[Qemu-devel] [RFC v1 00/23] Add RISC-V TCG backend support

[Qemu-devel] [RFC v1 00/23] Add RISC-V TCG backend support

[Alistair Francis]

This patch set adds RISC-V backend support to QEMU. This is based on
Michael Clark's original work with some patches ontop.

This has been slightly tested and can run other architecture softmmu
code for a number of instructions but eventually QEMU will either
seg fault or generate an illigal instruction (depending on the guest
architecture).

I haven't tested linux user support at all yet. I think Michael had that
working reliably though and hopefully my changes haven't broken it. I'll
test both a lot more before I send a full patchset.

My hope of submitting an RFC is that some extra eyes on the code might
help catch what is wrong. Comparing the guest CPU state to a working
version hasn't given any hints as the states match, even up until the
generated code segfaults.

This branch can be found here:
https://github.com/alistair23/qemu/tree/mainline/alistair/tcg-backend-upstream.next

The working version with Michael's orignal patch and work ontop can be
found here:
https://github.com/alistair23/qemu/tree/mainline/alistair/tcg-backend.next


Alistair Francis (23):
  elf.h: Add the RISCV ELF magic numbers
  linux-user: Add host dependency for RISC-V 32-bit
  linux-user: Add host dependency for RISC-V 64-bit
  exec: Add RISC-V GCC poison macro
  riscv: Add the tcg-target header file
  riscv: Add the tcg target registers
  riscv: tcg-target: Regiser the JIT
  riscv: tcg-target: Add support for the constraints
  riscv: tcg-target: Add the immediate encoders
  riscv: tcg-target: Add the instruction emitters
  riscv: tcg-target: Add the relocation functions
  riscv: tcg-target: Add the mov and movi instruction
  riscv: tcg-target: Add the extract instructions
  riscv: tcg-target: Add the out load and store instructions
  riscv: tcg-target: Add branch and jump instructions
  riscv: tcg-target: Add slowpath load and store instructions
  riscv: tcg-target: Add direct load and store instructions
  riscv: tcg-target: Add the out op decoder
  riscv: tcg-target: Add the prologue generation
  riscv: tcg-target: Add the target init code
  tcg: Add RISC-V cpu signal handler
  dias: Add RISC-V support
  configure: Add support for building RISC-V host

 accel/tcg/user-exec.c             |   48 +
 configure                         |   12 +-
 disas.c                           |   10 +-
 include/elf.h                     |   55 +
 include/exec/poison.h             |    1 +
 linux-user/host/riscv32/hostdep.h |   11 +
 linux-user/host/riscv64/hostdep.h |   11 +
 tcg/riscv/tcg-target.h            |  173 +++
 tcg/riscv/tcg-target.inc.c        | 1728 +++++++++++++++++++++++++++++
 9 files changed, 2045 insertions(+), 4 deletions(-)
 create mode 100644 linux-user/host/riscv32/hostdep.h
 create mode 100644 linux-user/host/riscv64/hostdep.h
 create mode 100644 tcg/riscv/tcg-target.h
 create mode 100644 tcg/riscv/tcg-target.inc.c

-- 
2.19.1

[Qemu-devel] [RFC v1 01/23] elf.h: Add the RISCV ELF magic numbers

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 include/elf.h | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 55 insertions(+)

diff --git a/include/elf.h b/include/elf.h
index c151164b63..0ac7911b7b 100644
--- a/include/elf.h
+++ b/include/elf.h
@@ -1338,6 +1338,61 @@ typedef struct {
 #define R_IA64_DTPREL64LSB	0xb7	/* @dtprel(sym + add), data8 LSB */
 #define R_IA64_LTOFF_DTPREL22	0xba	/* @ltoff(@dtprel(s+a)), imm22 */
 
+/* RISC-V relocations.  */
+#define R_RISCV_NONE          0
+#define R_RISCV_32            1
+#define R_RISCV_64            2
+#define R_RISCV_RELATIVE      3
+#define R_RISCV_COPY          4
+#define R_RISCV_JUMP_SLOT     5
+#define R_RISCV_TLS_DTPMOD32  6
+#define R_RISCV_TLS_DTPMOD64  7
+#define R_RISCV_TLS_DTPREL32  8
+#define R_RISCV_TLS_DTPREL64  9
+#define R_RISCV_TLS_TPREL32   10
+#define R_RISCV_TLS_TPREL64   11
+#define R_RISCV_BRANCH        16
+#define R_RISCV_JAL           17
+#define R_RISCV_CALL          18
+#define R_RISCV_CALL_PLT      19
+#define R_RISCV_GOT_HI20      20
+#define R_RISCV_TLS_GOT_HI20  21
+#define R_RISCV_TLS_GD_HI20   22
+#define R_RISCV_PCREL_HI20    23
+#define R_RISCV_PCREL_LO12_I  24
+#define R_RISCV_PCREL_LO12_S  25
+#define R_RISCV_HI20          26
+#define R_RISCV_LO12_I        27
+#define R_RISCV_LO12_S        28
+#define R_RISCV_TPREL_HI20    29
+#define R_RISCV_TPREL_LO12_I  30
+#define R_RISCV_TPREL_LO12_S  31
+#define R_RISCV_TPREL_ADD     32
+#define R_RISCV_ADD8          33
+#define R_RISCV_ADD16         34
+#define R_RISCV_ADD32         35
+#define R_RISCV_ADD64         36
+#define R_RISCV_SUB8          37
+#define R_RISCV_SUB16         38
+#define R_RISCV_SUB32         39
+#define R_RISCV_SUB64         40
+#define R_RISCV_GNU_VTINHERIT 41
+#define R_RISCV_GNU_VTENTRY   42
+#define R_RISCV_ALIGN         43
+#define R_RISCV_RVC_BRANCH    44
+#define R_RISCV_RVC_JUMP      45
+#define R_RISCV_RVC_LUI       46
+#define R_RISCV_GPREL_I       47
+#define R_RISCV_GPREL_S       48
+#define R_RISCV_TPREL_I       49
+#define R_RISCV_TPREL_S       50
+#define R_RISCV_RELAX         51
+#define R_RISCV_SUB6          52
+#define R_RISCV_SET6          53
+#define R_RISCV_SET8          54
+#define R_RISCV_SET16         55
+#define R_RISCV_SET32         56
+
 typedef struct elf32_rel {
   Elf32_Addr	r_offset;
   Elf32_Word	r_info;
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 01/23] elf.h: Add the RISCV ELF magic numbers

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  include/elf.h | 55 +++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 55 insertions(+)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

[Qemu-devel] [RFC v1 02/23] linux-user: Add host dependency for RISC-V 32-bit

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 linux-user/host/riscv32/hostdep.h | 11 +++++++++++
 1 file changed, 11 insertions(+)
 create mode 100644 linux-user/host/riscv32/hostdep.h

diff --git a/linux-user/host/riscv32/hostdep.h b/linux-user/host/riscv32/hostdep.h
new file mode 100644
index 0000000000..adf9edbf2d
--- /dev/null
+++ b/linux-user/host/riscv32/hostdep.h
@@ -0,0 +1,11 @@
+/*
+ * hostdep.h : things which are dependent on the host architecture
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef RISCV32_HOSTDEP_H
+#define RISCV32_HOSTDEP_H
+
+#endif
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 02/23] linux-user: Add host dependency for RISC-V 32-bit

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  linux-user/host/riscv32/hostdep.h | 11 +++++++++++
>  1 file changed, 11 insertions(+)
>  create mode 100644 linux-user/host/riscv32/hostdep.h

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

Re: [Qemu-devel] [RFC v1 02/23] linux-user: Add host dependency for RISC-V 32-bit

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  linux-user/host/riscv32/hostdep.h | 11 +++++++++++
>  1 file changed, 11 insertions(+)
>  create mode 100644 linux-user/host/riscv32/hostdep.h

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

[Qemu-devel] [RFC v1 03/23] linux-user: Add host dependency for RISC-V 64-bit

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 linux-user/host/riscv64/hostdep.h | 11 +++++++++++
 1 file changed, 11 insertions(+)
 create mode 100644 linux-user/host/riscv64/hostdep.h

diff --git a/linux-user/host/riscv64/hostdep.h b/linux-user/host/riscv64/hostdep.h
new file mode 100644
index 0000000000..28467ba00b
--- /dev/null
+++ b/linux-user/host/riscv64/hostdep.h
@@ -0,0 +1,11 @@
+/*
+ * hostdep.h : things which are dependent on the host architecture
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef RISCV64_HOSTDEP_H
+#define RISCV64_HOSTDEP_H
+
+#endif
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 03/23] linux-user: Add host dependency for RISC-V 64-bit

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  linux-user/host/riscv64/hostdep.h | 11 +++++++++++
>  1 file changed, 11 insertions(+)
>  create mode 100644 linux-user/host/riscv64/hostdep.h


Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

[Qemu-devel] [RFC v1 04/23] exec: Add RISC-V GCC poison macro

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 include/exec/poison.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/include/exec/poison.h b/include/exec/poison.h
index 32d53789f8..ecdc83c147 100644
--- a/include/exec/poison.h
+++ b/include/exec/poison.h
@@ -79,6 +79,7 @@
 #pragma GCC poison CONFIG_MOXIE_DIS
 #pragma GCC poison CONFIG_NIOS2_DIS
 #pragma GCC poison CONFIG_PPC_DIS
+#pragma GCC poison CONFIG_RISCV_DIS
 #pragma GCC poison CONFIG_S390_DIS
 #pragma GCC poison CONFIG_SH4_DIS
 #pragma GCC poison CONFIG_SPARC_DIS
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 04/23] exec: Add RISC-V GCC poison macro

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  include/exec/poison.h | 1 +
>  1 file changed, 1 insertion(+)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

[Qemu-devel] [RFC v1 05/23] riscv: Add the tcg-target header file

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.h | 173 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 173 insertions(+)
 create mode 100644 tcg/riscv/tcg-target.h

diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
new file mode 100644
index 0000000000..53ab485983
--- /dev/null
+++ b/tcg/riscv/tcg-target.h
@@ -0,0 +1,173 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2018 SiFive, Inc
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef RISCV_TCG_TARGET_H
+#define RISCV_TCG_TARGET_H
+
+#if __riscv_xlen == 32
+# define TCG_TARGET_REG_BITS 32
+#elif __riscv_xlen == 64
+# define TCG_TARGET_REG_BITS 64
+#endif
+
+#define TCG_TARGET_INSN_UNIT_SIZE 4
+#define TCG_TARGET_TLB_DISPLACEMENT_BITS 20
+#define TCG_TARGET_NB_REGS 32
+
+typedef enum {
+    TCG_REG_ZERO,
+    TCG_REG_RA,
+    TCG_REG_SP,
+    TCG_REG_GP,
+    TCG_REG_TP,
+    TCG_REG_T0,
+    TCG_REG_T1,
+    TCG_REG_T2,
+    TCG_REG_S0,
+    TCG_REG_S1,
+    TCG_REG_A0,
+    TCG_REG_A1,
+    TCG_REG_A2,
+    TCG_REG_A3,
+    TCG_REG_A4,
+    TCG_REG_A5,
+    TCG_REG_A6,
+    TCG_REG_A7,
+    TCG_REG_S2,
+    TCG_REG_S3,
+    TCG_REG_S4,
+    TCG_REG_S5,
+    TCG_REG_S6,
+    TCG_REG_S7,
+    TCG_REG_S8,
+    TCG_REG_S9,
+    TCG_REG_S10,
+    TCG_REG_S11,
+    TCG_REG_T3,
+    TCG_REG_T4,
+    TCG_REG_T5,
+    TCG_REG_T6,
+
+    /* aliases */
+    TCG_AREG0          = TCG_REG_S0,
+    TCG_GUEST_BASE_REG = TCG_REG_S1,
+    TCG_REG_TMP0       = TCG_REG_T6,
+    TCG_REG_TMP1       = TCG_REG_T5,
+    TCG_REG_TMP2       = TCG_REG_T4,
+    TCG_REG_L0         = TCG_REG_A6,
+    TCG_REG_L1         = TCG_REG_A7,
+} TCGReg;
+
+/* used for function call generation */
+#define TCG_REG_CALL_STACK              TCG_REG_SP
+#define TCG_TARGET_STACK_ALIGN          16
+#define TCG_TARGET_CALL_ALIGN_ARGS      1
+#define TCG_TARGET_CALL_STACK_OFFSET    0
+
+/* optional instructions */
+#define TCG_TARGET_HAS_goto_ptr         1
+#define TCG_TARGET_HAS_movcond_i32      0
+#define TCG_TARGET_HAS_div_i32          1
+#define TCG_TARGET_HAS_rem_i32          1
+#define TCG_TARGET_HAS_div2_i32         0
+#define TCG_TARGET_HAS_rot_i32          0
+#define TCG_TARGET_HAS_deposit_i32      0
+#define TCG_TARGET_HAS_extract_i32      0
+#define TCG_TARGET_HAS_sextract_i32     0
+#define TCG_TARGET_HAS_add2_i32         0
+#define TCG_TARGET_HAS_sub2_i32         0
+#define TCG_TARGET_HAS_mulu2_i32        0
+#define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        (TCG_TARGET_REG_BITS == 32)
+#define TCG_TARGET_HAS_mulsh_i32        (TCG_TARGET_REG_BITS == 32)
+#define TCG_TARGET_HAS_ext8s_i32        1
+#define TCG_TARGET_HAS_ext16s_i32       1
+#define TCG_TARGET_HAS_ext8u_i32        1
+#define TCG_TARGET_HAS_ext16u_i32       1
+#define TCG_TARGET_HAS_bswap16_i32      0
+#define TCG_TARGET_HAS_bswap32_i32      0
+#define TCG_TARGET_HAS_not_i32          1
+#define TCG_TARGET_HAS_neg_i32          1
+#define TCG_TARGET_HAS_andc_i32         0
+#define TCG_TARGET_HAS_orc_i32          0
+#define TCG_TARGET_HAS_eqv_i32          0
+#define TCG_TARGET_HAS_nand_i32         0
+#define TCG_TARGET_HAS_nor_i32          0
+#define TCG_TARGET_HAS_clz_i32          0
+#define TCG_TARGET_HAS_ctz_i32          0
+#define TCG_TARGET_HAS_ctpop_i32        0
+#define TCG_TARGET_HAS_direct_jump      1
+
+#if TCG_TARGET_REG_BITS == 64
+#define TCG_TARGET_HAS_movcond_i64      0
+#define TCG_TARGET_HAS_div_i64          1
+#define TCG_TARGET_HAS_rem_i64          1
+#define TCG_TARGET_HAS_div2_i64         0
+#define TCG_TARGET_HAS_rot_i64          0
+#define TCG_TARGET_HAS_deposit_i64      0
+#define TCG_TARGET_HAS_extract_i64      0
+#define TCG_TARGET_HAS_sextract_i64     0
+#define TCG_TARGET_HAS_extrl_i64_i32    0
+#define TCG_TARGET_HAS_extrh_i64_i32    0
+#define TCG_TARGET_HAS_ext8s_i64        1
+#define TCG_TARGET_HAS_ext16s_i64       1
+#define TCG_TARGET_HAS_ext32s_i64       1
+#define TCG_TARGET_HAS_ext8u_i64        1
+#define TCG_TARGET_HAS_ext16u_i64       1
+#define TCG_TARGET_HAS_ext32u_i64       1
+#define TCG_TARGET_HAS_bswap16_i64      0
+#define TCG_TARGET_HAS_bswap32_i64      0
+#define TCG_TARGET_HAS_bswap64_i64      0
+#define TCG_TARGET_HAS_not_i64          1
+#define TCG_TARGET_HAS_neg_i64          1
+#define TCG_TARGET_HAS_andc_i64         0
+#define TCG_TARGET_HAS_orc_i64          0
+#define TCG_TARGET_HAS_eqv_i64          0
+#define TCG_TARGET_HAS_nand_i64         0
+#define TCG_TARGET_HAS_nor_i64          0
+#define TCG_TARGET_HAS_clz_i64          0
+#define TCG_TARGET_HAS_ctz_i64          0
+#define TCG_TARGET_HAS_ctpop_i64        0
+#define TCG_TARGET_HAS_add2_i64         0
+#define TCG_TARGET_HAS_sub2_i64         0
+#define TCG_TARGET_HAS_mulu2_i64        0
+#define TCG_TARGET_HAS_muls2_i64        0
+#define TCG_TARGET_HAS_muluh_i64        1
+#define TCG_TARGET_HAS_mulsh_i64        1
+#endif
+
+static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+{
+    __builtin___clear_cache((char *)start, (char *)stop);
+}
+
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+
+#define TCG_TARGET_DEFAULT_MO (0)
+
+#ifdef CONFIG_SOFTMMU
+#define TCG_TARGET_NEED_LDST_LABELS
+#endif
+
+#endif
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 05/23] riscv: Add the tcg-target header file

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> +#define TCG_TARGET_HAS_add2_i32         0
> +#define TCG_TARGET_HAS_sub2_i32         0

You're not allowed to not implement this for riscv-32,
along with brcond2 and setcond2.

> +#define TCG_TARGET_HAS_movcond_i32      0

The architecture manual suggests that implementations recognize short
branch-over and convert internally to predication.  Do any do that?

Even if they don't, you might implement it that way anyway, as the code size
savings are large vs the inline fallback (6 insns).  But you can definitely put
that off for a later patch.

> +#define TCG_TARGET_HAS_ext16u_i32       1

Really?  I thought this required two insns...  Or do you expand to shifts while
the default is x & 0xffff, which would take 3 insns?  I guess I'll find out as
I keep reading.  ;-)


r~

Re: [Qemu-devel] [RFC v1 05/23] riscv: Add the tcg-target header file

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> +#define TCG_TARGET_HAS_extrl_i64_i32    0
> +#define TCG_TARGET_HAS_extrh_i64_i32    0

You need these two for rv64 to preserve the invariant that 32-bit values are
sign-extended in a 64-bit register.  Otherwise _i32 comparisons won't work
reliably.

The first is yet another alias for exts32_i64; the second is an arithmetic
right shift of 32.  Their existence changes how code is emitted in tcg-op.c.


r~

[Qemu-devel] [RFC v1 06/23] riscv: Add the tcg target registers

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 120 +++++++++++++++++++++++++++++++++++++
 1 file changed, 120 insertions(+)
 create mode 100644 tcg/riscv/tcg-target.inc.c

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
new file mode 100644
index 0000000000..e5eca4bbbe
--- /dev/null
+++ b/tcg/riscv/tcg-target.inc.c
@@ -0,0 +1,120 @@
+/*
+ * Tiny Code Generator for QEMU
+ *
+ * Copyright (c) 2018 SiFive, Inc
+ * Copyright (c) 2008-2009 Arnaud Patard <arnaud.patard@rtp-net.org>
+ * Copyright (c) 2009 Aurelien Jarno <aurelien@aurel32.net>
+ * Copyright (c) 2008 Fabrice Bellard
+ *
+ * Based on i386/tcg-target.c and mips/tcg-target.c
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifdef CONFIG_DEBUG_TCG
+static const char * const tcg_target_reg_names[TCG_TARGET_NB_REGS] = {
+    "zero",
+    "ra",
+    "sp",
+    "gp",
+    "tp",
+    "t0",
+    "t1",
+    "t2",
+    "s0",
+    "s1",
+    "a0",
+    "a1",
+    "a2",
+    "a3",
+    "a4",
+    "a5",
+    "a6",
+    "a7",
+    "s2",
+    "s3",
+    "s4",
+    "s5",
+    "s6",
+    "s7",
+    "s8",
+    "s9",
+    "s10",
+    "s11",
+    "t3",
+    "t4",
+    "t5",
+    "t6"
+};
+#endif
+
+static const int tcg_target_reg_alloc_order[] = {
+    /* Call saved registers */
+    /* TCG_REG_S0 reservered for TCG_AREG0 */
+    TCG_REG_S1,
+    TCG_REG_S2,
+    TCG_REG_S3,
+    TCG_REG_S4,
+    TCG_REG_S5,
+    TCG_REG_S6,
+    TCG_REG_S7,
+    TCG_REG_S8,
+    TCG_REG_S9,
+    TCG_REG_S10,
+    TCG_REG_S11,
+
+    /* Call clobbered registers */
+    TCG_REG_T0,
+    TCG_REG_T1,
+    TCG_REG_T2,
+    TCG_REG_T3,
+    TCG_REG_T4,
+    TCG_REG_T5,
+    TCG_REG_T6,
+
+    /* Argument registers */
+    TCG_REG_A0,
+    TCG_REG_A1,
+    TCG_REG_A2,
+    TCG_REG_A3,
+    TCG_REG_A4,
+    TCG_REG_A5,
+    TCG_REG_A6,
+    TCG_REG_A7,
+};
+
+static const int tcg_target_call_iarg_regs[] = {
+    TCG_REG_A0,
+    TCG_REG_A1,
+    TCG_REG_A2,
+    TCG_REG_A3,
+    TCG_REG_A4,
+    TCG_REG_A5,
+    TCG_REG_A6,
+    TCG_REG_A7,
+};
+
+static const int tcg_target_call_oarg_regs[] = {
+    TCG_REG_A0,
+    TCG_REG_A1,
+};
+
+#define TCG_CT_CONST_ZERO  0x100
+#define TCG_CT_CONST_S12   0x200
+#define TCG_CT_CONST_N12   0x400
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 06/23] riscv: Add the tcg target registers

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> +
> +#define TCG_CT_CONST_ZERO  0x100
> +#define TCG_CT_CONST_S12   0x200
> +#define TCG_CT_CONST_N12   0x400

Logically this would go with patch 8.  But,

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

[Qemu-devel] [RFC v1 07/23] riscv: tcg-target: Regiser the JIT

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 45 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 45 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index e5eca4bbbe..a9c57493a0 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -118,3 +118,48 @@ static const int tcg_target_call_oarg_regs[] = {
 #define TCG_CT_CONST_ZERO  0x100
 #define TCG_CT_CONST_S12   0x200
 #define TCG_CT_CONST_N12   0x400
+
+typedef struct {
+    DebugFrameHeader h;
+    uint8_t fde_def_cfa[4];
+    uint8_t fde_reg_ofs[ARRAY_SIZE(tcg_target_callee_save_regs) * 2];
+} DebugFrame;
+
+#define ELF_HOST_MACHINE EM_RISCV
+
+static const DebugFrame debug_frame = {
+    .h.cie.len = sizeof(DebugFrameCIE) - 4, /* length after .len member */
+    .h.cie.id = -1,
+    .h.cie.version = 1,
+    .h.cie.code_align = 1,
+    .h.cie.data_align = -(TCG_TARGET_REG_BITS / 8) & 0x7f, /* sleb128 */
+    .h.cie.return_column = TCG_REG_RA,
+
+    /* Total FDE size does not include the "len" member.  */
+    .h.fde.len = sizeof(DebugFrame) - offsetof(DebugFrame, h.fde.cie_offset),
+
+    .fde_def_cfa = {
+        12, TCG_REG_SP,                 /* DW_CFA_def_cfa sp, ... */
+        (FRAME_SIZE & 0x7f) | 0x80,     /* ... uleb128 FRAME_SIZE */
+        (FRAME_SIZE >> 7)
+    },
+    .fde_reg_ofs = {
+        0x80 + 9,  12,                  /* DW_CFA_offset, s1,  -96 */
+        0x80 + 18, 11,                  /* DW_CFA_offset, s2,  -88 */
+        0x80 + 19, 10,                  /* DW_CFA_offset, s3,  -80 */
+        0x80 + 20, 9,                   /* DW_CFA_offset, s4,  -72 */
+        0x80 + 21, 8,                   /* DW_CFA_offset, s5,  -64 */
+        0x80 + 22, 7,                   /* DW_CFA_offset, s6,  -56 */
+        0x80 + 23, 6,                   /* DW_CFA_offset, s7,  -48 */
+        0x80 + 24, 5,                   /* DW_CFA_offset, s8,  -40 */
+        0x80 + 25, 4,                   /* DW_CFA_offset, s9,  -32 */
+        0x80 + 26, 3,                   /* DW_CFA_offset, s10, -24 */
+        0x80 + 27, 2,                   /* DW_CFA_offset, s11, -16 */
+        0x80 + 1 , 1,                   /* DW_CFA_offset, ra,  -8 */
+    }
+};
+
+void tcg_register_jit(void *buf, size_t buf_size)
+{
+    tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
+}
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 07/23] riscv: tcg-target: Regiser the JIT

[Richard Henderson]

On 11/15/18 11:34 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  tcg/riscv/tcg-target.inc.c | 45 ++++++++++++++++++++++++++++++++++++++
>  1 file changed, 45 insertions(+)

I think this whole patch should be merged with the prologue.  They are
interrelated and it's hard to verify this without the other.

Also, FRAME_SIZE isn't defined yet, so this patch isn't even self-contained.


r~

[Qemu-devel] [RFC v1 08/23] riscv: tcg-target: Add support for the constraints

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 139 +++++++++++++++++++++++++++++++++++++
 1 file changed, 139 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index a9c57493a0..e585740870 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -119,6 +119,145 @@ static const int tcg_target_call_oarg_regs[] = {
 #define TCG_CT_CONST_S12   0x200
 #define TCG_CT_CONST_N12   0x400
 
+/* parse target specific constraints */
+static const char *target_parse_constraint(TCGArgConstraint *ct,
+                                           const char *ct_str, TCGType type)
+{
+    switch (*ct_str++) {
+    case 'r':
+        ct->ct |= TCG_CT_REG;
+        ct->u.regs = 0xffffffff;
+        break;
+    case 'L':
+        /* qemu_ld/qemu_st constraint */
+        ct->ct |= TCG_CT_REG;
+        ct->u.regs = 0xffffffff;
+        /* qemu_ld/qemu_st uses TCG_REG_TMP0 */
+#if defined(CONFIG_SOFTMMU)
+        /* tcg_out_tlb_load uses TCG_REG_TMP0/TMP1 and TCG_REG_L0/L1 */
+        /* tcg_regset_reset_reg(ct->u.regs, TCG_REG_TMP0); */
+        /* tcg_regset_reset_reg(ct->u.regs, TCG_REG_TMP1); */
+        tcg_regset_reset_reg(ct->u.regs, TCG_REG_TMP2);
+
+        tcg_regset_reset_reg(ct->u.regs, tcg_target_call_iarg_regs[0]);
+        tcg_regset_reset_reg(ct->u.regs, tcg_target_call_iarg_regs[1]);
+        tcg_regset_reset_reg(ct->u.regs, tcg_target_call_iarg_regs[2]);
+        tcg_regset_reset_reg(ct->u.regs, tcg_target_call_iarg_regs[3]);
+        tcg_regset_reset_reg(ct->u.regs, tcg_target_call_iarg_regs[4]);
+#endif
+        break;
+    case 'I':
+        ct->ct |= TCG_CT_CONST_S12;
+        break;
+    case 'N':
+        ct->ct |= TCG_CT_CONST_N12;
+        break;
+    case 'Z':
+        /* we can use a zero immediate as a zero register argument. */
+        ct->ct |= TCG_CT_CONST_ZERO;
+        break;
+    default:
+        return NULL;
+    }
+    return ct_str;
+}
+
+/* test if a constant matches the constraint */
+static int tcg_target_const_match(tcg_target_long val, TCGType type,
+                                  const TCGArgConstraint *arg_ct)
+{
+    int ct = arg_ct->ct;
+    if (ct & TCG_CT_CONST) {
+        return 1;
+    }
+    if ((ct & TCG_CT_CONST_ZERO) && val == 0) {
+        return 1;
+    }
+    if ((ct & TCG_CT_CONST_S12) && val >= -2048 && val <= 2047) {
+        return 1;
+    }
+    if ((ct & TCG_CT_CONST_N12) && val >= -2047 && val <= 2048) {
+        return 1;
+    }
+    return 0;
+}
+
+/*
+ * RISC-V Base ISA opcodes (IM)
+ */
+
+typedef enum {
+    OPC_ADD = 0x33,
+    OPC_ADDI = 0x13,
+    OPC_ADDIW = 0x1b,
+    OPC_ADDW = 0x3b,
+    OPC_AND = 0x7033,
+    OPC_ANDI = 0x7013,
+    OPC_AUIPC = 0x17,
+    OPC_BEQ = 0x63,
+    OPC_BGE = 0x5063,
+    OPC_BGEU = 0x7063,
+    OPC_BLT = 0x4063,
+    OPC_BLTU = 0x6063,
+    OPC_BNE = 0x1063,
+    OPC_DIV = 0x2004033,
+    OPC_DIVU = 0x2005033,
+    OPC_DIVUW = 0x200503b,
+    OPC_DIVW = 0x200403b,
+    OPC_JAL = 0x6f,
+    OPC_JALR = 0x67,
+    OPC_LB = 0x3,
+    OPC_LBU = 0x4003,
+    OPC_LD = 0x3003,
+    OPC_LH = 0x1003,
+    OPC_LHU = 0x5003,
+    OPC_LUI = 0x37,
+    OPC_LW = 0x2003,
+    OPC_LWU = 0x6003,
+    OPC_MUL = 0x2000033,
+    OPC_MULH = 0x2001033,
+    OPC_MULHSU = 0x2002033,
+    OPC_MULHU = 0x2003033,
+    OPC_MULW = 0x200003b,
+    OPC_OR = 0x6033,
+    OPC_ORI = 0x6013,
+    OPC_REM = 0x2006033,
+    OPC_REMU = 0x2007033,
+    OPC_REMUW = 0x200703b,
+    OPC_REMW = 0x200603b,
+    OPC_SB = 0x23,
+    OPC_SD = 0x3023,
+    OPC_SH = 0x1023,
+    OPC_SLL = 0x1033,
+    OPC_SLLI = 0x1013,
+    OPC_SLLIW = 0x101b,
+    OPC_SLLW = 0x103b,
+    OPC_SLT = 0x2033,
+    OPC_SLTI = 0x2013,
+    OPC_SLTIU = 0x3013,
+    OPC_SLTU = 0x3033,
+    OPC_SRA = 0x40005033,
+    OPC_SRAI = 0x40005013,
+    OPC_SRAIW = 0x4000501b,
+    OPC_SRAW = 0x4000503b,
+    OPC_SRL = 0x5033,
+    OPC_SRLI = 0x5013,
+    OPC_SRLIW = 0x501b,
+    OPC_SRLW = 0x503b,
+    OPC_SUB = 0x40000033,
+    OPC_SUBW = 0x4000003b,
+    OPC_SW = 0x2023,
+    OPC_XOR = 0x4033,
+    OPC_XORI = 0x4013,
+    OPC_FENCE_RW_RW = 0x0330000f,
+    OPC_FENCE_R_R = 0x0220000f,
+    OPC_FENCE_W_R = 0x0120000f,
+    OPC_FENCE_R_W = 0x0210000f,
+    OPC_FENCE_W_W = 0x0110000f,
+    OPC_FENCE_R_RW = 0x0230000f,
+    OPC_FENCE_RW_W = 0x0310000f,
+} RISCVInsn;
+
 typedef struct {
     DebugFrameHeader h;
     uint8_t fde_def_cfa[4];
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 08/23] riscv: tcg-target: Add support for the constraints

[Richard Henderson]

On 11/15/18 11:35 PM, Alistair Francis wrote:
> +    case 'L':
> +        /* qemu_ld/qemu_st constraint */
> +        ct->ct |= TCG_CT_REG;
> +        ct->u.regs = 0xffffffff;
> +        /* qemu_ld/qemu_st uses TCG_REG_TMP0 */
> +#if defined(CONFIG_SOFTMMU)
> +        /* tcg_out_tlb_load uses TCG_REG_TMP0/TMP1 and TCG_REG_L0/L1 */
> +        /* tcg_regset_reset_reg(ct->u.regs, TCG_REG_TMP0); */
> +        /* tcg_regset_reset_reg(ct->u.regs, TCG_REG_TMP1); */
> +        tcg_regset_reset_reg(ct->u.regs, TCG_REG_TMP2);

Normally TMP registers are reserved and do not need to be mentioned in
constraints at all.

> +    if ((ct & TCG_CT_CONST_S12) && val >= -2048 && val <= 2047) {
> +        return 1;
> +    }
> +    if ((ct & TCG_CT_CONST_N12) && val >= -2047 && val <= 2048) {
> +        return 1;
> +    }

Clearer as val == sextracttl(val, 0, 12)
and -val == sextracttl(-val, 0, 12)?


r~

[Qemu-devel] [RFC v1 09/23] riscv: tcg-target: Add the immediate encoders

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 81 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 81 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index e585740870..5319f7ade5 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -258,6 +258,87 @@ typedef enum {
     OPC_FENCE_RW_W = 0x0310000f,
 } RISCVInsn;
 
+/*
+ * RISC-V immediate and instruction encoders (excludes 16-bit RVC)
+ */
+
+/* Type-R */
+
+static int32_t encode_r(RISCVInsn opc, TCGReg rd, TCGReg rs1, TCGReg rs2)
+{
+    return opc | (rd & 0x1f) << 7 | (rs1 & 0x1f) << 15 | (rs2 & 0x1f) << 20;
+}
+
+/* Type-I */
+
+static int32_t encode_imm12(uint32_t imm)
+{
+    return (imm & 0xfff) << 20;
+}
+
+static int32_t encode_i(RISCVInsn opc, TCGReg rd, TCGReg rs1, uint32_t imm)
+{
+    return opc | (rd & 0x1f) << 7 | (rs1 & 0x1f) << 15 | encode_imm12(imm);
+}
+
+/* Type-S */
+
+static int32_t encode_simm12(uint32_t imm)
+{
+    return ((imm << 20) >> 25) << 25 | ((imm << 27) >> 27) << 7;
+}
+
+static int32_t encode_s(RISCVInsn opc, TCGReg rs1, TCGReg rs2, uint32_t imm)
+{
+    return opc | (rs1 & 0x1f) << 15 | (rs2 & 0x1f) << 20 | encode_simm12(imm);
+}
+
+/* Type-SB */
+
+static int32_t encode_sbimm12(uint32_t imm)
+{
+    return ((imm << 19) >> 31) << 31 | ((imm << 21) >> 26) << 25 |
+           ((imm << 27) >> 28) << 8 | ((imm << 20) >> 31) << 7;
+}
+
+static int32_t encode_sb(RISCVInsn opc, TCGReg rs1, TCGReg rs2, uint32_t imm)
+{
+    return opc | (rs1 & 0x1f) << 15 | (rs2 & 0x1f) << 20 | encode_sbimm12(imm);
+}
+
+/* Type-U */
+
+static int32_t encode_uimm20(uint32_t imm)
+{
+    return (imm >> 12) << 12;
+}
+
+static int32_t encode_u(RISCVInsn opc, TCGReg rd, uint32_t imm)
+{
+    return opc | (rd & 0x1f) << 7 | encode_uimm20(imm);
+}
+
+/* Type-UJ */
+
+static int32_t encode_ujimm12(uint32_t imm)
+{
+    return ((imm << 11) >> 31) << 31 | ((imm << 21) >> 22) << 21 |
+           ((imm << 20) >> 31) << 20 | ((imm << 12) >> 24) << 12;
+}
+
+static int32_t encode_uj(RISCVInsn opc, TCGReg rd, uint32_t imm)
+{
+    return opc | (rd & 0x1f) << 7 | encode_ujimm12(imm);
+}
+
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
+                              uintptr_t addr)
+{
+    /* Note: jump target patching should be atomic */
+    reloc_call((tcg_insn_unit *)jmp_addr, (tcg_insn_unit*)addr);
+    flush_icache_range(jmp_addr, jmp_addr + 8);
+}
+
 typedef struct {
     DebugFrameHeader h;
     uint8_t fde_def_cfa[4];
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 09/23] riscv: tcg-target: Add the immediate encoders

[Richard Henderson]

On 11/15/18 11:35 PM, Alistair Francis wrote:
> +static int32_t encode_simm12(uint32_t imm)
> +{
> +    return ((imm << 20) >> 25) << 25 | ((imm << 27) >> 27) << 7;
> +}

I'm not fond of triple shifts like this, as it makes me count.
I'd be ok with a formulation like

    return ((imm & 0x1f) << 7) | ((imm & 0xfe) << (25 - 5));

but perhaps best as

    uint32_t ret = 0;
    ret = deposit32(ret, imm, 7, 5);
    ret = deposit32(ret, imm >> 5, 25, 7);

Similarly with encode_sbimm12, encode_uimm20 (which is a simple &), encode_ujimm12.


r~

[Qemu-devel] [RFC v1 10/23] riscv: tcg-target: Add the instruction emitters

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 40 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index 5319f7ade5..d402e48cbf 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -331,6 +331,46 @@ static int32_t encode_uj(RISCVInsn opc, TCGReg rd, uint32_t imm)
     return opc | (rd & 0x1f) << 7 | encode_ujimm12(imm);
 }
 
+/*
+ * RISC-V instruction emitters
+ */
+
+static void tcg_out_opc_reg(TCGContext *s, RISCVInsn opc,
+                            TCGReg rd, TCGReg rs1, TCGReg rs2)
+{
+    tcg_out32(s, encode_r(opc, rd, rs1, rs2));
+}
+
+static void tcg_out_opc_imm(TCGContext *s, RISCVInsn opc,
+                            TCGReg rd, TCGReg rs1, TCGArg imm)
+{
+    tcg_out32(s, encode_i(opc, rd, rs1, imm));
+}
+
+static void tcg_out_opc_store(TCGContext *s, RISCVInsn opc,
+                              TCGReg rs1, TCGReg rs2, uint32_t imm)
+{
+    tcg_out32(s, encode_s(opc, rs1, rs2, imm));
+}
+
+static void tcg_out_opc_branch(TCGContext *s, RISCVInsn opc,
+                               TCGReg rs1, TCGReg rs2, uint32_t imm)
+{
+    tcg_out32(s, encode_sb(opc, rs1, rs2, imm));
+}
+
+static void tcg_out_opc_upper(TCGContext *s, RISCVInsn opc,
+                              TCGReg rd, uint32_t imm)
+{
+    tcg_out32(s, encode_u(opc, rd, imm));
+}
+
+static void tcg_out_opc_jump(TCGContext *s, RISCVInsn opc,
+                             TCGReg rd, uint32_t imm)
+{
+    tcg_out32(s, encode_uj(opc, rd, imm));
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 10/23] riscv: tcg-target: Add the instruction emitters

[Richard Henderson]

On 11/15/18 11:35 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  tcg/riscv/tcg-target.inc.c | 40 ++++++++++++++++++++++++++++++++++++++
>  1 file changed, 40 insertions(+)


Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

[Qemu-devel] [RFC v1 11/23] riscv: tcg-target: Add the relocation functions

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 51 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index d402e48cbf..475feca906 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -371,6 +371,57 @@ static void tcg_out_opc_jump(TCGContext *s, RISCVInsn opc,
     tcg_out32(s, encode_uj(opc, rd, imm));
 }
 
+/*
+ * Relocations
+ */
+
+static void reloc_sbimm12(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+{
+    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
+    tcg_debug_assert(offset == sextract32(offset, 1, 12) << 1);
+
+    code_ptr[0] |= encode_sbimm12(offset);
+}
+
+static void reloc_jimm20(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+{
+    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
+    tcg_debug_assert(offset == sextract32(offset, 1, 20) << 1);
+
+    code_ptr[0] |= encode_ujimm12(offset);
+}
+
+static void reloc_call(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+{
+    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
+    tcg_debug_assert(offset == (int32_t)offset);
+
+    int32_t hi20 = ((offset + 0x800) >> 12) << 12;
+    int32_t lo12 = offset - hi20;
+
+    code_ptr[0] |= encode_uimm20(hi20);
+    code_ptr[1] |= encode_imm12(lo12);
+}
+
+static void patch_reloc(tcg_insn_unit *code_ptr, int type,
+                        intptr_t value, intptr_t addend)
+{
+    tcg_debug_assert(addend == 0);
+    switch (type) {
+    case R_RISCV_BRANCH:
+        reloc_sbimm12(code_ptr, (tcg_insn_unit *)value);
+        break;
+    case R_RISCV_JAL:
+        reloc_jimm20(code_ptr, (tcg_insn_unit *)value);
+        break;
+    case R_RISCV_CALL:
+        reloc_call(code_ptr, (tcg_insn_unit *)value);
+        break;
+    default:
+        tcg_abort();
+    }
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 11/23] riscv: tcg-target: Add the relocation functions

[Richard Henderson]

On 11/15/18 11:35 PM, Alistair Francis wrote:
> +static void reloc_sbimm12(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
> +{
> +    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
> +    tcg_debug_assert(offset == sextract32(offset, 1, 12) << 1);
> +
> +    code_ptr[0] |= encode_sbimm12(offset);
> +}

FYI, I have an in-flight patch for 4.0 that will make patch_reloc return a bool
for relocation success, which will move these asserts.

http://lists.nongnu.org/archive/html/qemu-devel/2018-11/msg02237.html


> +static void reloc_call(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
> +{
> +    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
> +    tcg_debug_assert(offset == (int32_t)offset);
> +
> +    int32_t hi20 = ((offset + 0x800) >> 12) << 12;
> +    int32_t lo12 = offset - hi20;
> +
> +    code_ptr[0] |= encode_uimm20(hi20);
> +    code_ptr[1] |= encode_imm12(lo12);
> +}
> +

This is ok for patching during generation, but it is not ok for
tb_target_set_jmp_target from patch 9.

Will the riscv-32 compiler use a FSTD insn to implement atomic_set for 64-bit?
 If not, you may be just better off using the indirect method.


r~

Re: [Qemu-devel] [RFC v1 11/23] riscv: tcg-target: Add the relocation functions

[Alistair Francis]

On Fri, Nov 16, 2018 at 12:33 AM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/15/18 11:35 PM, Alistair Francis wrote:
> > +static void reloc_sbimm12(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
> > +{
> > +    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
> > +    tcg_debug_assert(offset == sextract32(offset, 1, 12) << 1);
> > +
> > +    code_ptr[0] |= encode_sbimm12(offset);
> > +}
>
> FYI, I have an in-flight patch for 4.0 that will make patch_reloc return a bool
> for relocation success, which will move these asserts.
>
> http://lists.nongnu.org/archive/html/qemu-devel/2018-11/msg02237.html

Thanks, I'll keep an eye on this.

>
>
> > +static void reloc_call(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
> > +{
> > +    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
> > +    tcg_debug_assert(offset == (int32_t)offset);
> > +
> > +    int32_t hi20 = ((offset + 0x800) >> 12) << 12;
> > +    int32_t lo12 = offset - hi20;
> > +
> > +    code_ptr[0] |= encode_uimm20(hi20);
> > +    code_ptr[1] |= encode_imm12(lo12);
> > +}
> > +
>
> This is ok for patching during generation, but it is not ok for
> tb_target_set_jmp_target from patch 9.
>
> Will the riscv-32 compiler use a FSTD insn to implement atomic_set for 64-bit?
>  If not, you may be just better off using the indirect method.

I'm not sure. Is the indirect method just using atomic set, because
that is what I have now?

Alistair

>
>
> r~

Re: [Qemu-devel] [RFC v1 11/23] riscv: tcg-target: Add the relocation functions

[Richard Henderson]

On 11/21/18 2:15 AM, Alistair Francis wrote:
>> Will the riscv-32 compiler use a FSTD insn to implement atomic_set for 64-bit?
>>  If not, you may be just better off using the indirect method.
> 
> I'm not sure. Is the indirect method just using atomic set, because
> that is what I have now?

It's controlled by TCG_TARGET_HAS_direct_jump, and goes through here:

> +    case INDEX_op_goto_tb:
> +        if (s->tb_jmp_insn_offset) {
> +            /* direct jump method */
> +            s->tb_jmp_insn_offset[a0] = tcg_current_code_size(s);
> +            /* should align on 64-bit boundary for atomic patching */
> +            tcg_out_opc_upper(s, OPC_AUIPC, TCG_REG_TMP0, 0);
> +            tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, TCG_REG_TMP0, 0);
> +        } else {
> +            /* indirect jump method */
> +            tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP0, TCG_REG_ZERO,
> +                       (uintptr_t)(s->tb_jmp_target_addr + a0));
> +            tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, TCG_REG_TMP0, 0);
> +        }
> +        s->tb_jmp_reset_offset[a0] = tcg_current_code_size(s);
> +        break;

where as you've correctly implemented, the indirect jump loads a destination
from memory (atomically, via a single load insn), and then branches to it.

The direct jump method avoids the load from memory, but then one has to be able
to modify the insn(s) that perform the jump with a single atomic_set.  Which in
this case, means that the two insns must be aligned so that we can perform a
single aligned 64-bit store.

I recommend at least starting with the indirect method because it's easier.


r~

Re: [Qemu-devel] [RFC v1 11/23] riscv: tcg-target: Add the relocation functions

[Palmer Dabbelt]

On Tue, 20 Nov 2018 17:15:11 PST (-0800), alistair23@gmail.com wrote:
> On Fri, Nov 16, 2018 at 12:33 AM Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> On 11/15/18 11:35 PM, Alistair Francis wrote:
>> > +static void reloc_sbimm12(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
>> > +{
>> > +    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
>> > +    tcg_debug_assert(offset == sextract32(offset, 1, 12) << 1);
>> > +
>> > +    code_ptr[0] |= encode_sbimm12(offset);
>> > +}
>>
>> FYI, I have an in-flight patch for 4.0 that will make patch_reloc return a bool
>> for relocation success, which will move these asserts.
>>
>> http://lists.nongnu.org/archive/html/qemu-devel/2018-11/msg02237.html
>
> Thanks, I'll keep an eye on this.
>
>>
>>
>> > +static void reloc_call(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
>> > +{
>> > +    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
>> > +    tcg_debug_assert(offset == (int32_t)offset);
>> > +
>> > +    int32_t hi20 = ((offset + 0x800) >> 12) << 12;
>> > +    int32_t lo12 = offset - hi20;
>> > +
>> > +    code_ptr[0] |= encode_uimm20(hi20);
>> > +    code_ptr[1] |= encode_imm12(lo12);
>> > +}
>> > +
>>
>> This is ok for patching during generation, but it is not ok for
>> tb_target_set_jmp_target from patch 9.
>>
>> Will the riscv-32 compiler use a FSTD insn to implement atomic_set for 64-bit?
>>  If not, you may be just better off using the indirect method.
>
> I'm not sure. Is the indirect method just using atomic set, because
> that is what I have now?

Per the memory model chapter (which is being ratified now), FSD is not atomic on rv32i:

    An FLD or FSD instruction for which XLEN<64 may also be decomposed into a 
    set of component memory operations of any granularity.

So they should not be used on RV32I hosts to provide atomicity, even if they 
may be atomic on some microarchitectures.

Re: [Qemu-devel] [RFC v1 11/23] riscv: tcg-target: Add the relocation functions

[Richard Henderson]

On 11/21/18 4:53 PM, Palmer Dabbelt wrote:
> 
> Per the memory model chapter (which is being ratified now), FSD is not atomic
> on rv32i:
> 
>    An FLD or FSD instruction for which XLEN<64 may also be decomposed into a   
> set of component memory operations of any granularity.

Ah, thanks.  I knew I should have checked the manual.  Anyway, that means
*only* the indirect goto_tb method is viable for rv32.


r~

[Qemu-devel] [RFC v1 12/23] riscv: tcg-target: Add the mov and movi instruction

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 62 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index 475feca906..0e891e24c9 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -422,6 +422,68 @@ static void patch_reloc(tcg_insn_unit *code_ptr, int type,
     }
 }
 
+/*
+ * TCG intrinsics
+ */
+
+static void tcg_out_mov(TCGContext *s, TCGType type, TCGReg ret, TCGReg arg)
+{
+    if (ret == arg) {
+        return;
+    }
+    switch (type) {
+    case TCG_TYPE_I32:
+    case TCG_TYPE_I64:
+        tcg_out_opc_imm(s, OPC_ADDI, ret, arg, 0);
+        break;
+    default:
+        g_assert_not_reached();
+    }
+}
+
+static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg rd,
+                         tcg_target_long val)
+{
+    tcg_target_long lo = sextract32(val, 0, 12);
+    tcg_target_long hi = val - lo;
+
+    RISCVInsn add32_op = TCG_TARGET_REG_BITS == 64 ? OPC_ADDIW : OPC_ADDI;
+
+#if TCG_TARGET_REG_BITS == 64
+    ptrdiff_t offset = tcg_pcrel_diff(s, (void *)val);
+#endif
+
+    if (val == lo) {
+        tcg_out_opc_imm(s, OPC_ADDI, rd, TCG_REG_ZERO, val);
+    } else if (val && !(val & (val - 1))) {
+        /* power of 2 */
+        tcg_out_opc_imm(s, OPC_ADDI, rd, TCG_REG_ZERO, 1);
+        tcg_out_opc_imm(s, OPC_SLLI, rd, rd, ctz64(val));
+    } else if (TCG_TARGET_REG_BITS == 64 &&
+               !(val >> 31 == 0 || val >> 31 == -1)) {
+        int shift = 12 + ctz64(hi >> 12);
+        hi >>= shift;
+        tcg_out_movi(s, type, rd, hi);
+        tcg_out_opc_imm(s, OPC_SLLI, rd, rd, shift);
+        if (lo != 0) {
+            tcg_out_opc_imm(s, OPC_ADDI, rd, rd, lo);
+        }
+#if TCG_TARGET_REG_BITS == 64
+    } else if (offset == sextract32(offset, 1, 31) << 1) {
+        tcg_out_opc_upper(s, OPC_AUIPC, rd, 0);
+        tcg_out_opc_imm(s, OPC_ADDI, rd, rd, 0);
+        reloc_call(s->code_ptr - 2, (tcg_insn_unit *)val);
+#endif
+    } else {
+        if (hi != 0) {
+            tcg_out_opc_upper(s, OPC_LUI, rd, hi);
+        }
+        if (lo != 0) {
+            tcg_out_opc_imm(s, add32_op, rd, hi == 0 ? TCG_REG_ZERO : rd, lo);
+        }
+    }
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 12/23] riscv: tcg-target: Add the mov and movi instruction

[Richard Henderson]

On 11/15/18 11:35 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  tcg/riscv/tcg-target.inc.c | 62 ++++++++++++++++++++++++++++++++++++++
>  1 file changed, 62 insertions(+)
> 
> diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
> index 475feca906..0e891e24c9 100644
> --- a/tcg/riscv/tcg-target.inc.c
> +++ b/tcg/riscv/tcg-target.inc.c
> @@ -422,6 +422,68 @@ static void patch_reloc(tcg_insn_unit *code_ptr, int type,
>      }
>  }
>  
> +/*
> + * TCG intrinsics
> + */
> +
> +static void tcg_out_mov(TCGContext *s, TCGType type, TCGReg ret, TCGReg arg)
> +{
> +    if (ret == arg) {
> +        return;
> +    }
> +    switch (type) {
> +    case TCG_TYPE_I32:
> +    case TCG_TYPE_I64:
> +        tcg_out_opc_imm(s, OPC_ADDI, ret, arg, 0);
> +        break;
> +    default:
> +        g_assert_not_reached();
> +    }
> +}
> +
> +static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg rd,
> +                         tcg_target_long val)
> +{
> +    tcg_target_long lo = sextract32(val, 0, 12);

sextract64, otherwise you'll make wrong decisions for rv64.
(Although it might be worthwhile to add a local alias so that rv32 doesn't do
more work than necessary.)

> +    tcg_target_long hi = val - lo;
> +
> +    RISCVInsn add32_op = TCG_TARGET_REG_BITS == 64 ? OPC_ADDIW : OPC_ADDI;
> +
> +#if TCG_TARGET_REG_BITS == 64
> +    ptrdiff_t offset = tcg_pcrel_diff(s, (void *)val);
> +#endif
> +
> +    if (val == lo) {
> +        tcg_out_opc_imm(s, OPC_ADDI, rd, TCG_REG_ZERO, val);
           return;
       }

Should match

    if (TCG_TARGET_REG_BITS == 32 || val == (int32_t)val) {
        tcg_out_opc_upper(s, OPC_LUI, rd, hi);
        if (lo != 0) {
            tcg_out_opc_imm(s, add32_op, rd, rd, lo);
        }
        return;
    }

here.

(1) Almost all values requested are 32-bit constants, so check the most common
cases first.  (2) You know hi != 0 because you just eliminated val == lo.  (3)
This handles the cases where LUI alone can load the constant, e.g. 0x1000,
which would otherwise have been matched by your power-of-two test.


> +    } else if (val && !(val & (val - 1))) {
> +        /* power of 2 */
> +        tcg_out_opc_imm(s, OPC_ADDI, rd, TCG_REG_ZERO, 1);
> +        tcg_out_opc_imm(s, OPC_SLLI, rd, rd, ctz64(val));

There's no reason to restrict this to powers of 2 and a shift of the constant 1:

    shift = ctz64(val);
    tmp = val >> shift;
    if (tmp == sextract64(tmp, 0, 12))

> +    } else if (TCG_TARGET_REG_BITS == 64 &&
> +               !(val >> 31 == 0 || val >> 31 == -1)) {
> +        int shift = 12 + ctz64(hi >> 12);

This is just ctz64(hi), since you've already cleared the lo 12 bits.

> +        hi >>= shift;
> +        tcg_out_movi(s, type, rd, hi);
> +        tcg_out_opc_imm(s, OPC_SLLI, rd, rd, shift);
> +        if (lo != 0) {
> +            tcg_out_opc_imm(s, OPC_ADDI, rd, rd, lo);
> +        }
> +#if TCG_TARGET_REG_BITS == 64
> +    } else if (offset == sextract32(offset, 1, 31) << 1) {

sextract64.

> +        tcg_out_opc_upper(s, OPC_AUIPC, rd, 0);
> +        tcg_out_opc_imm(s, OPC_ADDI, rd, rd, 0);
> +        reloc_call(s->code_ptr - 2, (tcg_insn_unit *)val);
> +#endif

Move this pc-rel case above the fully general case and then you can make the
fully general case unconditional.  Also, that preserves an invariant of
increasing order of complexity of the cases.  No need for the ifdef, since this
code should be removed as dead for rv32 (which saw the lui+addi case as
unconditional).


r~

[Qemu-devel] [RFC v1 13/23] riscv: tcg-target: Add the extract instructions

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 34 ++++++++++++++++++++++++++++++++++
 1 file changed, 34 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index 0e891e24c9..646a4d3ebd 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -484,6 +484,40 @@ static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg rd,
     }
 }
 
+static void tcg_out_ext8u(TCGContext *s, TCGReg ret, TCGReg arg)
+{
+    tcg_out_opc_imm(s, OPC_ANDI, ret, arg, 0xff);
+}
+
+static void tcg_out_ext16u(TCGContext *s, TCGReg ret, TCGReg arg)
+{
+    tcg_out_opc_imm(s, OPC_SLLI, ret, arg, TCG_TARGET_REG_BITS - 16);
+    tcg_out_opc_imm(s, OPC_SRLI, ret, ret, TCG_TARGET_REG_BITS - 16);
+}
+
+static void tcg_out_ext32u(TCGContext *s, TCGReg ret, TCGReg arg)
+{
+    tcg_out_opc_imm(s, OPC_SLLI, ret, arg, 32);
+    tcg_out_opc_imm(s, OPC_SRLI, ret, ret, 32);
+}
+
+static void tcg_out_ext8s(TCGContext *s, TCGReg ret, TCGReg arg)
+{
+    tcg_out_opc_imm(s, OPC_SLLI, ret, arg, TCG_TARGET_REG_BITS - 8);
+    tcg_out_opc_imm(s, OPC_SRAI, ret, ret, TCG_TARGET_REG_BITS - 8);
+}
+
+static void tcg_out_ext16s(TCGContext *s, TCGReg ret, TCGReg arg)
+{
+    tcg_out_opc_imm(s, OPC_SLLI, ret, arg, TCG_TARGET_REG_BITS - 16);
+    tcg_out_opc_imm(s, OPC_SRAI, ret, ret, TCG_TARGET_REG_BITS - 16);
+}
+
+static void tcg_out_ext32s(TCGContext *s, TCGReg ret, TCGReg arg)
+{
+    tcg_out_opc_imm(s, OPC_ADDIW, ret, arg, 0);
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 13/23] riscv: tcg-target: Add the extract instructions

[Richard Henderson]

On 11/15/18 11:35 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  tcg/riscv/tcg-target.inc.c | 34 ++++++++++++++++++++++++++++++++++
>  1 file changed, 34 insertions(+)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~

[Qemu-devel] [RFC v1 14/23] riscv: tcg-target: Add the out load and store instructions

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 56 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 56 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index 646a4d3ebd..bc433170c4 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -518,6 +518,62 @@ static void tcg_out_ext32s(TCGContext *s, TCGReg ret, TCGReg arg)
     tcg_out_opc_imm(s, OPC_ADDIW, ret, arg, 0);
 }
 
+static void tcg_out_ldst(TCGContext *s, RISCVInsn opc, TCGReg data,
+                         TCGReg addr, intptr_t offset)
+{
+    int32_t imm12 = sextract32(offset, 0, 12);
+    if (offset != imm12) {
+        tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_TMP2, offset - imm12);
+        if (addr != TCG_REG_ZERO) {
+            tcg_out_opc_reg(s, OPC_ADD, TCG_REG_TMP2, TCG_REG_TMP2, addr);
+        }
+        addr = TCG_REG_TMP2;
+    }
+    switch (opc) {
+    case OPC_SB:
+    case OPC_SH:
+    case OPC_SW:
+    case OPC_SD:
+        tcg_out_opc_store(s, opc, addr, data, imm12);
+        break;
+    case OPC_LB:
+    case OPC_LBU:
+    case OPC_LH:
+    case OPC_LHU:
+    case OPC_LW:
+    case OPC_LWU:
+    case OPC_LD:
+        tcg_out_opc_imm(s, opc, data, addr, imm12);
+        break;
+    default:
+        g_assert_not_reached();
+    }
+}
+
+static void tcg_out_ld(TCGContext *s, TCGType type, TCGReg arg,
+                       TCGReg arg1, intptr_t arg2)
+{
+    bool is32bit = (TCG_TARGET_REG_BITS == 32 || type == TCG_TYPE_I32);
+    tcg_out_ldst(s, is32bit ? OPC_LW : OPC_LD, arg, arg1, arg2);
+}
+
+static void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg,
+                       TCGReg arg1, intptr_t arg2)
+{
+    bool is32bit = (TCG_TARGET_REG_BITS == 32 || type == TCG_TYPE_I32);
+    tcg_out_ldst(s, is32bit ? OPC_SW : OPC_SD, arg, arg1, arg2);
+}
+
+static bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val,
+                        TCGReg base, intptr_t ofs)
+{
+    if (val == 0) {
+        tcg_out_st(s, type, TCG_REG_ZERO, base, ofs);
+        return true;
+    }
+    return false;
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 14/23] riscv: tcg-target: Add the out load and store instructions

[Richard Henderson]

On 11/15/18 11:36 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  tcg/riscv/tcg-target.inc.c | 56 ++++++++++++++++++++++++++++++++++++++
>  1 file changed, 56 insertions(+)
> 
> diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
> index 646a4d3ebd..bc433170c4 100644
> --- a/tcg/riscv/tcg-target.inc.c
> +++ b/tcg/riscv/tcg-target.inc.c
> @@ -518,6 +518,62 @@ static void tcg_out_ext32s(TCGContext *s, TCGReg ret, TCGReg arg)
>      tcg_out_opc_imm(s, OPC_ADDIW, ret, arg, 0);
>  }
>  
> +static void tcg_out_ldst(TCGContext *s, RISCVInsn opc, TCGReg data,
> +                         TCGReg addr, intptr_t offset)
> +{
> +    int32_t imm12 = sextract32(offset, 0, 12);

sextract64 for rv64.

Otherwise,
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

[Qemu-devel] [RFC v1 15/23] riscv: tcg-target: Add branch and jump instructions

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 144 +++++++++++++++++++++++++++++++++++++
 1 file changed, 144 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index bc433170c4..b449e17295 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -574,6 +574,150 @@ static bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val,
     return false;
 }
 
+static const struct {
+    RISCVInsn op;
+    bool swap;
+} tcg_brcond_to_riscv[] = {
+    [TCG_COND_EQ] =  { OPC_BEQ,  false },
+    [TCG_COND_NE] =  { OPC_BNE,  false },
+    [TCG_COND_LT] =  { OPC_BLT,  false },
+    [TCG_COND_GE] =  { OPC_BGE,  false },
+    [TCG_COND_LE] =  { OPC_BGE,  true  },
+    [TCG_COND_GT] =  { OPC_BLT,  true  },
+    [TCG_COND_LTU] = { OPC_BLTU, false },
+    [TCG_COND_GEU] = { OPC_BGEU, false },
+    [TCG_COND_LEU] = { OPC_BGEU, true  },
+    [TCG_COND_GTU] = { OPC_BLTU, true  }
+};
+
+static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
+                           TCGReg arg2, TCGLabel *l)
+{
+    RISCVInsn op = tcg_brcond_to_riscv[cond].op;
+    bool swap = tcg_brcond_to_riscv[cond].swap;
+
+    tcg_out_opc_branch(s, op, swap ? arg2 : arg1, swap ? arg1 : arg2, 0);
+
+    if (l->has_value) {
+        reloc_sbimm12(s->code_ptr - 1, l->u.value_ptr);
+    } else {
+        tcg_out_reloc(s, s->code_ptr - 1, R_RISCV_BRANCH, l, 0);
+    }
+}
+
+static void tcg_out_setcond(TCGContext *s, TCGCond cond, TCGReg ret,
+                            TCGReg arg1, TCGReg arg2)
+{
+    switch (cond) {
+    case TCG_COND_EQ:
+        tcg_out_opc_reg(s, OPC_SUB, ret, arg1, arg2);
+        tcg_out_opc_imm(s, OPC_SLTIU, ret, ret, 1);
+        break;
+    case TCG_COND_NE:
+        tcg_out_opc_reg(s, OPC_SUB, ret, arg1, arg2);
+        tcg_out_opc_reg(s, OPC_SLTU, ret, TCG_REG_ZERO, ret);
+        break;
+    case TCG_COND_LT:
+        tcg_out_opc_reg(s, OPC_SLT, ret, arg1, arg2);
+        break;
+    case TCG_COND_GE:
+        tcg_out_opc_reg(s, OPC_SLT, ret, arg1, arg2);
+        tcg_out_opc_imm(s, OPC_XORI, ret, ret, 1);
+        break;
+    case TCG_COND_LE:
+        tcg_out_opc_reg(s, OPC_SLT, ret, arg2, arg1);
+        tcg_out_opc_imm(s, OPC_XORI, ret, ret, 1);
+        break;
+    case TCG_COND_GT:
+        tcg_out_opc_reg(s, OPC_SLT, ret, arg2, arg1);
+        break;
+    case TCG_COND_LTU:
+        tcg_out_opc_reg(s, OPC_SLTU, ret, arg1, arg2);
+        break;
+    case TCG_COND_GEU:
+        tcg_out_opc_reg(s, OPC_SLTU, ret, arg1, arg2);
+        tcg_out_opc_imm(s, OPC_XORI, ret, ret, 1);
+        break;
+    case TCG_COND_LEU:
+        tcg_out_opc_reg(s, OPC_SLTU, ret, arg2, arg1);
+        tcg_out_opc_imm(s, OPC_XORI, ret, ret, 1);
+        break;
+    case TCG_COND_GTU:
+        tcg_out_opc_reg(s, OPC_SLTU, ret, arg2, arg1);
+        break;
+    default:
+         g_assert_not_reached();
+         break;
+     }
+}
+
+static void tcg_out_brcond2(TCGContext *s, TCGCond cond, TCGReg al, TCGReg ah,
+                            TCGReg bl, TCGReg bh, TCGLabel *l)
+{
+    /* todo */
+    g_assert_not_reached();
+}
+
+static void tcg_out_setcond2(TCGContext *s, TCGCond cond, TCGReg ret,
+                             TCGReg al, TCGReg ah, TCGReg bl, TCGReg bh)
+{
+    /* todo */
+    g_assert_not_reached();
+}
+
+static inline void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
+{
+    ptrdiff_t offset = tcg_pcrel_diff(s, target);
+    tcg_debug_assert(offset == sextract64(offset, 0, 26));
+    tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
+}
+
+static inline void tcg_out_goto_long(TCGContext *s, tcg_insn_unit *target)
+{
+    ptrdiff_t offset = tcg_pcrel_diff(s, target);
+
+    if (offset == sextract64(offset, 0, 26)) {
+        tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
+    } else {
+        tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP0, (intptr_t)target);
+        tcg_out_opc_jump(s, OPC_JAL, TCG_REG_TMP0, 0);
+    }
+}
+
+static void tcg_out_call_int(TCGContext *s, tcg_insn_unit *arg, bool tail)
+{
+    TCGReg link = tail ? TCG_REG_ZERO : TCG_REG_RA;
+    ptrdiff_t offset = tcg_pcrel_diff(s, arg);
+    if (offset == sextract32(offset, 1, 20) << 1) {
+        /* short jump: -2097150 to 2097152 */
+        tcg_out_opc_jump(s, OPC_JAL, link, offset);
+    } else if (TCG_TARGET_REG_BITS == 32 ||
+        offset == sextract32(offset, 1, 31) << 1) {
+        /* long jump: -2147483646 to 2147483648 */
+        tcg_out_opc_upper(s, OPC_AUIPC, TCG_REG_TMP0, 0);
+        tcg_out_opc_imm(s, OPC_JALR, link, TCG_REG_TMP0, 0);
+        reloc_call(s->code_ptr - 2, arg);
+    } else if (TCG_TARGET_REG_BITS == 64) {
+        /* far jump: 64-bit */
+        tcg_target_long imm = sextract32((tcg_target_long)arg, 0, 12);
+        tcg_target_long base = (tcg_target_long)arg - imm;
+        tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_TMP0, base);
+        tcg_out_opc_imm(s, OPC_JALR, link, TCG_REG_TMP0, imm);
+    } else {
+        g_assert_not_reached();
+    }
+}
+
+static void tcg_out_tail(TCGContext *s, tcg_insn_unit *arg)
+{
+    tcg_out_call_int(s, arg, true);
+}
+
+static void tcg_out_call(TCGContext *s, tcg_insn_unit *arg)
+{
+    tcg_out_call_int(s, arg, false);
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 15/23] riscv: tcg-target: Add branch and jump instructions

[Richard Henderson]

On 11/15/18 11:36 PM, Alistair Francis wrote:
> +static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
> +                           TCGReg arg2, TCGLabel *l)
> +{
> +    RISCVInsn op = tcg_brcond_to_riscv[cond].op;
> +    bool swap = tcg_brcond_to_riscv[cond].swap;
> +
> +    tcg_out_opc_branch(s, op, swap ? arg2 : arg1, swap ? arg1 : arg2, 0);

You might want to tcg_debug_assert(op != 0) here.

> +    if (l->has_value) {
> +        reloc_sbimm12(s->code_ptr - 1, l->u.value_ptr);

I'm concerned about the conditional branch range.  +-4K isn't much to work
with.  The minimum we have for other hosts is +-32K.

We have two options: (1) greatly reduce the max size of the TB for this host;
(2) be prepared to emit a 2 insn sequence: conditional branch across
unconditional branch, with forward branches that turn out to be small patched
with a nop.

FWIW, the first case would be done via modification of tcg_op_buf_full.  You
might have to go as low as 500 opcodes, I'm not sure.

> +static inline void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
> +{
> +    ptrdiff_t offset = tcg_pcrel_diff(s, target);
> +    tcg_debug_assert(offset == sextract64(offset, 0, 26));
> +    tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
> +}
> +
> +static inline void tcg_out_goto_long(TCGContext *s, tcg_insn_unit *target)
> +{
> +    ptrdiff_t offset = tcg_pcrel_diff(s, target);
> +
> +    if (offset == sextract64(offset, 0, 26)) {
> +        tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
> +    } else {
> +        tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP0, (intptr_t)target);
> +        tcg_out_opc_jump(s, OPC_JAL, TCG_REG_TMP0, 0);
> +    }
> +}

How are these to be used?  I guess I'll find out...

> +static void tcg_out_call_int(TCGContext *s, tcg_insn_unit *arg, bool tail)
> +{
> +    TCGReg link = tail ? TCG_REG_ZERO : TCG_REG_RA;
> +    ptrdiff_t offset = tcg_pcrel_diff(s, arg);
> +    if (offset == sextract32(offset, 1, 20) << 1) {

sextract64.

> +        /* short jump: -2097150 to 2097152 */
> +        tcg_out_opc_jump(s, OPC_JAL, link, offset);
> +    } else if (TCG_TARGET_REG_BITS == 32 ||
> +        offset == sextract32(offset, 1, 31) << 1) {

sextract64.

> +        /* long jump: -2147483646 to 2147483648 */
> +        tcg_out_opc_upper(s, OPC_AUIPC, TCG_REG_TMP0, 0);
> +        tcg_out_opc_imm(s, OPC_JALR, link, TCG_REG_TMP0, 0);
> +        reloc_call(s->code_ptr - 2, arg);
> +    } else if (TCG_TARGET_REG_BITS == 64) {
> +        /* far jump: 64-bit */
> +        tcg_target_long imm = sextract32((tcg_target_long)arg, 0, 12);
> +        tcg_target_long base = (tcg_target_long)arg - imm;
> +        tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_TMP0, base);
> +        tcg_out_opc_imm(s, OPC_JALR, link, TCG_REG_TMP0, imm);


r~

Re: [Qemu-devel] [RFC v1 15/23] riscv: tcg-target: Add branch and jump instructions

[Alistair Francis]

On Fri, Nov 16, 2018 at 1:14 AM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/15/18 11:36 PM, Alistair Francis wrote:
> > +static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
> > +                           TCGReg arg2, TCGLabel *l)
> > +{
> > +    RISCVInsn op = tcg_brcond_to_riscv[cond].op;
> > +    bool swap = tcg_brcond_to_riscv[cond].swap;
> > +
> > +    tcg_out_opc_branch(s, op, swap ? arg2 : arg1, swap ? arg1 : arg2, 0);
>
> You might want to tcg_debug_assert(op != 0) here.
>
> > +    if (l->has_value) {
> > +        reloc_sbimm12(s->code_ptr - 1, l->u.value_ptr);
>
> I'm concerned about the conditional branch range.  +-4K isn't much to work
> with.  The minimum we have for other hosts is +-32K.
>
> We have two options: (1) greatly reduce the max size of the TB for this host;
> (2) be prepared to emit a 2 insn sequence: conditional branch across
> unconditional branch, with forward branches that turn out to be small patched
> with a nop.
>
> FWIW, the first case would be done via modification of tcg_op_buf_full.  You
> might have to go as low as 500 opcodes, I'm not sure.

How do we do option 2?

>
> > +static inline void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
> > +{
> > +    ptrdiff_t offset = tcg_pcrel_diff(s, target);
> > +    tcg_debug_assert(offset == sextract64(offset, 0, 26));
> > +    tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
> > +}
> > +
> > +static inline void tcg_out_goto_long(TCGContext *s, tcg_insn_unit *target)
> > +{
> > +    ptrdiff_t offset = tcg_pcrel_diff(s, target);
> > +
> > +    if (offset == sextract64(offset, 0, 26)) {
> > +        tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
> > +    } else {
> > +        tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP0, (intptr_t)target);
> > +        tcg_out_opc_jump(s, OPC_JAL, TCG_REG_TMP0, 0);
> > +    }
> > +}
>
> How are these to be used?  I guess I'll find out...

It's called on a exit TB op. I'm not sure if that's correct though.

Alistair

>
> > +static void tcg_out_call_int(TCGContext *s, tcg_insn_unit *arg, bool tail)
> > +{
> > +    TCGReg link = tail ? TCG_REG_ZERO : TCG_REG_RA;
> > +    ptrdiff_t offset = tcg_pcrel_diff(s, arg);
> > +    if (offset == sextract32(offset, 1, 20) << 1) {
>
> sextract64.
>
> > +        /* short jump: -2097150 to 2097152 */
> > +        tcg_out_opc_jump(s, OPC_JAL, link, offset);
> > +    } else if (TCG_TARGET_REG_BITS == 32 ||
> > +        offset == sextract32(offset, 1, 31) << 1) {
>
> sextract64.
>
> > +        /* long jump: -2147483646 to 2147483648 */
> > +        tcg_out_opc_upper(s, OPC_AUIPC, TCG_REG_TMP0, 0);
> > +        tcg_out_opc_imm(s, OPC_JALR, link, TCG_REG_TMP0, 0);
> > +        reloc_call(s->code_ptr - 2, arg);
> > +    } else if (TCG_TARGET_REG_BITS == 64) {
> > +        /* far jump: 64-bit */
> > +        tcg_target_long imm = sextract32((tcg_target_long)arg, 0, 12);
> > +        tcg_target_long base = (tcg_target_long)arg - imm;
> > +        tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_TMP0, base);
> > +        tcg_out_opc_imm(s, OPC_JALR, link, TCG_REG_TMP0, imm);
>
>
> r~

Re: [Qemu-devel] [RFC v1 15/23] riscv: tcg-target: Add branch and jump instructions

[Richard Henderson]

On 11/21/18 12:49 AM, Alistair Francis wrote:
> On Fri, Nov 16, 2018 at 1:14 AM Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> On 11/15/18 11:36 PM, Alistair Francis wrote:
>>> +static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
>>> +                           TCGReg arg2, TCGLabel *l)
>>> +{
>>> +    RISCVInsn op = tcg_brcond_to_riscv[cond].op;
>>> +    bool swap = tcg_brcond_to_riscv[cond].swap;
>>> +
>>> +    tcg_out_opc_branch(s, op, swap ? arg2 : arg1, swap ? arg1 : arg2, 0);
>>
>> You might want to tcg_debug_assert(op != 0) here.
>>
>>> +    if (l->has_value) {
>>> +        reloc_sbimm12(s->code_ptr - 1, l->u.value_ptr);
>>
>> I'm concerned about the conditional branch range.  +-4K isn't much to work
>> with.  The minimum we have for other hosts is +-32K.
>>
>> We have two options: (1) greatly reduce the max size of the TB for this host;
>> (2) be prepared to emit a 2 insn sequence: conditional branch across
>> unconditional branch, with forward branches that turn out to be small patched
>> with a nop.
>>
>> FWIW, the first case would be done via modification of tcg_op_buf_full.  You
>> might have to go as low as 500 opcodes, I'm not sure.
> 
> How do we do option 2?

If l->has_value, just check the actual range.  But of course backward branching
isn't that common in tcg generated code.  Most branches within the TB are short
forward branches, but we also don't know how short is short.

But every TB begins with a test of env->exit_code and a conditional branch to
the end of the block, where we place some code to return to the main loop and
return the pointer to the TB at which we exited.  Thus every TB has a branch
that spans the size of the entire TB.

So, invent (or repurpose) an R_RISCV_FOO value.  It doesn't matter which
because it's private within tcg/riscv/.  Just add some commentary.  (See e.g.
tcg/sparc/ and its use of R_SPARC_13.)

While generating code, emit the conditional branch as normal; leave the unknown
destination 0 for now.  Emit a nop as the second insn.

When resolving R_RISCV_FOO, if the conditional branch is in range, great!  Just
patch it.  If it is out of range, then you need to edit the conditional branch
to reverse the condition (insn ^ (1 << 12)) and branch to pc+8, i.e. over the
next instruction.  Which was a nop during generation, but you will now install
jal r0,dest going to the real destination.


r~

Re: [Qemu-devel] [RFC v1 15/23] riscv: tcg-target: Add branch and jump instructions

[Alistair Francis]

On Tue, Nov 20, 2018 at 11:40 PM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/21/18 12:49 AM, Alistair Francis wrote:
> > On Fri, Nov 16, 2018 at 1:14 AM Richard Henderson
> > <richard.henderson@linaro.org> wrote:
> >>
> >> On 11/15/18 11:36 PM, Alistair Francis wrote:
> >>> +static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
> >>> +                           TCGReg arg2, TCGLabel *l)
> >>> +{
> >>> +    RISCVInsn op = tcg_brcond_to_riscv[cond].op;
> >>> +    bool swap = tcg_brcond_to_riscv[cond].swap;
> >>> +
> >>> +    tcg_out_opc_branch(s, op, swap ? arg2 : arg1, swap ? arg1 : arg2, 0);
> >>
> >> You might want to tcg_debug_assert(op != 0) here.
> >>
> >>> +    if (l->has_value) {
> >>> +        reloc_sbimm12(s->code_ptr - 1, l->u.value_ptr);
> >>
> >> I'm concerned about the conditional branch range.  +-4K isn't much to work
> >> with.  The minimum we have for other hosts is +-32K.
> >>
> >> We have two options: (1) greatly reduce the max size of the TB for this host;
> >> (2) be prepared to emit a 2 insn sequence: conditional branch across
> >> unconditional branch, with forward branches that turn out to be small patched
> >> with a nop.
> >>
> >> FWIW, the first case would be done via modification of tcg_op_buf_full.  You
> >> might have to go as low as 500 opcodes, I'm not sure.
> >
> > How do we do option 2?
>
> If l->has_value, just check the actual range.  But of course backward branching
> isn't that common in tcg generated code.  Most branches within the TB are short
> forward branches, but we also don't know how short is short.
>
> But every TB begins with a test of env->exit_code and a conditional branch to
> the end of the block, where we place some code to return to the main loop and
> return the pointer to the TB at which we exited.  Thus every TB has a branch
> that spans the size of the entire TB.
>
> So, invent (or repurpose) an R_RISCV_FOO value.  It doesn't matter which
> because it's private within tcg/riscv/.  Just add some commentary.  (See e.g.
> tcg/sparc/ and its use of R_SPARC_13.)
>
> While generating code, emit the conditional branch as normal; leave the unknown
> destination 0 for now.  Emit a nop as the second insn.
>
> When resolving R_RISCV_FOO, if the conditional branch is in range, great!  Just
> patch it.  If it is out of range, then you need to edit the conditional branch
> to reverse the condition (insn ^ (1 << 12)) and branch to pc+8, i.e. over the
> next instruction.  Which was a nop during generation, but you will now install
> jal r0,dest going to the real destination.

Ok, I think I have done this correctly. I have something that compiles and runs.

I'll send a second RFC out sometime this week. It won't be based on
your latest patches and it's still missing some things but I want to
keep this moving along.

Thanks so much for your help Richard!

Alistair

>
>
> r~

[Qemu-devel] [RFC v1 16/23] riscv: tcg-target: Add slowpath load and store instructions

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 240 +++++++++++++++++++++++++++++++++++++
 1 file changed, 240 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index b449e17295..5fe6935e24 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -718,6 +718,246 @@ static void tcg_out_call(TCGContext *s, tcg_insn_unit *arg)
     tcg_out_call_int(s, arg, false);
 }
 
+static void tcg_out_mb(TCGContext *s, TCGArg a0)
+{
+    static const RISCVInsn fence[] = {
+        [0 ... TCG_MO_ALL] = OPC_FENCE_RW_RW,
+        [TCG_MO_LD_LD]     = OPC_FENCE_R_R,
+        [TCG_MO_ST_LD]     = OPC_FENCE_W_R,
+        [TCG_MO_LD_ST]     = OPC_FENCE_R_W,
+        [TCG_MO_ST_ST]     = OPC_FENCE_W_W,
+        [TCG_BAR_LDAQ]     = OPC_FENCE_R_RW,
+        [TCG_BAR_STRL]     = OPC_FENCE_RW_W,
+        [TCG_BAR_SC]       = OPC_FENCE_RW_RW,
+    };
+    tcg_out32(s, fence[a0 & TCG_MO_ALL]);
+}
+
+/*
+ * Load/store and TLB
+ */
+
+#if defined(CONFIG_SOFTMMU)
+#include "tcg-ldst.inc.c"
+
+/* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr,
+ *                                     TCGMemOpIdx oi, uintptr_t ra)
+ */
+static void * const qemu_ld_helpers[16] = {
+    [MO_UB]   = helper_ret_ldub_mmu,
+    [MO_SB]   = helper_ret_ldsb_mmu,
+    [MO_LEUW] = helper_le_lduw_mmu,
+    [MO_LESW] = helper_le_ldsw_mmu,
+    [MO_LEUL] = helper_le_ldul_mmu,
+    [MO_LESL] = helper_le_ldsl_mmu,
+    [MO_LEQ]  = helper_le_ldq_mmu,
+    [MO_BEUW] = helper_be_lduw_mmu,
+    [MO_BESW] = helper_be_ldsw_mmu,
+    [MO_BEUL] = helper_be_ldul_mmu,
+    [MO_BESL] = helper_be_ldsl_mmu,
+    [MO_BEQ]  = helper_be_ldq_mmu,
+};
+
+/* helper signature: helper_ret_st_mmu(CPUState *env, target_ulong addr,
+ *                                     uintxx_t val, TCGMemOpIdx oi,
+ *                                     uintptr_t ra)
+ */
+static void * const qemu_st_helpers[16] = {
+    [MO_UB]   = helper_ret_stb_mmu,
+    [MO_LEUW] = helper_le_stw_mmu,
+    [MO_LEUL] = helper_le_stl_mmu,
+    [MO_LEQ]  = helper_le_stq_mmu,
+    [MO_BEUW] = helper_be_stw_mmu,
+    [MO_BEUL] = helper_be_stl_mmu,
+    [MO_BEQ]  = helper_be_stq_mmu,
+};
+
+static void tcg_out_tlb_load(TCGContext *s, TCGReg addrl,
+                             TCGReg addrh, TCGMemOpIdx oi,
+                             tcg_insn_unit **label_ptr, bool is_load)
+{
+    TCGMemOp opc = get_memop(oi);
+    unsigned s_bits = opc & MO_SIZE;
+    unsigned a_bits = get_alignment_bits(opc);
+    target_ulong mask;
+    int mem_index = get_mmuidx(oi);
+    int cmp_off
+        = (is_load
+           ? offsetof(CPUArchState, tlb_table[mem_index][0].addr_read)
+           : offsetof(CPUArchState, tlb_table[mem_index][0].addr_write));
+    int add_off = offsetof(CPUArchState, tlb_table[mem_index][0].addend);
+    int addend_offset = (offsetof(CPUTLBEntry, addend)) -
+                            (is_load ? offsetof(CPUTLBEntry, addr_read)
+                                     : offsetof(CPUTLBEntry, addr_write));
+    RISCVInsn load_cmp_op = (TARGET_LONG_BITS == 64 ? OPC_LD :
+                             TCG_TARGET_REG_BITS == 64 ? OPC_LWU : OPC_LW);
+    RISCVInsn load_add_op = TCG_TARGET_REG_BITS == 64 ? OPC_LD : OPC_LW;
+    TCGReg base = TCG_AREG0;
+    TCGReg cmpr;
+
+    /* We don't support oversize guests */
+    if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) {
+        g_assert_not_reached();
+    }
+
+    /* We don't support unaligned accesses. */
+    if (a_bits < s_bits) {
+        a_bits = s_bits;
+    }
+    mask = (target_ulong)TARGET_PAGE_MASK | ((1 << a_bits) - 1);
+
+
+    /* Compensate for very large offsets.  */
+    if (add_off >= 0x1000) {
+        int adj;
+        base = TCG_REG_TMP2;
+        if (cmp_off <= 2 * 0xfff) {
+            adj = 0xfff;
+            tcg_out_opc_imm(s, OPC_ADDI, base, TCG_AREG0, adj);
+        } else {
+            adj = cmp_off - sextract32(cmp_off, 0, 12);
+            tcg_debug_assert(add_off - adj >= -0x1000
+                             && add_off - adj < 0x1000);
+
+            tcg_out_opc_upper(s, OPC_LUI, base, adj);
+            tcg_out_opc_reg(s, OPC_ADD, base, TCG_REG_ZERO, TCG_AREG0);
+        }
+        add_off -= adj;
+        cmp_off -= adj;
+    }
+
+    /* Extract the page index.  */
+    if (CPU_TLB_BITS + CPU_TLB_ENTRY_BITS < 12) {
+        tcg_out_opc_imm(s, OPC_SRLI, TCG_REG_TMP0, addrl,
+                        TARGET_PAGE_BITS - CPU_TLB_ENTRY_BITS);
+        tcg_out_opc_imm(s, OPC_ANDI, TCG_REG_TMP0, TCG_REG_TMP0,
+                        MAKE_64BIT_MASK(CPU_TLB_ENTRY_BITS, CPU_TLB_BITS));
+    } else {
+        tcg_out_opc_imm(s, OPC_SRLI, TCG_REG_TMP0, addrl, TARGET_PAGE_BITS);
+        tcg_out_opc_imm(s, OPC_ANDI, TCG_REG_TMP0, TCG_REG_TMP0,
+                        MAKE_64BIT_MASK(0, CPU_TLB_BITS));
+        tcg_out_opc_imm(s, OPC_SLLI, TCG_REG_TMP0, TCG_REG_TMP0,
+                        CPU_TLB_ENTRY_BITS);
+    }
+
+    /* Add that to the base address to index the tlb.  */
+    tcg_out_opc_reg(s, OPC_ADD, TCG_REG_TMP2, base, TCG_REG_TMP0);
+    base = TCG_REG_TMP2;
+
+    /* Load the tlb comparator and the addend.  */
+    tcg_out_ldst(s, load_cmp_op, TCG_REG_TMP0, base, cmp_off);
+    tcg_out_ldst(s, load_cmp_op, TCG_REG_TMP2, base, add_off);
+
+    /* Clear the non-page, non-alignment bits from the address.  */
+    if (mask == sextract64(mask, 0, 12)) {
+        tcg_out_opc_imm(s, OPC_ANDI, TCG_REG_TMP1, addrl, mask);
+    } else {
+        tcg_out_movi(s, TCG_TYPE_REG, TCG_REG_TMP1, mask);
+        tcg_out_opc_reg(s, OPC_AND, TCG_REG_TMP1, TCG_REG_TMP1, addrl);
+     }
+
+    /* Compare masked address with the TLB entry. */
+    label_ptr[0] = s->code_ptr;
+    tcg_out_opc_branch(s, OPC_BNE, TCG_REG_TMP0, TCG_REG_TMP1, 0);
+
+    /* TLB Hit - translate address using addend.  */
+    if (TCG_TARGET_REG_BITS > TARGET_LONG_BITS) {
+        tcg_out_ext32u(s, TCG_REG_TMP0, addrl);
+        addrl = TCG_REG_TMP0;
+    }
+    tcg_out_opc_reg(s, OPC_ADD, TCG_REG_L0, TCG_REG_TMP2, addrl);
+}
+
+static void add_qemu_ldst_label(TCGContext *s, int is_ld, TCGMemOpIdx oi,
+                                TCGType ext,
+                                TCGReg datalo, TCGReg datahi,
+                                TCGReg addrlo, TCGReg addrhi,
+                                void *raddr, tcg_insn_unit **label_ptr)
+{
+    TCGLabelQemuLdst *label = new_ldst_label(s);
+
+    label->is_ld = is_ld;
+    label->oi = oi;
+    label->type = ext;
+    label->datalo_reg = datalo;
+    label->datahi_reg = datahi;
+    label->addrlo_reg = addrlo;
+    label->addrhi_reg = addrhi;
+    label->raddr = raddr;
+    label->label_ptr[0] = label_ptr[0];
+}
+
+static void tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
+{
+    TCGMemOpIdx oi = l->oi;
+    TCGMemOp opc = get_memop(oi);
+    TCGReg a0 = tcg_target_call_iarg_regs[0];
+    TCGReg a1 = tcg_target_call_iarg_regs[1];
+    TCGReg a2 = tcg_target_call_iarg_regs[2];
+    TCGReg a3 = tcg_target_call_iarg_regs[3];
+
+    /* We don't support oversize guests */
+    if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) {
+        g_assert_not_reached();
+    }
+
+    /* resolve label address */
+    reloc_sbimm12(l->label_ptr[0], s->code_ptr);
+
+    /* call load helper */
+    tcg_out_mov(s, TCG_TYPE_PTR, a0, TCG_AREG0);
+    tcg_out_mov(s, TCG_TYPE_PTR, a1, l->addrlo_reg);
+    tcg_out_movi(s, TCG_TYPE_PTR, a2, oi);
+    tcg_out_movi(s, TCG_TYPE_PTR, a3, (tcg_target_long)l->raddr);
+
+    tcg_out_call(s, qemu_ld_helpers[opc & (MO_BSWAP | MO_SSIZE)]);
+    tcg_out_mov(s, (opc & MO_SIZE) == MO_64, l->datalo_reg, a0);
+
+    tcg_out_goto(s, l->raddr);
+}
+
+static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
+{
+    TCGMemOpIdx oi = l->oi;
+    TCGMemOp opc = get_memop(oi);
+    TCGMemOp s_bits = opc & MO_SIZE;
+    TCGReg a0 = tcg_target_call_iarg_regs[0];
+    TCGReg a1 = tcg_target_call_iarg_regs[1];
+    TCGReg a2 = tcg_target_call_iarg_regs[2];
+    TCGReg a3 = tcg_target_call_iarg_regs[3];
+    TCGReg a4 = tcg_target_call_iarg_regs[4];
+
+    /* We don't support oversize guests */
+    if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) {
+        g_assert_not_reached();
+    }
+
+    /* resolve label address */
+    reloc_sbimm12(l->label_ptr[0], s->code_ptr);
+
+    /* call store helper */
+    tcg_out_mov(s, TCG_TYPE_PTR, a0, TCG_AREG0);
+    tcg_out_mov(s, TCG_TYPE_PTR, a1, l->addrlo_reg);
+    tcg_out_mov(s, TCG_TYPE_PTR, a2, l->datalo_reg);
+    switch (s_bits) {
+    case MO_8:
+        tcg_out_ext8u(s, a2, a2);
+        break;
+    case MO_16:
+        tcg_out_ext16u(s, a2, a2);
+        break;
+    default:
+        break;
+    }
+    tcg_out_movi(s, TCG_TYPE_PTR, a3, oi);
+    tcg_out_movi(s, TCG_TYPE_PTR, a4, (tcg_target_long)l->raddr);
+
+    tcg_out_call(s, qemu_st_helpers[opc & (MO_BSWAP | MO_SSIZE)]);
+
+    tcg_out_goto(s, l->raddr);
+}
+#endif /* CONFIG_SOFTMMU */
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 16/23] riscv: tcg-target: Add slowpath load and store instructions

[Richard Henderson]

On 11/15/18 11:36 PM, Alistair Francis wrote:
> +static void tcg_out_mb(TCGContext *s, TCGArg a0)
> +{
> +    static const RISCVInsn fence[] = {
> +        [0 ... TCG_MO_ALL] = OPC_FENCE_RW_RW,
> +        [TCG_MO_LD_LD]     = OPC_FENCE_R_R,
> +        [TCG_MO_ST_LD]     = OPC_FENCE_W_R,
> +        [TCG_MO_LD_ST]     = OPC_FENCE_R_W,
> +        [TCG_MO_ST_ST]     = OPC_FENCE_W_W,
> +        [TCG_BAR_LDAQ]     = OPC_FENCE_R_RW,
> +        [TCG_BAR_STRL]     = OPC_FENCE_RW_W,
> +        [TCG_BAR_SC]       = OPC_FENCE_RW_RW,
> +    };
> +    tcg_out32(s, fence[a0 & TCG_MO_ALL]);
> +}
> +

TCG_MO_* and TCG_BAR_* are two different bitmasks, or'ed together.
Which you've separated by "& TCG_MO_ALL".  Thus the TCG_BAR_* constants should
not appear in this table.


> +static void * const qemu_ld_helpers[16] = {
> +    [MO_UB]   = helper_ret_ldub_mmu,
> +    [MO_SB]   = helper_ret_ldsb_mmu,
> +    [MO_LEUW] = helper_le_lduw_mmu,
> +    [MO_LESW] = helper_le_ldsw_mmu,
> +    [MO_LEUL] = helper_le_ldul_mmu,
> +    [MO_LESL] = helper_le_ldsl_mmu,
> +    [MO_LEQ]  = helper_le_ldq_mmu,
> +    [MO_BEUW] = helper_be_lduw_mmu,
> +    [MO_BESW] = helper_be_ldsw_mmu,
> +    [MO_BEUL] = helper_be_ldul_mmu,
> +    [MO_BESL] = helper_be_ldsl_mmu,
> +    [MO_BEQ]  = helper_be_ldq_mmu,
> +};

The LESL and BESL functions will not be present for rv32 -> link error.  Here
you do need an ifdef.

> +        } else {
> +            adj = cmp_off - sextract32(cmp_off, 0, 12);
> +            tcg_debug_assert(add_off - adj >= -0x1000
> +                             && add_off - adj < 0x1000);
> +
> +            tcg_out_opc_upper(s, OPC_LUI, base, adj);
> +            tcg_out_opc_reg(s, OPC_ADD, base, TCG_REG_ZERO, TCG_AREG0);

base, base, TCG_AREG0.

> +    /* Compare masked address with the TLB entry. */
> +    label_ptr[0] = s->code_ptr;
> +    tcg_out_opc_branch(s, OPC_BNE, TCG_REG_TMP0, TCG_REG_TMP1, 0);

Another case of a potential out-of-range branch.

It might be worthwhile to move all of this out-of-line from the start, where
that branch will always be short.  See

http://lists.nongnu.org/archive/html/qemu-devel/2018-11/msg02234.html


r~

Re: [Qemu-devel] [RFC v1 16/23] riscv: tcg-target: Add slowpath load and store instructions

[Alistair Francis]

On Fri, Nov 16, 2018 at 1:24 AM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/15/18 11:36 PM, Alistair Francis wrote:
> > +static void tcg_out_mb(TCGContext *s, TCGArg a0)
> > +{
> > +    static const RISCVInsn fence[] = {
> > +        [0 ... TCG_MO_ALL] = OPC_FENCE_RW_RW,
> > +        [TCG_MO_LD_LD]     = OPC_FENCE_R_R,
> > +        [TCG_MO_ST_LD]     = OPC_FENCE_W_R,
> > +        [TCG_MO_LD_ST]     = OPC_FENCE_R_W,
> > +        [TCG_MO_ST_ST]     = OPC_FENCE_W_W,
> > +        [TCG_BAR_LDAQ]     = OPC_FENCE_R_RW,
> > +        [TCG_BAR_STRL]     = OPC_FENCE_RW_W,
> > +        [TCG_BAR_SC]       = OPC_FENCE_RW_RW,
> > +    };
> > +    tcg_out32(s, fence[a0 & TCG_MO_ALL]);
> > +}
> > +
>
> TCG_MO_* and TCG_BAR_* are two different bitmasks, or'ed together.
> Which you've separated by "& TCG_MO_ALL".  Thus the TCG_BAR_* constants should
> not appear in this table.
>
>
> > +static void * const qemu_ld_helpers[16] = {
> > +    [MO_UB]   = helper_ret_ldub_mmu,
> > +    [MO_SB]   = helper_ret_ldsb_mmu,
> > +    [MO_LEUW] = helper_le_lduw_mmu,
> > +    [MO_LESW] = helper_le_ldsw_mmu,
> > +    [MO_LEUL] = helper_le_ldul_mmu,
> > +    [MO_LESL] = helper_le_ldsl_mmu,
> > +    [MO_LEQ]  = helper_le_ldq_mmu,
> > +    [MO_BEUW] = helper_be_lduw_mmu,
> > +    [MO_BESW] = helper_be_ldsw_mmu,
> > +    [MO_BEUL] = helper_be_ldul_mmu,
> > +    [MO_BESL] = helper_be_ldsl_mmu,
> > +    [MO_BEQ]  = helper_be_ldq_mmu,
> > +};
>
> The LESL and BESL functions will not be present for rv32 -> link error.  Here
> you do need an ifdef.
>
> > +        } else {
> > +            adj = cmp_off - sextract32(cmp_off, 0, 12);
> > +            tcg_debug_assert(add_off - adj >= -0x1000
> > +                             && add_off - adj < 0x1000);
> > +
> > +            tcg_out_opc_upper(s, OPC_LUI, base, adj);
> > +            tcg_out_opc_reg(s, OPC_ADD, base, TCG_REG_ZERO, TCG_AREG0);
>
> base, base, TCG_AREG0.
>
> > +    /* Compare masked address with the TLB entry. */
> > +    label_ptr[0] = s->code_ptr;
> > +    tcg_out_opc_branch(s, OPC_BNE, TCG_REG_TMP0, TCG_REG_TMP1, 0);
>
> Another case of a potential out-of-range branch.
>
> It might be worthwhile to move all of this out-of-line from the start, where
> that branch will always be short.  See
>
> http://lists.nongnu.org/archive/html/qemu-devel/2018-11/msg02234.html

That does look cool, but it's not in the tree yet.

Otherwise can we directly just call tcg_out_brcond()?

PS: Thanks for your review. I have gone through most of your comments.
I now don't see any segfaults when running. My guest still doesn't
boot, but it's getting further then it used to :)

Alistair

>
>
> r~

Re: [Qemu-devel] [RFC v1 16/23] riscv: tcg-target: Add slowpath load and store instructions

[Richard Henderson]

On 11/21/18 1:18 AM, Alistair Francis wrote:
> On Fri, Nov 16, 2018 at 1:24 AM Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> On 11/15/18 11:36 PM, Alistair Francis wrote:
>>> +static void tcg_out_mb(TCGContext *s, TCGArg a0)
>>> +{
>>> +    static const RISCVInsn fence[] = {
>>> +        [0 ... TCG_MO_ALL] = OPC_FENCE_RW_RW,
>>> +        [TCG_MO_LD_LD]     = OPC_FENCE_R_R,
>>> +        [TCG_MO_ST_LD]     = OPC_FENCE_W_R,
>>> +        [TCG_MO_LD_ST]     = OPC_FENCE_R_W,
>>> +        [TCG_MO_ST_ST]     = OPC_FENCE_W_W,
>>> +        [TCG_BAR_LDAQ]     = OPC_FENCE_R_RW,
>>> +        [TCG_BAR_STRL]     = OPC_FENCE_RW_W,
>>> +        [TCG_BAR_SC]       = OPC_FENCE_RW_RW,
>>> +    };
>>> +    tcg_out32(s, fence[a0 & TCG_MO_ALL]);
>>> +}
>>> +
>>
>> TCG_MO_* and TCG_BAR_* are two different bitmasks, or'ed together.
>> Which you've separated by "& TCG_MO_ALL".  Thus the TCG_BAR_* constants should
>> not appear in this table.
>>
>>
>>> +static void * const qemu_ld_helpers[16] = {
>>> +    [MO_UB]   = helper_ret_ldub_mmu,
>>> +    [MO_SB]   = helper_ret_ldsb_mmu,
>>> +    [MO_LEUW] = helper_le_lduw_mmu,
>>> +    [MO_LESW] = helper_le_ldsw_mmu,
>>> +    [MO_LEUL] = helper_le_ldul_mmu,
>>> +    [MO_LESL] = helper_le_ldsl_mmu,
>>> +    [MO_LEQ]  = helper_le_ldq_mmu,
>>> +    [MO_BEUW] = helper_be_lduw_mmu,
>>> +    [MO_BESW] = helper_be_ldsw_mmu,
>>> +    [MO_BEUL] = helper_be_ldul_mmu,
>>> +    [MO_BESL] = helper_be_ldsl_mmu,
>>> +    [MO_BEQ]  = helper_be_ldq_mmu,
>>> +};
>>
>> The LESL and BESL functions will not be present for rv32 -> link error.  Here
>> you do need an ifdef.
>>
>>> +        } else {
>>> +            adj = cmp_off - sextract32(cmp_off, 0, 12);
>>> +            tcg_debug_assert(add_off - adj >= -0x1000
>>> +                             && add_off - adj < 0x1000);
>>> +
>>> +            tcg_out_opc_upper(s, OPC_LUI, base, adj);
>>> +            tcg_out_opc_reg(s, OPC_ADD, base, TCG_REG_ZERO, TCG_AREG0);
>>
>> base, base, TCG_AREG0.
>>
>>> +    /* Compare masked address with the TLB entry. */
>>> +    label_ptr[0] = s->code_ptr;
>>> +    tcg_out_opc_branch(s, OPC_BNE, TCG_REG_TMP0, TCG_REG_TMP1, 0);
>>
>> Another case of a potential out-of-range branch.
>>
>> It might be worthwhile to move all of this out-of-line from the start, where
>> that branch will always be short.  See
>>
>> http://lists.nongnu.org/archive/html/qemu-devel/2018-11/msg02234.html
> 
> That does look cool, but it's not in the tree yet.

No, but it'll probably be in tree before this code is.  ;-)
I'll put it into my tcg-next-for-4.0 branch, and you can base off that.

> Otherwise can we directly just call tcg_out_brcond()?

Not quite, because there's no label structure.  But you can break out
subroutines and use those.

> PS: Thanks for your review. I have gone through most of your comments.
> I now don't see any segfaults when running. My guest still doesn't
> boot, but it's getting further then it used to :)

Excellent.


r~

[Qemu-devel] [RFC v1 17/23] riscv: tcg-target: Add direct load and store instructions

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 145 +++++++++++++++++++++++++++++++++++++
 1 file changed, 145 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index 5fe6935e24..c4a013a962 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -958,6 +958,151 @@ static void tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
 }
 #endif /* CONFIG_SOFTMMU */
 
+static void tcg_out_qemu_ld_direct(TCGContext *s, TCGReg lo, TCGReg hi,
+                                   TCGReg base, TCGMemOp opc, bool is_64)
+{
+    const TCGMemOp bswap = opc & MO_BSWAP;
+
+    /* TODO: Handle byte swapping */
+
+    switch (opc & (MO_SSIZE)) {
+    case MO_UB:
+        tcg_out_opc_imm(s, OPC_LBU, lo, base, 0);
+        break;
+    case MO_SB:
+        tcg_out_opc_imm(s, OPC_LB, lo, base, 0);
+        break;
+    case MO_UW:
+        tcg_out_opc_imm(s, OPC_LHU, lo, base, 0);
+        break;
+    case MO_SW:
+        tcg_out_opc_imm(s, OPC_LH, lo, base, 0);
+        break;
+    case MO_UL:
+        if (TCG_TARGET_REG_BITS == 64 && is_64) {
+            tcg_out_opc_imm(s, OPC_LWU, lo, base, 0);
+            break;
+        }
+        /* FALLTHRU */
+    case MO_SL:
+        tcg_out_opc_imm(s, OPC_LW, lo, base, 0);
+        break;
+    case MO_Q:
+        /* Prefer to load from offset 0 first, but allow for overlap.  */
+        if (TCG_TARGET_REG_BITS == 64) {
+            tcg_out_opc_imm(s, OPC_LD, lo, base, 0);
+        } else {
+            tcg_out_opc_imm(s, OPC_LW, lo, base, 0);
+            tcg_out_opc_imm(s, OPC_LW, hi, base, 4);
+        }
+        break;
+    default:
+        g_assert_not_reached();
+    }
+}
+
+static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args, bool is_64)
+{
+    TCGReg addr_regl, addr_regh __attribute__((unused));
+    TCGReg data_regl, data_regh;
+    TCGMemOpIdx oi;
+    TCGMemOp opc;
+#if defined(CONFIG_SOFTMMU)
+    tcg_insn_unit *label_ptr[1];
+#endif
+    TCGReg base = TCG_REG_TMP0;
+
+    data_regl = *args++;
+    data_regh = (TCG_TARGET_REG_BITS == 32 && is_64 ? *args++ : 0);
+    addr_regl = *args++;
+    addr_regh = (TCG_TARGET_REG_BITS < TARGET_LONG_BITS ? *args++ : 0);
+    oi = *args++;
+    opc = get_memop(oi);
+
+#if defined(CONFIG_SOFTMMU)
+    tcg_out_tlb_load(s, addr_regl, addr_regh, oi, label_ptr, 1);
+    tcg_out_qemu_ld_direct(s, data_regl, data_regh, base, opc, is_64);
+    add_qemu_ldst_label(s, 1, oi,
+                        (is_64 ? TCG_TYPE_I64 : TCG_TYPE_I32),
+                        data_regl, data_regh, addr_regl, addr_regh,
+                        s->code_ptr, label_ptr);
+#else
+    if (TCG_TARGET_REG_BITS > TARGET_LONG_BITS) {
+        tcg_out_ext32u(s, base, addr_regl);
+        addr_regl = base;
+    }
+    tcg_out_opc_reg(s, OPC_ADD, base, TCG_GUEST_BASE_REG, addr_regl);
+    tcg_out_qemu_ld_direct(s, data_regl, data_regh, base, opc, is_64);
+#endif
+}
+
+static void tcg_out_qemu_st_direct(TCGContext *s, TCGReg lo, TCGReg hi,
+                                   TCGReg base, TCGMemOp opc)
+{
+    const TCGMemOp bswap = opc & MO_BSWAP;
+
+    /* TODO: Handle byte swapping */
+
+    switch (opc & (MO_SSIZE)) {
+    case MO_8:
+        tcg_out_opc_store(s, OPC_SB, base, lo, 0);
+        break;
+    case MO_16:
+        tcg_out_opc_store(s, OPC_SH, base, lo, 0);
+        break;
+    case MO_32:
+        tcg_out_opc_store(s, OPC_SW, base, lo, 0);
+        break;
+    case MO_64:
+        if (TCG_TARGET_REG_BITS == 64) {
+            tcg_out_opc_store(s, OPC_SD, base, lo, 0);
+        } else {
+            tcg_out_opc_store(s, OPC_SW, base, lo, 0);
+            tcg_out_opc_store(s, OPC_SW, base, hi, 4);
+        }
+        break;
+    default:
+        g_assert_not_reached();
+    }
+}
+
+static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, bool is_64)
+{
+    TCGReg addr_regl, addr_regh __attribute__((unused));
+    TCGReg data_regl, data_regh;
+    TCGMemOpIdx oi;
+    TCGMemOp opc;
+#if defined(CONFIG_SOFTMMU)
+    tcg_insn_unit *label_ptr[1];
+    TCGReg base = TCG_REG_L0;
+#else
+    TCGReg base = TCG_REG_TMP0;
+#endif
+
+    data_regl = *args++;
+    data_regh = (TCG_TARGET_REG_BITS == 32 && is_64 ? *args++ : 0);
+    addr_regl = *args++;
+    addr_regh = (TCG_TARGET_REG_BITS < TARGET_LONG_BITS ? *args++ : 0);
+    oi = *args++;
+    opc = get_memop(oi);
+
+#if defined(CONFIG_SOFTMMU)
+    tcg_out_tlb_load(s, addr_regl, addr_regh, oi, label_ptr, 0);
+    tcg_out_qemu_st_direct(s, data_regl, data_regh, base, opc);
+    add_qemu_ldst_label(s, 0, oi,
+                        (is_64 ? TCG_TYPE_I64 : TCG_TYPE_I32),
+                        data_regl, data_regh, addr_regl, addr_regh,
+                        s->code_ptr, label_ptr);
+#else
+    if (TCG_TARGET_REG_BITS > TARGET_LONG_BITS) {
+        tcg_out_ext32u(s, base, addr_regl);
+        addr_regl = base;
+    }
+    tcg_out_opc_reg(s, OPC_ADD, base, TCG_GUEST_BASE_REG, addr_regl);
+    tcg_out_qemu_st_direct(s, data_regl, data_regh, base, opc);
+#endif
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 17/23] riscv: tcg-target: Add direct load and store instructions

[Richard Henderson]

On 11/15/18 11:36 PM, Alistair Francis wrote:
> +    tcg_out_opc_reg(s, OPC_ADD, base, TCG_GUEST_BASE_REG, addr_regl);

Should avoid this when guest_base == 0, which happens fairly regularly for a
64-bit guest.

> +        /* Prefer to load from offset 0 first, but allow for overlap.  */
> +        if (TCG_TARGET_REG_BITS == 64) {
> +            tcg_out_opc_imm(s, OPC_LD, lo, base, 0);
> +        } else {
> +            tcg_out_opc_imm(s, OPC_LW, lo, base, 0);
> +            tcg_out_opc_imm(s, OPC_LW, hi, base, 4);
> +        }

Comment sounds like two lines of code that's missing.

> +    const TCGMemOp bswap = opc & MO_BSWAP;
> +
> +    /* TODO: Handle byte swapping */

Should assert rather than emit bad code.

I do still plan to change tcg to allow backends to *not* handle byte swapping
if they don't want.  This will make the i386 and arm32 backends less icky.


r~

Re: [Qemu-devel] [RFC v1 17/23] riscv: tcg-target: Add direct load and store instructions

[Alistair Francis]

On Fri, Nov 16, 2018 at 9:10 AM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/15/18 11:36 PM, Alistair Francis wrote:
> > +    tcg_out_opc_reg(s, OPC_ADD, base, TCG_GUEST_BASE_REG, addr_regl);
>
> Should avoid this when guest_base == 0, which happens fairly regularly for a
> 64-bit guest.
>
> > +        /* Prefer to load from offset 0 first, but allow for overlap.  */
> > +        if (TCG_TARGET_REG_BITS == 64) {
> > +            tcg_out_opc_imm(s, OPC_LD, lo, base, 0);
> > +        } else {
> > +            tcg_out_opc_imm(s, OPC_LW, lo, base, 0);
> > +            tcg_out_opc_imm(s, OPC_LW, hi, base, 4);
> > +        }
>
> Comment sounds like two lines of code that's missing.

I can't figure out what this comment should be for. Why would we want
to prefer loading with an offset 0?

Alistair

>
> > +    const TCGMemOp bswap = opc & MO_BSWAP;
> > +
> > +    /* TODO: Handle byte swapping */
>
> Should assert rather than emit bad code.
>
> I do still plan to change tcg to allow backends to *not* handle byte swapping
> if they don't want.  This will make the i386 and arm32 backends less icky.
>
>
> r~

Re: [Qemu-devel] [RFC v1 17/23] riscv: tcg-target: Add direct load and store instructions

[Richard Henderson]

On 11/20/18 12:06 AM, Alistair Francis wrote:
> On Fri, Nov 16, 2018 at 9:10 AM Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> On 11/15/18 11:36 PM, Alistair Francis wrote:
>>> +    tcg_out_opc_reg(s, OPC_ADD, base, TCG_GUEST_BASE_REG, addr_regl);
>>
>> Should avoid this when guest_base == 0, which happens fairly regularly for a
>> 64-bit guest.
>>
>>> +        /* Prefer to load from offset 0 first, but allow for overlap.  */
>>> +        if (TCG_TARGET_REG_BITS == 64) {
>>> +            tcg_out_opc_imm(s, OPC_LD, lo, base, 0);
>>> +        } else {
>>> +            tcg_out_opc_imm(s, OPC_LW, lo, base, 0);
>>> +            tcg_out_opc_imm(s, OPC_LW, hi, base, 4);
>>> +        }
>>
>> Comment sounds like two lines of code that's missing.
> 
> I can't figure out what this comment should be for. Why would we want
> to prefer loading with an offset 0?

Perhaps to help the memory controler; perhaps no reason at all.
But "allow for overlap" suggests

  } else if (lo != base) {
      tcg_out_opc_imm(s, OPC_LW, lo, base, 0);
      tcg_out_opc_imm(s, OPC_LW, hi, base, 4);
  } else {
      tcg_out_opc_imm(s, OPC_LW, hi, base, 4);
      tcg_out_opc_imm(s, OPC_LW, lo, base, 0);
  }


r~

[Qemu-devel] [RFC v1 18/23] riscv: tcg-target: Add the out op decoder

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 472 +++++++++++++++++++++++++++++++++++++
 1 file changed, 472 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index c4a013a962..bf3b04f7dc 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -1103,6 +1103,478 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, bool is_64)
 #endif
 }
 
+static tcg_insn_unit *tb_ret_addr;
+
+static void tcg_out_op(TCGContext *s, TCGOpcode opc,
+                       const TCGArg *args, const int *const_args)
+{
+    TCGArg a0 = args[0];
+    TCGArg a1 = args[1];
+    TCGArg a2 = args[2];
+    int c2 = const_args[2];
+    const bool is32bit = TCG_TARGET_REG_BITS == 32;
+
+    switch (opc) {
+    case INDEX_op_exit_tb:
+        /* Reuse the zeroing that exists for goto_ptr.  */
+        if (a0 == 0) {
+            tcg_out_goto_long(s, s->code_gen_epilogue);
+        } else {
+            tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_A0, a0);
+            tcg_out_goto_long(s, tb_ret_addr);
+        }
+        break;
+
+    case INDEX_op_goto_tb:
+        if (s->tb_jmp_insn_offset) {
+            /* direct jump method */
+            s->tb_jmp_insn_offset[a0] = tcg_current_code_size(s);
+            /* should align on 64-bit boundary for atomic patching */
+            tcg_out_opc_upper(s, OPC_AUIPC, TCG_REG_TMP0, 0);
+            tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, TCG_REG_TMP0, 0);
+        } else {
+            /* indirect jump method */
+            tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_TMP0, TCG_REG_ZERO,
+                       (uintptr_t)(s->tb_jmp_target_addr + a0));
+            tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, TCG_REG_TMP0, 0);
+        }
+        s->tb_jmp_reset_offset[a0] = tcg_current_code_size(s);
+        break;
+
+    case INDEX_op_goto_ptr:
+        tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, a0, 0);
+        break;
+
+    case INDEX_op_br:
+        tcg_out_reloc(s, s->code_ptr, R_RISCV_JAL, arg_label(a0), 0);
+        tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, 0);
+        break;
+
+    case INDEX_op_ld8u_i32:
+    case INDEX_op_ld8u_i64:
+        tcg_out_ldst(s, OPC_LBU, a0, a1, a2);
+        break;
+    case INDEX_op_ld8s_i32:
+    case INDEX_op_ld8s_i64:
+        tcg_out_ldst(s, OPC_LB, a0, a1, a2);
+        break;
+    case INDEX_op_ld16u_i32:
+    case INDEX_op_ld16u_i64:
+        tcg_out_ldst(s, OPC_LHU, a0, a1, a2);
+        break;
+    case INDEX_op_ld16s_i32:
+    case INDEX_op_ld16s_i64:
+        tcg_out_ldst(s, OPC_LH, a0, a1, a2);
+        break;
+    case INDEX_op_ld32u_i64:
+        tcg_out_ldst(s, OPC_LWU, a0, a1, a2);
+        break;
+    case INDEX_op_ld_i32:
+    case INDEX_op_ld32s_i64:
+        tcg_out_ldst(s, OPC_LW, a0, a1, a2);
+        break;
+    case INDEX_op_ld_i64:
+        tcg_out_ldst(s, OPC_LD, a0, a1, a2);
+        break;
+
+    case INDEX_op_st8_i32:
+    case INDEX_op_st8_i64:
+        tcg_out_ldst(s, OPC_SB, a0, a1, a2);
+        break;
+    case INDEX_op_st16_i32:
+    case INDEX_op_st16_i64:
+        tcg_out_ldst(s, OPC_SH, a0, a1, a2);
+        break;
+    case INDEX_op_st_i32:
+    case INDEX_op_st32_i64:
+        tcg_out_ldst(s, OPC_SW, a0, a1, a2);
+        break;
+    case INDEX_op_st_i64:
+        tcg_out_ldst(s, OPC_SD, a0, a1, a2);
+        break;
+
+    case INDEX_op_add_i32:
+        if (c2) {
+            tcg_out_opc_imm(s, is32bit ? OPC_ADDI : OPC_ADDIW, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, is32bit ? OPC_ADD : OPC_ADDW, a0, a1, a2);
+        }
+        break;
+    case INDEX_op_add_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_ADDI, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_ADD, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_sub_i32:
+        if (c2) {
+            tcg_out_opc_imm(s, is32bit ? OPC_ADDI : OPC_ADDIW, a0, a1, -a2);
+        } else {
+            tcg_out_opc_reg(s, is32bit ? OPC_SUB : OPC_SUBW, a0, a1, a2);
+        }
+        break;
+    case INDEX_op_sub_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_ADDI, a0, a1, -a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_SUB, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_and_i32:
+    case INDEX_op_and_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_ANDI, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_AND, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_or_i32:
+    case INDEX_op_or_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_ORI, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_OR, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_xor_i32:
+    case INDEX_op_xor_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_XORI, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_XOR, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_not_i32:
+    case INDEX_op_not_i64:
+        tcg_out_opc_imm(s, OPC_XORI, a0, a1, -1);
+        break;
+
+    case INDEX_op_neg_i32:
+        tcg_out_opc_reg(s, is32bit ? OPC_SUB : OPC_SUBW, a0, TCG_REG_ZERO, a1);
+        break;
+    case INDEX_op_neg_i64:
+        tcg_out_opc_imm(s, OPC_SUB, a0, TCG_REG_ZERO, a1);
+        break;
+
+    case INDEX_op_mul_i32:
+        tcg_out_opc_reg(s, is32bit ? OPC_MUL : OPC_MULW, a0, a1, a2);
+        break;
+    case INDEX_op_mul_i64:
+        tcg_out_opc_reg(s, OPC_MUL, a0, a1, a2);
+        break;
+
+    case INDEX_op_div_i32:
+        tcg_out_opc_reg(s, is32bit ? OPC_DIV : OPC_DIVW, a0, a1, a2);
+        break;
+    case INDEX_op_div_i64:
+        tcg_out_opc_reg(s, OPC_DIV, a0, a1, a2);
+        break;
+
+    case INDEX_op_divu_i32:
+        tcg_out_opc_reg(s, is32bit ? OPC_DIVU : OPC_DIVUW, a0, a1, a2);
+        break;
+    case INDEX_op_divu_i64:
+        tcg_out_opc_reg(s, OPC_DIVU, a0, a1, a2);
+        break;
+
+    case INDEX_op_rem_i32:
+        tcg_out_opc_reg(s, is32bit ? OPC_REM : OPC_REMW, a0, a1, a2);
+        break;
+    case INDEX_op_rem_i64:
+        tcg_out_opc_reg(s, OPC_REM, a0, a1, a2);
+        break;
+
+    case INDEX_op_remu_i32:
+        tcg_out_opc_reg(s, is32bit ? OPC_REMU : OPC_REMUW, a0, a1, a2);
+        break;
+    case INDEX_op_remu_i64:
+        tcg_out_opc_reg(s, OPC_REMU, a0, a1, a2);
+        break;
+
+    case INDEX_op_shl_i32:
+        if (c2) {
+            tcg_out_opc_imm(s, is32bit ? OPC_SLLI : OPC_SLLIW, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, is32bit ? OPC_SLL : OPC_SLLW, a0, a1, a2);
+        }
+        break;
+    case INDEX_op_shl_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_SLLI, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_SLL, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_shr_i32:
+        if (c2) {
+            tcg_out_opc_imm(s, is32bit ? OPC_SRLI : OPC_SRLIW, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, is32bit ? OPC_SRL : OPC_SRLW, a0, a1, a2);
+        }
+        break;
+    case INDEX_op_shr_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_SRLI, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_SRL, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_sar_i32:
+        if (c2) {
+            tcg_out_opc_imm(s, is32bit ? OPC_SRAI : OPC_SRAIW, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, is32bit ? OPC_SRA : OPC_SRAW, a0, a1, a2);
+        }
+        break;
+    case INDEX_op_sar_i64:
+        if (c2) {
+            tcg_out_opc_imm(s, OPC_SRAI, a0, a1, a2);
+        } else {
+            tcg_out_opc_reg(s, OPC_SRA, a0, a1, a2);
+        }
+        break;
+
+    case INDEX_op_brcond_i32:
+    case INDEX_op_brcond_i64:
+        tcg_out_brcond(s, a2, a0, a1, arg_label(args[3]));
+        break;
+    case INDEX_op_brcond2_i32:
+        tcg_out_brcond2(s, args[4], a0, a1, a2, args[3], arg_label(args[5]));
+        break;
+
+    case INDEX_op_setcond_i32:
+    case INDEX_op_setcond_i64:
+        tcg_out_setcond(s, args[3], a0, a1, a2);
+        break;
+    case INDEX_op_setcond2_i32:
+        tcg_out_setcond2(s, args[5], a0, a1, a2, args[3], args[4]);
+        break;
+
+    case INDEX_op_qemu_ld_i32:
+        tcg_out_qemu_ld(s, args, false);
+        break;
+    case INDEX_op_qemu_ld_i64:
+        tcg_out_qemu_ld(s, args, true);
+        break;
+    case INDEX_op_qemu_st_i32:
+        tcg_out_qemu_st(s, args, false);
+        break;
+    case INDEX_op_qemu_st_i64:
+        tcg_out_qemu_st(s, args, true);
+        break;
+
+    case INDEX_op_ext8u_i32:
+    case INDEX_op_ext8u_i64:
+        tcg_out_ext8u(s, a0, a1);
+        break;
+
+    case INDEX_op_ext16u_i32:
+    case INDEX_op_ext16u_i64:
+        tcg_out_ext16u(s, a0, a1);
+        break;
+
+    case INDEX_op_ext32u_i64:
+    case INDEX_op_extu_i32_i64:
+        tcg_out_ext32u(s, a0, a1);
+        break;
+
+    case INDEX_op_ext8s_i32:
+    case INDEX_op_ext8s_i64:
+        tcg_out_ext8s(s, a0, a1);
+        break;
+
+    case INDEX_op_ext16s_i32:
+    case INDEX_op_ext16s_i64:
+        tcg_out_ext16s(s, a0, a1);
+        break;
+
+    case INDEX_op_ext32s_i64:
+    case INDEX_op_ext_i32_i64:
+        tcg_out_ext32s(s, a0, a1);
+        break;
+
+    case INDEX_op_mulsh_i32:
+    case INDEX_op_mulsh_i64:
+        tcg_out_opc_imm(s, OPC_MULH, a0, a1, a2);
+        break;
+
+    case INDEX_op_muluh_i32:
+    case INDEX_op_muluh_i64:
+        tcg_out_opc_imm(s, OPC_MULHU, a0, a1, a2);
+        break;
+
+    case INDEX_op_mb:
+        tcg_out_mb(s, a0);
+        break;
+
+    case INDEX_op_mov_i32:  /* Always emitted via tcg_out_mov.  */
+    case INDEX_op_mov_i64:
+    case INDEX_op_movi_i32: /* Always emitted via tcg_out_movi.  */
+    case INDEX_op_movi_i64:
+    case INDEX_op_call:     /* Always emitted via tcg_out_call.  */
+    default:
+        g_assert_not_reached();
+    }
+}
+
+static const TCGTargetOpDef *tcg_target_op_def(TCGOpcode op)
+{
+    static const TCGTargetOpDef r
+        = { .args_ct_str = { "r" } };
+    static const TCGTargetOpDef r_r
+        = { .args_ct_str = { "r", "r" } };
+    static const TCGTargetOpDef rZ_r
+        = { .args_ct_str = { "rZ", "r" } };
+    static const TCGTargetOpDef rZ_rZ
+        = { .args_ct_str = { "rZ", "rZ" } };
+    static const TCGTargetOpDef rZ_rZ_rZ_rZ
+        = { .args_ct_str = { "rZ", "rZ", "rZ", "rZ" } };
+    static const TCGTargetOpDef r_r_ri
+        = { .args_ct_str = { "r", "r", "ri" } };
+    static const TCGTargetOpDef r_r_rI
+        = { .args_ct_str = { "r", "r", "rI" } };
+    static const TCGTargetOpDef r_rZ_rN
+        = { .args_ct_str = { "r", "rZ", "rN" } };
+    static const TCGTargetOpDef r_rZ_rZ
+        = { .args_ct_str = { "r", "rZ", "rZ" } };
+    static const TCGTargetOpDef r_rZ_rZ_rZ_rZ
+        = { .args_ct_str = { "r", "rZ", "rZ", "rZ", "rZ" } };
+    static const TCGTargetOpDef r_L
+        = { .args_ct_str = { "r", "L" } };
+    static const TCGTargetOpDef r_r_L
+        = { .args_ct_str = { "r", "r", "L" } };
+    static const TCGTargetOpDef r_L_L
+        = { .args_ct_str = { "r", "L", "L" } };
+    static const TCGTargetOpDef r_r_L_L
+        = { .args_ct_str = { "r", "r", "L", "L" } };
+    static const TCGTargetOpDef LZ_L
+        = { .args_ct_str = { "LZ", "L" } };
+    static const TCGTargetOpDef LZ_L_L
+        = { .args_ct_str = { "LZ", "L", "L" } };
+    static const TCGTargetOpDef LZ_LZ_L
+        = { .args_ct_str = { "LZ", "LZ", "L" } };
+    static const TCGTargetOpDef LZ_LZ_L_L
+        = { .args_ct_str = { "LZ", "LZ", "L", "L" } };
+
+    switch (op) {
+    case INDEX_op_goto_ptr:
+        return &r;
+
+    case INDEX_op_ld8u_i32:
+    case INDEX_op_ld8s_i32:
+    case INDEX_op_ld16u_i32:
+    case INDEX_op_ld16s_i32:
+    case INDEX_op_ld_i32:
+    case INDEX_op_not_i32:
+    case INDEX_op_neg_i32:
+    case INDEX_op_ld8u_i64:
+    case INDEX_op_ld8s_i64:
+    case INDEX_op_ld16u_i64:
+    case INDEX_op_ld16s_i64:
+    case INDEX_op_ld32s_i64:
+    case INDEX_op_ld32u_i64:
+    case INDEX_op_ld_i64:
+    case INDEX_op_not_i64:
+    case INDEX_op_neg_i64:
+    case INDEX_op_ext8u_i32:
+    case INDEX_op_ext8u_i64:
+    case INDEX_op_ext16u_i32:
+    case INDEX_op_ext16u_i64:
+    case INDEX_op_ext32u_i64:
+    case INDEX_op_extu_i32_i64:
+    case INDEX_op_ext8s_i32:
+    case INDEX_op_ext8s_i64:
+    case INDEX_op_ext16s_i32:
+    case INDEX_op_ext16s_i64:
+    case INDEX_op_ext32s_i64:
+    case INDEX_op_ext_i32_i64:
+        return &r_r;
+
+    case INDEX_op_st8_i32:
+    case INDEX_op_st16_i32:
+    case INDEX_op_st_i32:
+    case INDEX_op_st8_i64:
+    case INDEX_op_st16_i64:
+    case INDEX_op_st32_i64:
+    case INDEX_op_st_i64:
+        return &rZ_r;
+
+    case INDEX_op_add_i32:
+    case INDEX_op_and_i32:
+    case INDEX_op_or_i32:
+    case INDEX_op_xor_i32:
+    case INDEX_op_add_i64:
+    case INDEX_op_and_i64:
+    case INDEX_op_or_i64:
+    case INDEX_op_xor_i64:
+        return &r_r_rI;
+
+    case INDEX_op_sub_i32:
+    case INDEX_op_sub_i64:
+        return &r_rZ_rN;
+
+    case INDEX_op_mul_i32:
+    case INDEX_op_mulsh_i32:
+    case INDEX_op_muluh_i32:
+    case INDEX_op_div_i32:
+    case INDEX_op_divu_i32:
+    case INDEX_op_rem_i32:
+    case INDEX_op_remu_i32:
+    case INDEX_op_setcond_i32:
+    case INDEX_op_mul_i64:
+    case INDEX_op_mulsh_i64:
+    case INDEX_op_muluh_i64:
+    case INDEX_op_div_i64:
+    case INDEX_op_divu_i64:
+    case INDEX_op_rem_i64:
+    case INDEX_op_remu_i64:
+    case INDEX_op_setcond_i64:
+        return &r_rZ_rZ;
+
+    case INDEX_op_shl_i32:
+    case INDEX_op_shr_i32:
+    case INDEX_op_sar_i32:
+    case INDEX_op_shl_i64:
+    case INDEX_op_shr_i64:
+    case INDEX_op_sar_i64:
+        return &r_r_ri;
+
+    case INDEX_op_brcond_i32:
+    case INDEX_op_brcond_i64:
+        return &rZ_rZ;
+
+    case INDEX_op_brcond2_i32:
+        return &rZ_rZ_rZ_rZ;
+
+    case INDEX_op_setcond2_i32:
+        return &r_rZ_rZ_rZ_rZ;
+
+    case INDEX_op_qemu_ld_i32:
+        return TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? &r_L : &r_L_L;
+    case INDEX_op_qemu_st_i32:
+        return TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? &LZ_L : &LZ_L_L;
+    case INDEX_op_qemu_ld_i64:
+        return TCG_TARGET_REG_BITS == 64 ? &r_L
+               : TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? &r_r_L
+               : &r_r_L_L;
+    case INDEX_op_qemu_st_i64:
+        return TCG_TARGET_REG_BITS == 64 ? &LZ_L
+               : TARGET_LONG_BITS <= TCG_TARGET_REG_BITS ? &LZ_LZ_L
+               : &LZ_LZ_L_L;
+
+    default:
+        return NULL;
+    }
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 18/23] riscv: tcg-target: Add the out op decoder

[Richard Henderson]

On 11/15/18 11:36 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  tcg/riscv/tcg-target.inc.c | 472 +++++++++++++++++++++++++++++++++++++
>  1 file changed, 472 insertions(+)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~

[Qemu-devel] [RFC v1 19/23] riscv: tcg-target: Add the prologue generation

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 66 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index bf3b04f7dc..4b8f65f367 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -1575,6 +1575,72 @@ static const TCGTargetOpDef *tcg_target_op_def(TCGOpcode op)
     }
 }
 
+static const int tcg_target_callee_save_regs[] = {
+    TCG_REG_S0,       /* used for the global env (TCG_AREG0) */
+    TCG_REG_S1,
+    TCG_REG_S2,
+    TCG_REG_S3,
+    TCG_REG_S4,
+    TCG_REG_S5,
+    TCG_REG_S6,
+    TCG_REG_S7,
+    TCG_REG_S8,
+    TCG_REG_S9,
+    TCG_REG_S10,
+    TCG_REG_S11,
+    TCG_REG_RA,       /* should be last for ABI compliance */
+};
+
+/* Stack frame parameters.  */
+#define REG_SIZE   (TCG_TARGET_REG_BITS / 8)
+#define SAVE_SIZE  ((int)ARRAY_SIZE(tcg_target_callee_save_regs) * REG_SIZE)
+#define TEMP_SIZE  (CPU_TEMP_BUF_NLONGS * (int)sizeof(long))
+#define FRAME_SIZE ((TCG_STATIC_CALL_ARGS_SIZE + TEMP_SIZE + SAVE_SIZE \
+                     + TCG_TARGET_STACK_ALIGN - 1) \
+                    & -TCG_TARGET_STACK_ALIGN)
+#define SAVE_OFS   (TCG_STATIC_CALL_ARGS_SIZE + TEMP_SIZE)
+
+/* We're expecting to be able to use an immediate for frame allocation.  */
+QEMU_BUILD_BUG_ON(FRAME_SIZE > 0x7ff);
+
+/* Generate global QEMU prologue and epilogue code */
+static void tcg_target_qemu_prologue(TCGContext *s)
+{
+    int i;
+
+    tcg_set_frame(s, TCG_REG_SP, TCG_STATIC_CALL_ARGS_SIZE, TEMP_SIZE);
+
+    /* TB prologue */
+    tcg_out_opc_imm(s, OPC_ADDI, TCG_REG_SP, TCG_REG_SP, -FRAME_SIZE);
+    for (i = 0; i < ARRAY_SIZE(tcg_target_callee_save_regs); i++) {
+        tcg_out_st(s, TCG_TYPE_REG, tcg_target_callee_save_regs[i],
+                   TCG_REG_SP, SAVE_OFS + i * REG_SIZE);
+    }
+
+#if !defined(CONFIG_SOFTMMU)
+    tcg_out_movi(s, TCG_TYPE_PTR, TCG_GUEST_BASE_REG, guest_base);
+    tcg_regset_set_reg(s->reserved_regs, TCG_GUEST_BASE_REG);
+#endif
+
+    /* Call generated code */
+    tcg_out_mov(s, TCG_TYPE_PTR, TCG_AREG0, tcg_target_call_iarg_regs[0]);
+    tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, tcg_target_call_iarg_regs[1], 0);
+
+    /* Return path for goto_ptr. Set return value to 0 */
+    s->code_gen_epilogue = s->code_ptr;
+    tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_A0, TCG_REG_ZERO);
+
+    /* TB epilogue */
+    tb_ret_addr = s->code_ptr;
+    for (i = 0; i < ARRAY_SIZE(tcg_target_callee_save_regs); i++) {
+        tcg_out_ld(s, TCG_TYPE_REG, tcg_target_callee_save_regs[i],
+                   TCG_REG_SP, SAVE_OFS + i * REG_SIZE);
+    }
+
+    tcg_out_opc_imm(s, OPC_ADDI, TCG_REG_SP, TCG_REG_SP, FRAME_SIZE);
+    tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, TCG_REG_RA, 0);
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 19/23] riscv: tcg-target: Add the prologue generation

[Richard Henderson]

On 11/15/18 11:36 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  tcg/riscv/tcg-target.inc.c | 66 ++++++++++++++++++++++++++++++++++++++
>  1 file changed, 66 insertions(+)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~

[Qemu-devel] [RFC v1 20/23] riscv: tcg-target: Add the target init code

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 tcg/riscv/tcg-target.inc.c | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/tcg/riscv/tcg-target.inc.c b/tcg/riscv/tcg-target.inc.c
index 4b8f65f367..dfc4e8b44c 100644
--- a/tcg/riscv/tcg-target.inc.c
+++ b/tcg/riscv/tcg-target.inc.c
@@ -1641,6 +1641,39 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, TCG_REG_RA, 0);
 }
 
+static void tcg_target_init(TCGContext *s)
+{
+    tcg_target_available_regs[TCG_TYPE_I32] = 0xffffffff;
+    if (TCG_TARGET_REG_BITS == 64) {
+        tcg_target_available_regs[TCG_TYPE_I64] = 0xffffffff;
+    }
+
+    tcg_target_call_clobber_regs = -1u;
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S0);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S1);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S2);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S3);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S4);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S5);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S6);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S7);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S8);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S9);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S10);
+    tcg_regset_reset_reg(tcg_target_call_clobber_regs, TCG_REG_S11);
+
+    s->reserved_regs = 0;
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_ZERO);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_TMP0);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_TMP1);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L0);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L1);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RA);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_SP);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_GP);
+    tcg_regset_set_reg(s->reserved_regs, TCG_REG_TP);
+}
+
 void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 20/23] riscv: tcg-target: Add the target init code

[Richard Henderson]

On 11/15/18 11:36 PM, Alistair Francis wrote:
> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L0);
> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L1);
> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RA);

Why are these three reserved?


r~

Re: [Qemu-devel] [RFC v1 20/23] riscv: tcg-target: Add the target init code

[Alistair Francis]

On Fri, Nov 16, 2018 at 9:26 AM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/15/18 11:36 PM, Alistair Francis wrote:
> > +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L0);
> > +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L1);
> > +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RA);
>
> Why are these three reserved?

Do these not need to be? I thought we had to reserve them.

Alistair

>
>
> r~

Re: [Qemu-devel] [RFC v1 20/23] riscv: tcg-target: Add the target init code

[Richard Henderson]

On 11/20/18 12:04 AM, Alistair Francis wrote:
> On Fri, Nov 16, 2018 at 9:26 AM Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> On 11/15/18 11:36 PM, Alistair Francis wrote:
>>> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L0);
>>> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L1);
>>> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RA);
>>
>> Why are these three reserved?
> 
> Do these not need to be? I thought we had to reserve them.

The return address, I presume, has been saved by the prologue.  I see no reason
why it can't be yet another call-clobbered register.

As for the other two... what are they supposed to be?


r~

Re: [Qemu-devel] [RFC v1 20/23] riscv: tcg-target: Add the target init code

[Alistair Francis]

On Mon, Nov 19, 2018 at 10:55 PM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/20/18 12:04 AM, Alistair Francis wrote:
> > On Fri, Nov 16, 2018 at 9:26 AM Richard Henderson
> > <richard.henderson@linaro.org> wrote:
> >>
> >> On 11/15/18 11:36 PM, Alistair Francis wrote:
> >>> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L0);
> >>> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_L1);
> >>> +    tcg_regset_set_reg(s->reserved_regs, TCG_REG_RA);
> >>
> >> Why are these three reserved?
> >
> > Do these not need to be? I thought we had to reserve them.
>
> The return address, I presume, has been saved by the prologue.  I see no reason
> why it can't be yet another call-clobbered register.

Ok, removed.

>
> As for the other two... what are they supposed to be?

They were old temp registers. L1 wasn't even used and I have replace
L0 with TMP0 as we have enough tmp registers.

Alistair

>
>
> r~

[Qemu-devel] [RFC v1 21/23] tcg: Add RISC-V cpu signal handler

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 accel/tcg/user-exec.c | 48 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)

diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c
index cd75829cf2..bb693484ed 100644
--- a/accel/tcg/user-exec.c
+++ b/accel/tcg/user-exec.c
@@ -571,6 +571,54 @@ int cpu_signal_handler(int host_signum, void *pinfo,
     return handle_cpu_signal(pc, info, is_write, &uc->uc_sigmask);
 }
 
+#elif defined(__riscv)
+
+int cpu_signal_handler(int host_signum, void *pinfo,
+                       void *puc)
+{
+    siginfo_t *info = pinfo;
+    ucontext_t *uc = puc;
+    greg_t pc = uc->uc_mcontext.__gregs[REG_PC];
+
+    /* Detect store by reading the instruction at the program
+       counter. Note: we currently only generate 32-bit
+       instructions so we thus only detect 32-bit stores */
+    uint32_t insn = *(uint32_t *)pc;
+    int is_write = 0;
+    switch (((insn >> 0) & 0b11)) {
+    case 3:
+        switch (((insn >> 2) & 0b11111)) {
+        case 8:
+            switch (((insn >> 12) & 0b111)) {
+            case 0: /* sb */
+            case 1: /* sh */
+            case 2: /* sw */
+            case 3: /* sd */
+            case 4: /* sq */
+                is_write = 1;
+                break;
+            default:
+                break;
+            }
+            break;
+        case 9:
+            switch (((insn >> 12) & 0b111)) {
+            case 2: /* fsw */
+            case 3: /* fsd */
+            case 4: /* fsq */
+                is_write = 1;
+                break;
+            default:
+                break;
+            }
+            break;
+        default:
+            break;
+        }
+    }
+    return handle_cpu_signal(pc, info, is_write, &uc->uc_sigmask);
+}
+
 #else
 
 #error host CPU specific signal handler needed
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 21/23] tcg: Add RISC-V cpu signal handler

[Richard Henderson]

On 11/15/18 11:37 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  accel/tcg/user-exec.c | 48 +++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 48 insertions(+)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~

Re: [Qemu-devel] [RFC v1 21/23] tcg: Add RISC-V cpu signal handler

[Richard Henderson]

On 11/15/18 11:37 PM, Alistair Francis wrote:
> +    /* Detect store by reading the instruction at the program
> +       counter. Note: we currently only generate 32-bit
> +       instructions so we thus only detect 32-bit stores */

Actually, you need to handle what the compiler generates too.  So, if
__riscv_compressed is defined, you need to handle it.


r~

[Qemu-devel] [RFC v1 22/23] dias: Add RISC-V support

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 disas.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/disas.c b/disas.c
index 5325b7e6be..82a408f272 100644
--- a/disas.c
+++ b/disas.c
@@ -522,8 +522,14 @@ void disas(FILE *out, void *code, unsigned long size)
 # ifdef _ARCH_PPC64
     s.info.cap_mode = CS_MODE_64;
 # endif
-#elif defined(__riscv__)
-    print_insn = print_insn_riscv;
+#elif defined(__riscv) && defined(CONFIG_RISCV_DIS)
+#if defined(_ILP32)
+    print_insn = print_insn_riscv32;
+#elif defined(_LP64)
+    print_insn = print_insn_riscv64;
+#else
+#error unsupported RISC-V ABI
+#endif
 #elif defined(__aarch64__) && defined(CONFIG_ARM_A64_DIS)
     print_insn = print_insn_arm_a64;
     s.info.cap_arch = CS_ARCH_ARM64;
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 22/23] dias: Add RISC-V support

[Richard Henderson]

On 11/15/18 11:37 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  disas.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~

[Qemu-devel] [RFC v1 23/23] configure: Add support for building RISC-V host

[Alistair Francis]

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Michael Clark <mjc@sifive.com>
---
 configure | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/configure b/configure
index 74e313a810..8a7b9fc933 100755
--- a/configure
+++ b/configure
@@ -700,6 +700,12 @@ elif check_define __s390__ ; then
   else
     cpu="s390"
   fi
+elif check_define __riscv ; then
+  if check_define _LP64 ; then
+    cpu="riscv64"
+  elif check_define _ILP32 ; then
+    cpu="riscv32"
+  fi
 elif check_define __arm__ ; then
   cpu="arm"
 elif check_define __aarch64__ ; then
@@ -712,7 +718,7 @@ ARCH=
 # Normalise host CPU name and set ARCH.
 # Note that this case should only have supported host CPUs, not guests.
 case "$cpu" in
-  ppc|ppc64|s390|s390x|sparc64|x32)
+  ppc|ppc64|s390|s390x|sparc64|x32|riscv32|riscv64)
     cpu="$cpu"
     supported_cpu="yes"
     eval "cross_cc_${cpu}=\$host_cc"
@@ -6809,6 +6815,8 @@ elif test "$ARCH" = "x86_64" -o "$ARCH" = "x32" ; then
   QEMU_INCLUDES="-iquote \$(SRC_PATH)/tcg/i386 $QEMU_INCLUDES"
 elif test "$ARCH" = "ppc64" ; then
   QEMU_INCLUDES="-iquote \$(SRC_PATH)/tcg/ppc $QEMU_INCLUDES"
+elif test "$ARCH" = "riscv32" -o "$ARCH" = "riscv64" ; then
+  QEMU_INCLUDES="-I\$(SRC_PATH)/tcg/riscv $QEMU_INCLUDES"
 else
   QEMU_INCLUDES="-iquote \$(SRC_PATH)/tcg/\$(ARCH) $QEMU_INCLUDES"
 fi
@@ -7306,7 +7314,7 @@ for i in $ARCH $TARGET_BASE_ARCH ; do
   ppc*)
     disas_config "PPC"
   ;;
-  riscv)
+  riscv*)
     disas_config "RISCV"
   ;;
   s390*)
-- 
2.19.1

Re: [Qemu-devel] [RFC v1 23/23] configure: Add support for building RISC-V host

[Richard Henderson]

On 11/15/18 11:37 PM, Alistair Francis wrote:
> Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Michael Clark <mjc@sifive.com>
> ---
>  configure | 12 ++++++++++--
>  1 file changed, 10 insertions(+), 2 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~

Re: [Qemu-devel] [RFC v1 00/23] Add RISC-V TCG backend support

[no-reply]

Hi,

This series seems to have some coding style problems. See output below for
more information:

Message-id: cover.1542321076.git.alistair.francis@wdc.com
Type: series
Subject: [Qemu-devel] [RFC v1 00/23]  Add RISC-V TCG backend support

=== TEST SCRIPT BEGIN ===
#!/bin/bash

BASE=base
n=1
total=$(git log --oneline $BASE.. | wc -l)
failed=0

git config --local diff.renamelimit 0
git config --local diff.renames True
git config --local diff.algorithm histogram

commits="$(git log --format=%H --reverse $BASE..)"
for c in $commits; do
    echo "Checking PATCH $n/$total: $(git log -n 1 --format=%s $c)..."
    if ! git show $c --format=email | ./scripts/checkpatch.pl --mailback -; then
        failed=1
        echo
    fi
    n=$((n+1))
done

exit $failed
=== TEST SCRIPT END ===

Updating 3c8cf5a9c21ff8782164d1def7f44bd888713384
Switched to a new branch 'test'
e0946be configure: Add support for building RISC-V host
769ab2a dias: Add RISC-V support
88c2947 tcg: Add RISC-V cpu signal handler
19e47a6 riscv: tcg-target: Add the target init code
506445c riscv: tcg-target: Add the prologue generation
496bf6b riscv: tcg-target: Add the out op decoder
7d5359d riscv: tcg-target: Add direct load and store instructions
25c5371 riscv: tcg-target: Add slowpath load and store instructions
4ca5f0c riscv: tcg-target: Add branch and jump instructions
cd8821b riscv: tcg-target: Add the out load and store instructions
58bb3e0 riscv: tcg-target: Add the extract instructions
a9dc143 riscv: tcg-target: Add the mov and movi instruction
4cf20d8 riscv: tcg-target: Add the relocation functions
78c1eb9 riscv: tcg-target: Add the instruction emitters
e277371 riscv: tcg-target: Add the immediate encoders
dcdb8bc riscv: tcg-target: Add support for the constraints
6a691a5 riscv: tcg-target: Regiser the JIT
b2165c4 riscv: Add the tcg target registers
243a714 riscv: Add the tcg-target header file
c2feca77 exec: Add RISC-V GCC poison macro
19ca815 linux-user: Add host dependency for RISC-V 64-bit
e84813f linux-user: Add host dependency for RISC-V 32-bit
a62d116 elf.h: Add the RISCV ELF magic numbers

=== OUTPUT BEGIN ===
Checking PATCH 1/23: elf.h: Add the RISCV ELF magic numbers...
Checking PATCH 2/23: linux-user: Add host dependency for RISC-V 32-bit...
WARNING: added, moved or deleted file(s), does MAINTAINERS need updating?
#11: 
new file mode 100644

total: 0 errors, 1 warnings, 11 lines checked

Your patch has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
Checking PATCH 3/23: linux-user: Add host dependency for RISC-V 64-bit...
WARNING: added, moved or deleted file(s), does MAINTAINERS need updating?
#11: 
new file mode 100644

total: 0 errors, 1 warnings, 11 lines checked

Your patch has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
Checking PATCH 4/23: exec: Add RISC-V GCC poison macro...
Checking PATCH 5/23: riscv: Add the tcg-target header file...
WARNING: added, moved or deleted file(s), does MAINTAINERS need updating?
#11: 
new file mode 100644

WARNING: architecture specific defines should be avoided
#43: FILE: tcg/riscv/tcg-target.h:28:
+#if __riscv_xlen == 32

total: 0 errors, 2 warnings, 173 lines checked

Your patch has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
Checking PATCH 6/23: riscv: Add the tcg target registers...
WARNING: added, moved or deleted file(s), does MAINTAINERS need updating?
#11: 
new file mode 100644

total: 0 errors, 1 warnings, 120 lines checked

Your patch has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
Checking PATCH 7/23: riscv: tcg-target: Regiser the JIT...
Checking PATCH 8/23: riscv: tcg-target: Add support for the constraints...
Checking PATCH 9/23: riscv: tcg-target: Add the immediate encoders...
Checking PATCH 10/23: riscv: tcg-target: Add the instruction emitters...
Checking PATCH 11/23: riscv: tcg-target: Add the relocation functions...
Checking PATCH 12/23: riscv: tcg-target: Add the mov and movi instruction...
Checking PATCH 13/23: riscv: tcg-target: Add the extract instructions...
Checking PATCH 14/23: riscv: tcg-target: Add the out load and store instructions...
Checking PATCH 15/23: riscv: tcg-target: Add branch and jump instructions...
Checking PATCH 16/23: riscv: tcg-target: Add slowpath load and store instructions...
Checking PATCH 17/23: riscv: tcg-target: Add direct load and store instructions...
ERROR: spaces required around that '*' (ctx:WxV)
#68: FILE: tcg/riscv/tcg-target.inc.c:1011:
+    tcg_insn_unit *label_ptr[1];
                   ^

ERROR: spaces required around that '*' (ctx:WxV)
#133: FILE: tcg/riscv/tcg-target.inc.c:1076:
+    tcg_insn_unit *label_ptr[1];
                   ^

total: 2 errors, 0 warnings, 151 lines checked

Your patch has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.

Checking PATCH 18/23: riscv: tcg-target: Add the out op decoder...
Checking PATCH 19/23: riscv: tcg-target: Add the prologue generation...
Checking PATCH 20/23: riscv: tcg-target: Add the target init code...
Checking PATCH 21/23: tcg: Add RISC-V cpu signal handler...
Checking PATCH 22/23: dias: Add RISC-V support...
Checking PATCH 23/23: configure: Add support for building RISC-V host...
=== OUTPUT END ===

Test command exited with code: 1


---
Email generated automatically by Patchew [http://patchew.org/].
Please send your feedback to patchew-devel@redhat.com