Bug 715237 - Core dump occurs when doing dd in the guest
Summary: Core dump occurs when doing dd in the guest
Alias: None
Product: Red Hat Enterprise Linux 5
Classification: Red Hat
Component: kvm
Version: 5.7
Hardware: x86_64
OS: Linux
Target Milestone: rc
: ---
Assignee: Kevin Wolf
QA Contact: Virtualization Bugs
Depends On:
Blocks: Rhel5KvmTier2
TreeView+ depends on / blocked
Reported: 2011-06-22 10:31 UTC by FuXiangChun
Modified: 2011-07-29 07:59 UTC (History)
6 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Last Closed: 2011-07-29 07:59:42 UTC

Attachments (Terms of Use)
backtrace (10.24 KB, application/octet-stream)
2011-06-22 10:40 UTC, FuXiangChun
no flags Details

Description FuXiangChun 2011-06-22 10:31:19 UTC
Description of problem:
as subject, this is error message in monitor.
(qemu) qemu-kvm: block-qcow2.c:1621: qcow_aio_write_cb: Assertion `(acb->cluster_offset & 511) == 0' failed.
Aborted (core dumped)

backtrace file is in attachment.

Version-Release number of selected component (if applicable):
host info:
# uname -r
#rpm -qa|grep kvm

guest info:

How reproducible:

Steps to Reproduce:
1./usr/libexec/qemu-kvm -M rhel5.6.0 -m 4096 -smp 4,sockets=4,cores=1,threads=1 -name RHEL5u7 -uuid 13bd47ff-7458-a214-9c43-d311ed5ca5a3 -monitor stdio -no-kvm-pit-reinjection -boot c -drive file=/dev/vg0/data1,if=virtio,format=qcow2,cache=none,boot=on,werror=stop -net nic,macaddr=54:52:00:52:ed:62,vlan=0,model=e1000 -net tap,script=/etc/qemu-ifup,downscript=no,vlan=0 -vnc :1 -balloon none -notify all -no-hpet -soundhw ac97

2.run below script in guest
for ((;;))
dd if=/dev/zero of=test1 bs=1M count=10000

3. after several minutes, monitor will show below core dumped message.
(qemu) qemu-kvm: block-qcow2.c:1621: qcow_aio_write_cb: Assertion `(acb->cluster_offset & 511) == 0' failed.
Aborted (core dumped)
Actual results:
core dumped

Expected results:
guest work well

Additional info:
kvm storage: block device(iscsi)

Comment 1 FuXiangChun 2011-06-22 10:38:30 UTC
repeat 3 times for above steps, 100% reproduce.

Comment 2 FuXiangChun 2011-06-22 10:40:03 UTC
Created attachment 505964 [details]

Comment 3 FuXiangChun 2011-06-23 10:08:03 UTC
since only LV(/dev/vg0/data1) can reproduce this issue. in order to reproduce this bug i did below two testing:  
1. i reinstall guest on LV(/dev/vg0/data1) and run dd command in guest 
2. use other LV to start vm and run dd command in guest

but all can not reproduce this bug, i am not sure how to trigger it.

Comment 4 Kevin Wolf 2011-06-27 09:36:35 UTC
Can you please try qemu-img check on this image? Do you know if there have been I/O errors?

Comment 5 FuXiangChun 2011-06-28 01:22:30 UTC
for block device qemu-img don't support check option. 
# qemu-img check /dev/vg0/data1
qemu-img: This image format does not support checks

Comment 6 Kevin Wolf 2011-06-28 11:24:27 UTC
Try the -f option:

# qemu-img check -f qcow2 /dev/vg0/data1

Comment 7 FuXiangChun 2011-06-29 01:47:35 UTC
hi kevin

   as i suspect guest maybe have problems that lead to core dump, so i reinstalled guest and try to reproduce it on lv data1. but after reinstall guest this bug can not be reproduced. now lv data1 haven't any errors and guest work well. 
# qemu-img check -f qcow2 /dev/vg0/data1
No errors were found on the image.

Comment 8 Kevin Wolf 2011-07-29 07:59:42 UTC
There seems to be a real bug somewhere, but it hasn't been reproduced so far and we don't have any useful information about the broken image. Closing.

Note You need to log in before you can comment on or make changes to this bug.