[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [PULL 06/39] tcg/i386: Make direct jump patching thread-saf
From: |
Richard Henderson |
Subject: |
[Qemu-devel] [PULL 06/39] tcg/i386: Make direct jump patching thread-safe |
Date: |
Thu, 12 May 2016 14:13:07 -1000 |
From: Sergey Fedorov <address@hidden>
Ensure direct jump patching in i386 is atomic by:
* naturally aligning a location of direct jump address;
* using atomic_read()/atomic_set() for code patching.
tcg_out_nopn() implementation:
Suggested-by: Richard Henderson <address@hidden>.
Signed-off-by: Sergey Fedorov <address@hidden>
Signed-off-by: Sergey Fedorov <address@hidden>
Message-Id: <address@hidden>
Signed-off-by: Richard Henderson <address@hidden>
---
include/exec/exec-all.h | 2 +-
tcg/i386/tcg-target.inc.c | 23 +++++++++++++++++++++++
2 files changed, 24 insertions(+), 1 deletion(-)
diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h
index d49befd..0ab7803 100644
--- a/include/exec/exec-all.h
+++ b/include/exec/exec-all.h
@@ -313,7 +313,7 @@ void ppc_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t
addr);
static inline void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr)
{
/* patch the branch destination */
- stl_le_p((void*)jmp_addr, addr - (jmp_addr + 4));
+ atomic_set((int32_t *)jmp_addr, addr - (jmp_addr + 4));
/* no need to flush icache explicitly */
}
#elif defined(__s390x__)
diff --git a/tcg/i386/tcg-target.inc.c b/tcg/i386/tcg-target.inc.c
index 007407c..8d242a6 100644
--- a/tcg/i386/tcg-target.inc.c
+++ b/tcg/i386/tcg-target.inc.c
@@ -1123,6 +1123,21 @@ static void tcg_out_jmp(TCGContext *s, tcg_insn_unit
*dest)
tcg_out_branch(s, 0, dest);
}
+static void tcg_out_nopn(TCGContext *s, int n)
+{
+ int i;
+ /* Emit 1 or 2 operand size prefixes for the standard one byte nop,
+ * "xchg %eax,%eax", forming "xchg %ax,%ax". All cores accept the
+ * duplicate prefix, and all of the interesting recent cores can
+ * decode and discard the duplicates in a single cycle.
+ */
+ tcg_debug_assert(n >= 1);
+ for (i = 1; i < n; ++i) {
+ tcg_out8(s, 0x66);
+ }
+ tcg_out8(s, 0x90);
+}
+
#if defined(CONFIG_SOFTMMU)
/* helper signature: helper_ret_ld_mmu(CPUState *env, target_ulong addr,
* int mmu_idx, uintptr_t ra)
@@ -1777,6 +1792,14 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode
opc,
case INDEX_op_goto_tb:
if (s->tb_jmp_offset) {
/* direct jump method */
+ int gap;
+ /* jump displacement must be aligned for atomic patching;
+ * see if we need to add extra nops before jump
+ */
+ gap = tcg_pcrel_diff(s, QEMU_ALIGN_PTR_UP(s->code_ptr + 1, 4));
+ if (gap != 1) {
+ tcg_out_nopn(s, gap - 1);
+ }
tcg_out8(s, OPC_JMP_long); /* jmp im */
s->tb_jmp_offset[args[0]] = tcg_current_code_size(s);
tcg_out32(s, 0);
--
2.5.5
- [Qemu-devel] [PULL 00/39] tcg-next patch queue, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 02/39] include/qemu/osdep.h: Add a macro to check for alignment, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 03/39] include/qemu/osdep.h: Add macros for pointer alignment, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 01/39] tb: consistently use uint32_t for tb->flags, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 04/39] tci: Make direct jump patching thread-safe, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 06/39] tcg/i386: Make direct jump patching thread-safe,
Richard Henderson <=
- [Qemu-devel] [PULL 08/39] tcg/arm: Make direct jump patching thread-safe, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 10/39] tcg/sparc: Make direct jump patching thread-safe, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 05/39] tcg/ppc: Make direct jump patching thread-safe, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 07/39] tcg/s390: Make direct jump patching thread-safe, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 09/39] tcg/aarch64: Make direct jump patching thread-safe, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 11/39] tcg/mips: Make direct jump patching thread-safe, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 12/39] tcg: Note requirement on atomic direct jump patching, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 13/39] translate-all: remove redundant setting of tcg_ctx.code_gen_buffer_size, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 15/39] translate-all: Adjust 256mb testing for mips64, Richard Henderson, 2016/05/12
- [Qemu-devel] [PULL 14/39] translate-all: add missing munmap of the code_gen guard page for MIPS, Richard Henderson, 2016/05/12