Re: [PATCH 1/2] target-i386: tcg: Accept clwb instruction

2015-11-06 Thread Richard Henderson

On 11/04/2015 10:24 PM, Eduardo Habkost wrote:

Accept the clwb instruction (66 0F AE /6) if its corresponding feature
flag is enabled on CPUID[7].

Signed-off-by: Eduardo Habkost 
---
  target-i386/translate.c | 13 -
  1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/target-i386/translate.c b/target-i386/translate.c
index b400d24..bac1685 100644
--- a/target-i386/translate.c
+++ b/target-i386/translate.c
@@ -7716,10 +7716,21 @@ static target_ulong disas_insn(CPUX86State *env, 
DisasContext *s,
  }
  break;
  case 5: /* lfence */
-case 6: /* mfence */
  if ((modrm & 0xc7) != 0xc0 || !(s->cpuid_features & CPUID_SSE2))
  goto illegal_op;
  break;
+case 6: /* mfence/clwb */
+if (s->prefix & PREFIX_DATA) {
+/* clwb */
+if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_CLWB))
+goto illegal_op;
+gen_lea_modrm(env, s, modrm);


You should use gen_nop_modrm here, since we're not going to do anything with 
the address.  Otherwise,


Reviewed-by: Richard Henderson 


r~
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


Re: [PATCH 1/2] target-i386: tcg: Accept clwb instruction

2015-11-06 Thread Eduardo Habkost
On Fri, Nov 06, 2015 at 12:04:56PM +0100, Richard Henderson wrote:
> On 11/04/2015 10:24 PM, Eduardo Habkost wrote:
> >Accept the clwb instruction (66 0F AE /6) if its corresponding feature
> >flag is enabled on CPUID[7].
> >
> >Signed-off-by: Eduardo Habkost 
> >---
> >  target-i386/translate.c | 13 -
> >  1 file changed, 12 insertions(+), 1 deletion(-)
> >
> >diff --git a/target-i386/translate.c b/target-i386/translate.c
> >index b400d24..bac1685 100644
> >--- a/target-i386/translate.c
> >+++ b/target-i386/translate.c
> >@@ -7716,10 +7716,21 @@ static target_ulong disas_insn(CPUX86State *env, 
> >DisasContext *s,
> >  }
> >  break;
> >  case 5: /* lfence */
> >-case 6: /* mfence */
> >  if ((modrm & 0xc7) != 0xc0 || !(s->cpuid_features & 
> > CPUID_SSE2))
> >  goto illegal_op;
> >  break;
> >+case 6: /* mfence/clwb */
> >+if (s->prefix & PREFIX_DATA) {
> >+/* clwb */
> >+if (!(s->cpuid_7_0_ebx_features & CPUID_7_0_EBX_CLWB))
> >+goto illegal_op;
> >+gen_lea_modrm(env, s, modrm);
> 
> You should use gen_nop_modrm here, since we're not going to do anything with
> the address.  Otherwise,
> 
> Reviewed-by: Richard Henderson 

Thanks! BTW, clflush uses gen_lea_modrm() too, does it do anything with
the address somewhere else?

-- 
Eduardo
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


Re: [PATCH 1/2] target-i386: tcg: Accept clwb instruction

2015-11-06 Thread Richard Henderson

On 11/06/2015 02:59 PM, Eduardo Habkost wrote:

Thanks! BTW, clflush uses gen_lea_modrm() too, does it do anything with
the address somewhere else?


Nope.  It probably pre-dates the introduction of gen_nop_modrm.


r~
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html