Re: [Qemu-devel] [PATCH v4 2/5] target/i386: Populate AMD Processor Cache Information

[Moger, Babu]

> -----Original Message-----
> From: Eduardo Habkost <address@hidden>
> Sent: Wednesday, March 21, 2018 12:10 PM
> To: Moger, Babu <address@hidden>
> Cc: address@hidden; address@hidden; address@hidden;
> Lendacky, Thomas <address@hidden>; Singh, Brijesh
> <address@hidden>; address@hidden; address@hidden;
> address@hidden; Hook, Gary <address@hidden>; qemu-
> address@hidden
> Subject: Re: [Qemu-devel] [PATCH v4 2/5] target/i386: Populate AMD
> Processor Cache Information
> 
> On Wed, Mar 21, 2018 at 03:58:41PM +0000, Moger, Babu wrote:
> > Hi Eduardo,
> >
> > > -----Original Message-----
> > > From: Eduardo Habkost <address@hidden>
> > > Sent: Tuesday, March 20, 2018 12:54 PM
> > > To: Moger, Babu <address@hidden>
> > > Cc: address@hidden; address@hidden; address@hidden;
> > > Lendacky, Thomas <address@hidden>; Singh, Brijesh
> > > <address@hidden>; address@hidden; address@hidden;
> > > address@hidden; Hook, Gary <address@hidden>; qemu-
> > > address@hidden
> > > Subject: Re: [Qemu-devel] [PATCH v4 2/5] target/i386: Populate AMD
> > > Processor Cache Information
> > >
> > > On Tue, Mar 20, 2018 at 05:25:52PM +0000, Moger, Babu wrote:
> > > > Hi Eduardo, Thanks for the comments. Please see the response inline.
> > > >
> > > > > -----Original Message-----
> > > > > From: Eduardo Habkost <address@hidden>
> > > > > Sent: Friday, March 16, 2018 1:00 PM
> > > > > To: Moger, Babu <address@hidden>
> > > > > Cc: address@hidden; address@hidden; address@hidden;
> > > > > Lendacky, Thomas <address@hidden>; Singh, Brijesh
> > > > > <address@hidden>; address@hidden;
> address@hidden;
> > > > > address@hidden; Hook, Gary <address@hidden>; qemu-
> > > > > address@hidden
> > > > > Subject: Re: [Qemu-devel] [PATCH v4 2/5] target/i386: Populate AMD
> > > > > Processor Cache Information
> > > > >
> > > > > On Mon, Mar 12, 2018 at 05:00:46PM -0400, Babu Moger wrote:
> > > > > > From: Stanislav Lanci <address@hidden>
> > > > > >
> > > > > > Add information for cpuid 0x8000001D leaf. Populate cache topology
> > > > > information
> > > > > > for different cache types(Data Cache, Instruction Cache, L2 and L3)
> > > > > supported
> > > > > > by 0x8000001D leaf. Please refer Processor Programming Reference
> > > (PPR)
> > > > > for AMD
> > > > > > Family 17h Model for more details.
> > > > > >
> > > > > > Signed-off-by: Stanislav Lanci <address@hidden>
> > > > > > Signed-off-by: Babu Moger <address@hidden>
> > > > >
> > > > > The new CPUID leaves don't seem to match the existing AMD cache
> > > > > information
> > > > > leaves.  Is this intentional?  Why?
> > > >
> > > > It is not intentional. These values are from older family of processors.
> > > These values have changed from Family 14  or later.
> > > > The latest one is Family 17. You can see the differences here.
> > > >  https://support.amd.com/TechDocs/41131.pdf
> > > >
> > >
> https://support.amd.com/TechDocs/55072_AMD_Family_15h_Models_70h-
> > > 7Fh_BKDG.pdf
> > > >
> > >
> https://support.amd.com/TechDocs/54945_PPR_Family_17h_Models_00h-
> > > 0Fh.pdf
> > > >
> > > > Some of these are bugs in our code. For some we need to add checks
> for
> > > the family and correct these values.
> > > > You understand the code much better than me. Correct me if I missed
> > > something.
> > > >
> > > > Note that older family of processors don't support topology extensions.
> > >
> > > If you want to make the cache size/topology look different
> > > depending on the CPU model/options, this would require more work,
> > > but it would be an interesting feature.
> > >
> > > The "i386: Helpers to encode cache information consistently"
> > > patch I sent last week might be a useful starting point for that.
> > >
> > > If you plan to implement that, please keep in mind that existing
> > > CPUID cache info needs to be kept on previous machine-types (this
> > > is implemented by adding QOM properties that can be used to
> > > enable the old behavior, and by setting them at
> > > MachineClass::compat_props).
> >
> > Wanted to get some confirmation what you meant by setting
> MachineClass::compat_props.
> > Here is the patch I created to add new property for cpu. Now, I can use
> enable_topoext to display
> > new  change properly based on family.  Is that what you meant ?
> 
> If the only change you introduce in the defaults is enabling
> topoext but keeping the same default cache sizes, the example
> following would make sense (but see comments below about
> implementation details).
> 
> But if you also want to change the default cache size, you will
> also need properties that control the cache size.

Yes. We will need cpu family information.  Let me see if that information is
 already available or we need to add that as well.

> 
> (In theory, you could make the "topoext" property control the
> cache size too, but I don't see why the cache size should be
> coupled to topoext)
> 
> >
> >
> > diff --git a/include/hw/i386/pc.h b/include/hw/i386/pc.h
> > index ffee841..d1ee053 100644
> > --- a/include/hw/i386/pc.h
> > +++ b/include/hw/i386/pc.h
> > @@ -369,6 +369,11 @@ bool e820_get_entry(int, uint32_t, uint64_t *,
> uint64_t *);
> >      HW_COMPAT_2_7 \
> >      {\
> >          .driver   = TYPE_X86_CPU,\
> > +        .property = "topoext",\
> > +        .value    = "off",\
> 
> This makes sense if you want to change a CPU model to enable
> topoext by default.  But I suggest setting it for
> ("EPYC" "-" TYPE_X86_CPU) only, not TYPE_X86_CPU, as EPYC is the
> only CPU model affected by patch 4/5.

As Kash already mentioned, right now we have two family's  EPYC and EPYC-IBPB 
supporting 
this new cache sizes/info. So, ("EPYC" "-" TYPE_X86_CPU) may not work. Will 
keep it as old one.

> 
> However, the property registration can be simpler:
> 
> > +    },\
> > +    {\
> > +        .driver   = TYPE_X86_CPU,\
> >          .property = "l3-cache",\
> >          .value    = "off",\
> >      },\
> > diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> > index 84d64de..557a2d6 100644
> > --- a/target/i386/cpu.c
> > +++ b/target/i386/cpu.c
> > @@ -5080,6 +5080,7 @@ static Property x86_cpu_properties[] = {
> >                       false),
> >      DEFINE_PROP_BOOL("vmware-cpuid-freq", X86CPU,
> vmware_cpuid_freq, true),
> >      DEFINE_PROP_BOOL("tcg-cpuid", X86CPU, expose_tcg, true),
> > +    DEFINE_PROP_BOOL("topoext", X86CPU, enable_topoext, true),
> 
> You don't need to manually add a X86CPU field + property if you
> are just controlling a CPUID feature flag.  In this case, you can
> just add "topoext" to
> feature_word_info[FEAT_8000_0001_ECX].feat_names[22], and
> a "topoext" QOM property will be available automatically.

Actually "topoext" feature is already there as cupid feature. I should have not 
used topoext
to control the cache sizes. I will come up with a new name. 
 

> 
> >
> >      /*
> >       * From "Requirements for Implementing the Microsoft
> > diff --git a/target/i386/cpu.h b/target/i386/cpu.h
> > index 2e2bab5..3d3caa1 100644
> > --- a/target/i386/cpu.h
> > +++ b/target/i386/cpu.h
> > @@ -1333,6 +1333,11 @@ struct X86CPU {
> >       */
> >      bool enable_l3_cache;
> >
> > +    /* Compatibility bits for old machine types.
> > +     * If true present the new cache topology information
> > +     */
> > +    bool enable_topoext;
> > +
> >      /* Compatibility bits for old machine types: */
> >      bool enable_cpuid_0xb;
> >
> >
> >
> > >
> > > --
> > > Eduardo
> 
> --
> Eduardo