Re: [Qemu-devel] [PATCH v3 4/4] x86: Set physical address bits based on host

[Eduardo Habkost]

On Tue, Jul 05, 2016 at 08:03:18PM +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.
> 
> We can also use the value we read from the host to check the users
> explicitly set value and warn them if it doesn't match.
> 
> Signed-off-by: Dr. David Alan Gilbert <address@hidden>
> ---
>  target-i386/cpu.c | 53 +++++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 53 insertions(+)
> 
> diff --git a/target-i386/cpu.c b/target-i386/cpu.c
> index f33cf58..6ebd26b 100644
> --- a/target-i386/cpu.c
> +++ b/target-i386/cpu.c
> @@ -2952,7 +2952,60 @@ 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) {
> +        if (kvm_enabled()) {
> +            /* Read the hosts physical address size, and compare it to what 
> we
> +             * were asked for; note old machine types default to 40 bits
> +             */

Isn't the "old machine types default to 40 bits" part obsolete,
now that you use 9999 to indicate it was not set explicitly?

Also, the observation makes me confused: I know the old machine
types default to 40 bits, but I don't know why I need to know
that to understand the host_cpuid() logic below.

> +            uint32_t eax;
> +            uint32_t host_phys_bits = 0;
> +            static bool warned;
> +
> +            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;
> +            }

I would love to see this logic moved inside a
x86_host_phys_bits() function.

> +
> +            if (cpu->phys_bits == 0) {
> +                /* The user asked for us to use the host physical bits */
> +                cpu->phys_bits = host_phys_bits;
> +            }
> +
> +            /* Print a warning if the user set it to a value that's not the
> +             * host value.
> +             */
> +            if (cpu->phys_bits != host_phys_bits && cpu->phys_bits != 9999 &&
> +                !warned) {
> +                error_report("Warning: Host physical bits (%u)"
> +                                 " does not match phys_bits (%u)",
> +                                 host_phys_bits, cpu->phys_bits);
> +                warned = true;

The name of the user-visible property is "phys-bits", not
phys_bits. Maybe we could say "does not match phys-bits
property".

> +            }
> +        } else {
> +            if (cpu->phys_bits == 0) {
> +                error_setg(errp, "phys_bits can not be read from the host in"
> +                                 " TCG mode");
> +                return;
> +            }
> +        }
>          /* 9999 is a special meaning 'use the old default',
>           */
>          if (cpu->phys_bits == 9999) {
> -- 
> 2.7.4
> 

-- 
Eduardo