On 07/13/2015 03:32 PM, Paolo Bonzini wrote:
I'm seeing a significant regression in boot performance on Intel
hardware with assigned devices that bisects back to this patch. There's
a long delay with Seabios between the version splash and execution of
option ROMs, and a _very_ long delay with OVMF before the display is
initialized. The delay is long enough that users are reporting their
previously working VM is hung with 100% CPU usage on v4.2-rc1. Thanks,
Alex, thanks for your report. I will try to reproduce and fix it asap.
I have reproduced the issue and I think Bandan and Paolo is correct.
The code that Bandan pointed out
+ /* MTRR is completely disabled, use UC for all of physical memory. */
+ if (!(mtrr_state->enabled & 0x2))
+ return MTRR_TYPE_UNCACHABLE;
actually disappears in commit fa61213746a7 (KVM: MTRR: simplify
kvm_mtrr_get_guest_memory_type, 2015-06-15). Should mtrr_default_type
actually be something like this:
:(
Based on the SDM, UC is applied to all memory rather than default-type
if MTRR is disabled.
If i changed the code to:
if (!(mtrr_state->enabled & 0x2))
return mtrr_state->def_type;
the result is the same as before.
However, fast boot came back if "return 0xFF" here. So fast boot expects
that the memory type is WB.
static u8 mtrr_default_type(struct kvm_mtrr *mtrr_state)
{
if (mtrr_is_enabled(mtrr_state))
return mtrr_state->deftype & IA32_MTRR_DEF_TYPE_TYPE_MASK;
else
return MTRR_TYPE_UNCACHABLE;
}
? Then it's easy to add a quirk that makes the default WRITEBACK until
MTRRs are enabled.
It is the wrong configure in OVMF... shall we need to adjust KVM to satisfy
OVMF?
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
