On 19/12/2014 13:42, Igor Mammedov wrote:
> On Fri, 19 Dec 2014 13:02:24 +0100
> Paolo Bonzini <pbonz...@redhat.com> wrote:
>
>>
>>
>> On 19/12/2014 12:32, Igor Mammedov wrote:
>>> There is one more commit that breaks it, this time with subsection
>>> 6c3bff0 "exec: Save CPUState::exception_index field"
>>>
>>> qemu: warning: error while loading state for instance 0x0 of device
>>> 'cpu_common'
>>> the same reproducer with -parallel none
>>
>> Patch sent, thanks.
>>
>> Paolo
>
> one more breakage:
> a28fe7e pckbd: adding new fields to vmstate
>
> source:
> qemu-system-x86_64 -monitor stdio -M pc-i440fx-1.7 -parallel none
> xpsp3x86.qcow2
>
> switch to text screen where you could select Safe Mode and stop/migrate at
> this point
>
> target:
> qemu-system-x86_64-1.7 -monitor stdio -M pc-i440fx-1.7 -incoming "exec: gzip
> -c -d STATEFILE.gz" xpsp3x86.qcow2
>
> qemu: warning: error while loading state for instance 0x0 of device 'pckbd'
Looks like Windows (all versions) writes 0xDD and 0xDF to the outport
port of the keyboard controller when it enables/disables the A20 line.
There are two possibilities:
1) disable this if you care about backwards-migration
2) apply upstream a patch like this:
diff --git a/hw/input/pckbd.c b/hw/input/pckbd.c
index 2b0cd3d..cb4a8be 100644
--- a/hw/input/pckbd.c
+++ b/hw/input/pckbd.c
@@ -373,7 +373,7 @@ static void kbd_reset(void *opaque)
static uint8_t kbd_outport_default(KBDState *s)
{
- return KBD_OUT_RESET | KBD_OUT_A20
+ return (0xdf & ~KBD_OUT_OBF & ~KBD_OUT_MOUSE_OBF)
| (s->status & KBD_STAT_OBF ? KBD_OUT_OBF : 0)
| (s->status & KBD_STAT_MOUSE_OBF ? KBD_OUT_MOUSE_OBF : 0);
}
3) do both, so upstream works better
Paolo
