[ipxe-devel] [PATCH 2/2] [build] Disable TIVOLI_VMM_WORKAROUND in the qemu configuration

[Laszlo Ersek]

On 10/27/16 08:27, Ján ONDREJ (SAL) wrote:
> Hello,
> 
>   I have troubles with this FXSAVE/FXRSTOR not only on qemu/kvm,
> but with some older hardware too (HP server).
> 
>   Maybe this tivoli workaround should be disabled for all and enabled
> only for tivoli.

Hmmm. QEMU has a named configuration, and "bundling iPXE for QEMU" is a
regular activity that the QEMU developer community does.

I don't think "iPXE for Tivoli" is a similarly well-defined environment
or use case, so a build-time config for that seems unlikely.

If the FXSAVE/FXRSTOR stuff breaks on real hardware as well, then I'm
afraid the original analysis of FXSAVE/FXRSTOR availability, in commit
71560d185475, was inexact:

> The use of "fxsave" and "fxrstor" introduces an implicit assumption
> that the CPU supports SSE instructions (even though we make no
> assumption about whether or not SSE is currently enabled).  SSE was
> introduced in 1999 with the Pentium III (and added by AMD in 2001),
> and is an architectural requirement for x86_64.  Experimentation with
> current versions of gcc suggest that it may generate SSE instructions
> even when using "-m32", unless an explicit "-march=i386" or "-mno-sse"
> is used to inhibit this.  It therefore seems reasonable to assume that
> SSE will be supported on any hardware that might realistically be used
> with new iPXE builds.

If this assumption is unsafe after all, then I bet Michael will want to
make FXSAVE/FXRSTOR dependent on dynamic feature detection (CPUID or
similar -- maybe even detect the Tivoli VMM somehow, at runtime, like
you said).

Thanks
Laszlo


> On Thu, Oct 27, 2016 at 12:13:51AM +0200, Laszlo Ersek wrote:
>> This prevents KVM guests from crashing that run iPXE on host CPUs without
>> "unrestricted_guest" support.
>>
>> Once KVM gets the FXSAVE / FXRSTOR emulation feature (*), and the feature
>> becomes widely available to users, we can back out this change from iPXE.
>>
>> (*) Already in progress by Radim:
>>
>>     [PATCH 0/2] KVM: x86: emulate fxsave and fxrstor
>>
>>     https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1258895.html
>>     https://www.spinics.net/lists/kernel/msg2370327.html
>>
>> Cc: Bandan Das <bsd at redhat.com>
>> Cc: Gerd Hoffmann <kraxel at redhat.com>
>> Cc: Greg <rollenwiese at yahoo.com>
>> Cc: Michael Brown <mcb30 at ipxe.org>
>> Cc: Michael Prokop <launchpad at michael-prokop.at>
>> Cc: Paolo Bonzini <pbonzini at redhat.com>
>> Cc: Peter Pickford <arch at netremedies.ca>
>> Cc: Radim Krčmář <rkrcmar at redhat.com>
>> Ref: https://bugs.archlinux.org/task/50778
>> Ref: https://bugs.launchpad.net/qemu/+bug/1623276
>> Ref: https://bugzilla.proxmox.com/show_bug.cgi?id=1182
>> Ref: https://bugzilla.redhat.com/show_bug.cgi?id=1356762
>> Signed-off-by: Laszlo Ersek <lersek at redhat.com>
>> ---
>>  src/config/qemu/general.h | 5 +++++
>>  1 file changed, 5 insertions(+)
>>
>> diff --git a/src/config/qemu/general.h b/src/config/qemu/general.h
>> index 30f60d3f704a..a084497359fe 100644
>> --- a/src/config/qemu/general.h
>> +++ b/src/config/qemu/general.h
>> @@ -8,3 +8,8 @@
>>  
>>  /* Work around missing EFI_PXE_BASE_CODE_PROTOCOL */
>>  #define EFI_DOWNGRADE_UX
>> +
>> +/* The Tivoli VMM workaround causes a KVM emulation failure on hosts
>> + * without unrestricted_guest support
>> + */
>> +#undef TIVOLI_VMM_WORKAROUND
>> -- 
>> 2.9.2
>>
>> _______________________________________________
>> ipxe-devel mailing list
>> ipxe-devel at lists.ipxe.org
>> https://lists.ipxe.org/mailman/listinfo.cgi/ipxe-devel