| Message ID | 20110525110938.03bdc229.takuya.yoshikawa@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 05/25/2011 05:09 AM, Takuya Yoshikawa wrote: > From: Takuya Yoshikawa<yoshikawa.takuya@oss.ntt.co.jp> > > During tracing the emulator, we noticed that init_emulate_ctxt() > sometimes took a bit longer time than we expected. > > This patch is for mitigating the problem by some degree. > > By looking into the function, we soon notice that it clears the whole > decode_cache whose size is about 2.5K bytes now. Furthermore, most of > the bytes are taken for the two read_cache arrays, which are used only > by a few instructions. > > Considering the fact that we are not assuming the cache arrays have > been cleared when we store actual data, we do not need to clear the > arrays: 2K bytes elimination. In addition, we can avoid clearing the > fetch_cache and regs arrays. > > This patch changes the initialization not to clear the arrays. > > On our 64-bit host, init_emulate_ctxt() becomes 0.3 to 0.5us faster with > this patch applied. > Thanks, applied. It strikes me that initializing the emulator in x86.c is the wrong thing. We should move the entire thing to x86_decode_insn(). We'll need a few more callbacks for that, though (register, eflags); eventually we can read just registers that are used and write only registers that were updated.
On Thu, May 26, 2011 at 11:19:03AM +0300, Avi Kivity wrote: > On 05/25/2011 05:09 AM, Takuya Yoshikawa wrote: > >From: Takuya Yoshikawa<yoshikawa.takuya@oss.ntt.co.jp> > > > >During tracing the emulator, we noticed that init_emulate_ctxt() > >sometimes took a bit longer time than we expected. > > > >This patch is for mitigating the problem by some degree. > > > >By looking into the function, we soon notice that it clears the whole > >decode_cache whose size is about 2.5K bytes now. Furthermore, most of > >the bytes are taken for the two read_cache arrays, which are used only > >by a few instructions. > > > >Considering the fact that we are not assuming the cache arrays have > >been cleared when we store actual data, we do not need to clear the > >arrays: 2K bytes elimination. In addition, we can avoid clearing the > >fetch_cache and regs arrays. > > > >This patch changes the initialization not to clear the arrays. > > > >On our 64-bit host, init_emulate_ctxt() becomes 0.3 to 0.5us faster with > >this patch applied. > > > > Thanks, applied. > > It strikes me that initializing the emulator in x86.c is the wrong > thing. We should move the entire thing to x86_decode_insn(). > We initialize it in task switch and interrupt injection code too. > We'll need a few more callbacks for that, though (register, eflags); > eventually we can read just registers that are used and write only > registers that were updated. > -- Gleb. -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/arch/x86/include/asm/kvm_emulate.h b/arch/x86/include/asm/kvm_emulate.h index ab09ba2..c0f77e0 100644 --- a/arch/x86/include/asm/kvm_emulate.h +++ b/arch/x86/include/asm/kvm_emulate.h @@ -246,8 +246,6 @@ struct decode_cache { unsigned int d; int (*execute)(struct x86_emulate_ctxt *ctxt); int (*check_perm)(struct x86_emulate_ctxt *ctxt); - unsigned long regs[NR_VCPU_REGS]; - unsigned long eip; /* modrm */ u8 modrm; u8 modrm_mod; @@ -255,6 +253,9 @@ struct decode_cache { u8 modrm_rm; u8 modrm_seg; bool rip_relative; + unsigned long eip; + /* Fields above regs are cleared together. */ + unsigned long regs[NR_VCPU_REGS]; struct fetch_cache fetch; struct read_cache io_read; struct read_cache mem_read; diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index e7d337e..57f4160 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -4506,6 +4506,20 @@ static void inject_emulated_exception(struct kvm_vcpu *vcpu) kvm_queue_exception(vcpu, ctxt->exception.vector); } +static void init_decode_cache(struct decode_cache *c, + const unsigned long *regs) +{ + memset(c, 0, offsetof(struct decode_cache, regs)); + memcpy(c->regs, regs, sizeof(c->regs)); + + c->fetch.start = 0; + c->fetch.end = 0; + c->io_read.pos = 0; + c->io_read.end = 0; + c->mem_read.pos = 0; + c->mem_read.end = 0; +} + static void init_emulate_ctxt(struct kvm_vcpu *vcpu) { struct x86_emulate_ctxt *ctxt = &vcpu->arch.emulate_ctxt; @@ -4531,8 +4545,7 @@ static void init_emulate_ctxt(struct kvm_vcpu *vcpu) X86EMUL_MODE_PROT16; ctxt->guest_mode = is_guest_mode(vcpu); - memset(c, 0, sizeof(struct decode_cache)); - memcpy(c->regs, vcpu->arch.regs, sizeof c->regs); + init_decode_cache(c, vcpu->arch.regs); vcpu->arch.emulate_regs_need_sync_from_vcpu = false; }