qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Jing Liu <jing2.liu@linux.intel.com>
To: Paolo Bonzini <pbonzini@redhat.com>, qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH v1] x86: Intel AVX512_BF16 feature enabling
Date: Fri, 19 Jul 2019 15:20:05 +0800	[thread overview]
Message-ID: <0ffa32db-3615-16e0-71bf-a8c552c2fcfb@linux.intel.com> (raw)
In-Reply-To: <0f9763fe-af69-e057-9fb1-00527e636fe4@redhat.com>



On 7/18/2019 4:15 PM, Paolo Bonzini wrote:
> On 18/07/19 06:55, Jing Liu wrote:
>>>
>>> +            *eax = kvm_arch_get_supported_cpuid(cs->kvm_state, 0x7,
>>> +                                                count, R_EAX);
>> This needs to be firstly checked as follows, otherwise some
>> architectures would fail to compile.
>>
>> What about hvf and tcg CPUID 07 EAX value?
>>
>> +            /* Maximum ECX value for sub-leaves */
>> +            if (kvm_enabled()) {
>> +                *eax = kvm_arch_get_supported_cpuid(cs->kvm_state, 0x7,
>> +                                                    count, R_EAX);
>> +            } else if (hvf_enabled()) {
>> +                *eax = hvf_get_supported_cpuid(0x7, count, R_EAX);
>> +            } else {
>> +                *eax = 0;
>> +            }
>>
> 
> Good question.  You need to add a new property, for example
> cpuid_level_func7, whose code would be modeled around cpuid_level (and a
> field cpuid_min_level_func7 whose code would be modeled around
> cpuid_min_level).
> 
> Then CPUID[7,0].EAX is set automatically to 0 or 1 depending on whether
> BF16 is enabled or not.

Could I ask why don't we directly check BF16 enabling when
cpu_x86_cpuid(env, 7, 0, ...) during kvm_arch_init_vcpu ?

What is the use of the two new properties? Are they used for users
setting parameters when boot up guest, and why we need users setting 
func7 level?

I tried to implement the code as follows.

@@ -4293,13 +4313,19 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t 
index, uint32_t count,
      case 7:
          /* Structured Extended Feature Flags Enumeration Leaf */
          if (count == 0) {
-            *eax = 0; /* Maximum ECX value for sub-leaves */
+            /* Maximum ECX value for sub-leaves */
+            *eax = env->cpuid_level_func7;
[...]
+        } else if (count == 1) {
+            *eax = env->features[FEAT_7_1_EAX];
+            *ebx = 0;
+            *ecx = 0;
+            *edx = 0;
[...]
@@ -5075,6 +5101,10 @@ static void x86_cpu_expand_features(X86CPU *cpu, 
Error **errp)
          x86_cpu_adjust_feat_level(cpu, FEAT_SVM);
          x86_cpu_adjust_feat_level(cpu, FEAT_XSAVE);

+       if ((env->features[FEAT_7_1_EAX] & CPUID_7_1_EAX_AVX512_BF16) &&
+            kvm_enabled()) {
+            x86_cpu_adjust_level(cpu, &env->cpuid_min_level_func7, 1);
+        }
          /* Intel Processor Trace requires CPUID[0x14] */
          if ((env->features[FEAT_7_0_EBX] & CPUID_7_0_EBX_INTEL_PT) &&
               kvm_enabled() && cpu->intel_pt_auto_level) {
@@ -5098,6 +5128,9 @@ static void x86_cpu_expand_features(X86CPU *cpu, 
Error **errp)
      }

      /* Set cpuid_*level* based on cpuid_min_*level, if not explicitly 
set */
+    if (env->cpuid_level_func7 == UINT32_MAX) {
+        env->cpuid_level_func7 = env->cpuid_min_level_func7;
+    }
      if (env->cpuid_level == UINT32_MAX) {
          env->cpuid_level = env->cpuid_min_level;
      }
@@ -5869,9 +5902,11 @@ static Property x86_cpu_properties[] = {
      DEFINE_PROP_BOOL("host-phys-bits", X86CPU, host_phys_bits, false),
      DEFINE_PROP_UINT8("host-phys-bits-limit", X86CPU, 
host_phys_bits_limit, 0),
      DEFINE_PROP_BOOL("fill-mtrr-mask", X86CPU, fill_mtrr_mask, true),
+    DEFINE_PROP_UINT32("level-func7", X86CPU, env.cpuid_level_func7, 
UINT32_MAX),
      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),
+    DEFINE_PROP_UINT32("min-level-func7", X86CPU, 
env.cpuid_min_level_func7, 0),
[...]

Thanks,
Jing


> 
> Paolo
> 


  reply	other threads:[~2019-07-19  7:20 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-07-11  5:38 [Qemu-devel] [PATCH v1] x86: Intel AVX512_BF16 feature enabling Jing Liu
2019-07-11  5:47 ` no-reply
2019-07-18  4:55 ` Jing Liu
2019-07-18  8:15   ` Paolo Bonzini
2019-07-19  7:20     ` Jing Liu [this message]
2019-07-19  8:10       ` Paolo Bonzini
2019-07-22  2:59         ` Jing Liu
2019-07-22 11:50           ` Paolo Bonzini
2019-07-24 12:05             ` Jing Liu
2019-07-24 12:12               ` Paolo Bonzini

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=0ffa32db-3615-16e0-71bf-a8c552c2fcfb@linux.intel.com \
    --to=jing2.liu@linux.intel.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).