
+		__vmclrmap((void *)vaddr, PAGE_SIZE);
+}
+
+/*
+ * Flush TLB entries associated with a kernel address range.
+ * Like flush range, but without the check on the vma->vm_mm.
+ */
+void flush_tlb_kernel_range(unsigned long start, unsigned long end)
+{
+		__vmclrmap((void *)start, end - start);
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 27/36] Hexagon: Provide DMA implementation

[Richard Kuo]

[-- Attachment #1: dma.diff --]
[-- Type: text/plain, Size: 11193 bytes --]

Basic DMA implementation for our architecture.

Added sync_single_for_* routines; hexagon_map_page() still seems to be a
convenient place to flush/invalidate.


Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/dma-mapping.h |  101 +++++++++++++++
 arch/hexagon/include/asm/dma.h         |   29 ++++
 arch/hexagon/kernel/dma.c              |  220 +++++++++++++++++++++++++++++++++
 3 files changed, 350 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/dma.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/dma.h	2011-09-03 20:14:59.764981503 -0500
@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_DMA_H
+#define _ASM_DMA_H
+
+#include <asm/io.h>
+
+#define MAX_DMA_CHANNELS 1
+#define MAX_DMA_ADDRESS  (PAGE_OFFSET)
+
+extern size_t hexagon_coherent_pool_size;
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/dma-mapping.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/dma-mapping.h	2011-09-03 20:14:59.764981503 -0500
@@ -0,0 +1,101 @@
+/*
+ * DMA operations for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_DMA_MAPPING_H
+#define _ASM_DMA_MAPPING_H
+
+#include <linux/types.h>
+#include <linux/cache.h>
+#include <linux/mm.h>
+#include <linux/scatterlist.h>
+#include <linux/dma-mapping.h>
+#include <linux/dma-debug.h>
+#include <linux/dma-attrs.h>
+#include <asm/io.h>
+
+struct device;
+extern int bad_dma_address;
+
+extern struct dma_map_ops *dma_ops;
+
+#define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f)
+#define dma_free_noncoherent(d, s, v, h) dma_free_coherent(d, s, v, h)
+
+static inline struct dma_map_ops *get_dma_ops(struct device *dev)
+{
+	if (unlikely(dev == NULL))
+		return NULL;
+
+	return dma_ops;
+}
+
+extern int dma_supported(struct device *dev, u64 mask);
+extern int dma_set_mask(struct device *dev, u64 mask);
+extern int dma_is_consistent(struct device *dev, dma_addr_t dma_handle);
+extern void dma_cache_sync(struct device *dev, void *vaddr, size_t size,
+			   enum dma_data_direction direction);
+
+#include <asm-generic/dma-mapping-common.h>
+
+static inline bool dma_capable(struct device *dev, dma_addr_t addr, size_t size)
+{
+	if (!dev->dma_mask)
+		return 0;
+	return addr + size - 1 <= *dev->dma_mask;
+}
+
+static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
+{
+	struct dma_map_ops *dma_ops = get_dma_ops(dev);
+
+	if (dma_ops->mapping_error)
+		return dma_ops->mapping_error(dev, dma_addr);
+
+	return (dma_addr == bad_dma_address);
+}
+
+static inline void *dma_alloc_coherent(struct device *dev, size_t size,
+				       dma_addr_t *dma_handle, gfp_t flag)
+{
+	void *ret;
+	struct dma_map_ops *ops = get_dma_ops(dev);
+
+	BUG_ON(!dma_ops);
+
+	ret = ops->alloc_coherent(dev, size, dma_handle, flag);
+
+	debug_dma_alloc_coherent(dev, size, *dma_handle, ret);
+
+	return ret;
+}
+
+static inline void dma_free_coherent(struct device *dev, size_t size,
+				     void *cpu_addr, dma_addr_t dma_handle)
+{
+	struct dma_map_ops *dma_ops = get_dma_ops(dev);
+
+	BUG_ON(!dma_ops);
+
+	dma_ops->free_coherent(dev, size, cpu_addr, dma_handle);
+
+	debug_dma_free_coherent(dev, size, cpu_addr, dma_handle);
+}
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/kernel/dma.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/dma.c	2011-09-07 13:00:31.040646956 -0500
@@ -0,0 +1,220 @@
+/*
+ * DMA implementation for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/dma-mapping.h>
+#include <linux/bootmem.h>
+#include <linux/genalloc.h>
+#include <asm/dma-mapping.h>
+
+struct dma_map_ops *dma_ops;
+EXPORT_SYMBOL(dma_ops);
+
+int bad_dma_address;  /*  globals are automatically initialized to zero  */
+
+int dma_supported(struct device *dev, u64 mask)
+{
+	if (mask == DMA_BIT_MASK(32))
+		return 1;
+	else
+		return 0;
+}
+EXPORT_SYMBOL(dma_supported);
+
+int dma_set_mask(struct device *dev, u64 mask)
+{
+	if (!dev->dma_mask || !dma_supported(dev, mask))
+		return -EIO;
+
+	*dev->dma_mask = mask;
+
+	return 0;
+}
+EXPORT_SYMBOL(dma_set_mask);
+
+static struct gen_pool *coherent_pool;
+
+
+/* Allocates from a pool of uncached memory that was reserved at boot time */
+
+void *hexagon_dma_alloc_coherent(struct device *dev, size_t size,
+				 dma_addr_t *dma_addr, gfp_t flag)
+{
+	void *ret;
+
+	if (coherent_pool == NULL) {
+		coherent_pool = gen_pool_create(PAGE_SHIFT, -1);
+
+		if (coherent_pool == NULL)
+			panic("Can't create %s() memory pool!", __func__);
+		else
+			gen_pool_add(coherent_pool,
+				(PAGE_OFFSET + (max_low_pfn << PAGE_SHIFT)),
+				hexagon_coherent_pool_size, -1);
+	}
+
+	ret = (void *) gen_pool_alloc(coherent_pool, size);
+
+	if (ret) {
+		memset(ret, 0, size);
+		*dma_addr = (dma_addr_t) (ret - PAGE_OFFSET);
+	} else
+		*dma_addr = ~0;
+
+	return ret;
+}
+
+static void hexagon_free_coherent(struct device *dev, size_t size, void *vaddr,
+				  dma_addr_t dma_addr)
+{
+	gen_pool_free(coherent_pool, (unsigned long) vaddr, size);
+}
+
+static int check_addr(const char *name, struct device *hwdev,
+		      dma_addr_t bus, size_t size)
+{
+	if (hwdev && hwdev->dma_mask && !dma_capable(hwdev, bus, size)) {
+		if (*hwdev->dma_mask >= DMA_BIT_MASK(32))
+			printk(KERN_ERR
+				"%s: overflow %Lx+%zu of device mask %Lx\n",
+				name, (long long)bus, size,
+				(long long)*hwdev->dma_mask);
+		return 0;
+	}
+	return 1;
+}
+
+static int hexagon_map_sg(struct device *hwdev, struct scatterlist *sg,
+			  int nents, enum dma_data_direction dir,
+			  struct dma_attrs *attrs)
+{
+	struct scatterlist *s;
+	int i;
+
+	WARN_ON(nents == 0 || sg[0].length == 0);
+
+	for_each_sg(sg, s, nents, i) {
+		s->dma_address = sg_phys(s);
+		if (!check_addr("map_sg", hwdev, s->dma_address, s->length))
+			return 0;
+
+		s->dma_length = s->length;
+
+		flush_dcache_range(PAGE_OFFSET + s->dma_address,
+				   PAGE_OFFSET + s->dma_address + s->length);
+	}
+
+	return nents;
+}
+
+/*
+ * address is virtual
+ */
+static inline void dma_sync(void *addr, size_t size,
+			    enum dma_data_direction dir)
+{
+	switch (dir) {
+	case DMA_TO_DEVICE:
+		hexagon_clean_dcache_range((unsigned long) addr,
+		(unsigned long) addr + size);
+		break;
+	case DMA_FROM_DEVICE:
+		hexagon_inv_dcache_range((unsigned long) addr,
+		(unsigned long) addr + size);
+		break;
+	case DMA_BIDIRECTIONAL:
+		flush_dcache_range((unsigned long) addr,
+		(unsigned long) addr + size);
+		break;
+	default:
+		BUG();
+	}
+}
+
+static inline void *dma_addr_to_virt(dma_addr_t dma_addr)
+{
+	return phys_to_virt((unsigned long) dma_addr);
+}
+
+/**
+ * hexagon_map_page() - maps an address for device DMA
+ * @dev:	pointer to DMA device
+ * @page:	pointer to page struct of DMA memory
+ * @offset:	offset within page
+ * @size:	size of memory to map
+ * @dir:	transfer direction
+ * @attrs:	pointer to DMA attrs (not used)
+ *
+ * Called to map a memory address to a DMA address prior
+ * to accesses to/from device.
+ *
+ * We don't particularly have many hoops to jump through
+ * so far.  Straight translation between phys and virtual.
+ *
+ * DMA is not cache coherent so sync is necessary; this
+ * seems to be a convenient place to do it.
+ *
+ */
+static dma_addr_t hexagon_map_page(struct device *dev, struct page *page,
+				   unsigned long offset, size_t size,
+				   enum dma_data_direction dir,
+				   struct dma_attrs *attrs)
+{
+	dma_addr_t bus = page_to_phys(page) + offset;
+	WARN_ON(size == 0);
+
+	if (!check_addr("map_single", dev, bus, size))
+		return bad_dma_address;
+
+	dma_sync(dma_addr_to_virt(bus), size, dir);
+
+	return bus;
+}
+
+static void hexagon_sync_single_for_cpu(struct device *dev,
+					dma_addr_t dma_handle, size_t size,
+					enum dma_data_direction dir)
+{
+	dma_sync(dma_addr_to_virt(dma_handle), size, dir);
+}
+
+static void hexagon_sync_single_for_device(struct device *dev,
+					dma_addr_t dma_handle, size_t size,
+					enum dma_data_direction dir)
+{
+	dma_sync(dma_addr_to_virt(dma_handle), size, dir);
+}
+
+struct dma_map_ops hexagon_dma_ops = {
+	.alloc_coherent	= hexagon_dma_alloc_coherent,
+	.free_coherent	= hexagon_free_coherent,
+	.map_sg		= hexagon_map_sg,
+	.map_page	= hexagon_map_page,
+	.sync_single_for_cpu = hexagon_sync_single_for_cpu,
+	.sync_single_for_device = hexagon_sync_single_for_device,
+	.is_phys	= 1,
+};
+
+void __init hexagon_dma_init(void)
+{
+	if (dma_ops)
+		return;
+
+	dma_ops = &hexagon_dma_ops;
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 28/36] Hexagon: Add ioremap support

[Richard Kuo]

[-- Attachment #1: ioremap.diff --]
[-- Type: text/plain, Size: 2354 bytes --]

Removed redundant flush_cache_vmap.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

---
 arch/hexagon/mm/ioremap.c |   56 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 56 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/mm/ioremap.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/ioremap.c	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,56 @@
+/*
+ * I/O remap functions for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/io.h>
+#include <linux/vmalloc.h>
+
+void __iomem *ioremap_nocache(unsigned long phys_addr, unsigned long size)
+{
+	unsigned long last_addr, addr;
+	unsigned long offset = phys_addr & ~PAGE_MASK;
+	struct vm_struct *area;
+
+	pgprot_t prot = __pgprot(_PAGE_PRESENT|_PAGE_READ|_PAGE_WRITE
+					|(__HEXAGON_C_DEV << 6));
+
+	last_addr = phys_addr + size - 1;
+
+	/*  Wrapping not allowed  */
+	if (!size || (last_addr < phys_addr))
+		return NULL;
+
+	/*  Rounds up to next page size, including whole-page offset */
+	size = PAGE_ALIGN(offset + size);
+
+	area = get_vm_area(size, VM_IOREMAP);
+	addr = (unsigned long)area->addr;
+
+	if (ioremap_page_range(addr, addr+size, phys_addr, prot)) {
+		vunmap((void *)addr);
+		return NULL;
+	}
+
+	return (void __iomem *) (offset + addr);
+}
+
+void __iounmap(const volatile void __iomem *addr)
+{
+	vunmap((void *) ((unsigned long) addr & PAGE_MASK));
+}

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 28/36] Hexagon: Add ioremap support

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:15 Richard Kuo wrote:
> Removed redundant flush_cache_vmap.
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

Acked-by: Arnd Bergmann <arnd@arndb.de>

[patch v3 29/36] Hexagon: Add page table header files & etc.

[Richard Kuo]

[-- Attachment #1: pgtables-highmem.diff --]
[-- Type: text/plain, Size: 35006 bytes --]

Removed highmem more highmem stuff.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>

---
 arch/hexagon/include/asm/fixmap.h      |   73 ++++
 arch/hexagon/include/asm/mmu.h         |   37 ++
 arch/hexagon/include/asm/mmu_context.h |  100 ++++++
 arch/hexagon/include/asm/page.h        |  157 ++++++++++
 arch/hexagon/include/asm/pgalloc.h     |  146 +++++++++
 arch/hexagon/include/asm/pgtable.h     |  518 +++++++++++++++++++++++++++++++++
 arch/hexagon/mm/pgalloc.c              |   25 +
 7 files changed, 1056 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/fixmap.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/fixmap.h	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,73 @@
+/*
+ * Fixmap support for Hexagon - enough to support highmem features
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_FIXMAP_H
+#define _ASM_FIXMAP_H
+
+/*
+ * A lot of the fixmap info is already in mem-layout.h
+ */
+#include <asm/mem-layout.h>
+
+/*
+ * Full fixmap support involves set_fixmap() functions, but
+ * these may not be needed if all we're after is an area for
+ * highmem kernel mappings.
+ */
+#define	__fix_to_virt(x)	(FIXADDR_TOP - ((x) << PAGE_SHIFT))
+#define	__virt_to_fix(x)	((FIXADDR_TOP - ((x)&PAGE_MASK)) >> PAGE_SHIFT)
+
+extern void __this_fixmap_does_not_exist(void);
+
+/**
+ * fix_to_virt -- "index to address" translation.
+ *
+ * If anyone tries to use the idx directly without translation,
+ * we catch the bug with a NULL-deference kernel oops. Illegal
+ * ranges of incoming indices are caught too.
+ */
+static inline unsigned long fix_to_virt(const unsigned int idx)
+{
+	/*
+	 * This branch gets completely eliminated after inlining,
+	 * except when someone tries to use fixaddr indices in an
+	 * illegal way. (such as mixing up address types or using
+	 * out-of-range indices).
+	 *
+	 * If it doesn't get removed, the linker will complain
+	 * loudly with a reasonably clear error message..
+	 */
+	if (idx >= __end_of_fixed_addresses)
+		__this_fixmap_does_not_exist();
+
+	return __fix_to_virt(idx);
+}
+
+static inline unsigned long virt_to_fix(const unsigned long vaddr)
+{
+	BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START);
+	return __virt_to_fix(vaddr);
+}
+
+#define kmap_get_fixmap_pte(vaddr) \
+	pte_offset_kernel(pmd_offset(pud_offset(pgd_offset_k(vaddr), \
+				(vaddr)), (vaddr)), (vaddr))
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/mmu_context.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/mmu_context.h	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,100 @@
+/*
+ * MM context support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_MMU_CONTEXT_H
+#define _ASM_MMU_CONTEXT_H
+
+#include <asm/setup.h>
+#include <asm/page.h>
+#include <asm/pgalloc.h>
+#include <asm/mem-layout.h>
+
+static inline void destroy_context(struct mm_struct *mm)
+{
+}
+
+/*
+ * VM port hides all TLB management, so "lazy TLB" isn't very
+ * meaningful.  Even for ports to architectures with visble TLBs,
+ * this is almost invariably a null function.
+ */
+static inline void enter_lazy_tlb(struct mm_struct *mm,
+	struct task_struct *tsk)
+{
+}
+
+/*
+ * Architecture-specific actions, if any, for memory map deactivation.
+ */
+static inline void deactivate_mm(struct task_struct *tsk,
+	struct mm_struct *mm)
+{
+}
+
+/**
+ * init_new_context - initialize context related info for new mm_struct instance
+ * @tsk: pointer to a task struct
+ * @mm: pointer to a new mm struct
+ */
+static inline int init_new_context(struct task_struct *tsk,
+					struct mm_struct *mm)
+{
+	/* mm->context is set up by pgd_alloc */
+	return 0;
+}
+
+/*
+ *  Switch active mm context
+ */
+static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next,
+				struct task_struct *tsk)
+{
+	int l1;
+
+	/*
+	 * For virtual machine, we have to update system map if it's been
+	 * touched.
+	 */
+	if (next->context.generation < prev->context.generation) {
+		for (l1 = MIN_KERNEL_SEG; l1 <= max_kernel_seg; l1++)
+			next->pgd[l1] = init_mm.pgd[l1];
+
+		next->context.generation = prev->context.generation;
+	}
+
+	__vmnewmap((void *)next->context.ptbase);
+}
+
+/*
+ *  Activate new memory map for task
+ */
+static inline void activate_mm(struct mm_struct *prev, struct mm_struct *next)
+{
+	unsigned long flags;
+
+	local_irq_save(flags);
+	switch_mm(prev, next, current_thread_info()->task);
+	local_irq_restore(flags);
+}
+
+/*  Generic hooks for arch_dup_mmap and arch_exit_mmap  */
+#include <asm-generic/mm_hooks.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/mmu.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/mmu.h	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,37 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_MMU_H
+#define _ASM_MMU_H
+
+#include <asm/vdso.h>
+
+/*
+ * Architecture-specific state for a mm_struct.
+ * For the Hexagon Virtual Machine, it can be a copy
+ * of the pointer to the page table base.
+ */
+struct mm_context {
+	unsigned long long generation;
+	unsigned long ptbase;
+	struct hexagon_vdso *vdso;
+};
+
+typedef struct mm_context mm_context_t;
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/page.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/page.h	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,157 @@
+/*
+ * Page management definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_PAGE_H
+#define _ASM_PAGE_H
+
+#include <linux/const.h>
+
+/*  This is probably not the most graceful way to handle this.  */
+
+#ifdef CONFIG_PAGE_SIZE_4KB
+#define PAGE_SHIFT 12
+#define HEXAGON_L1_PTE_SIZE __HVM_PDE_S_4KB
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_16KB
+#define PAGE_SHIFT 14
+#define HEXAGON_L1_PTE_SIZE __HVM_PDE_S_16KB
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_64KB
+#define PAGE_SHIFT 16
+#define HEXAGON_L1_PTE_SIZE __HVM_PDE_S_64KB
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_256KB
+#define PAGE_SHIFT 18
+#define HEXAGON_L1_PTE_SIZE __HVM_PDE_S_256KB
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_1MB
+#define PAGE_SHIFT 20
+#define HEXAGON_L1_PTE_SIZE __HVM_PDE_S_1MB
+#endif
+
+/*
+ *  These should be defined in hugetlb.h, but apparently not.
+ *  "Huge" for us should be 4MB or 16MB, which are both represented
+ *  in L1 PTE's.  Right now, it's set up for 4MB.
+ */
+#ifdef CONFIG_HUGETLB_PAGE
+#define HPAGE_SHIFT 22
+#define HPAGE_SIZE (1UL << HPAGE_SHIFT)
+#define HPAGE_MASK (~(HPAGE_SIZE-1))
+#define HUGETLB_PAGE_ORDER (HPAGE_SHIFT-PAGE_SHIFT)
+#define HVM_HUGEPAGE_SIZE 0x5
+#endif
+
+#define PAGE_SIZE  (1UL << PAGE_SHIFT)
+#define PAGE_MASK  (~((1 << PAGE_SHIFT) - 1))
+
+#ifdef __KERNEL__
+#ifndef __ASSEMBLY__
+
+/*
+ * This is for PFN_DOWN, which mm.h needs.  Seems the right place to pull it in.
+ */
+#include <linux/pfn.h>
+
+/*
+ * We implement a two-level architecture-specific page table structure.
+ * Null intermediate page table level (pmd, pud) definitions will come from
+ * asm-generic/pagetable-nopmd.h and asm-generic/pagetable-nopud.h
+ */
+typedef struct { unsigned long pte; } pte_t;
+typedef struct { unsigned long pgd; } pgd_t;
+typedef struct { unsigned long pgprot; } pgprot_t;
+typedef struct page *pgtable_t;
+
+#define pte_val(x)     ((x).pte)
+#define pgd_val(x)     ((x).pgd)
+#define pgprot_val(x)  ((x).pgprot)
+#define __pte(x)       ((pte_t) { (x) })
+#define __pgd(x)       ((pgd_t) { (x) })
+#define __pgprot(x)    ((pgprot_t) { (x) })
+
+/*
+ * We need a __pa and a __va routine for kernel space.
+ * MIPS says they're only used during mem_init.
+ * also, check if we need a PHYS_OFFSET.
+ */
+#define __pa(x) ((unsigned long)(x) - PAGE_OFFSET)
+#define __va(x) ((void *)((unsigned long)(x) + PAGE_OFFSET))
+
+/* The "page frame" descriptor is defined in linux/mm.h */
+struct page;
+
+/* Returns page frame descriptor for virtual address. */
+#define virt_to_page(kaddr) pfn_to_page(PFN_DOWN(__pa(kaddr)))
+
+/* Default vm area behavior is non-executable.  */
+#define VM_DATA_DEFAULT_FLAGS (VM_READ | VM_WRITE | \
+				VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
+
+#define pfn_valid(pfn) ((pfn) < max_mapnr)
+#define virt_addr_valid(kaddr) pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
+
+/*  Need to not use a define for linesize; may move this to another file.  */
+static inline void clear_page(void *page)
+{
+	/*  This can only be done on pages with L1 WB cache */
+	asm volatile(
+		"	loop0(1f,%1);\n"
+		"1:	{ dczeroa(%0);\n"
+		"	  %0 = add(%0,#32); }:endloop0\n"
+		: "+r" (page)
+		: "r" (PAGE_SIZE/32)
+		: "lc0", "sa0", "memory"
+	);
+}
+
+#define copy_page(to, from)	memcpy((to), (from), PAGE_SIZE)
+
+/*
+ * Under assumption that kernel always "sees" user map...
+ */
+#define clear_user_page(page, vaddr, pg)	clear_page(page)
+#define copy_user_page(to, from, vaddr, pg)	copy_page(to, from)
+
+/*
+ * page_to_phys - convert page to physical address
+ * @page - pointer to page entry in mem_map
+ */
+#define page_to_phys(page)      (page_to_pfn(page) << PAGE_SHIFT)
+
+/*
+ * For port to Hexagon Virtual Machine, MAYBE we check for attempts
+ * to reference reserved HVM space, but in any case, the VM will be
+ * protected.
+ */
+#define kern_addr_valid(addr)   (1)
+
+#include <asm-generic/memory_model.h>
+/* XXX Todo: implement assembly-optimized version of getorder. */
+#include <asm-generic/getorder.h>
+
+#endif /* ifdef __ASSEMBLY__ */
+#endif /* ifdef __KERNEL__ */
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/pgalloc.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/pgalloc.h	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,146 @@
+/*
+ * Page table support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_PGALLOC_H
+#define _ASM_PGALLOC_H
+
+#include <asm/mem-layout.h>
+#include <asm/atomic.h>
+
+#define check_pgt_cache() do {} while (0)
+
+extern unsigned long long kmap_generation;
+
+/*
+ * Page table creation interface
+ */
+static inline pgd_t *pgd_alloc(struct mm_struct *mm)
+{
+	pgd_t *pgd;
+
+	pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
+
+	/*
+	 * There may be better ways to do this, but to ensure
+	 * that new address spaces always contain the kernel
+	 * base mapping, and to ensure that the user area is
+	 * initially marked invalid, initialize the new map
+	 * map with a copy of the kernel's persistent map.
+	 */
+
+	memcpy(pgd, swapper_pg_dir, PTRS_PER_PGD*sizeof(pgd_t *));
+	mm->context.generation = kmap_generation;
+
+	/* Physical version is what is passed to virtual machine on switch */
+	mm->context.ptbase = __pa(pgd);
+
+	return pgd;
+}
+
+static inline void pgd_free(struct mm_struct *mm, pgd_t *pgd)
+{
+	free_page((unsigned long) pgd);
+}
+
+static inline struct page *pte_alloc_one(struct mm_struct *mm,
+					 unsigned long address)
+{
+	struct page *pte;
+
+	pte = alloc_page(GFP_KERNEL | __GFP_REPEAT | __GFP_ZERO);
+
+	if (pte)
+		pgtable_page_ctor(pte);
+
+	return pte;
+}
+
+/* _kernel variant gets to use a different allocator */
+static inline pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
+					  unsigned long address)
+{
+	gfp_t flags =  GFP_KERNEL | __GFP_REPEAT | __GFP_ZERO;
+	return (pte_t *) __get_free_page(flags);
+}
+
+static inline void pte_free(struct mm_struct *mm, struct page *pte)
+{
+	pgtable_page_dtor(pte);
+	__free_page(pte);
+}
+
+static inline void pte_free_kernel(struct mm_struct *mm, pte_t *pte)
+{
+	free_page((unsigned long)pte);
+}
+
+static inline void pmd_populate(struct mm_struct *mm, pmd_t *pmd,
+				pgtable_t pte)
+{
+	/*
+	 * Conveniently, zero in 3 LSB means indirect 4K page table.
+	 * Not so convenient when you're trying to vary the page size.
+	 */
+	set_pmd(pmd, __pmd(((unsigned long)page_to_pfn(pte) << PAGE_SHIFT) |
+		HEXAGON_L1_PTE_SIZE));
+}
+
+/*
+ * Other architectures seem to have ways of making all processes
+ * share the same pmd's for their kernel mappings, but the v0.3
+ * Hexagon VM spec has a "monolithic" L1 table for user and kernel
+ * segments.  We track "generations" of the kernel map to minimize
+ * overhead, and update the "slave" copies of the kernel mappings
+ * as part of switch_mm.  However, we still need to update the
+ * kernel map of the active thread who's calling pmd_populate_kernel...
+ */
+static inline void pmd_populate_kernel(struct mm_struct *mm, pmd_t *pmd,
+				       pte_t *pte)
+{
+	extern spinlock_t kmap_gen_lock;
+	pmd_t *ppmd;
+	int pmdindex;
+
+	spin_lock(&kmap_gen_lock);
+	kmap_generation++;
+	mm->context.generation = kmap_generation;
+	current->active_mm->context.generation = kmap_generation;
+	spin_unlock(&kmap_gen_lock);
+
+	set_pmd(pmd, __pmd(((unsigned long)__pa(pte)) | HEXAGON_L1_PTE_SIZE));
+
+	/*
+	 * Now the "slave" copy of the current thread.
+	 * This is pointer arithmetic, not byte addresses!
+	 */
+	pmdindex = (pgd_t *)pmd - mm->pgd;
+	ppmd = (pmd_t *)current->active_mm->pgd + pmdindex;
+	set_pmd(ppmd, __pmd(((unsigned long)__pa(pte)) | HEXAGON_L1_PTE_SIZE));
+	if (pmdindex > max_kernel_seg)
+		max_kernel_seg = pmdindex;
+}
+
+#define __pte_free_tlb(tlb, pte, addr)		\
+do {						\
+	pgtable_page_dtor((pte));		\
+	tlb_remove_page((tlb), (pte));		\
+} while (0)
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/include/asm/pgtable.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/pgtable.h	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,518 @@
+/*
+ * Page table support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_PGTABLE_H
+#define _ASM_PGTABLE_H
+
+/*
+ * Page table definitions for Qualcomm Hexagon processor.
+ */
+#include <linux/swap.h>
+#include <asm/page.h>
+#include <asm-generic/pgtable-nopmd.h>
+
+/* A handy thing to have if one has the RAM. Declared in head.S */
+extern unsigned long empty_zero_page;
+extern unsigned long zero_page_mask;
+
+/*
+ * The PTE model described here is that of the Hexagon Virtual Machine,
+ * which autonomously walks 2-level page tables.  At a lower level, we
+ * also describe the RISCish software-loaded TLB entry structure of
+ * the underlying Hexagon processor. A kernel built to run on the
+ * virtual machine has no need to know about the underlying hardware.
+ */
+#include <asm/vm_mmu.h>
+
+/*
+ * To maximize the comfort level for the PTE manipulation macros,
+ * define the "well known" architecture-specific bits.
+ */
+#define _PAGE_READ	__HVM_PTE_R
+#define _PAGE_WRITE	__HVM_PTE_W
+#define _PAGE_EXECUTE	__HVM_PTE_X
+#define _PAGE_USER	__HVM_PTE_U
+
+/*
+ * We have a total of 4 "soft" bits available in the abstract PTE.
+ * The two mandatory software bits are Dirty and Accessed.
+ * To make nonlinear swap work according to the more recent
+ * model, we want a low order "Present" bit to indicate whether
+ * the PTE describes MMU programming or swap space.
+ */
+#define _PAGE_PRESENT	(1<<0)
+#define _PAGE_DIRTY	(1<<1)
+#define _PAGE_ACCESSED	(1<<2)
+
+/*
+ * _PAGE_FILE is only meaningful if _PAGE_PRESENT is false, while
+ * _PAGE_DIRTY is only meaningful if _PAGE_PRESENT is true.
+ * So we can overload the bit...
+ */
+#define _PAGE_FILE	_PAGE_DIRTY /* set:  pagecache, unset = swap */
+
+/*
+ * For now, let's say that Valid and Present are the same thing.
+ * Alternatively, we could say that it's the "or" of R, W, and X
+ * permissions.
+ */
+#define _PAGE_VALID	_PAGE_PRESENT
+
+/*
+ * We're not defining _PAGE_GLOBAL here, since there's no concept
+ * of global pages or ASIDs exposed to the Hexagon Virtual Machine,
+ * and we want to use the same page table structures and macros in
+ * the native kernel as we do in the virtual machine kernel.
+ * So we'll put up with a bit of inefficiency for now...
+ */
+
+/*
+ * Top "FOURTH" level (pgd), which for the Hexagon VM is really
+ * only the second from the bottom, pgd and pud both being collapsed.
+ * Each entry represents 4MB of virtual address space, 4K of table
+ * thus maps the full 4GB.
+ */
+#define PGDIR_SHIFT 22
+#define PTRS_PER_PGD 1024
+
+#define PGDIR_SIZE (1UL << PGDIR_SHIFT)
+#define PGDIR_MASK (~(PGDIR_SIZE-1))
+
+#ifdef CONFIG_PAGE_SIZE_4KB
+#define PTRS_PER_PTE 1024
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_16KB
+#define PTRS_PER_PTE 256
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_64KB
+#define PTRS_PER_PTE 64
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_256KB
+#define PTRS_PER_PTE 16
+#endif
+
+#ifdef CONFIG_PAGE_SIZE_1MB
+#define PTRS_PER_PTE 4
+#endif
+
+/*  Any bigger and the PTE disappears.  */
+#define pgd_ERROR(e) \
+	printk(KERN_ERR "%s:%d: bad pgd %08lx.\n", __FILE__, __LINE__,\
+		pgd_val(e))
+
+/*
+ * Page Protection Constants. Includes (in this variant) cache attributes.
+ */
+extern unsigned long _dflt_cache_att;
+
+#define PAGE_NONE	__pgprot(_PAGE_PRESENT | _PAGE_USER | \
+				_dflt_cache_att)
+#define PAGE_READONLY	__pgprot(_PAGE_PRESENT | _PAGE_USER | \
+				_PAGE_READ | _PAGE_EXECUTE | _dflt_cache_att)
+#define PAGE_COPY	PAGE_READONLY
+#define PAGE_EXEC	__pgprot(_PAGE_PRESENT | _PAGE_USER | \
+				_PAGE_READ | _PAGE_EXECUTE | _dflt_cache_att)
+#define PAGE_COPY_EXEC	PAGE_EXEC
+#define PAGE_SHARED	__pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_READ | \
+				_PAGE_EXECUTE | _PAGE_WRITE | _dflt_cache_att)
+#define PAGE_KERNEL	__pgprot(_PAGE_PRESENT | _PAGE_READ | \
+				_PAGE_WRITE | _PAGE_EXECUTE | _dflt_cache_att)
+
+
+/*
+ * Aliases for mapping mmap() protection bits to page protections.
+ * These get used for static initialization, so using the _dflt_cache_att
+ * variable for the default cache attribute isn't workable. If the
+ * default gets changed at boot time, the boot option code has to
+ * update data structures like the protaction_map[] array.
+ */
+#define CACHEDEF	(CACHE_DEFAULT << 6)
+
+/* Private (copy-on-write) page protections. */
+#define __P000 __pgprot(_PAGE_PRESENT | _PAGE_USER | CACHEDEF)
+#define __P001 __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_READ | CACHEDEF)
+#define __P010 __P000	/* Write-only copy-on-write */
+#define __P011 __P001	/* Read/Write copy-on-write */
+#define __P100 __pgprot(_PAGE_PRESENT | _PAGE_USER | \
+			_PAGE_EXECUTE | CACHEDEF)
+#define __P101 __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_EXECUTE | \
+			_PAGE_READ | CACHEDEF)
+#define __P110 __P100	/* Write/execute copy-on-write */
+#define __P111 __P101	/* Read/Write/Execute, copy-on-write */
+
+/* Shared page protections. */
+#define __S000 __P000
+#define __S001 __P001
+#define __S010 __pgprot(_PAGE_PRESENT | _PAGE_USER | \
+			_PAGE_WRITE | CACHEDEF)
+#define __S011 __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_READ | \
+			_PAGE_WRITE | CACHEDEF)
+#define __S100 __pgprot(_PAGE_PRESENT | _PAGE_USER | \
+			_PAGE_EXECUTE | CACHEDEF)
+#define __S101 __P101
+#define __S110 __pgprot(_PAGE_PRESENT | _PAGE_USER | \
+			_PAGE_EXECUTE | _PAGE_WRITE | CACHEDEF)
+#define __S111 __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_READ | \
+			_PAGE_EXECUTE | _PAGE_WRITE | CACHEDEF)
+
+extern pgd_t swapper_pg_dir[PTRS_PER_PGD];  /* located in head.S */
+
+/* Seems to be zero even in architectures where the zero page is firewalled? */
+#define FIRST_USER_ADDRESS 0
+#define pte_special(pte)	0
+#define pte_mkspecial(pte)	(pte)
+
+/*  HUGETLB not working currently  */
+#ifdef CONFIG_HUGETLB_PAGE
+#define pte_mkhuge(pte) __pte((pte_val(pte) & ~0x3) | HVM_HUGEPAGE_SIZE)
+#endif
+
+/*
+ * For now, assume that higher-level code will do TLB/MMU invalidations
+ * and don't insert that overhead into this low-level function.
+ */
+extern void sync_icache_dcache(pte_t pte);
+
+#define pte_present_exec_user(pte) \
+	((pte_val(pte) & (_PAGE_EXECUTE | _PAGE_USER)) == \
+	(_PAGE_EXECUTE | _PAGE_USER))
+
+static inline void set_pte(pte_t *ptep, pte_t pteval)
+{
+	/*  should really be using pte_exec, if it weren't declared later. */
+	if (pte_present_exec_user(pteval))
+		sync_icache_dcache(pteval);
+
+	*ptep = pteval;
+}
+
+/*
+ * For the Hexagon Virtual Machine MMU (or its emulation), a null/invalid
+ * L1 PTE (PMD/PGD) has 7 in the least significant bits. For the L2 PTE
+ * (Linux PTE), the key is to have bits 11..9 all zero.  We'd use 0x7
+ * as a universal null entry, but some of those least significant bits
+ * are interpreted by software.
+ */
+#define _NULL_PMD	0x7
+#define _NULL_PTE	0x0
+
+static inline void pmd_clear(pmd_t *pmd_entry_ptr)
+{
+	 pmd_val(*pmd_entry_ptr) = _NULL_PMD;
+}
+
+/*
+ * Conveniently, a null PTE value is invalid.
+ */
+static inline void pte_clear(struct mm_struct *mm, unsigned long addr,
+				pte_t *ptep)
+{
+	pte_val(*ptep) = _NULL_PTE;
+}
+
+#ifdef NEED_PMD_INDEX_DESPITE_BEING_2_LEVEL
+/**
+ * pmd_index - returns the index of the entry in the PMD page
+ * which would control the given virtual address
+ */
+#define pmd_index(address) (((address) >> PMD_SHIFT) & (PTRS_PER_PMD-1))
+
+#endif
+
+/**
+ * pgd_index - returns the index of the entry in the PGD page
+ * which would control the given virtual address
+ *
+ * This returns the *index* for the address in the pgd_t
+ */
+#define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD-1))
+
+/*
+ * pgd_offset - find an offset in a page-table-directory
+ */
+#define pgd_offset(mm, addr) ((mm)->pgd + pgd_index(addr))
+
+/*
+ * pgd_offset_k - get kernel (init_mm) pgd entry pointer for addr
+ */
+#define pgd_offset_k(address) pgd_offset(&init_mm, address)
+
+/**
+ * pmd_none - check if pmd_entry is mapped
+ * @pmd_entry:  pmd entry
+ *
+ * MIPS checks it against that "invalid pte table" thing.
+ */
+static inline int pmd_none(pmd_t pmd)
+{
+	return pmd_val(pmd) == _NULL_PMD;
+}
+
+/**
+ * pmd_present - is there a page table behind this?
+ * Essentially the inverse of pmd_none.  We maybe
+ * save an inline instruction by defining it this
+ * way, instead of simply "!pmd_none".
+ */
+static inline int pmd_present(pmd_t pmd)
+{
+	return pmd_val(pmd) != (unsigned long)_NULL_PMD;
+}
+
+/**
+ * pmd_bad - check if a PMD entry is "bad". That might mean swapped out.
+ * As we have no known cause of badness, it's null, as it is for many
+ * architectures.
+ */
+static inline int pmd_bad(pmd_t pmd)
+{
+	return 0;
+}
+
+/*
+ * pmd_page - converts a PMD entry to a page pointer
+ */
+#define pmd_page(pmd)  (pfn_to_page(pmd_val(pmd) >> PAGE_SHIFT))
+#define pmd_pgtable(pmd) pmd_page(pmd)
+
+/**
+ * pte_none - check if pte is mapped
+ * @pte: pte_t entry
+ */
+static inline int pte_none(pte_t pte)
+{
+	return pte_val(pte) == _NULL_PTE;
+};
+
+/*
+ * pte_present - check if page is present
+ */
+static inline int pte_present(pte_t pte)
+{
+	return pte_val(pte) & _PAGE_PRESENT;
+}
+
+/* mk_pte - make a PTE out of a page pointer and protection bits */
+#define mk_pte(page, pgprot) pfn_pte(page_to_pfn(page), (pgprot))
+
+/* pte_page - returns a page (frame pointer/descriptor?) based on a PTE */
+#define pte_page(x) pfn_to_page(pte_pfn(x))
+
+/* pte_mkold - mark PTE as not recently accessed */
+static inline pte_t pte_mkold(pte_t pte)
+{
+	pte_val(pte) &= ~_PAGE_ACCESSED;
+	return pte;
+}
+
+/* pte_mkyoung - mark PTE as recently accessed */
+static inline pte_t pte_mkyoung(pte_t pte)
+{
+	pte_val(pte) |= _PAGE_ACCESSED;
+	return pte;
+}
+
+/* pte_mkclean - mark page as in sync with backing store */
+static inline pte_t pte_mkclean(pte_t pte)
+{
+	pte_val(pte) &= ~_PAGE_DIRTY;
+	return pte;
+}
+
+/* pte_mkdirty - mark page as modified */
+static inline pte_t pte_mkdirty(pte_t pte)
+{
+	pte_val(pte) |= _PAGE_DIRTY;
+	return pte;
+}
+
+/* pte_young - "is PTE marked as accessed"? */
+static inline int pte_young(pte_t pte)
+{
+	return pte_val(pte) & _PAGE_ACCESSED;
+}
+
+/* pte_dirty - "is PTE dirty?" */
+static inline int pte_dirty(pte_t pte)
+{
+	return pte_val(pte) & _PAGE_DIRTY;
+}
+
+/* pte_modify - set protection bits on PTE */
+static inline pte_t pte_modify(pte_t pte, pgprot_t prot)
+{
+	pte_val(pte) &= PAGE_MASK;
+	pte_val(pte) |= pgprot_val(prot);
+	return pte;
+}
+
+/* pte_wrprotect - mark page as not writable */
+static inline pte_t pte_wrprotect(pte_t pte)
+{
+	pte_val(pte) &= ~_PAGE_WRITE;
+	return pte;
+}
+
+/* pte_mkwrite - mark page as writable */
+static inline pte_t pte_mkwrite(pte_t pte)
+{
+	pte_val(pte) |= _PAGE_WRITE;
+	return pte;
+}
+
+/* pte_mkexec - mark PTE as executable */
+static inline pte_t pte_mkexec(pte_t pte)
+{
+	pte_val(pte) |= _PAGE_EXECUTE;
+	return pte;
+}
+
+/* pte_read - "is PTE marked as readable?" */
+static inline int pte_read(pte_t pte)
+{
+	return pte_val(pte) & _PAGE_READ;
+}
+
+/* pte_write - "is PTE marked as writable?" */
+static inline int pte_write(pte_t pte)
+{
+	return pte_val(pte) & _PAGE_WRITE;
+}
+
+
+/* pte_exec - "is PTE marked as executable?" */
+static inline int pte_exec(pte_t pte)
+{
+	return pte_val(pte) & _PAGE_EXECUTE;
+}
+
+/* __pte_to_swp_entry - extract swap entry from PTE */
+#define __pte_to_swp_entry(pte) ((swp_entry_t) { pte_val(pte) })
+
+/* __swp_entry_to_pte - extract PTE from swap entry */
+#define __swp_entry_to_pte(x) ((pte_t) { (x).val })
+
+/* pfn_pte - convert page number and protection value to page table entry */
+#define pfn_pte(pfn, pgprot) __pte((pfn << PAGE_SHIFT) | pgprot_val(pgprot))
+
+/* pte_pfn - convert pte to page frame number */
+#define pte_pfn(pte) (pte_val(pte) >> PAGE_SHIFT)
+#define set_pmd(pmdptr, pmdval) (*(pmdptr) = (pmdval))
+
+/*
+ * set_pte_at - update page table and do whatever magic may be
+ * necessary to make the underlying hardware/firmware take note.
+ *
+ * VM may require a virtual instruction to alert the MMU.
+ */
+#define set_pte_at(mm, addr, ptep, pte) set_pte(ptep, pte)
+
+/*
+ * May need to invoke the virtual machine as well...
+ */
+#define pte_unmap(pte)		do { } while (0)
+#define pte_unmap_nested(pte)	do { } while (0)
+
+/*
+ * pte_offset_map - returns the linear address of the page table entry
+ * corresponding to an address
+ */
+#define pte_offset_map(dir, address)                                    \
+	((pte_t *)page_address(pmd_page(*(dir))) + __pte_offset(address))
+
+#define pte_offset_map_nested(pmd, addr) pte_offset_map(pmd, addr)
+
+/* pte_offset_kernel - kernel version of pte_offset */
+#define pte_offset_kernel(dir, address) \
+	((pte_t *) (unsigned long) __va(pmd_val(*dir) & PAGE_MASK) \
+				+  __pte_offset(address))
+
+/* ZERO_PAGE - returns the globally shared zero page */
+#define ZERO_PAGE(vaddr) (virt_to_page(&empty_zero_page))
+
+#define __pte_offset(address) (((address) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
+
+/* Nothing special about IO remapping at this point */
+#define io_remap_pfn_range(vma, vaddr, pfn, size, prot) \
+	remap_pfn_range(vma, vaddr, pfn, size, prot)
+
+/*  I think this is in case we have page table caches; needed by init/main.c  */
+#define pgtable_cache_init()    do { } while (0)
+
+/*
+ * Swap/file PTE definitions.  If _PAGE_PRESENT is zero, the rest of the
+ * PTE is interpreted as swap information.  Depending on the _PAGE_FILE
+ * bit, the remaining free bits are eitehr interpreted as a file offset
+ * or a swap type/offset tuple.  Rather than have the TLB fill handler
+ * test _PAGE_PRESENT, we're going to reserve the permissions bits
+ * and set them to all zeros for swap entries, which speeds up the
+ * miss handler at the cost of 3 bits of offset.  That trade-off can
+ * be revisited if necessary, but Hexagon processor architecture and
+ * target applications suggest a lot of TLB misses and not much swap space.
+ *
+ * Format of swap PTE:
+ *	bit	0:	Present (zero)
+ *	bit	1:	_PAGE_FILE (zero)
+ *	bits	2-6:	swap type (arch independent layer uses 5 bits max)
+ *	bits	7-9:	bits 2:0 of offset
+ *	bits 10-12:	effectively _PAGE_PROTNONE (all zero)
+ *	bits 13-31:  bits 21:3 of swap offset
+ *
+ * Format of file PTE:
+ *	bit	0:	Present (zero)
+ *	bit	1:	_PAGE_FILE (zero)
+ *	bits	2-9:	bits 7:0 of offset
+ *	bits 10-12:	effectively _PAGE_PROTNONE (all zero)
+ *	bits 13-31:  bits 26:8 of swap offset
+ *
+ * The split offset makes some of the following macros a little gnarly,
+ * but there's plenty of precedent for this sort of thing.
+ */
+#define PTE_FILE_MAX_BITS     27
+
+/* Used for swap PTEs */
+#define __swp_type(swp_pte)		(((swp_pte).val >> 2) & 0x1f)
+
+#define __swp_offset(swp_pte) \
+	((((swp_pte).val >> 7) & 0x7) | (((swp_pte).val >> 10) & 0x003ffff8))
+
+#define __swp_entry(type, offset) \
+	((swp_entry_t)	{ \
+		((type << 2) | \
+		 ((offset & 0x3ffff8) << 10) | ((offset & 0x7) << 7)) })
+
+/* Used for file PTEs */
+#define pte_file(pte) \
+	((pte_val(pte) & (_PAGE_FILE | _PAGE_PRESENT)) == _PAGE_FILE)
+
+#define pte_to_pgoff(pte) \
+	(((pte_val(pte) >> 2) & 0xff) | ((pte_val(pte) >> 5) & 0x07ffff00))
+
+#define pgoff_to_pte(off) \
+	((pte_t) { ((((off) & 0x7ffff00) << 5) | (((off) & 0xff) << 2)\
+	| _PAGE_FILE) })
+
+/*  Oh boy.  There are a lot of possible arch overrides found in this file.  */
+#include <asm-generic/pgtable.h>
+
+#endif
Index: linux-hexagon-kernel/arch/hexagon/mm/pgalloc.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/pgalloc.c	2011-09-03 20:14:59.784981559 -0500
@@ -0,0 +1,25 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/init.h>
+
+void __init pgtable_cache_init(void)
+{
+}
+
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 29/36] Hexagon: Add page table header files & etc.

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:16 Richard Kuo wrote:
> Removed highmem more highmem stuff.
> 
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
> Signed-off-by: Linas Vepstas <linas@codeaurora.org>
> 

Acked-by: Arnd Bergmann <arnd@arndb.de>

[patch v3 30/36] Hexagon: Add page-fault support.

[Richard Kuo]

[-- Attachment #1: mem-fault.diff --]
[-- Type: text/plain, Size: 18695 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/mem-layout.h |  112 +++++++++++++
 arch/hexagon/include/asm/vm_fault.h   |   26 +++
 arch/hexagon/mm/init.c                |  276 ++++++++++++++++++++++++++++++++++
 arch/hexagon/mm/vm_fault.c            |  188 +++++++++++++++++++++++
 4 files changed, 602 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/mem-layout.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/mem-layout.h	2011-09-07 13:00:30.930646788 -0500
@@ -0,0 +1,112 @@
+/*
+ * Memory layout definitions for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_HEXAGON_MEM_LAYOUT_H
+#define _ASM_HEXAGON_MEM_LAYOUT_H
+
+#include <linux/const.h>
+
+/*
+ * Have to do this for ginormous numbers, else they get printed as
+ * negative numbers, which the linker no likey when you try to
+ * assign it to the location counter.
+ */
+
+#define PAGE_OFFSET			_AC(0xc0000000, UL)
+
+/*
+ * LOAD_ADDRESS is the physical/linear address of where in memory
+ * the kernel gets loaded. The 12 least significant bits must be zero (0)
+ * due to limitations on setting the EVB
+ *
+ */
+
+#ifndef LOAD_ADDRESS
+#define LOAD_ADDRESS			0x00000000
+#endif
+
+#define TASK_SIZE			(PAGE_OFFSET)
+
+/*  not sure how these are used yet  */
+#define STACK_TOP			TASK_SIZE
+#define STACK_TOP_MAX			TASK_SIZE
+
+#ifndef __ASSEMBLY__
+enum fixed_addresses {
+	FIX_KMAP_BEGIN,
+	FIX_KMAP_END,  /*  check for per-cpuism  */
+	__end_of_fixed_addresses
+};
+
+#define MIN_KERNEL_SEG 0x300   /* From 0xc0000000 */
+extern int max_kernel_seg;
+
+/*
+ * Start of vmalloc virtual address space for kernel;
+ * supposed to be based on the amount of physical memory available
+ */
+
+#define VMALLOC_START (PAGE_OFFSET + VMALLOC_OFFSET + \
+	(unsigned long)high_memory)
+
+/* Gap between physical ram and vmalloc space for guard purposes. */
+#define VMALLOC_OFFSET PAGE_SIZE
+
+/*
+ * Create the space between VMALLOC_START and FIXADDR_TOP backwards
+ * from the ... "top".
+ *
+ * Permanent IO mappings will live at 0xfexx_xxxx
+ * Hypervisor occupies the last 16MB page at 0xffxxxxxx
+ */
+
+#define FIXADDR_TOP     0xfe000000
+#define FIXADDR_SIZE    (__end_of_fixed_addresses << PAGE_SHIFT)
+#define FIXADDR_START   (FIXADDR_TOP - FIXADDR_SIZE)
+
+/*
+ * "permanent kernel mappings", defined as long-lasting mappings of
+ * high-memory page frames into the kernel address space.
+ */
+
+#define LAST_PKMAP	PTRS_PER_PTE
+#define LAST_PKMAP_MASK	(LAST_PKMAP - 1)
+#define PKMAP_NR(virt)	((virt - PKMAP_BASE) >> PAGE_SHIFT)
+#define PKMAP_ADDR(nr)	(PKMAP_BASE + ((nr) << PAGE_SHIFT))
+
+/*
+ * To the "left" of the fixed map space is the kmap space
+ *
+ * "Permanent Kernel Mappings"; fancy (or less fancy) PTE table
+ * that looks like it's actually walked.
+ * Need to check the alignment/shift usage; some archs use
+ * PMD_MASK on this value
+ */
+#define PKMAP_BASE (FIXADDR_START-PAGE_SIZE*LAST_PKMAP)
+
+/*
+ * 2 pages of guard gap between where vmalloc area ends
+ * and pkmap_base begins.
+ */
+#define VMALLOC_END (PKMAP_BASE-PAGE_SIZE*2)
+#endif /*  !__ASSEMBLY__  */
+
+
+#endif /* _ASM_HEXAGON_MEM_LAYOUT_H */
Index: linux-hexagon-kernel/arch/hexagon/mm/init.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/init.c	2011-09-07 13:00:31.210647214 -0500
@@ -0,0 +1,276 @@
+/*
+ * Memory subsystem initialization for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/init.h>
+#include <linux/mm.h>
+#include <linux/bootmem.h>
+#include <asm/atomic.h>
+#include <linux/highmem.h>
+#include <asm/tlb.h>
+#include <asm/sections.h>
+#include <asm/vm_mmu.h>
+
+/*
+ * Define a startpg just past the end of the kernel image and a lastpg
+ * that corresponds to the end of real or simulated platform memory.
+ */
+#define bootmem_startpg (PFN_UP(((unsigned long) _end) - PAGE_OFFSET))
+
+unsigned long bootmem_lastpg;  /*  Should be set by platform code  */
+
+/*  Set as variable to limit PMD copies  */
+int max_kernel_seg = 0x303;
+
+/*  think this should be (page_size-1) the way it's used...*/
+unsigned long zero_page_mask;
+
+/*  indicate pfn's of high memory  */
+unsigned long highstart_pfn, highend_pfn;
+
+/* struct mmu_gather defined in asm-generic.h;  */
+DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
+
+/* Default cache attribute for newly created page tables */
+unsigned long _dflt_cache_att = CACHEDEF;
+
+/*
+ * The current "generation" of kernel map, which should not roll
+ * over until Hell freezes over.  Actual bound in years needs to be
+ * calculated to confirm.
+ */
+DEFINE_SPINLOCK(kmap_gen_lock);
+
+/*  checkpatch says don't init this to 0.  */
+unsigned long long kmap_generation;
+
+/*
+ * mem_init - initializes memory
+ *
+ * Frees up bootmem
+ * Fixes up more stuff for HIGHMEM
+ * Calculates and displays memory available/used
+ */
+void __init mem_init(void)
+{
+	/*  No idea where this is actually declared.  Seems to evade LXR.  */
+	totalram_pages += free_all_bootmem();
+	num_physpages = bootmem_lastpg;	/*  seriously, what?  */
+
+	printk(KERN_INFO "totalram_pages = %ld\n", totalram_pages);
+
+	/*
+	 *  To-Do:  someone somewhere should wipe out the bootmem map
+	 *  after we're done?
+	 */
+
+	/*
+	 * This can be moved to some more virtual-memory-specific
+	 * initialization hook at some point.  Set the init_mm
+	 * descriptors "context" value to point to the initial
+	 * kernel segment table's physical address.
+	 */
+	init_mm.context.ptbase = __pa(init_mm.pgd);
+}
+
+/*
+ * free_initmem - frees memory used by stuff declared with __init
+ *
+ * Todo:  free pages between __init_begin and __init_end; possibly
+ * some devtree related stuff as well.
+ */
+void __init_refok free_initmem(void)
+{
+}
+
+/*
+ * free_initrd_mem - frees...  initrd memory.
+ * @start - start of init memory
+ * @end - end of init memory
+ *
+ * Apparently has to be passed the address of the initrd memory.
+ *
+ * Wrapped by #ifdef CONFIG_BLKDEV_INITRD
+ */
+void free_initrd_mem(unsigned long start, unsigned long end)
+{
+}
+
+void sync_icache_dcache(pte_t pte)
+{
+	unsigned long addr;
+	struct page *page;
+
+	page = pte_page(pte);
+	addr = (unsigned long) page_address(page);
+
+	__vmcache_idsync(addr, PAGE_SIZE);
+}
+
+/*
+ * In order to set up page allocator "nodes",
+ * somebody has to call free_area_init() for UMA.
+ *
+ * In this mode, we only have one pg_data_t
+ * structure: contig_mem_data.
+ */
+void __init paging_init(void)
+{
+	unsigned long zones_sizes[MAX_NR_ZONES] = {0, };
+
+	/*
+	 *  This is not particularly well documented anywhere, but
+	 *  give ZONE_NORMAL all the memory, including the big holes
+	 *  left by the kernel+bootmem_map which are already left as reserved
+	 *  in the bootmem_map; free_area_init should see those bits and
+	 *  adjust accordingly.
+	 */
+
+	zones_sizes[ZONE_NORMAL] = max_low_pfn;
+
+	free_area_init(zones_sizes);  /*  sets up the zonelists and mem_map  */
+
+	/*
+	 * Start of high memory area.  Will probably need something more
+	 * fancy if we...  get more fancy.
+	 */
+	high_memory = (void *)((bootmem_lastpg + 1) << PAGE_SHIFT);
+}
+
+#ifndef DMA_RESERVE
+#define DMA_RESERVE		(4)
+#endif
+
+#define DMA_CHUNKSIZE		(1<<22)
+#define DMA_RESERVED_BYTES	(DMA_RESERVE * DMA_CHUNKSIZE)
+
+/*
+ * Pick out the memory size.  We look for mem=size,
+ * where size is "size[KkMm]"
+ */
+static int __init early_mem(char *p)
+{
+	unsigned long size;
+	char *endp;
+
+	size = memparse(p, &endp);
+
+	bootmem_lastpg = PFN_DOWN(size);
+
+	return 0;
+}
+early_param("mem", early_mem);
+
+size_t hexagon_coherent_pool_size = (size_t) (DMA_RESERVE << 22);
+
+void __init setup_arch_memory(void)
+{
+	int bootmap_size;
+	/*  XXX Todo: this probably should be cleaned up  */
+	u32 *segtable = (u32 *) &swapper_pg_dir[0];
+	u32 *segtable_end;
+
+	/*
+	 * Set up boot memory allocator
+	 *
+	 * The Gorman book also talks about these functions.
+	 * This needs to change for highmem setups.
+	 */
+
+	/* Memory size needs to be a multiple of 16M */
+	bootmem_lastpg = PFN_DOWN((bootmem_lastpg << PAGE_SHIFT) &
+		~((BIG_KERNEL_PAGE_SIZE) - 1));
+
+	/*
+	 * Reserve the top DMA_RESERVE bytes of RAM for DMA (uncached)
+	 * memory allocation
+	 */
+	bootmap_size = init_bootmem(bootmem_startpg, bootmem_lastpg -
+				    PFN_DOWN(DMA_RESERVED_BYTES));
+
+	printk(KERN_INFO "bootmem_startpg:  0x%08lx\n", bootmem_startpg);
+	printk(KERN_INFO "bootmem_lastpg:  0x%08lx\n", bootmem_lastpg);
+	printk(KERN_INFO "bootmap_size:  %d\n", bootmap_size);
+	printk(KERN_INFO "max_low_pfn:  0x%08lx\n", max_low_pfn);
+
+	/*
+	 * The default VM page tables (will be) populated with
+	 * VA=PA+PAGE_OFFSET mapping.  We go in and invalidate entries
+	 * higher than what we have memory for.
+	 */
+
+	/*  this is pointer arithmetic; each entry covers 4MB  */
+	segtable = segtable + (PAGE_OFFSET >> 22);
+
+	/*  this actually only goes to the end of the first gig  */
+	segtable_end = segtable + (1<<(30-22));
+
+	/*  Move forward to the start of empty pages  */
+	segtable += bootmem_lastpg >> (22-PAGE_SHIFT);
+
+	{
+	    int i;
+
+	    for (i = 1 ; i <= DMA_RESERVE ; i++)
+		segtable[-i] = ((segtable[-i] & __HVM_PTE_PGMASK_4MB)
+				| __HVM_PTE_R | __HVM_PTE_W | __HVM_PTE_X
+				| __HEXAGON_C_UNC << 6
+				| __HVM_PDE_S_4MB);
+	}
+
+	printk(KERN_INFO "clearing segtable from %p to %p\n", segtable,
+		segtable_end);
+	while (segtable < (segtable_end-8))
+		*(segtable++) = __HVM_PDE_S_INVALID;
+	/* stop the pointer at the device I/O 4MB page  */
+
+	printk(KERN_INFO "segtable = %p (should be equal to _K_io_map)\n",
+		segtable);
+
+#if 0
+	/*  Other half of the early device table from vm_init_segtable. */
+	printk(KERN_INFO "&_K_init_devicetable = 0x%08x\n",
+		(unsigned long) _K_init_devicetable-PAGE_OFFSET);
+	*segtable = ((u32) (unsigned long) _K_init_devicetable-PAGE_OFFSET) |
+		__HVM_PDE_S_4KB;
+	printk(KERN_INFO "*segtable = 0x%08x\n", *segtable);
+#endif
+
+	/*
+	 * Free all the memory that wasn't taken up by the bootmap, the DMA
+	 * reserve, or kernel itself.
+	 */
+	free_bootmem(PFN_PHYS(bootmem_startpg)+bootmap_size,
+		     PFN_PHYS(bootmem_lastpg - bootmem_startpg) - bootmap_size -
+		     DMA_RESERVED_BYTES);
+
+	/*
+	 *  The bootmem allocator seemingly just lives to feed memory
+	 *  to the paging system
+	 */
+	printk(KERN_INFO "PAGE_SIZE=%lu\n", PAGE_SIZE);
+	paging_init();  /*  See Gorman Book, 2.3  */
+
+	/*
+	 *  At this point, the page allocator is kind of initialized, but
+	 *  apparently no pages are available (just like with the bootmem
+	 *  allocator), and need to be freed themselves via mem_init(),
+	 *  which is called by start_kernel() later on in the process
+	 */
+}
Index: linux-hexagon-kernel/arch/hexagon/mm/vm_fault.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/vm_fault.c	2011-09-03 20:14:59.804981615 -0500
@@ -0,0 +1,188 @@
+/*
+ * Memory fault handling for Hexagon
+ *
+ * Copyright (c) 2010-2011 Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * Page fault handling for the Hexagon Virtual Machine.
+ * Can also be called by a native port emulating the HVM
+ * execptions.
+ */
+
+#include <asm/pgtable.h>
+#include <asm/traps.h>
+#include <asm/uaccess.h>
+#include <linux/mm.h>
+#include <linux/signal.h>
+#include <linux/module.h>
+#include <linux/hardirq.h>
+
+/*
+ * Decode of hardware exception sends us to one of several
+ * entry points.  At each, we generate canonical arguments
+ * for handling by the abstract memory management code.
+ */
+#define FLT_IFETCH     -1
+#define FLT_LOAD        0
+#define FLT_STORE       1
+
+
+/*
+ * Canonical page fault handler
+ */
+void do_page_fault(unsigned long address, long cause, struct pt_regs *regs)
+{
+	struct vm_area_struct *vma;
+	struct mm_struct *mm = current->mm;
+	siginfo_t info;
+	int si_code;
+	int fault;
+	const struct exception_table_entry *fixup;
+
+	si_code = SEGV_MAPERR;
+	local_irq_enable();
+
+	/*
+	 * If we're in an interrupt or have no user context,
+	 * then must not take the fault.
+	 */
+	if (unlikely(in_interrupt() || !mm))
+		goto no_context;
+
+	down_read(&mm->mmap_sem);
+	vma = find_vma(mm, address);
+	if (!vma)
+		goto bad_area;
+
+	if (vma->vm_start <= address)
+		goto good_area;
+
+	if (!(vma->vm_flags & VM_GROWSDOWN))
+		goto bad_area;
+
+	if (expand_stack(vma, address))
+		goto bad_area;
+
+good_area:
+	/* Address space is OK.  Now check access rights. */
+	si_code = SEGV_ACCERR;
+
+	switch (cause) {
+	case FLT_IFETCH:
+		if (!(vma->vm_flags & VM_EXEC))
+			goto bad_area;
+		break;
+	case FLT_LOAD:
+		if (!(vma->vm_flags & VM_READ))
+			goto bad_area;
+		break;
+	case FLT_STORE:
+		if (!(vma->vm_flags & VM_WRITE))
+			goto bad_area;
+		break;
+	}
+
+	fault = handle_mm_fault(mm, vma, address, (cause > 0));
+
+	/* The most common case -- we are done. */
+	if (likely(!(fault & VM_FAULT_ERROR))) {
+		if (fault & VM_FAULT_MAJOR)
+			current->maj_flt++;
+		else
+			current->min_flt++;
+
+		up_read(&mm->mmap_sem);
+		return;
+	}
+
+	up_read(&mm->mmap_sem);
+
+	/* Handle copyin/out exception cases */
+	if (!user_mode(regs))
+		goto no_context;
+
+	if (fault & VM_FAULT_OOM) {
+		pagefault_out_of_memory();
+		return;
+	}
+
+	/* User-mode address is in the memory map, but we are
+	 * unable to fix up the page fault.
+	 */
+	if (fault & VM_FAULT_SIGBUS) {
+		info.si_signo = SIGBUS;
+		info.si_code = BUS_ADRERR;
+	}
+	/* Address is not in the memory map */
+	else {
+		info.si_signo = SIGSEGV;
+		info.si_code = SEGV_ACCERR;
+	}
+	info.si_errno = 0;
+	info.si_addr = (void __user *)address;
+	force_sig_info(info.si_code, &info, current);
+	return;
+
+bad_area:
+	up_read(&mm->mmap_sem);
+
+	if (user_mode(regs)) {
+		info.si_signo = SIGSEGV;
+		info.si_errno = 0;
+		info.si_code = si_code;
+		info.si_addr = (void *)address;
+		force_sig_info(SIGSEGV, &info, current);
+		return;
+	}
+	/* Kernel-mode fault falls through */
+
+no_context:
+	fixup = search_exception_tables(pt_elr(regs));
+	if (fixup) {
+		pt_set_elr(regs, fixup->fixup);
+		return;
+	}
+
+	/* Things are looking very, very bad now */
+	bust_spinlocks(1);
+	printk(KERN_EMERG "Unable to handle kernel paging request at "
+		"virtual address 0x%08lx, regs %p\n", address, regs);
+	die("Bad Kernel VA", regs, SIGKILL);
+}
+
+
+void read_protection_fault(struct pt_regs *regs)
+{
+	unsigned long badvadr = pt_badva(regs);
+
+	do_page_fault(badvadr, FLT_LOAD, regs);
+}
+
+void write_protection_fault(struct pt_regs *regs)
+{
+	unsigned long badvadr = pt_badva(regs);
+
+	do_page_fault(badvadr, FLT_STORE, regs);
+}
+
+void execute_protection_fault(struct pt_regs *regs)
+{
+	unsigned long badvadr = pt_badva(regs);
+
+	do_page_fault(badvadr, FLT_IFETCH, regs);
+}
Index: linux-hexagon-kernel/arch/hexagon/include/asm/vm_fault.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/vm_fault.h	2011-09-03 20:14:59.804981615 -0500
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef _ASM_HEXAGON_VM_FAULT_H
+#define _ASM_HEXAGON_VM_FAULT_H
+
+extern void execute_protection_fault(struct pt_regs *);
+extern void write_protection_fault(struct pt_regs *);
+extern void read_protection_fault(struct pt_regs *);
+
+#endif

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 30/36] Hexagon: Add page-fault support.

[Benjamin Herrenschmidt]

On Thu, 2011-09-08 at 20:09 -0500, Richard Kuo wrote:

> +/*
> + * Canonical page fault handler
> + */
> +void do_page_fault(unsigned long address, long cause, struct pt_regs *regs)
> +{
> +	struct vm_area_struct *vma;
> +	struct mm_struct *mm = current->mm;
> +	siginfo_t info;
> +	int si_code;
> +	int fault;
> +	const struct exception_table_entry *fixup;
> +
> +	si_code = SEGV_MAPERR;
> +	local_irq_enable();
> +
> +	/*
> +	 * If we're in an interrupt or have no user context,
> +	 * then must not take the fault.
> +	 */
> +	if (unlikely(in_interrupt() || !mm))
> +		goto no_context;

Are you sure about enabling interrupts before going to "no_context" ?

If somebody does a user access "in atomic" with interrupt disabled (with
the expectation that a fault would be caught by search_exception_tables
and returned as -EFAULT), such code probably doesn't expect interrupts
to be re-enabled implicitely.

Cheers,
Ben.

Re: [patch v3 30/36] Hexagon: Add page-fault support.

[Richard Kuo]

On Sun, Sep 11, 2011 at 12:08:49PM -0300, Benjamin Herrenschmidt wrote:
> Are you sure about enabling interrupts before going to "no_context" ?
> 
> If somebody does a user access "in atomic" with interrupt disabled (with
> the expectation that a fault would be caught by search_exception_tables
> and returned as -EFAULT), such code probably doesn't expect interrupts
> to be re-enabled implicitely.

I hadn't considered that...  I'll review this and make sure this isn't what
we intended (it probably isn't) and get it fixed.


Thanks,
Richard Kuo


-- 

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 31/36] Hexagon: kgdb support files

[Richard Kuo]

[-- Attachment #1: kgdb.diff --]
[-- Type: text/plain, Size: 10386 bytes --]

Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/include/asm/kgdb.h |   39 ++++++
 arch/hexagon/kernel/kgdb.c      |  241 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 280 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/include/asm/kgdb.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/include/asm/kgdb.h	2011-09-03 20:14:59.814981643 -0500
@@ -0,0 +1,39 @@
+/*
+ * arch/hexagon/include/asm/kgdb.h - Hexagon KGDB Support
+ *
+ * Copyright (c) 2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#ifndef __HEXAGON_KGDB_H__
+#define __HEXAGON_KGDB_H__
+
+#define BREAK_INSTR_SIZE 4
+#define CACHE_FLUSH_IS_SAFE   1
+#define BUFMAX       ((NUMREGBYTES * 2) + 512)
+
+static inline void arch_kgdb_breakpoint(void)
+{
+	asm("trap0(#0xDB)");
+}
+
+/* Registers:
+ * 32 gpr + sa0/1 + lc0/1 + m0/1 + gp + ugp + pred + pc = 42 total.
+ */
+#define NUMREGBYTES  (42*4)
+#define DBG_MAX_REG_NUM 42
+
+#endif /* __HEXAGON_KGDB_H__ */
Index: linux-hexagon-kernel/arch/hexagon/kernel/kgdb.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/kgdb.c	2011-09-03 20:14:59.814981643 -0500
@@ -0,0 +1,241 @@
+/*
+ * arch/hexagon/kernel/kgdb.c - Hexagon KGDB Support
+ *
+ * Copyright (c) 2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/kdebug.h>
+#include <linux/kgdb.h>
+
+/* All registers are 4 bytes, for now */
+#define GDB_SIZEOF_REG 4
+
+/* The register names are used during printing of the regs;
+ * Keep these at three letters to pretty-print. */
+struct dbg_reg_def_t dbg_reg_def[DBG_MAX_REG_NUM] = {
+	{ " r0", GDB_SIZEOF_REG, offsetof(struct pt_regs, r00)},
+	{ " r1", GDB_SIZEOF_REG, offsetof(struct pt_regs, r01)},
+	{ " r2", GDB_SIZEOF_REG, offsetof(struct pt_regs, r02)},
+	{ " r3", GDB_SIZEOF_REG, offsetof(struct pt_regs, r03)},
+	{ " r4", GDB_SIZEOF_REG, offsetof(struct pt_regs, r04)},
+	{ " r5", GDB_SIZEOF_REG, offsetof(struct pt_regs, r05)},
+	{ " r6", GDB_SIZEOF_REG, offsetof(struct pt_regs, r06)},
+	{ " r7", GDB_SIZEOF_REG, offsetof(struct pt_regs, r07)},
+	{ " r8", GDB_SIZEOF_REG, offsetof(struct pt_regs, r08)},
+	{ " r9", GDB_SIZEOF_REG, offsetof(struct pt_regs, r09)},
+	{ "r10", GDB_SIZEOF_REG, offsetof(struct pt_regs, r10)},
+	{ "r11", GDB_SIZEOF_REG, offsetof(struct pt_regs, r11)},
+	{ "r12", GDB_SIZEOF_REG, offsetof(struct pt_regs, r12)},
+	{ "r13", GDB_SIZEOF_REG, offsetof(struct pt_regs, r13)},
+	{ "r14", GDB_SIZEOF_REG, offsetof(struct pt_regs, r14)},
+	{ "r15", GDB_SIZEOF_REG, offsetof(struct pt_regs, r15)},
+	{ "r16", GDB_SIZEOF_REG, offsetof(struct pt_regs, r16)},
+	{ "r17", GDB_SIZEOF_REG, offsetof(struct pt_regs, r17)},
+	{ "r18", GDB_SIZEOF_REG, offsetof(struct pt_regs, r18)},
+	{ "r19", GDB_SIZEOF_REG, offsetof(struct pt_regs, r19)},
+	{ "r20", GDB_SIZEOF_REG, offsetof(struct pt_regs, r20)},
+	{ "r21", GDB_SIZEOF_REG, offsetof(struct pt_regs, r21)},
+	{ "r22", GDB_SIZEOF_REG, offsetof(struct pt_regs, r22)},
+	{ "r23", GDB_SIZEOF_REG, offsetof(struct pt_regs, r23)},
+	{ "r24", GDB_SIZEOF_REG, offsetof(struct pt_regs, r24)},
+	{ "r25", GDB_SIZEOF_REG, offsetof(struct pt_regs, r25)},
+	{ "r26", GDB_SIZEOF_REG, offsetof(struct pt_regs, r26)},
+	{ "r27", GDB_SIZEOF_REG, offsetof(struct pt_regs, r27)},
+	{ "r28", GDB_SIZEOF_REG, offsetof(struct pt_regs, r28)},
+	{ "r29", GDB_SIZEOF_REG, offsetof(struct pt_regs, r29)},
+	{ "r30", GDB_SIZEOF_REG, offsetof(struct pt_regs, r30)},
+	{ "r31", GDB_SIZEOF_REG, offsetof(struct pt_regs, r31)},
+
+	{ "usr", GDB_SIZEOF_REG, offsetof(struct pt_regs, usr)},
+	{ "preds", GDB_SIZEOF_REG, offsetof(struct pt_regs, preds)},
+	{ " m0", GDB_SIZEOF_REG, offsetof(struct pt_regs, m0)},
+	{ " m1", GDB_SIZEOF_REG, offsetof(struct pt_regs, m1)},
+	{ "sa0", GDB_SIZEOF_REG, offsetof(struct pt_regs, sa0)},
+	{ "sa1", GDB_SIZEOF_REG, offsetof(struct pt_regs, sa1)},
+	{ "lc0", GDB_SIZEOF_REG, offsetof(struct pt_regs, lc0)},
+	{ "lc1", GDB_SIZEOF_REG, offsetof(struct pt_regs, lc1)},
+	{ " gp", GDB_SIZEOF_REG, offsetof(struct pt_regs, gp)},
+	{ "ugp", GDB_SIZEOF_REG, offsetof(struct pt_regs, ugp)},
+};
+
+struct kgdb_arch arch_kgdb_ops = {
+	/* trap0(#0xDB) 0x0cdb0054 */
+	.gdb_bpt_instr = {0x54, 0x00, 0xdb, 0x0c},
+};
+
+char *dbg_get_reg(int regno, void *mem, struct pt_regs *regs)
+{
+	if (regno >= DBG_MAX_REG_NUM || regno < 0)
+		return NULL;
+
+	*((unsigned long *) mem) = *((unsigned long *) ((void *)regs +
+		dbg_reg_def[regno].offset));
+
+	return dbg_reg_def[regno].name;
+}
+
+int dbg_set_reg(int regno, void *mem, struct pt_regs *regs)
+{
+	if (regno >= DBG_MAX_REG_NUM || regno < 0)
+		return -EINVAL;
+
+	*((unsigned long *) ((void *)regs + dbg_reg_def[regno].offset)) =
+		*((unsigned long *) mem);
+
+	return 0;
+}
+
+void kgdb_arch_set_pc(struct pt_regs *regs, unsigned long pc)
+{
+	instruction_pointer(regs) = pc;
+}
+
+#ifdef CONFIG_SMP
+
+/**
+ * kgdb_roundup_cpus - Get other CPUs into a holding pattern
+ * @flags: Current IRQ state
+ *
+ * On SMP systems, we need to get the attention of the other CPUs
+ * and get them be in a known state.  This should do what is needed
+ * to get the other CPUs to call kgdb_wait(). Note that on some arches,
+ * the NMI approach is not used for rounding up all the CPUs. For example,
+ * in case of MIPS, smp_call_function() is used to roundup CPUs. In
+ * this case, we have to make sure that interrupts are enabled before
+ * calling smp_call_function(). The argument to this function is
+ * the flags that will be used when restoring the interrupts. There is
+ * local_irq_save() call before kgdb_roundup_cpus().
+ *
+ * On non-SMP systems, this is not called.
+ */
+
+static void hexagon_kgdb_nmi_hook(void *ignored)
+{
+	kgdb_nmicallback(raw_smp_processor_id(), get_irq_regs());
+}
+
+void kgdb_roundup_cpus(unsigned long flags)
+{
+	smp_call_function(hexagon_kgdb_nmi_hook, NULL, 0);
+}
+#endif
+
+
+/*  Not yet working  */
+void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs,
+				 struct task_struct *task)
+{
+	struct pt_regs *thread_regs;
+
+	if (task == NULL)
+		return;
+
+	/* Initialize to zero */
+	memset(gdb_regs, 0, NUMREGBYTES);
+
+	/* Otherwise, we have only some registers from switch_to() */
+	thread_regs = task_pt_regs(task);
+	gdb_regs[0] = thread_regs->r00;
+}
+
+/**
+ * kgdb_arch_handle_exception - Handle architecture specific GDB packets.
+ * @vector: The error vector of the exception that happened.
+ * @signo: The signal number of the exception that happened.
+ * @err_code: The error code of the exception that happened.
+ * @remcom_in_buffer: The buffer of the packet we have read.
+ * @remcom_out_buffer: The buffer of %BUFMAX bytes to write a packet into.
+ * @regs: The &struct pt_regs of the current process.
+ *
+ * This function MUST handle the 'c' and 's' command packets,
+ * as well packets to set / remove a hardware breakpoint, if used.
+ * If there are additional packets which the hardware needs to handle,
+ * they are handled here.  The code should return -1 if it wants to
+ * process more packets, and a %0 or %1 if it wants to exit from the
+ * kgdb callback.
+ *
+ * Not yet working.
+ */
+int kgdb_arch_handle_exception(int vector, int signo, int err_code,
+			       char *remcom_in_buffer, char *remcom_out_buffer,
+			       struct pt_regs *linux_regs)
+{
+	switch (remcom_in_buffer[0]) {
+	case 's':
+	case 'c':
+		return 0;
+	}
+	/* Stay in the debugger. */
+	return -1;
+}
+
+static int __kgdb_notify(struct die_args *args, unsigned long cmd)
+{
+	struct pt_regs *regs = args->regs;
+#if 0
+	ret = kgdb_handle_exception(args->trapnr & 0xff, args->signr, args->err,
+				    regs);
+	if (ret)
+		return NOTIFY_DONE;
+#endif
+	return NOTIFY_STOP;
+}
+
+static int
+kgdb_notify(struct notifier_block *self, unsigned long cmd, void *ptr)
+{
+	unsigned long flags;
+	int ret;
+
+	local_irq_save(flags);
+	ret = __kgdb_notify(ptr, cmd);
+	local_irq_restore(flags);
+
+	return ret;
+}
+
+static struct notifier_block kgdb_notifier = {
+	.notifier_call = kgdb_notify,
+
+	/*
+	 * Lowest-prio notifier priority, we want to be notified last:
+	 */
+	.priority = -INT_MAX,
+};
+
+/**
+ * kgdb_arch_init - Perform any architecture specific initalization.
+ *
+ * This function will handle the initalization of any architecture
+ * specific callbacks.
+ */
+int kgdb_arch_init(void)
+{
+	return register_die_notifier(&kgdb_notifier);
+}
+
+/**
+ * kgdb_arch_exit - Perform any architecture specific uninitalization.
+ *
+ * This function will handle the uninitalization of any architecture
+ * specific callbacks, for dynamic registration and unregistration.
+ */
+void kgdb_arch_exit(void)
+{
+	unregister_die_notifier(&kgdb_notifier);
+}
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 32/36] Hexagon: Comet platform support

[Richard Kuo]

[-- Attachment #1: platform-comet.diff --]
[-- Type: text/plain, Size: 2981 bytes --]

Used the shortened version of the config; other platform stuff moved out until
it's ready.

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/configs/comet_defconfig |   85 +++++++++++++++++++++++++++++++++++
 1 file changed, 85 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/configs/comet_defconfig
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/configs/comet_defconfig	2011-09-07 13:00:30.860646683 -0500
@@ -0,0 +1,85 @@
+CONFIG_SMP=y
+CONFIG_DEFAULT_MMAP_MIN_ADDR=0
+CONFIG_HZ_100=y
+CONFIG_EXPERIMENTAL=y
+CONFIG_CROSS_COMPILE="hexagon-"
+CONFIG_LOCALVERSION="-smp"
+# CONFIG_LOCALVERSION_AUTO is not set
+CONFIG_SYSVIPC=y
+CONFIG_POSIX_MQUEUE=y
+CONFIG_BSD_PROCESS_ACCT=y
+CONFIG_BSD_PROCESS_ACCT_V3=y
+CONFIG_TASKSTATS=y
+CONFIG_TASK_DELAY_ACCT=y
+CONFIG_IKCONFIG=y
+CONFIG_IKCONFIG_PROC=y
+CONFIG_LOG_BUF_SHIFT=18
+CONFIG_BLK_DEV_INITRD=y
+CONFIG_EMBEDDED=y
+# CONFIG_VM_EVENT_COUNTERS is not set
+# CONFIG_BLK_DEV_BSG is not set
+CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug"
+# CONFIG_STANDALONE is not set
+CONFIG_CONNECTOR=y
+CONFIG_BLK_DEV_LOOP=y
+CONFIG_BLK_DEV_NBD=y
+CONFIG_NETDEVICES=y
+CONFIG_MII=y
+CONFIG_PHYLIB=y
+CONFIG_NET_ETHERNET=y
+# CONFIG_NETDEV_1000 is not set
+# CONFIG_NETDEV_10000 is not set
+# CONFIG_INPUT_MOUSEDEV is not set
+# CONFIG_INPUT_KEYBOARD is not set
+# CONFIG_INPUT_MOUSE is not set
+# CONFIG_SERIO is not set
+# CONFIG_CONSOLE_TRANSLATIONS is not set
+CONFIG_LEGACY_PTY_COUNT=64
+# CONFIG_DEVKMEM is not set
+# CONFIG_HW_RANDOM is not set
+CONFIG_SPI=y
+CONFIG_SPI_DEBUG=y
+CONFIG_SPI_BITBANG=y
+# CONFIG_HWMON is not set
+# CONFIG_VGA_CONSOLE is not set
+# CONFIG_HID_SUPPORT is not set
+# CONFIG_USB_SUPPORT is not set
+CONFIG_EXT2_FS=y
+CONFIG_EXT2_FS_XATTR=y
+CONFIG_EXT2_FS_POSIX_ACL=y
+CONFIG_EXT2_FS_SECURITY=y
+CONFIG_EXT3_FS=y
+# CONFIG_EXT3_DEFAULTS_TO_ORDERED is not set
+CONFIG_EXT3_FS_POSIX_ACL=y
+CONFIG_EXT3_FS_SECURITY=y
+CONFIG_QUOTA=y
+CONFIG_PROC_KCORE=y
+CONFIG_TMPFS=y
+CONFIG_TMPFS_POSIX_ACL=y
+# CONFIG_MISC_FILESYSTEMS is not set
+CONFIG_NFS_FS=y
+CONFIG_NFS_V3=y
+CONFIG_NFS_V3_ACL=y
+# CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS is not set
+CONFIG_NET=y
+CONFIG_PACKET=y
+CONFIG_UNIX=y
+CONFIG_INET=y
+# CONFIG_INET_XFRM_MODE_TRANSPORT is not set
+# CONFIG_INET_XFRM_MODE_TUNNEL is not set
+# CONFIG_INET_XFRM_MODE_BEET is not set
+# CONFIG_INET_LRO is not set
+# CONFIG_INET_DIAG is not set
+# CONFIG_IPV6 is not set
+CONFIG_CRYPTO_MD5=y
+# CONFIG_CRYPTO_ANSI_CPRNG is not set
+# CONFIG_CRYPTO_HW is not set
+CONFIG_CRC_CCITT=y
+CONFIG_CRC16=y
+CONFIG_CRC_T10DIF=y
+CONFIG_LIBCRC32C=y
+CONFIG_FRAME_WARN=0
+CONFIG_MAGIC_SYSRQ=y
+CONFIG_DEBUG_FS=y
+# CONFIG_SCHED_DEBUG is not set
+CONFIG_DEBUG_INFO=y

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 33/36] Hexagon: Add configuration and makefiles for the Hexagon architecture.

[Richard Kuo]

[-- Attachment #1: linux-build.diff --]
[-- Type: text/plain, Size: 15687 bytes --]

Removed CONFIG_ARCH_HAS_KMAP

Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/Kconfig              |  218 ++++++++++++++++++++++++++++++++++++++
 arch/hexagon/Makefile             |   59 ++++++++++
 arch/hexagon/kernel/Makefile      |   19 +++
 arch/hexagon/kernel/asm-offsets.c |  104 ++++++++++++++++++
 arch/hexagon/kernel/vmlinux.lds.S |   93 ++++++++++++++++
 arch/hexagon/lib/Makefile         |    4 
 arch/hexagon/mm/Makefile          |    7 +
 7 files changed, 504 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/Kconfig
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/Kconfig	2011-09-07 13:01:00.740692085 -0500
@@ -0,0 +1,218 @@
+# Hexagon configuration
+comment "Linux Kernel Configuration for Hexagon"
+
+config HEXAGON
+	def_bool y
+	select HAVE_OPROFILE
+	select USE_GENERIC_SMP_HELPERS if SMP
+	# Other pending projects/to-do items.
+	# select HAVE_REGS_AND_STACK_ACCESS_API
+	# select HAVE_HW_BREAKPOINT if PERF_EVENTS
+	# select ARCH_HAS_CPU_IDLE_WAIT
+	# select ARCH_WANT_OPTIONAL_GPIOLIB
+	# select ARCH_REQUIRE_GPIOLIB
+	# select HAVE_CLK
+	# select IRQ_PER_CPU
+	select HAVE_IRQ_WORK
+	# select GENERIC_PENDING_IRQ if SMP
+	select GENERIC_ATOMIC64
+	select HAVE_PERF_EVENTS
+	select HAVE_GENERIC_HARDIRQS
+	select GENERIC_HARDIRQS_NO__DO_IRQ
+	select GENERIC_HARDIRQS_NO_DEPRECATED
+	# GENERIC_ALLOCATOR is used by dma_alloc_coherent()
+	select GENERIC_ALLOCATOR
+	select HAVE_ARCH_KGDB
+	select NO_IOPORT
+	# mostly generic routines, with some accelerated ones
+	---help---
+	  Qualcomm Hexagon is a processor architecture designed for high
+	  performance and low power across a wide variety of applications.
+
+config HEXAGON_ARCH_V1
+	bool
+
+config HEXAGON_ARCH_V2
+	bool
+
+config HEXAGON_ARCH_V3
+	bool
+
+config HEXAGON_ARCH_V4
+	bool
+
+config FRAME_POINTER
+	def_bool y
+
+config LOCKDEP_SUPPORT
+	def_bool y
+
+config PCI
+	def_bool n
+
+config EARLY_PRINTK
+	def_bool y
+
+config KTIME_SCALAR
+	def_bool y
+
+config MMU
+	def_bool y
+
+config TRACE_IRQFLAGS_SUPPORT
+	def_bool y
+
+config GENERIC_CSUM
+	def_bool y
+
+#
+# Use the generic interrupt handling code in kernel/irq/:
+#
+config GENERIC_IRQ_PROBE
+	def_bool y
+
+config GENERIC_IOMAP
+	def_bool y
+
+#config ZONE_DMA
+#	bool
+#	default y
+
+config HAS_DMA
+	bool
+	select HAVE_DMA_ATTRS
+	default y
+
+config NEED_SG_DMA_LENGTH
+	def_bool y
+
+config RWSEM_GENERIC_SPINLOCK
+	def_bool n
+
+config RWSEM_XCHGADD_ALGORITHM
+	def_bool y
+
+config GENERIC_FIND_NEXT_BIT
+	def_bool y
+
+config GENERIC_HWEIGHT
+	def_bool y
+
+config GENERIC_TIME
+	def_bool y
+
+config GENERIC_CLOCKEVENTS
+	def_bool y
+
+config GENERIC_CLOCKEVENTS_BROADCAST
+	def_bool y
+
+config STACKTRACE_SUPPORT
+	def_bool y
+	select STACKTRACE
+
+config GENERIC_BUG
+	def_bool y
+	depends on BUG
+
+config BUG
+	def_bool y
+
+menu "Machine selection"
+
+choice
+	prompt "System type"
+	default HEXAGON_ARCH_V2
+
+config HEXAGON_COMET
+	bool "Comet Board"
+	select HEXAGON_ARCH_V2
+	---help---
+	  Support for the Comet platform.
+
+endchoice
+
+config HEXAGON_VM
+	def_bool y
+
+config CMDLINE
+	string "Default kernel command string"
+	default ""
+	help
+	  On some platforms, there is currently no way for the boot loader
+	  to pass arguments to the kernel. For these, you should supply some
+	  command-line options at build time by entering them here.  At a
+	  minimum, you should specify the memory size and the root device
+	  (e.g., mem=64M root=/dev/nfs).
+
+config HEXAGON_ANGEL_TRAPS
+	bool "Use Angel Traps"
+	default n
+	---help---
+	  Enable angel debug traps (for printk's).
+
+config SMP
+	bool "Multi-Processing support"
+	---help---
+	  Enables SMP support in the kernel.  If unsure, say "Y"
+
+config NR_CPUS
+	int "Maximum number of CPUs" if SMP
+	range 2 6 if SMP
+	default "1" if !SMP
+	default "6" if SMP
+	---help---
+	  This allows you to specify the maximum number of CPUs which this
+	  kernel will support.  The maximum supported value is 6 and the
+	  minimum value which makes sense is 2.
+
+	  This is purely to save memory - each supported CPU adds
+	  approximately eight kilobytes to the kernel image.
+
+choice
+	prompt "Kernel page size"
+	default PAGE_SIZE_4KB
+	---help---
+	  Changes the default page size; use with caution.
+
+config PAGE_SIZE_4KB
+	bool "4KB"
+
+config PAGE_SIZE_16KB
+	bool "16KB"
+
+config PAGE_SIZE_64KB
+	bool "64KB"
+
+config PAGE_SIZE_256KB
+	bool "256KB"
+
+endchoice
+
+source "mm/Kconfig"
+
+source "kernel/Kconfig.hz"
+source "kernel/time/Kconfig"
+
+config GENERIC_GPIO
+	bool "Generic GPIO support"
+	default n
+
+endmenu
+
+source "init/Kconfig"
+source "drivers/Kconfig"
+source "fs/Kconfig"
+
+menu "Executable File Formats"
+source "fs/Kconfig.binfmt"
+endmenu
+
+source "net/Kconfig"
+source "security/Kconfig"
+source "crypto/Kconfig"
+source "lib/Kconfig"
+
+menu "Kernel hacking"
+source "lib/Kconfig.debug"
+endmenu
Index: linux-hexagon-kernel/arch/hexagon/kernel/asm-offsets.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/asm-offsets.c	2011-09-03 20:14:59.834981699 -0500
@@ -0,0 +1,104 @@
+/*
+ * Copyright (C) 1996 David S. Miller
+ * Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002, 2003 Ralf Baechle
+ * Copyright (C) 1999, 2000 Silicon Graphics, Inc.
+ * Kevin Kissell, kevink@mips.com and Carsten Langgaard, carstenl@mips.com
+ * Copyright (C) 2000 MIPS Technologies, Inc.
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/compat.h>
+#include <linux/types.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <linux/kbuild.h>
+#include <asm/ptrace.h>
+#include <asm/processor.h>
+
+/*  This file is used to produce asm/linkerscript constants from header
+    files typically used in c.  Specifically, it generates asm-offsets.h  */
+
+int main(void)
+{
+	COMMENT("This is a comment.");
+	/*  might get these from somewhere else.  */
+	DEFINE(_PAGE_SIZE, PAGE_SIZE);
+	DEFINE(_PAGE_SHIFT, PAGE_SHIFT);
+	BLANK();
+
+	COMMENT("Hexagon pt_regs definitions");
+	OFFSET(_PT_SYSCALL_NR, pt_regs, syscall_nr);
+	OFFSET(_PT_UGPGP, pt_regs, ugpgp);
+	OFFSET(_PT_R3130, pt_regs, r3130);
+	OFFSET(_PT_R2928, pt_regs, r2928);
+	OFFSET(_PT_R2726, pt_regs, r2726);
+	OFFSET(_PT_R2524, pt_regs, r2524);
+	OFFSET(_PT_R2322, pt_regs, r2322);
+	OFFSET(_PT_R2120, pt_regs, r2120);
+	OFFSET(_PT_R1918, pt_regs, r1918);
+	OFFSET(_PT_R1716, pt_regs, r1716);
+	OFFSET(_PT_R1514, pt_regs, r1514);
+	OFFSET(_PT_R1312, pt_regs, r1312);
+	OFFSET(_PT_R1110, pt_regs, r1110);
+	OFFSET(_PT_R0908, pt_regs, r0908);
+	OFFSET(_PT_R0706, pt_regs, r0706);
+	OFFSET(_PT_R0504, pt_regs, r0504);
+	OFFSET(_PT_R0302, pt_regs, r0302);
+	OFFSET(_PT_R0100, pt_regs, r0100);
+	OFFSET(_PT_LC0SA0, pt_regs, lc0sa0);
+	OFFSET(_PT_LC1SA1, pt_regs, lc1sa1);
+	OFFSET(_PT_M1M0, pt_regs, m1m0);
+	OFFSET(_PT_PREDSUSR, pt_regs, predsusr);
+	OFFSET(_PT_EVREC, pt_regs, hvmer);
+	OFFSET(_PT_ER_VMEL, pt_regs, hvmer.vmel);
+	OFFSET(_PT_ER_VMEST, pt_regs, hvmer.vmest);
+	OFFSET(_PT_ER_VMPSP, pt_regs, hvmer.vmpsp);
+	OFFSET(_PT_ER_VMBADVA, pt_regs, hvmer.vmbadva);
+	DEFINE(_PT_REGS_SIZE, sizeof(struct pt_regs));
+	BLANK();
+
+	COMMENT("Hexagon thread_info definitions");
+	OFFSET(_THREAD_INFO_FLAGS, thread_info, flags);
+	OFFSET(_THREAD_INFO_PT_REGS, thread_info, regs);
+	OFFSET(_THREAD_INFO_SP, thread_info, sp);
+	DEFINE(_THREAD_SIZE, THREAD_SIZE);
+	BLANK();
+
+	COMMENT("Hexagon hexagon_switch_stack definitions");
+	OFFSET(_SWITCH_R1716, hexagon_switch_stack, r1716);
+	OFFSET(_SWITCH_R1918, hexagon_switch_stack, r1918);
+	OFFSET(_SWITCH_R2120, hexagon_switch_stack, r2120);
+	OFFSET(_SWITCH_R2322, hexagon_switch_stack, r2322);
+
+	OFFSET(_SWITCH_R2524, hexagon_switch_stack, r2524);
+	OFFSET(_SWITCH_R2726, hexagon_switch_stack, r2726);
+	OFFSET(_SWITCH_FP, hexagon_switch_stack, fp);
+	OFFSET(_SWITCH_LR, hexagon_switch_stack, lr);
+	DEFINE(_SWITCH_STACK_SIZE, sizeof(struct hexagon_switch_stack));
+	BLANK();
+
+	COMMENT("Hexagon task_struct definitions");
+	OFFSET(_TASK_THREAD_INFO, task_struct, stack);
+	OFFSET(_TASK_STRUCT_THREAD, task_struct, thread);
+
+	COMMENT("Hexagon thread_struct definitions");
+	OFFSET(_THREAD_STRUCT_SWITCH_SP, thread_struct, switch_sp);
+
+	return 0;
+}
Index: linux-hexagon-kernel/arch/hexagon/kernel/vmlinux.lds.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/vmlinux.lds.S	2011-09-07 13:00:31.180647169 -0500
@@ -0,0 +1,93 @@
+/*
+ * Linker script for Hexagon kernel
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#define LOAD_OFFSET PAGE_OFFSET
+
+#include <asm-generic/vmlinux.lds.h>
+#include <asm/asm-offsets.h>	/*  Most of the kernel defines are here  */
+#include <asm/mem-layout.h>	/*  except for page_offset  */
+#include <asm/cache.h>		/*  and now we're pulling cache line size  */
+OUTPUT_ARCH(hexagon)
+ENTRY(stext)
+
+jiffies = jiffies_64;
+
+/*
+See asm-generic/vmlinux.lds.h for expansion of some of these macros.
+See asm-generic/sections.h for seemingly required labels.
+*/
+
+#define PAGE_SIZE _PAGE_SIZE
+
+/*  This LOAD_OFFSET is temporary for debugging on the simulator; it may change
+    for hypervisor pseudo-physical memory.  */
+
+
+SECTIONS
+{
+	. = PAGE_OFFSET + LOAD_ADDRESS;
+
+	__init_begin = .;
+	HEAD_TEXT_SECTION
+	INIT_TEXT_SECTION(PAGE_SIZE)
+	PERCPU_SECTION(L1_CACHE_BYTES)
+	__init_end = .;
+
+        . = ALIGN(_PAGE_SIZE);
+	_stext = .;
+	.text : AT(ADDR(.text) - LOAD_OFFSET) {
+		_text = .;
+		TEXT_TEXT
+		SCHED_TEXT
+		LOCK_TEXT
+		KPROBES_TEXT
+		*(.fixup)
+	}
+	_etext = .;
+
+	INIT_DATA_SECTION(PAGE_SIZE)
+
+	_sdata = .;
+		RW_DATA_SECTION(32,PAGE_SIZE,PAGE_SIZE)
+		RO_DATA_SECTION(PAGE_SIZE)
+	_edata = .;
+
+	EXCEPTION_TABLE(16)
+	NOTES
+
+	BSS_SECTION(_PAGE_SIZE, _PAGE_SIZE, _PAGE_SIZE)
+
+	_end = .;
+
+	/DISCARD/ : {
+		EXIT_TEXT
+		EXIT_DATA
+		EXIT_CALL
+	}
+
+	STABS_DEBUG
+	DWARF_DEBUG
+
+}
+
+
+
+
+
Index: linux-hexagon-kernel/arch/hexagon/lib/Makefile
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/lib/Makefile	2011-09-03 20:14:59.834981699 -0500
@@ -0,0 +1,4 @@
+#
+# Makefile for hexagon-specific library files.
+#
+obj-y = checksum.o io.o memcpy.o memset.o
Index: linux-hexagon-kernel/arch/hexagon/Makefile
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/Makefile	2011-09-07 13:00:30.850646667 -0500
@@ -0,0 +1,59 @@
+#  Makefile for the Hexagon arch
+
+KBUILD_DEFCONFIG = comet_defconfig
+
+# Do not use GP-relative jumps
+KBUILD_CFLAGS += -G0
+LDFLAGS_vmlinux += -G0
+
+# Do not use single-byte enums; these will overflow.
+KBUILD_CFLAGS += -fno-short-enums
+
+# Modules must use either long-calls, or use pic/plt.
+# Use long-calls for now, it's easier.  And faster.
+# CFLAGS_MODULE += -fPIC
+# LDFLAGS_MODULE += -shared
+CFLAGS_MODULE += -mlong-calls
+
+cflags-$(CONFIG_HEXAGON_ARCH_V1) += $(call cc-option,-mv1)
+cflags-$(CONFIG_HEXAGON_ARCH_V2) += $(call cc-option,-mv2)
+cflags-$(CONFIG_HEXAGON_ARCH_V3) += $(call cc-option,-mv3)
+cflags-$(CONFIG_HEXAGON_ARCH_V4) += $(call cc-option,-mv4)
+
+aflags-$(CONFIG_HEXAGON_ARCH_V1) += $(call cc-option,-mv1)
+aflags-$(CONFIG_HEXAGON_ARCH_V2) += $(call cc-option,-mv2)
+aflags-$(CONFIG_HEXAGON_ARCH_V3) += $(call cc-option,-mv3)
+aflags-$(CONFIG_HEXAGON_ARCH_V4) += $(call cc-option,-mv4)
+
+ldflags-$(CONFIG_HEXAGON_ARCH_V1) += $(call cc-option,-mv1)
+ldflags-$(CONFIG_HEXAGON_ARCH_V2) += $(call cc-option,-mv2)
+ldflags-$(CONFIG_HEXAGON_ARCH_V3) += $(call cc-option,-mv3)
+ldflags-$(CONFIG_HEXAGON_ARCH_V4) += $(call cc-option,-mv4)
+
+KBUILD_CFLAGS += $(cflags-y)
+KBUILD_AFLAGS += $(aflags-y)
+
+#  no KBUILD_LDFLAGS?
+LDFLAGS += $(ldflags-y)
+
+# Thread-info register will be r19.  This value is not configureable;
+# it is hard-coded in several files.
+TIR_NAME := r19
+KBUILD_CFLAGS += -ffixed-$(TIR_NAME) -DTHREADINFO_REG=$(TIR_NAME) -D__linux__
+KBUILD_AFLAGS += -DTHREADINFO_REG=$(TIR_NAME)
+
+LIBGCC := $(shell $(CC) $(KBUILD_CFLAGS) -print-libgcc-file-name)
+libs-y += $(LIBGCC)
+
+head-y := arch/hexagon/kernel/head.o \
+	arch/hexagon/kernel/init_task.o
+
+core-y += arch/hexagon/kernel/ \
+	arch/hexagon/mm/ \
+	arch/hexagon/lib/
+
+#	arch/hexagon/platform/common/
+#
+#core-$(CONFIG_HEXAGON_COMET)		+= arch/hexagon/platform/comet/
+#machine-$(CONFIG_HEXAGON_COMET)		:= comet
+
Index: linux-hexagon-kernel/arch/hexagon/mm/Makefile
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/mm/Makefile	2011-09-07 13:00:31.200647199 -0500
@@ -0,0 +1,7 @@
+#
+# Makefile for Hexagon memory management subsystem
+#
+
+obj-y := init.o pgalloc.o ioremap.o uaccess.o vm_fault.o cache.o
+obj-y += copy_to_user.o copy_from_user.o strnlen_user.o vm_tlb.o
+
Index: linux-hexagon-kernel/arch/hexagon/kernel/Makefile
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/Makefile	2011-09-07 13:00:31.030646941 -0500
@@ -0,0 +1,19 @@
+extra-y := head.o vmlinux.lds init_task.o
+
+obj-$(CONFIG_SMP) += smp.o topology.o
+
+obj-y += irq.o setup.o irq_cpu.o traps.o syscalltab.o signal.o time.o
+obj-y += process.o syscall.o trampoline.o reset.o ptrace.o
+obj-y += vdso.o
+
+obj-$(CONFIG_KGDB)    += kgdb.o
+obj-$(CONFIG_MODULES) += module.o hexagon_ksyms.o
+
+# Modules required to work with the Hexagon Virtual Machine
+obj-y += vm_entry.o vm_events.o vm_switch.o vm_ops.o vm_init_segtable.o
+obj-y += vm_vectors.o
+
+obj-$(CONFIG_HAS_DMA) += dma.o
+
+obj-$(CONFIG_STACKTRACE) += stacktrace.o
+

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 34/36] Hexagon: Add basic stacktrace functionality for Hexagon architecture.

[Richard Kuo]

[-- Attachment #1: stacktrace.diff --]
[-- Type: text/plain, Size: 2475 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 arch/hexagon/kernel/stacktrace.c |   66 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)

Index: linux-hexagon-kernel/arch/hexagon/kernel/stacktrace.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-hexagon-kernel/arch/hexagon/kernel/stacktrace.c	2011-09-07 13:00:31.140647106 -0500
@@ -0,0 +1,66 @@
+/*
+ * Stacktrace support for Hexagon
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#include <linux/sched.h>
+#include <linux/stacktrace.h>
+#include <linux/thread_info.h>
+#include <linux/module.h>
+
+register unsigned long current_frame_pointer asm("r30");
+
+struct stackframe {
+	unsigned long fp;
+	unsigned long rets;
+};
+
+/*
+ * Save stack-backtrace addresses into a stack_trace buffer.
+ */
+void save_stack_trace(struct stack_trace *trace)
+{
+	unsigned long low, high;
+	unsigned long fp;
+	struct stackframe *frame;
+	int skip = trace->skip;
+
+	low = (unsigned long)task_stack_page(current);
+	high = low + THREAD_SIZE;
+	fp = current_frame_pointer;
+
+	while (fp >= low && fp <= (high - sizeof(*frame))) {
+		frame = (struct stackframe *)fp;
+
+		if (skip) {
+			skip--;
+		} else {
+			trace->entries[trace->nr_entries++] = frame->rets;
+			if (trace->nr_entries >= trace->max_entries)
+				break;
+		}
+
+		/*
+		 * The next frame must be at a higher address than the
+		 * current frame.
+		 */
+		low = fp + sizeof(*frame);
+		fp = frame->fp;
+	}
+}
+EXPORT_SYMBOL_GPL(save_stack_trace);

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

[patch v3 35/36] Hexagon: Add self to MAINTAINERS

[Richard Kuo]

[-- Attachment #1: maintainer.diff --]
[-- Type: text/plain, Size: 831 bytes --]

Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

---
 MAINTAINERS |    6 ++++++
 1 file changed, 6 insertions(+)

Index: linux-hexagon-kernel/MAINTAINERS
===================================================================
--- linux-hexagon-kernel.orig/MAINTAINERS	2011-09-03 20:14:46.234943567 -0500
+++ linux-hexagon-kernel/MAINTAINERS	2011-09-07 13:00:30.630646332 -0500
@@ -5279,6 +5279,12 @@
 F:	include/linux/qnx4_fs.h
 F:	include/linux/qnxtypes.h
 
+QUALCOMM HEXAGON ARCHITECTURE
+M:	Richard Kuo <rkuo@codeaurora.org>
+L:	linux-hexagon@vger.kernel.org
+S:	Supported
+F:	arch/hexagon/
+
 RADOS BLOCK DEVICE (RBD)
 F:	include/linux/qnxtypes.h
 M:	Yehuda Sadeh <yehuda@hq.newdream.net>

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 35/36] Hexagon: Add self to MAINTAINERS

[Arnd Bergmann]

On Thursday 08 September 2011 20:09:22 Richard Kuo wrote:
> Signed-off-by: Richard Kuo <rkuo@codeaurora.org>

Acked-by: Arnd Bergmann <arnd@arndb.de>

[patch v3 36/36] Add extra arch overrides to asm-generic/checksum.h

[Richard Kuo]

[-- Attachment #1: generic-checksum.diff --]
[-- Type: text/plain, Size: 1619 bytes --]

There are plausible reasons for architectures to provide their own
versions of csum_partial_copy_nocheck and csum_tcpudp_magic. 
By protecting these, the architecture can still re-use the 
asm-generic checksum.h, instead of copying it.
 
Signed-off-by: Linas Vepstas <linas@codeaurora.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>

---
 include/asm-generic/checksum.h |    4 ++++
 1 file changed, 4 insertions(+)

Index: linux-hexagon-kernel/include/asm-generic/checksum.h
===================================================================
--- linux-hexagon-kernel.orig/include/asm-generic/checksum.h	2011-09-03 20:14:46.214943511 -0500
+++ linux-hexagon-kernel/include/asm-generic/checksum.h	2011-09-03 20:14:59.874981811 -0500
@@ -33,8 +33,10 @@
 extern __wsum csum_partial_copy_from_user(const void __user *src, void *dst,
 					int len, __wsum sum, int *csum_err);
 
+#ifndef csum_partial_copy_nocheck
 #define csum_partial_copy_nocheck(src, dst, len, sum)	\
 	csum_partial_copy((src), (dst), (len), (sum))
+#endif
 
 /*
  * This is a version of ip_compute_csum() optimized for IP headers,
@@ -63,12 +65,14 @@
 		unsigned short proto, __wsum sum);
 #endif
 
+#ifndef csum_tcpudp_magic
 static inline __sum16
 csum_tcpudp_magic(__be32 saddr, __be32 daddr, unsigned short len,
 		  unsigned short proto, __wsum sum)
 {
 	return csum_fold(csum_tcpudp_nofold(saddr, daddr, len, proto, sum));
 }
+#endif
 
 /*
  * this routine is used for miscellaneous IP-like checksums, mainly

--

Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: [patch v3 00/36] Hexagon: Add support for Qualcomm Hexagon architecture

[Arnd Bergmann]

On Thursday 08 September 2011 20:08:47 Richard Kuo wrote:
> This is the third version of the patchset to the Linux kernel to support
> the Qualcomm Hexagon architecture.  These patches have been organized like
> the last time, in a somewhat logically grouped fashion.
> 
> A summary of changes from the last patchset:
> * removed __delay
> * cleaned up checksum header to pull generic one
> * implemented vDSO so signal trampolines are no longer on stack
> * fixed cpu_idle()
> * removed init_IRQ() wrapper
> * cleaned up (most of) the wrongness in time.c
> * cleaned up smp.c to use more generic IPI handlers
> * filled in ins/outs routines in IO and other misc cleanup
> * removed redundant vunmap from ioremap
> * removed more highmem cruft
> * moved rwsem to asm-generic
> * bonus ifdefs added to generic checksum.h
> 
> Also, I have been compiling with sparse and it seems to pass.
> 
> These patches are available on Code Aurora:
> 
> https://www.codeaurora.org/patches/quic/hlk/hexagon-linux-20110908.tar.gz
> 
> Thanks again for everyone's feedback!

Ok, very good.

I haven't checked if it's already in linux-next, since you said you first
wanted to fix a few things. If it's not, please have Stephen add it now.

I wrote a small piece on the two new architectures for 3.2 in
http://lwn.net/Articles/457635/ , I hope I got the facts right. 

The ptrace interface is the only part that I'd ask you to definitely
fix before the merge window, because of the impact to the user space
ABI: We try to be strict about not breaking user-visible interfaces
for any code that has been in an official kernel release. It's of course
less important for a newly merged architecture when there are no existing
users, but I think it's best to always enforce this rule.

It would of course be very nice if you could get the board support based
on device tree probing ready as well by the same time, but I don't see
that as a show-stopper for inclusion.

One small comment about the way that you send patches: The changelog part
if each patch you send lists the changes that you have done since the first
submission. This is indeed helpful for the purpose of the review, but it's
not actually how you are supposed to use this. The body of the email
should instead start with a 'changeset comment' that describes the changes
made by the patch, followed by the 'Signed-off-by:' lines, and then after
the '---' line you should write the text that does not become part of
the permanent git history: the diffstat and any additional comments you have
for reviewers including what has changed since last time.

	Arnd