Make sure that we actually have a device when checking the aio
context. Otherwise guests could trigger QEMU crashes.
Signed-off-by: "Eugene (jno) Dvurechenski"
Reviewed-by: David Hildenbrand
---
hw/scsi/virtio-scsi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git
I've found the case where erratic guest was able to crash QEMU.
This patch prevents dereferencing of a NULL pointer (empty 'd'
in the assertion, no valid LUN was specified).
Eugene (jno) Dvurechenski (1):
virtio-scsi: don't crash without a valid device
hw/scsi/virtio-s
Yes, this patch doesn't make the code worse.
On 06/27/2014 05:11 PM, Christian Borntraeger wrote:
> On 27/06/14 15:04, Eugene "jno" Dvurechenski wrote:
>>
>>
>> On 06/27/2014 03:55 PM, Christian Borntraeger wrote:
>>>>> -const i
d preserve both checks.
In theory, we may catch a table that consumes all scratch space and
leave no unused entry.
Plus, this check for zero counter and last entry is for "continuation"
pointer, not for end-of-table by itself.
I think now, this code may need even few more checks to co
the real
> sector size?
This is the top limit for a table that takes entire block.
Real end-of-table is being handled by the code.
--
Best Regards,
Eugene "jno" Dvurechenski
zLinux (KVM) Development - Software Engineer
IBM Russia - Science & Technology Center
phone: +7 (495) 66
torage allocation (and similar
static calculations) now.
--
Best Regards,
Eugene "jno" Dvurechenski
zLinux (KVM) Development - Software Engineer
IBM Russia - Science & Technology Center
phone: +7 (495) 660 8940 ext. 1021
<>
signature.asc
Description: OpenPGP digital signature
unused_space(const void *p, size_t size)
+{
+size_t i;
--
Best Regards,
Eugene "jno" Dvurechenski
zLinux (KVM) Development - Software Engineer
IBM Russia - Science & Technology Center
phone: +7 (495) 660 8940 ext. 1021
<>
signature.asc
Description: OpenPGP digital signature
