@@ -330,15 +330,6 @@ static inline void tb_invalidate_phys_addr(AddressSpace *as, hwaddr addr)
#define CODE_GEN_AVG_BLOCK_SIZE 150
#endif
-#if defined(_ARCH_PPC) \
- || defined(__x86_64__) || defined(__i386__) \
- || defined(__sparc__) || defined(__aarch64__) \
- || defined(__s390x__) || defined(__mips__) \
- || defined(CONFIG_TCG_INTERPRETER)
-/* NOTE: Direct jump patching must be atomic to be thread-safe. */
-#define USE_DIRECT_JUMP
-#endif
-
struct TranslationBlock {
target_ulong pc; /* simulated PC corresponding to this block (EIP + CS base) */
target_ulong cs_base; /* CS base for this block */
@@ -376,11 +367,8 @@ struct TranslationBlock {
*/
uint16_t jmp_reset_offset[2]; /* offset of original jump target */
#define TB_JMP_RESET_OFFSET_INVALID 0xffff /* indicates no jump generated */
-#ifdef USE_DIRECT_JUMP
- uint16_t jmp_insn_offset[2]; /* offset of native jump instruction */
-#else
- uintptr_t jmp_target_addr[2]; /* target address for indirect jump */
-#endif
+ uintptr_t jmp_target_arg[2]; /* target address or offset */
+
/* Each TB has an assosiated circular list of TBs jumping to this one.
* jmp_list_first points to the first TB jumping to this one.
* jmp_list_next is used to point to the next TB in a list.
@@ -402,84 +390,7 @@ void tb_flush(CPUState *cpu);
void tb_phys_invalidate(TranslationBlock *tb, tb_page_addr_t page_addr);
TranslationBlock *tb_htable_lookup(CPUState *cpu, target_ulong pc,
target_ulong cs_base, uint32_t flags);
-
-#if defined(USE_DIRECT_JUMP)
-
-#if defined(CONFIG_TCG_INTERPRETER)
-static inline void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr)
-{
- /* patch the branch destination */
- atomic_set((int32_t *)jmp_addr, addr - (jmp_addr + 4));
- /* no need to flush icache explicitly */
-}
-#elif defined(_ARCH_PPC)
-void ppc_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr);
-#define tb_set_jmp_target1 ppc_tb_set_jmp_target
-#elif defined(__i386__) || defined(__x86_64__)
-static inline void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr)
-{
- /* patch the branch destination */
- atomic_set((int32_t *)jmp_addr, addr - (jmp_addr + 4));
- /* no need to flush icache explicitly */
-}
-#elif defined(__s390x__)
-static inline void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr)
-{
- /* patch the branch destination */
- intptr_t disp = addr - (jmp_addr - 2);
- atomic_set((int32_t *)jmp_addr, disp / 2);
- /* no need to flush icache explicitly */
-}
-#elif defined(__aarch64__)
-void aarch64_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr);
-#define tb_set_jmp_target1 aarch64_tb_set_jmp_target
-#elif defined(__sparc__) || defined(__mips__)
-void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr);
-#else
-#error tb_set_jmp_target1 is missing
-#endif
-
-static inline void tb_set_jmp_target(TranslationBlock *tb,
- int n, uintptr_t addr)
-{
- uint16_t offset = tb->jmp_insn_offset[n];
- tb_set_jmp_target1((uintptr_t)(tb->tc_ptr + offset), addr);
-}
-
-#else
-
-/* set the jump target */
-static inline void tb_set_jmp_target(TranslationBlock *tb,
- int n, uintptr_t addr)
-{
- tb->jmp_target_addr[n] = addr;
-}
-
-#endif
-
-/* Called with tb_lock held. */
-static inline void tb_add_jump(TranslationBlock *tb, int n,
- TranslationBlock *tb_next)
-{
- assert(n < ARRAY_SIZE(tb->jmp_list_next));
- if (tb->jmp_list_next[n]) {
- /* Another thread has already done this while we were
- * outside of the lock; nothing to do in this case */
- return;
- }
- qemu_log_mask_and_addr(CPU_LOG_EXEC, tb->pc,
- "Linking TBs %p [" TARGET_FMT_lx
- "] index %d -> %p [" TARGET_FMT_lx "]\n",
- tb->tc_ptr, tb->pc, n,
- tb_next->tc_ptr, tb_next->pc);
-
- /* patch the native jump address */
- tb_set_jmp_target(tb, n, (uintptr_t)tb_next->tc_ptr);
-
- /* add in TB jmp circular list */
- tb->jmp_list_next[n] = tb_next->jmp_list_first;
- tb_next->jmp_list_first = (uintptr_t)tb | n;
-}
+void tb_set_jmp_target(TranslationBlock *tb, int n, uintptr_t addr);
/* GETPC is the true target of the return instruction that we'll execute. */
#if defined(CONFIG_TCG_INTERPRETER)
@@ -111,10 +111,13 @@ typedef enum {
#define TCG_TARGET_HAS_muls2_i64 0
#define TCG_TARGET_HAS_muluh_i64 1
#define TCG_TARGET_HAS_mulsh_i64 1
+#define TCG_TARGET_HAS_direct_jump 1
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);
+
#endif /* AARCH64_TCG_TARGET_H */
@@ -124,6 +124,7 @@ extern bool use_idiv_instructions;
#define TCG_TARGET_HAS_div_i32 use_idiv_instructions
#define TCG_TARGET_HAS_rem_i32 0
#define TCG_TARGET_HAS_goto_ptr 1
+#define TCG_TARGET_HAS_direct_jump 0
enum {
TCG_AREG0 = TCG_REG_R6,
@@ -134,4 +135,7 @@ static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
__builtin___clear_cache((char *) start, (char *) stop);
}
+/* not defined -- call should be eliminated at compile time */
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+
#endif
@@ -108,6 +108,7 @@ extern bool have_popcnt;
#define TCG_TARGET_HAS_muluh_i32 0
#define TCG_TARGET_HAS_mulsh_i32 0
#define TCG_TARGET_HAS_goto_ptr 1
+#define TCG_TARGET_HAS_direct_jump 1
#if TCG_TARGET_REG_BITS == 64
#define TCG_TARGET_HAS_extrl_i64_i32 0
@@ -166,6 +167,14 @@ static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
{
}
+static inline void tb_target_set_jmp_target(uintptr_t tc_ptr,
+ uintptr_t jmp_addr, uintptr_t addr)
+{
+ /* patch the branch destination */
+ atomic_set((int32_t *)jmp_addr, addr - (jmp_addr + 4));
+ /* no need to flush icache explicitly */
+}
+
/* This defines the natural memory order supported by this
* architecture before guarantees made by various barrier
* instructions.
@@ -174,6 +174,7 @@ typedef enum {
#define TCG_TARGET_HAS_extrl_i64_i32 0
#define TCG_TARGET_HAS_extrh_i64_i32 0
#define TCG_TARGET_HAS_goto_ptr 0
+#define TCG_TARGET_HAS_direct_jump 0
#define TCG_TARGET_deposit_i32_valid(ofs, len) ((len) <= 16)
#define TCG_TARGET_deposit_i64_valid(ofs, len) ((len) <= 16)
@@ -195,4 +196,7 @@ static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
asm volatile (";;sync.i;;srlz.i;;");
}
+/* not defined -- call should be eliminated at compile time */
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+
#endif
@@ -131,6 +131,7 @@ extern bool use_mips32r2_instructions;
#define TCG_TARGET_HAS_mulsh_i32 1
#define TCG_TARGET_HAS_bswap32_i32 1
#define TCG_TARGET_HAS_goto_ptr 1
+#define TCG_TARGET_HAS_direct_jump 1
#if TCG_TARGET_REG_BITS == 64
#define TCG_TARGET_HAS_add2_i32 0
@@ -206,4 +207,6 @@ static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
cacheflush ((void *)start, stop-start, ICACHE);
}
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+
#endif
@@ -83,6 +83,7 @@ extern bool have_isa_3_00;
#define TCG_TARGET_HAS_muluh_i32 1
#define TCG_TARGET_HAS_mulsh_i32 1
#define TCG_TARGET_HAS_goto_ptr 1
+#define TCG_TARGET_HAS_direct_jump 1
#if TCG_TARGET_REG_BITS == 64
#define TCG_TARGET_HAS_add2_i32 0
@@ -124,5 +125,6 @@ extern bool have_isa_3_00;
#endif
void flush_icache_range(uintptr_t start, uintptr_t stop);
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
#endif
@@ -95,6 +95,7 @@ extern uint64_t s390_facilities;
#define TCG_TARGET_HAS_extrl_i64_i32 0
#define TCG_TARGET_HAS_extrh_i64_i32 0
#define TCG_TARGET_HAS_goto_ptr 1
+#define TCG_TARGET_HAS_direct_jump 1
#define TCG_TARGET_HAS_div2_i64 1
#define TCG_TARGET_HAS_rot_i64 1
@@ -143,4 +144,13 @@ static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
{
}
+static inline void tb_target_set_jmp_target(uintptr_t tc_ptr,
+ uintptr_t jmp_addr, uintptr_t addr)
+{
+ /* patch the branch destination */
+ intptr_t disp = addr - (jmp_addr - 2);
+ atomic_set((int32_t *)jmp_addr, disp / 2);
+ /* no need to flush icache explicitly */
+}
+
#endif
@@ -124,6 +124,7 @@ extern bool use_vis3_instructions;
#define TCG_TARGET_HAS_muluh_i32 0
#define TCG_TARGET_HAS_mulsh_i32 0
#define TCG_TARGET_HAS_goto_ptr 1
+#define TCG_TARGET_HAS_direct_jump 1
#define TCG_TARGET_HAS_extrl_i64_i32 1
#define TCG_TARGET_HAS_extrh_i64_i32 1
@@ -170,4 +171,6 @@ static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
}
}
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+
#endif
@@ -652,8 +652,8 @@ struct TCGContext {
/* goto_tb support */
tcg_insn_unit *code_buf;
uint16_t *tb_jmp_reset_offset; /* tb->jmp_reset_offset */
- uint16_t *tb_jmp_insn_offset; /* tb->jmp_insn_offset if USE_DIRECT_JUMP */
- uintptr_t *tb_jmp_target_addr; /* tb->jmp_target_addr if !USE_DIRECT_JUMP */
+ uintptr_t *tb_jmp_insn_offset; /* tb->jmp_target_arg if direct_jump */
+ uintptr_t *tb_jmp_target_addr; /* tb->jmp_target_arg if !direct_jump */
TCGRegSet reserved_regs;
intptr_t current_frame_offset;
@@ -86,6 +86,7 @@
#define TCG_TARGET_HAS_muluh_i32 0
#define TCG_TARGET_HAS_mulsh_i32 0
#define TCG_TARGET_HAS_goto_ptr 0
+#define TCG_TARGET_HAS_direct_jump 1
#if TCG_TARGET_REG_BITS == 64
#define TCG_TARGET_HAS_extrl_i64_i32 0
@@ -192,4 +193,12 @@ static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
{
}
+static inline void tb_target_set_jmp_target(uintptr_t tc_ptr,
+ uintptr_t jmp_addr, uintptr_t addr)
+{
+ /* patch the branch destination */
+ atomic_set((int32_t *)jmp_addr, addr - (jmp_addr + 4));
+ /* no need to flush icache explicitly */
+}
+
#endif /* TCG_TARGET_H */
@@ -329,6 +329,41 @@ TranslationBlock *tb_htable_lookup(CPUState *cpu, target_ulong pc,
return qht_lookup(&tcg_ctx.tb_ctx.htable, tb_cmp, &desc, h);
}
+void tb_set_jmp_target(TranslationBlock *tb, int n, uintptr_t addr)
+{
+ if (TCG_TARGET_HAS_direct_jump) {
+ uintptr_t offset = tb->jmp_target_arg[n];
+ uintptr_t tc_ptr = (uintptr_t)tb->tc_ptr;
+ tb_target_set_jmp_target(tc_ptr, tc_ptr + offset, addr);
+ } else {
+ tb->jmp_target_arg[n] = addr;
+ }
+}
+
+/* Called with tb_lock held. */
+static inline void tb_add_jump(TranslationBlock *tb, int n,
+ TranslationBlock *tb_next)
+{
+ assert(n < ARRAY_SIZE(tb->jmp_list_next));
+ if (tb->jmp_list_next[n]) {
+ /* Another thread has already done this while we were
+ * outside of the lock; nothing to do in this case */
+ return;
+ }
+ qemu_log_mask_and_addr(CPU_LOG_EXEC, tb->pc,
+ "Linking TBs %p [" TARGET_FMT_lx
+ "] index %d -> %p [" TARGET_FMT_lx "]\n",
+ tb->tc_ptr, tb->pc, n,
+ tb_next->tc_ptr, tb_next->pc);
+
+ /* patch the native jump address */
+ tb_set_jmp_target(tb, n, (uintptr_t)tb_next->tc_ptr);
+
+ /* add in TB jmp circular list */
+ tb->jmp_list_next[n] = tb_next->jmp_list_first;
+ tb_next->jmp_list_first = (uintptr_t)tb | n;
+}
+
static inline TranslationBlock *tb_find(CPUState *cpu,
TranslationBlock *last_tb,
int tb_exit)
@@ -1289,13 +1289,13 @@ TranslationBlock *tb_gen_code(CPUState *cpu,
tb->jmp_reset_offset[0] = TB_JMP_RESET_OFFSET_INVALID;
tb->jmp_reset_offset[1] = TB_JMP_RESET_OFFSET_INVALID;
tcg_ctx.tb_jmp_reset_offset = tb->jmp_reset_offset;
-#ifdef USE_DIRECT_JUMP
- tcg_ctx.tb_jmp_insn_offset = tb->jmp_insn_offset;
- tcg_ctx.tb_jmp_target_addr = NULL;
-#else
- tcg_ctx.tb_jmp_insn_offset = NULL;
- tcg_ctx.tb_jmp_target_addr = tb->jmp_target_addr;
-#endif
+ if (TCG_TARGET_HAS_direct_jump) {
+ tcg_ctx.tb_jmp_insn_offset = tb->jmp_target_arg;
+ tcg_ctx.tb_jmp_target_addr = NULL;
+ } else {
+ tcg_ctx.tb_jmp_insn_offset = NULL;
+ tcg_ctx.tb_jmp_target_addr = tb->jmp_target_arg;
+ }
#ifdef CONFIG_PROFILER
tcg_ctx.tb_count++;
@@ -871,9 +871,8 @@ static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *target)
}
}
-#ifdef USE_DIRECT_JUMP
-
-void aarch64_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
+ uintptr_t addr)
{
tcg_insn_unit i1, i2;
TCGType rt = TCG_TYPE_I64;
@@ -898,8 +897,6 @@ void aarch64_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr)
flush_icache_range(jmp_addr, jmp_addr + 8);
}
-#endif
-
static inline void tcg_out_goto_label(TCGContext *s, TCGLabel *l)
{
if (!l->has_value) {
@@ -1412,7 +1409,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
case INDEX_op_goto_tb:
if (s->tb_jmp_insn_offset != NULL) {
- /* USE_DIRECT_JUMP */
+ /* TCG_TARGET_HAS_direct_jump */
/* Ensure that ADRP+ADD are 8-byte aligned so that an atomic
write can be used to patch the target address. */
if ((uintptr_t)s->code_ptr & 7) {
@@ -1420,11 +1417,11 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
}
s->tb_jmp_insn_offset[a0] = tcg_current_code_size(s);
/* actual branch destination will be patched by
- aarch64_tb_set_jmp_target later. */
+ tb_target_set_jmp_target later. */
tcg_out_insn(s, 3406, ADRP, TCG_REG_TMP, 0);
tcg_out_insn(s, 3401, ADDI, TCG_TYPE_I64, TCG_REG_TMP, TCG_REG_TMP, 0);
} else {
- /* !USE_DIRECT_JUMP */
+ /* !TCG_TARGET_HAS_direct_jump */
tcg_debug_assert(s->tb_jmp_target_addr != NULL);
intptr_t offset = tcg_pcrel_diff(s, (s->tb_jmp_target_addr + a0)) >> 2;
tcg_out_insn(s, 3305, LDR, offset, TCG_REG_TMP);
@@ -2642,7 +2642,8 @@ static void tcg_target_init(TCGContext *s)
tcg_regset_set_reg(s->reserved_regs, TCG_REG_GP); /* global pointer */
}
-void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
+ uintptr_t addr)
{
atomic_set((uint32_t *)jmp_addr, deposit32(OPC_J, 0, 26, addr >> 2));
flush_icache_range(jmp_addr, jmp_addr + 4);
@@ -1296,7 +1296,8 @@ static void tcg_out_mb(TCGContext *s, TCGArg a0)
}
#ifdef __powerpc64__
-void ppc_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
+ uintptr_t addr)
{
tcg_insn_unit i1, i2;
uint64_t pair;
@@ -1328,7 +1329,8 @@ void ppc_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr)
flush_icache_range(jmp_addr, jmp_addr + 8);
}
#else
-void ppc_tb_set_jmp_target(uintptr_t jmp_addr, uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
+ uintptr_t addr)
{
intptr_t diff = addr - jmp_addr;
tcg_debug_assert(in_range_b(diff));
@@ -1708,7 +1708,8 @@ void tcg_register_jit(void *buf, size_t buf_size)
tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
}
-void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
+ uintptr_t addr)
{
uint32_t *ptr = (uint32_t *)jmp_addr;
uintptr_t disp = addr - jmp_addr;
Replace the USE_DIRECT_JUMP ifdef with a TCG_TARGET_HAS_direct_jump boolean test. Replace the tb_set_jmp_target1 ifdef with an unconditional function tb_target_set_jmp_target. While we're touching all backends, add a parameter for tb->tc_ptr; we're going to need it shortly for some backends. Move tb_set_jmp_target and tb_add_jump from exec-all.h to cpu-exec.c. This opens the possibility for TCG_TARGET_HAS_direct_jump to be a runtime decision -- based on host cpu capabilities, the size of code_gen_buffer, or a future debugging switch. Signed-off-by: Richard Henderson <rth@twiddle.net> --- include/exec/exec-all.h | 95 ++------------------------------------------ tcg/aarch64/tcg-target.h | 3 ++ tcg/arm/tcg-target.h | 4 ++ tcg/i386/tcg-target.h | 9 +++++ tcg/ia64/tcg-target.h | 4 ++ tcg/mips/tcg-target.h | 3 ++ tcg/ppc/tcg-target.h | 2 + tcg/s390/tcg-target.h | 10 +++++ tcg/sparc/tcg-target.h | 3 ++ tcg/tcg.h | 4 +- tcg/tci/tcg-target.h | 9 +++++ accel/tcg/cpu-exec.c | 35 ++++++++++++++++ accel/tcg/translate-all.c | 14 +++---- tcg/aarch64/tcg-target.inc.c | 13 +++--- tcg/mips/tcg-target.inc.c | 3 +- tcg/ppc/tcg-target.inc.c | 6 ++- tcg/sparc/tcg-target.inc.c | 3 +- 17 files changed, 107 insertions(+), 113 deletions(-)