Re: [Qemu-devel] [PATCH v2 4/6] x86: Set physical address bits based on host

[Dr. David Alan Gilbert]

* Eduardo Habkost (address@hidden) wrote:
> On Mon, Jul 04, 2016 at 08:16:07PM +0100, Dr. David Alan Gilbert (git) wrote:
> > From: "Dr. David Alan Gilbert" <address@hidden>
> > 
> > A special case based on the previous phys-bits property; if it's
> > the magic value 0 then use the hosts capabilities.
> > 
> > This becomes the default on new machine types.
> > 
> > Signed-off-by: Dr. David Alan Gilbert <address@hidden>
> > ---
> >  include/hw/i386/pc.h |  5 +++++
> >  target-i386/cpu.c    | 36 +++++++++++++++++++++++++++++++++++-
> >  2 files changed, 40 insertions(+), 1 deletion(-)
> > 
> > diff --git a/include/hw/i386/pc.h b/include/hw/i386/pc.h
> > index d85e924..bf31609 100644
> > --- a/include/hw/i386/pc.h
> > +++ b/include/hw/i386/pc.h
> > @@ -379,6 +379,11 @@ bool e820_get_entry(int, uint32_t, uint64_t *, 
> > uint64_t *);
> >          .driver = TYPE_X86_CPU,\
> >          .property = "fill-mtrr-mask",\
> >          .value = "off",\
> > +    },\
> > +    {\
> > +        .driver = TYPE_X86_CPU,\
> > +        .property = "phys-bits",\
> > +        .value = "40",\
> >      },
> >  
> >  #define PC_COMPAT_2_5 \
> > diff --git a/target-i386/cpu.c b/target-i386/cpu.c
> > index 5737aba..d45d2a6 100644
> > --- a/target-i386/cpu.c
> > +++ b/target-i386/cpu.c
> > @@ -2957,6 +2957,40 @@ static void x86_cpu_realizefn(DeviceState *dev, 
> > Error **errp)
> >             & CPUID_EXT2_AMD_ALIASES);
> >      }
> >  
> > +    /* For 64bit systems think about the number of physical bits to 
> > present.
> > +     * ideally this should be the same as the host; anything other than 
> > matching
> > +     * the host can cause incorrect guest behaviour.
> > +     * QEMU used to pick the magic value of 40 bits that corresponds to
> > +     * consumer AMD devices but nothing esle.
> > +     */
> > +    if (env->features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
> > +        uint32_t eax;
> > +        /* Read the hosts physical address size, and compare it to what we
> > +         * were asked for; note old machine types default to 40 bits
> > +         */
> > +        uint32_t host_phys_bits = 0;
> > +        host_cpuid(0x80000000, 0, &eax, NULL, NULL, NULL);
> > +        if (eax >= 0x80000008) {
> > +            host_cpuid(0x80000008, 0, &eax, NULL, NULL, NULL);
> > +            /* Note: According to AMD doc 25481 rev 2.34 they have a field
> > +             * at 23:16 that can specify a maximum physical address bits 
> > for
> > +             * the guest that can override this value; but I've not seen
> > +             * anything with that set.
> > +             */
> > +            host_phys_bits = eax & 0xff;
> > +        } else {
> > +            /* It's an odd 64 bit machine that doesn't have the leaf for
> > +             * physical address bits; fall back to 36 that's most older 
> > Intel.
> > +             */
> > +            host_phys_bits = 36;
> > +        }
> 
> Why do we need to calculate host_phys_bits when phys_bits is
> already set? Shouldn't we put all the code above after the "if
> (cpu->phys_bits)" check?

Because I reuse host_phys_bits to generate the warning if you've
explicitly set phys-bits and it doesn't match the host.

> > +
> > +        if (cpu->phys_bits == 0) {
> > +            /* The user asked for us to use the host physical bits */
> > +            cpu->phys_bits = host_phys_bits;
> > +
> > +        }
> > +    }
> >  
> >      cpu_exec_init(cs, &error_abort);
> >  
> > @@ -3259,7 +3293,7 @@ static Property x86_cpu_properties[] = {
> >      DEFINE_PROP_BOOL("enforce", X86CPU, enforce_cpuid, false),
> >      DEFINE_PROP_BOOL("kvm", X86CPU, expose_kvm, true),
> >      DEFINE_PROP_BOOL("fill-mtrr-mask", X86CPU, fill_mtrr_mask, true),
> > -    DEFINE_PROP_UINT32("phys-bits", X86CPU, phys_bits, 40),
> > +    DEFINE_PROP_UINT32("phys-bits", X86CPU, phys_bits, 0),
> 
> I would put this part (that sets the default to 0) and the
> PC_COMPAT_2_6 part in a separate patch. This way we can include
> the mechanism for setting phys-bits=0 even if we didn't reach a
> conclusion about the proper pc-2.7 default yet.

Will do.

Dave

> 
> >      DEFINE_PROP_UINT32("level", X86CPU, env.cpuid_level, 0),
> >      DEFINE_PROP_UINT32("xlevel", X86CPU, env.cpuid_xlevel, 0),
> >      DEFINE_PROP_UINT32("xlevel2", X86CPU, env.cpuid_xlevel2, 0),
> > -- 
> > 2.7.4
> > 
> 
> -- 
> Eduardo
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK