[ kvm-Bugs-2009439 ] data corruption with virtio-blk

[ kvm-Bugs-2009439 ] data corruption with virtio-blk

[SourceForge.net]

Bugs item #2009439, was opened at 2008-07-03 11:21
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Balaji Rao R (balajirrao)
Assigned to: Nobody/Anonymous (nobody)
Summary: data corruption with virtio-blk

Initial Comment:
kvm-userspace : kvm-70-138-g163308a

With a kernel from linux-2.6.git (v2.6.26-rc8-89-ge1441b9), data corruption is seen on block devices when exported through virtio. Once a block device is mounted and unmounted in a guest, the kernel   is not able to find a filesystem in the block device when  being remounted. This is seen from the host as well.


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599

Re: [ kvm-Bugs-2009439 ] data corruption with virtio-blk

[Mark McLoughlin]

Hi Avi,

On Thu, 2008-07-03 at 05:51 +0000, SourceForge.net wrote:
> Bugs item #2009439, was opened at 2008-07-03 11:21
> Message generated for change (Tracker Item Submitted) made by Item Submitter
> You can respond by visiting: 
> https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599
...
> Summary: data corruption with virtio-blk
> 
> Initial Comment:
> kvm-userspace : kvm-70-138-g163308a
> 
> With a kernel from linux-2.6.git (v2.6.26-rc8-89-ge1441b9), data
> corruption is seen on block devices when exported through virtio. Once
> a block device is mounted and unmounted in a guest, the kernel   is
> not able to find a filesystem in the block device when  being
> remounted. This is seen from the host as well.

I think the below fixes the data corrupter, but I'm still tracking down
another issue where the guest is hanging waiting for I/O to complete
with the latest virtio-blk backend.

Cheers,
Mark.

Subject: [PATCH] kvm: qemu: Fix virtio-blk data corrupter

Fix a typo in "avoid fragmented virtio-blk transfers by copying"
which causes data corruption.

See also:

  https://sf.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
---
 qemu/hw/virtio-blk.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/qemu/hw/virtio-blk.c b/qemu/hw/virtio-blk.c
index c88f477..88dc086 100644
--- a/qemu/hw/virtio-blk.c
+++ b/qemu/hw/virtio-blk.c
@@ -186,7 +186,7 @@ static void virtio_blk_handle_output(VirtIODevice *vdev, VirtQueue *vq)
 	    for (i = 1; i < req->elem.out_num; i++) {
 		size_t len;
 
-		len = MIN(req->elem.in_sg[i].iov_len,
+		len = MIN(req->elem.out_sg[i].iov_len,
 			  req->size - offset);
 		memcpy(req->buffer + offset,
 		       req->elem.out_sg[i].iov_base,
-- 
1.5.5.1

Re: [ kvm-Bugs-2009439 ] data corruption with virtio-blk

[Avi Kivity]

Mark McLoughlin wrote:
> Fix a typo in "avoid fragmented virtio-blk transfers by copying"
> which causes data corruption.
>
> See also:
>
>   https://sf.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599
>
>   

Applied, thanks.


-- 
error compiling committee.c: too many arguments to function

Re: [ kvm-Bugs-2009439 ] data corruption with virtio-blk

[Javier Guerra]

On Thu, Jul 3, 2008 at 3:03 AM, Mark McLoughlin <markmc@redhat.com> wrote:
> I think the below fixes the data corrupter, but I'm still tracking down
> another issue where the guest is hanging waiting for I/O to complete
> with the latest virtio-blk backend.

is that a common case? i haven't been able to finish a system install
on either virtio or scsi block devices.  both linux (Ubuntu JeOS 8.04)
and windows (xp sp2) freeze sometime while writing packages.  IDE
installs work perfectly

-- 
Javier

Re: [ kvm-Bugs-2009439 ] data corruption with virtio-blk

[Mark McLoughlin]

On Thu, 2008-07-03 at 09:58 -0500, Javier Guerra wrote:
> On Thu, Jul 3, 2008 at 3:03 AM, Mark McLoughlin <markmc@redhat.com> wrote:
> > I think the below fixes the data corrupter, but I'm still tracking down
> > another issue where the guest is hanging waiting for I/O to complete
> > with the latest virtio-blk backend.
> 
> is that a common case? i haven't been able to finish a system install
> on either virtio or scsi block devices.  both linux (Ubuntu JeOS 8.04)
> and windows (xp sp2) freeze sometime while writing packages.  IDE
> installs work perfectly

It's certainly possible the hangs you're seeing were caused by the
IOAPIC interrupt injection bug I just sent out a fix for - you could try
doing an install either running kvm-qemu -no-kvm-irqchip or booting the
guest with "noapic" on the kernel command line. If the IOAPIC bug is
your issue, you shouldn't see any freezes.

Cheers,
Mark.

[ kvm-Bugs-2009439 ] data corruption with virtio-blk

[SourceForge.net]

Bugs item #2009439, was opened at 2008-07-03 06:51
Message generated for change (Comment added) made by markmc
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Balaji Rao R (balajirrao)
Assigned to: Nobody/Anonymous (nobody)
Summary: data corruption with virtio-blk

Initial Comment:
kvm-userspace : kvm-70-138-g163308a

With a kernel from linux-2.6.git (v2.6.26-rc8-89-ge1441b9), data corruption is seen on block devices when exported through virtio. Once a block device is mounted and unmounted in a guest, the kernel   is not able to find a filesystem in the block device when  being remounted. This is seen from the host as well.


----------------------------------------------------------------------

Comment By: Mark McLoughlin (markmc)
Date: 2008-07-07 09:32

Message:
Logged In: YES 
user_id=116392
Originator: NO

Should be fixed by:

 
http://git.kernel.org/?p=virt/kvm/kvm-userspace.git;a=commit;h=b5a5e894a



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599

Re: [ kvm-Bugs-2009439 ] data corruption with virtio-blk

[Javier Guerra]

On Sat, Jul 5, 2008 at 5:29 AM, Mark McLoughlin <markmc@redhat.com> wrote:
> It's certainly possible the hangs you're seeing were caused by the
> IOAPIC interrupt injection bug I just sent out a fix for - you could try
> doing an install either running kvm-qemu -no-kvm-irqchip or booting the
> guest with "noapic" on the kernel command line. If the IOAPIC bug is
> your issue, you shouldn't see any freezes.

thanks! that works perfectly for virtio blockdevices; but still no
help for windows guests (with SCSI emulation). with -no-kvm-irqchipno
i don't get freezes anymore, just BSOD (and core dumps). also tried
adding -no-acpi, no difference.


-- 
Javier

[ kvm-Bugs-2009439 ] data corruption with virtio-blk

[SourceForge.net]

Bugs item #2009439, was opened at 2008-07-03 06:51
Message generated for change (Settings changed) made by markmc
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
>Status: Closed
>Resolution: Fixed
Priority: 5
Private: No
Submitted By: Balaji Rao R (balajirrao)
Assigned to: Nobody/Anonymous (nobody)
Summary: data corruption with virtio-blk

Initial Comment:
kvm-userspace : kvm-70-138-g163308a

With a kernel from linux-2.6.git (v2.6.26-rc8-89-ge1441b9), data corruption is seen on block devices when exported through virtio. Once a block device is mounted and unmounted in a guest, the kernel   is not able to find a filesystem in the block device when  being remounted. This is seen from the host as well.


----------------------------------------------------------------------

Comment By: Mark McLoughlin (markmc)
Date: 2008-07-07 09:32

Message:
Logged In: YES 
user_id=116392
Originator: NO

Should be fixed by:

 
http://git.kernel.org/?p=virt/kvm/kvm-userspace.git;a=commit;h=b5a5e894a



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2009439&group_id=180599