Bug 766216

Summary: libvirt: internal error unable to reserve PCI address
Product: Red Hat Enterprise Linux 6 Reporter: Vadym Chepkov <vchepkov>
Component: libvirtAssignee: Michal Privoznik <mprivozn>
Status: CLOSED DUPLICATE QA Contact: Virtualization Bugs <virt-bugs>
Severity: high Docs Contact:
Priority: unspecified    
Version: 6.2CC: acathrow, ajia, bgroslie, dallan, mzhan, rwu, weizhan
Target Milestone: rc   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2011-12-15 09:49:19 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Attachments:
Description Flags
VM xml
none
libvirtd log file none

Description Vadym Chepkov 2011-12-11 00:29:10 UTC 
After upgrade to Redhat 6.2 libvirt-0.9.4-23.el6_2.1.x86_64
libvirt fails to start any guest:

# virsh start mock5
error: Failed to start domain mock5
error: internal error unable to reserve PCI address 0:0:2.0

had to revert back to libvirt-0.8.7-18.el6.x86_64 to resolve the issue
Comment 2 Michal Privoznik 2011-12-13 09:31:49 UTC 
Can you please provide the daemon logs and mock5 XML?
That error message you are getting means that PCI address was already taken/reserved for other device.

To get logs, in /etc/libvirt/libvirtd.conf
log_level=1
MAY set log_filters="3:remote 4:event"
log_outputs="1:file:/var/log/libvirtd.log"
Comment 3 Vadym Chepkov 2011-12-13 13:43:17 UTC 
Created attachment 546206 [details]
VM xml
Comment 4 Vadym Chepkov 2011-12-13 13:44:01 UTC 
Created attachment 546207 [details]
libvirtd log file
Comment 5 Vadym Chepkov 2011-12-13 13:45:02 UTC 
There is only one VM running on this server
Comment 6 Michal Privoznik 2011-12-14 14:53:52 UTC 
Vadym,

your problem is related to https://bugzilla.redhat.com/show_bug.cgi?id=692355
In the fix for that bug, we had to unconditionally reserve PCI address 0:0:2.0 for a graphic card (even if there is currently none). See that bug.

To solve your problem you need to change PCI address of interface. So either edit <address> element and change slot to say '0x05' or remove the whole element and libvirt will generate it for you.

However, I am setting needinfo flag, so you can confirm it is working for you.
Comment 7 Vadym Chepkov 2011-12-14 18:51:48 UTC 
Yes, it did help, thank you.

I created a new VM just to compare configuration files and I see where slot 0x2 was added:

    <video>
      <model type='cirrus' vram='9216' heads='1'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/>
    </video>
Comment 8 Michal Privoznik 2011-12-15 09:49:19 UTC 
Since this issue is resolved, I am closing this. I agree that behavior during upgrade is not user friendly. However, it should be documented in errata for 692355 that PCI address 0:0:2.0 is reserved.
Comment 9 Dave Allan 2012-01-13 18:59:21 UTC 

*** This bug has been marked as a duplicate of bug 771603 ***