public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
To: akpm@linux-foundation.org
Cc: Ingo Molnar <mingo@elte.hu>,
	linux-kernel@vger.kernel.org, Andi Kleen <ak@muc.de>,
	Chuck Ebbert <cebbert@redhat.com>,
	Christoph Hellwig <hch@infradead.org>,
	Jeremy Fitzhardinge <jeremy@goop.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@redhat.com>,
	Rusty Russell <rusty@rustcorp.com.au>,
	"H. Peter Anvin" <hpa@zytor.com>
Subject: Re: [patch 14/24] Immediate Values - x86 Optimization (updated)
Date: Fri, 21 Dec 2007 08:19:32 -0500	[thread overview]
Message-ID: <20071221131932.GB30372@Krystal> (raw)
In-Reply-To: <476B335C.5080905@zytor.com>

x86 optimization of the immediate values which uses a movl with code patching
to set/unset the value used to populate the register used as variable source.

Changelog:
- Use text_poke_early with cr0 WP save/restore to patch the bypass. We are doing
  non atomic writes to a code region only touched by us (nobody can execute it
  since we are protected by the imv_mutex).
- Put imv_set and _imv_set in the architecture independent header.
- Use $0 instead of %2 with (0) operand.
- Add x86_64 support, ready for i386+x86_64 -> x86 merge.
- Use asm-x86/asm.h.

Ok, so the most flexible solution that I see, that should fit for both
x86 and x86_64 would be :
1 byte  :       "=q" : "a", "b", "c", or "d" register for the i386.  For
                       x86-64 it is equivalent to "r" class (for 8-bit
                       instructions that do not use upper halves).
2, 4, 8 bytes : "=r" : A register operand is allowed provided that it is in a
                       general register.

- "Redux" immediate values : no need to put a breakpoint, therefore, no
  need to know where the instruction starts. It's therefore OK to have a
  REX prefix.

- Bugfix : 8 bytes 64 bits immediate value was declared as "4 bytes" in the
  immediate structure.
- Change the immediate.c update code to support variable length opcodes.
- Vastly simplified, using a busy looping IPI with interrupts disabled.
  Does not protect against NMI nor MCE.
- Pack the __imv section. Use smallest types required for size (char).
- Use imv_* instead of immediate_*.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
CC: Andi Kleen <ak@muc.de>
CC: "H. Peter Anvin" <hpa@zytor.com>
CC: Chuck Ebbert <cebbert@redhat.com>
CC: Christoph Hellwig <hch@infradead.org>
CC: Jeremy Fitzhardinge <jeremy@goop.org>
CC: Thomas Gleixner <tglx@linutronix.de>
CC: Ingo Molnar <mingo@redhat.com>
CC: Rusty Russell <rusty@rustcorp.com.au>
---
 arch/x86/Kconfig            |    1 
 include/asm-x86/immediate.h |   77 ++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 78 insertions(+)

Index: linux-2.6-lttng.mm/include/asm-x86/immediate.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-2.6-lttng.mm/include/asm-x86/immediate.h	2007-12-20 18:55:00.000000000 -0500
@@ -0,0 +1,77 @@
+#ifndef _ASM_X86_IMMEDIATE_H
+#define _ASM_X86_IMMEDIATE_H
+
+/*
+ * Immediate values. x86 architecture optimizations.
+ *
+ * (C) Copyright 2006 Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
+ *
+ * This file is released under the GPLv2.
+ * See the file COPYING for more details.
+ */
+
+#include <asm/asm.h>
+
+/**
+ * imv_read - read immediate variable
+ * @name: immediate value name
+ *
+ * Reads the value of @name.
+ * Optimized version of the immediate.
+ * Do not use in __init and __exit functions. Use _imv_read() instead.
+ * If size is bigger than the architecture long size, fall back on a memory
+ * read.
+ *
+ * Make sure to populate the initial static 64 bits opcode with a value
+ * what will generate an instruction with 8 bytes immediate value (not the REX.W
+ * prefixed one that loads a sign extended 32 bits immediate value in a r64
+ * register).
+ */
+#define imv_read(name)							\
+	({								\
+		__typeof__(name##__imv) value;				\
+		BUILD_BUG_ON(sizeof(value) > 8);			\
+		switch (sizeof(value)) {				\
+		case 1:							\
+			asm(".section __imv,\"a\",@progbits\n\t"	\
+				_ASM_PTR "%c1, (3f)-%c2\n\t"		\
+				".byte %c2\n\t"				\
+				".previous\n\t"				\
+				"mov $0,%0\n\t"				\
+				"3:\n\t"				\
+				: "=q" (value)				\
+				: "i" (&name##__imv),			\
+				  "i" (sizeof(value)));			\
+			break;						\
+		case 2:							\
+		case 4:							\
+			asm(".section __imv,\"a\",@progbits\n\t"	\
+				_ASM_PTR "%c1, (3f)-%c2\n\t"		\
+				".byte %c2\n\t"				\
+				".previous\n\t"				\
+				"mov $0,%0\n\t"				\
+				"3:\n\t"				\
+				: "=r" (value)				\
+				: "i" (&name##__imv),			\
+				  "i" (sizeof(value)));			\
+			break;						\
+		case 8:							\
+			if (sizeof(long) < 8) {				\
+				value = name##__imv;			\
+				break;					\
+			}						\
+			asm(".section __imv,\"a\",@progbits\n\t"	\
+				_ASM_PTR "%c1, (3f)-%c2\n\t"		\
+				".byte %c2\n\t"				\
+				".previous\n\t"				\
+				"mov $0xFEFEFEFE01010101,%0\n\t" 	\
+				"3:\n\t"				\
+				: "=r" (value)				\
+				: "i" (&name##__imv),			\
+				  "i" (sizeof(value)));			\
+			break;						\
+		};							\
+		value;							\
+	})
+
+#endif /* _ASM_X86_IMMEDIATE_H */
Index: linux-2.6-lttng.mm/arch/x86/Kconfig
===================================================================
--- linux-2.6-lttng.mm.orig/arch/x86/Kconfig	2007-12-20 18:43:46.000000000 -0500
+++ linux-2.6-lttng.mm/arch/x86/Kconfig	2007-12-20 18:55:00.000000000 -0500
@@ -20,6 +20,7 @@ config X86
 	def_bool y
 	select HAVE_OPROFILE
 	select HAVE_KPROBES
+	select HAVE_IMMEDIATE
 
 config GENERIC_TIME
 	def_bool y

-- 
Mathieu Desnoyers
Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68

  parent reply	other threads:[~2007-12-21 13:24 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-12-21  1:54 [patch 00/24] Markers use immediate values, for 2.6.24-rc5-mm1 Mathieu Desnoyers
2007-12-21  1:54 ` [patch 01/24] Kprobes - use a mutex to protect the instruction pages list Mathieu Desnoyers
2007-12-21  1:54 ` [patch 02/24] Kprobes - do not use kprobes mutex in arch code Mathieu Desnoyers
2007-12-21  1:54 ` [patch 03/24] Kprobes - declare kprobe_mutex static Mathieu Desnoyers
2007-12-21  1:54 ` [patch 04/24] Add INIT_ARRAY() to kernel.h Mathieu Desnoyers
2007-12-21  1:54 ` [patch 05/24] Text Edit Lock - Architecture Independent Code Mathieu Desnoyers
2007-12-21  5:18   ` zhangxiliang
2007-12-21  6:01     ` zhangxiliang
2007-12-21 13:46     ` Mathieu Desnoyers
2007-12-21  1:54 ` [patch 06/24] Text Edit Lock - Alternative code for x86 Mathieu Desnoyers
2007-12-21  1:54 ` [patch 07/24] Text Edit Lock - kprobes architecture independent support Mathieu Desnoyers
2007-12-21  1:54 ` [patch 08/24] Text Edit Lock - kprobes x86_32 Mathieu Desnoyers
2007-12-21  1:54 ` [patch 09/24] Text Edit Lock - kprobes x86_64 Mathieu Desnoyers
2007-12-21  1:54 ` [patch 10/24] Text Edit Lock - x86_32 standardize debug rodata Mathieu Desnoyers
2007-12-21  1:54 ` [patch 11/24] Text Edit Lock - x86_64 " Mathieu Desnoyers
2007-12-21  1:54 ` [patch 12/24] Immediate Values - Architecture Independent Code Mathieu Desnoyers
2007-12-21  1:54 ` [patch 13/24] Immediate Values - Kconfig menu in EMBEDDED Mathieu Desnoyers
2007-12-21  1:54 ` [patch 14/24] Immediate Values - x86 Optimization Mathieu Desnoyers
2007-12-21  2:56   ` H. Peter Anvin
2007-12-21  3:19     ` Mathieu Desnoyers
2007-12-21  3:30       ` H. Peter Anvin
2007-12-21 13:16         ` [patch 14/24] Immediate Values - x86 Optimization (updated) Mathieu Desnoyers
2007-12-21 13:19         ` Mathieu Desnoyers [this message]
2007-12-21  1:54 ` [patch 15/24] Add text_poke and sync_core to powerpc Mathieu Desnoyers
2007-12-21  1:54 ` [patch 16/24] Immediate Values - Powerpc Optimization Mathieu Desnoyers
2007-12-21  1:54 ` [patch 17/24] Immediate Values - Documentation Mathieu Desnoyers
2007-12-21  1:54 ` [patch 18/24] Scheduler Profiling - Use Immediate Values Mathieu Desnoyers
2007-12-21  1:54 ` [patch 19/24] Immediate Values - Move Kprobes x86 restore_interrupt to kdebug.h Mathieu Desnoyers
2007-12-21  1:54 ` [patch 20/24] Add __discard section to x86 Mathieu Desnoyers
2007-12-21  1:54 ` [patch 21/24] Immediate Values - x86 Optimization NMI and MCE support Mathieu Desnoyers
2007-12-21 13:25   ` [patch 21/24] Immediate Values - x86 Optimization NMI and MCE support (updated) Mathieu Desnoyers
2007-12-21  1:55 ` [patch 22/24] Immediate Values - Powerpc Optimization NMI MCE support Mathieu Desnoyers
2007-12-21  1:55 ` [patch 23/24] Immediate Values Use Arch NMI and MCE Support Mathieu Desnoyers
2007-12-21  1:55 ` [patch 24/24] Linux Kernel Markers - Use Immediate Values Mathieu Desnoyers

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20071221131932.GB30372@Krystal \
    --to=mathieu.desnoyers@polymtl.ca \
    --cc=ak@muc.de \
    --cc=akpm@linux-foundation.org \
    --cc=cebbert@redhat.com \
    --cc=hch@infradead.org \
    --cc=hpa@zytor.com \
    --cc=jeremy@goop.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=mingo@redhat.com \
    --cc=rusty@rustcorp.com.au \
    --cc=tglx@linutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox