[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [PULL 09/12] tcg: Increase hit rate of lookup_tb_ptr
From: |
Richard Henderson |
Subject: |
[Qemu-devel] [PULL 09/12] tcg: Increase hit rate of lookup_tb_ptr |
Date: |
Mon, 19 Jun 2017 11:18:36 -0700 |
We can call tb_htable_lookup even when the tb_jmp_cache is completely
empty. Therefore, un-nest most of the code dependent on tb != NULL
from the read from the cache.
This improves the hit rate of lookup_tb_ptr; for instance, when booting
and immediately shutting down debian-arm, the hit rate improves from
93.2% to 99.4%.
Reviewed-by: Alex Bennée <address@hidden>
Signed-off-by: Emilio G. Cota <address@hidden>
Signed-off-by: Richard Henderson <address@hidden>
---
tcg-runtime.c | 28 ++++++++++++++--------------
1 file changed, 14 insertions(+), 14 deletions(-)
diff --git a/tcg-runtime.c b/tcg-runtime.c
index 7fa90ce..ec3a34e 100644
--- a/tcg-runtime.c
+++ b/tcg-runtime.c
@@ -149,23 +149,23 @@ void *HELPER(lookup_tb_ptr)(CPUArchState *env,
target_ulong addr)
CPUState *cpu = ENV_GET_CPU(env);
TranslationBlock *tb;
target_ulong cs_base, pc;
- uint32_t flags;
-
- tb = atomic_rcu_read(&cpu->tb_jmp_cache[tb_jmp_cache_hash_func(addr)]);
- if (likely(tb)) {
- cpu_get_tb_cpu_state(env, &pc, &cs_base, &flags);
- if (likely(tb->pc == addr && tb->cs_base == cs_base &&
- tb->flags == flags)) {
- goto found;
- }
+ uint32_t flags, addr_hash;
+
+ addr_hash = tb_jmp_cache_hash_func(addr);
+ tb = atomic_rcu_read(&cpu->tb_jmp_cache[addr_hash]);
+ cpu_get_tb_cpu_state(env, &pc, &cs_base, &flags);
+
+ if (unlikely(!(tb
+ && tb->pc == addr
+ && tb->cs_base == cs_base
+ && tb->flags == flags))) {
tb = tb_htable_lookup(cpu, addr, cs_base, flags);
- if (likely(tb)) {
- atomic_set(&cpu->tb_jmp_cache[tb_jmp_cache_hash_func(addr)], tb);
- goto found;
+ if (!tb) {
+ return tcg_ctx.code_gen_epilogue;
}
+ atomic_set(&cpu->tb_jmp_cache[addr_hash], tb);
}
- return tcg_ctx.code_gen_epilogue;
- found:
+
qemu_log_mask_and_addr(CPU_LOG_EXEC, addr,
"Chain %p [%d: " TARGET_FMT_lx "] %s\n",
tb->tc_ptr, cpu->cpu_index, addr,
--
2.9.4
- [Qemu-devel] [PULL 00/12] TCG queued patches, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 01/12] util: add cacheinfo, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 03/12] translate-all: consolidate tb init in tb_gen_code, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 04/12] tcg/aarch64: Use ADR in tcg_out_movi, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 05/12] tcg/arm: Use indirect branch for goto_tb, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 06/12] tcg/arm: Remove limit on code buffer size, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 02/12] tcg: allocate TB structs before the corresponding translated code, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 09/12] tcg: Increase hit rate of lookup_tb_ptr,
Richard Henderson <=
- [Qemu-devel] [PULL 07/12] tcg/arm: Try pc-relative addresses for movi, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 08/12] tcg/arm: Use ldr (literal) for goto_tb, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 10/12] target/alpha: Use tcg_gen_lookup_and_goto_ptr, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 11/12] target/s390x: Exit after changing PSW mask, Richard Henderson, 2017/06/19
- [Qemu-devel] [PULL 12/12] target/arm: Exit after clearing aarch64 interrupt mask, Richard Henderson, 2017/06/19
- Re: [Qemu-devel] [PULL 00/12] TCG queued patches, Peter Maydell, 2017/06/22