Upstream patch proposed:
https://www.redhat.com/archives/libvir-list/2013-June/msg01115.html

Fixed upstream:
commit 306c49ffd56a1c72b1892d50f2a75531c62f4a1d
Author:     Ján Tomko <jtomko>
AuthorDate: 2013-06-26 13:07:24 +0200
Commit:     Ján Tomko <jtomko>
CommitDate: 2013-06-26 15:05:43 +0200

    Fix invalid read in virCgroupGetValueStr
    
    Don't check for '\n' at the end of file if zero bytes were read.
    
    Found by valgrind:
    ==404== Invalid read of size 1
    ==404==    at 0x529B09F: virCgroupGetValueStr (vircgroup.c:540)
    ==404==    by 0x529AF64: virCgroupMoveTask (vircgroup.c:1079)
    ==404==    by 0x1EB475: qemuSetupCgroupForEmulator (qemu_cgroup.c:1061)
    ==404==    by 0x1D9489: qemuProcessStart (qemu_process.c:3801)
    ==404==    by 0x18557E: qemuDomainObjStart (qemu_driver.c:5787)
    ==404==    by 0x190FA4: qemuDomainCreateWithFlags (qemu_driver.c:5839)
    
    Introduced by 0d0b409.
    
    https://bugzilla.redhat.com/show_bug.cgi?id=978356

git describe: v1.1.0-rc1-25-g306c49f

Posted downstream:
http://post-office.corp.redhat.com/archives/rhvirt-patches/2013-June/msg00663.html

Test with: libvirt-0.10.2-19.el6
Steps same as comment 0 and no off-by-one error.
So it's verified.

Sorry i have to change this to ON_QA, i use the guest xml with <vcpu> and i can create the guest, but if i ignore this part, i can't create the guest, and i found a leak. i'll attach the valgrind msgs and needinfo dev.


virsh define /dev/stdin <<EOF
<domain type='qemu'>
  <name>duck</name>
  <memory unit='MiB'>32</memory>
  <vcpu placement='static' cpuset='1'>1</vcpu>  
  <os>
    <type arch='x86_64' machine='pc'>hvm</type>
  </os>
</domain>
EOF
Domain duck defined from /dev/stdin


virsh create /dev/stdin <<EOF
<domain type='qemu'>
  <name>duck</name>
  <memory unit='MiB'>32</memory>
  <os>
    <type arch='x86_64' machine='pc'>hvm</type>
  </os>
</domain>
EOF
error: Failed to create domain from /dev/stdin
error: internal error process exited while connecting to monitor:

Created attachment 771344 [details]
valgrind msg

I also can't boot a guest with <vcpu>
# virsh create /dev/stdin <<EOF
<domain type='qemu'>
  <name>duck</name>
  <memory unit='MiB'>32</memory>
  <vcpu placement='static' cpuset='1'>1</vcpu>  
  <os>
    <type arch='x86_64' machine='pc'>hvm</type>
  </os>
</domain>
EOF

error: Failed to create domain from /dev/stdin
error: internal error process exited while connecting to monitor:

It seems that <vcpu> is innocent. But I can define guest with/without <vcpu>.

So the problem is _start_ domain.

In valgrind log, we can see:
==15199== could not unlink /tmp/vgdb-pipe-from-vgdb-to-15199-by-root-on-zhpeng.example.com
==15199== could not unlink /tmp/vgdb-pipe-to-vgdb-from-15199-by-root-on-zhpeng.example.com
==15199== could not unlink /tmp/vgdb-pipe-shared-mem-vgdb-15199-by-root-on-zhpeng.example.com
==15199== execve(0x13e21890(/usr/libexec/qemu-kvm), 0x13e23370, 0x13b7ccb0) failed, errno 13
==15199== EXEC FAILED: I can't recover from execve() failing, so I'm dying.

Maybe valgrind break the start of domain. This is Valgrind-3.8.1.

@Jan, which version of valgrind you use?

I'm using valgrind-3.8.1-3.2.el6.x86_64 and can't start the domain in valgrind either, unless I disable selinux. And it should be valgrind's fault, because it works for me without it.

None of the leaks seem important to me.

Created attachment 773492 [details]
downgrade valgrind and test it again.

I tested it again with valgrind-3.6.0-5.el6.x86_64. And it's really a valgrind fault.
I can't find any off-by-one error anymore. So it's verified.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHBA-2013-1581.html