[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 1/4] target/i386: Use simple static property for
From: |
Igor Mammedov |
Subject: |
Re: [Qemu-devel] [PATCH 1/4] target/i386: Use simple static property for "model-id" |
Date: |
Tue, 18 Jul 2017 13:29:55 +0200 |
On Mon, 17 Jul 2017 14:18:27 -0300
Eduardo Habkost <address@hidden> wrote:
> On Mon, Jul 17, 2017 at 02:03:43PM +0200, Igor Mammedov wrote:
> > On Wed, 12 Jul 2017 13:20:55 -0300
> > Eduardo Habkost <address@hidden> wrote:
> >
> > > Instead of storing the raw CPUID data in CPUX86State and having
> > > to convert the data back and forth on the QOM getter/setter, use
> > > a simple static string property, and calculate CPUID data inside
> > > cpu_x86_cpuid().
> > wouldn't this slow down 'cpuid' instruction execution,
> > especially in tcg mode?
>
> It may add a few additional CPU cycles, but I really doubt we can
> find a workload where CPUID speed has measurable impact. See,
> for example, how expensive the kernel KVM CPUID code
> (kvm_cpuid(), kvm_find_cpuid_entry()) is.
I don't expect that it would affect KVM, but for TCG any instruction
execution is 'fast' path, so I'd leave current cpu_x86_cpuid()
not to loose those few cycles, it's not worth sacrifice for the sake of cleanup.
>
> Richard, what do you think?
>
>
> >
> > Why do you need this patch except of nice code reduction it gives?
>
> It's just for code reduction, and it is not needed for this
> series, if I recall correctly. It can be included later, if
> necessary.
>
> >
> > >
> > > Signed-off-by: Eduardo Habkost <address@hidden>
> > > ---
> > > target/i386/cpu.h | 2 +-
> > > target/i386/cpu.c | 62
> > > ++++++++++++++++---------------------------------------
> > > 2 files changed, 19 insertions(+), 45 deletions(-)
> > >
> > > diff --git a/target/i386/cpu.h b/target/i386/cpu.h
> > > index 7a228af..f56a3ea 100644
> > > --- a/target/i386/cpu.h
> > > +++ b/target/i386/cpu.h
> > > @@ -1151,7 +1151,7 @@ typedef struct CPUX86State {
> > > FeatureWordArray features;
> > > /* Features that were explicitly enabled/disabled */
> > > FeatureWordArray user_features;
> > > - uint32_t cpuid_model[12];
> > > + char *model_id;
> > >
> > > /* MTRRs */
> > > uint64_t mtrr_fixed[11];
> > > diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> > > index c571772..30b704c 100644
> > > --- a/target/i386/cpu.c
> > > +++ b/target/i386/cpu.c
> > > @@ -1828,43 +1828,6 @@ static void x86_cpuid_set_vendor(Object *obj,
> > > const char *value,
> > > }
> > > }
> > >
> > > -static char *x86_cpuid_get_model_id(Object *obj, Error **errp)
> > > -{
> > > - X86CPU *cpu = X86_CPU(obj);
> > > - CPUX86State *env = &cpu->env;
> > > - char *value;
> > > - int i;
> > > -
> > > - value = g_malloc(48 + 1);
> > > - for (i = 0; i < 48; i++) {
> > > - value[i] = env->cpuid_model[i >> 2] >> (8 * (i & 3));
> > > - }
> > > - value[48] = '\0';
> > > - return value;
> > > -}
> > > -
> > > -static void x86_cpuid_set_model_id(Object *obj, const char *model_id,
> > > - Error **errp)
> > > -{
> > > - X86CPU *cpu = X86_CPU(obj);
> > > - CPUX86State *env = &cpu->env;
> > > - int c, len, i;
> > > -
> > > - if (model_id == NULL) {
> > > - model_id = "";
> > > - }
> > > - len = strlen(model_id);
> > > - memset(env->cpuid_model, 0, 48);
> > > - for (i = 0; i < 48; i++) {
> > > - if (i >= len) {
> > > - c = '\0';
> > > - } else {
> > > - c = (uint8_t)model_id[i];
> > > - }
> > > - env->cpuid_model[i >> 2] |= c << (8 * (i & 3));
> > > - }
> > > -}
> > > -
> > > static void x86_cpuid_get_tsc_freq(Object *obj, Visitor *v, const char
> > > *name,
> > > void *opaque, Error **errp)
> > > {
> > > @@ -2899,10 +2862,23 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t
> > > index, uint32_t count,
> > > case 0x80000002:
> > > case 0x80000003:
> > > case 0x80000004:
> > > - *eax = env->cpuid_model[(index - 0x80000002) * 4 + 0];
> > > - *ebx = env->cpuid_model[(index - 0x80000002) * 4 + 1];
> > > - *ecx = env->cpuid_model[(index - 0x80000002) * 4 + 2];
> > > - *edx = env->cpuid_model[(index - 0x80000002) * 4 + 3];
> > > + {
> > > + int i;
> > > + int len = strlen(env->model_id);
> > > + uint32_t model[4] = { 0, 0, 0, 0 };
> > > +
> > > + for (i = 0; i < 16; i++) {
> > > + int p = (index - 0x80000002) * 16 + i;
> > > + if (p >= len) {
> > > + continue;
> > > + }
> > > + model[i >> 2] |= ((uint8_t)env->model_id[p]) << (8 * (i
> > > & 3));
> > > + }
> > > + *eax = model[0];
> > > + *ebx = model[1];
> > > + *ecx = model[2];
> > > + *edx = model[3];
> > > + }
> > > break;
> > > case 0x80000005:
> > > /* cache info (L1 cache) */
> > > @@ -3868,9 +3844,6 @@ static void x86_cpu_initfn(Object *obj)
> > > object_property_add_str(obj, "vendor",
> > > x86_cpuid_get_vendor,
> > > x86_cpuid_set_vendor, NULL);
> > > - object_property_add_str(obj, "model-id",
> > > - x86_cpuid_get_model_id,
> > > - x86_cpuid_set_model_id, NULL);
> > > object_property_add(obj, "tsc-frequency", "int",
> > > x86_cpuid_get_tsc_freq,
> > > x86_cpuid_set_tsc_freq, NULL, NULL, NULL);
> > > @@ -4004,6 +3977,7 @@ static Property x86_cpu_properties[] = {
> > > DEFINE_PROP_UINT32("phys-bits", X86CPU, phys_bits, 0),
> > > DEFINE_PROP_BOOL("host-phys-bits", X86CPU, host_phys_bits, false),
> > > DEFINE_PROP_BOOL("fill-mtrr-mask", X86CPU, fill_mtrr_mask, true),
> > > + DEFINE_PROP_STRING("model-id", X86CPU, env.model_id),
> > > DEFINE_PROP_UINT32("level", X86CPU, env.cpuid_level, UINT32_MAX),
> > > DEFINE_PROP_UINT32("xlevel", X86CPU, env.cpuid_xlevel, UINT32_MAX),
> > > DEFINE_PROP_UINT32("xlevel2", X86CPU, env.cpuid_xlevel2,
> > > UINT32_MAX),
> >
>
[Qemu-devel] [PATCH 2/4] target/i386: Use host_vendor_fms() in max_x86_cpu_initfn(), Eduardo Habkost, 2017/07/12
[Qemu-devel] [PATCH 3/4] target/i386: Define CPUID_MODEL_ID_SZ macro, Eduardo Habkost, 2017/07/12
[Qemu-devel] [PATCH 4/4] target/i386: Don't use x86_cpu_load_def() on "max" CPU model, Eduardo Habkost, 2017/07/12