summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorSergey Fedorov <serge.fdrv@gmail.com>2016-04-22 19:08:47 +0300
committerRichard Henderson <rth@twiddle.net>2016-05-12 14:06:41 -1000
commit0d07abf05e98903c7faf204a9a90f7d45b7554dc (patch)
tree1090615eecc65e59f17b1f236b47495aaecf728e
parent399f1648573d49b9411089e4e2ea62b6357b166e (diff)
downloadqemu-0d07abf05e98903c7faf204a9a90f7d45b7554dc.zip
tcg/i386: Make direct jump patching thread-safe
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 <rth@twiddle.net>. Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com> Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org> Message-Id: <1461341333-19646-6-git-send-email-sergey.fedorov@linaro.org> Signed-off-by: Richard Henderson <rth@twiddle.net>
-rw-r--r--include/exec/exec-all.h2
-rw-r--r--tcg/i386/tcg-target.inc.c23
2 files changed, 24 insertions, 1 deletions
diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h
index d49befdba5..0ab7803259 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 007407c3fc..8d242a6665 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);