Bug 1479674
Summary: | Start vm after remove some vPHBs will fail at first try | ||
---|---|---|---|
Product: | Red Hat Enterprise Linux 7 | Reporter: | Wayne Sun <gsun> |
Component: | libvirt | Assignee: | Andrea Bolognani <abologna> |
Status: | CLOSED CANTFIX | QA Contact: | Virtualization Bugs <virt-bugs> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 7.4-Alt | CC: | abologna, bugproxy, dgibson, dzheng, haizhao, hannsj_uhl, hhan, jsuchane, junli, rbalakri |
Target Milestone: | rc | ||
Target Release: | 7.4-Alt | ||
Hardware: | ppc64le | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | If docs needed, set a value | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2017-09-21 09:28:26 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: | |||
Bug Depends On: | |||
Bug Blocks: | 1440030 |
Description
Wayne Sun
2017-08-09 07:20:05 UTC
AFAICT what's happening here is that when you run the VM with the extra vPHBs, libvirt is assigning some devices on the second vPHB, and updating the XML to give those devices explicit addresses on the second vPHB. When it goes away, libvirt obviously can't place those devices on the second vPHB any more, hence the failure. (In reply to David Gibson from comment #2) > AFAICT what's happening here is that when you run the VM with the extra > vPHBs, libvirt is assigning some devices on the second vPHB, and updating > the XML to give those devices explicit addresses on the second vPHB. When > it goes away, libvirt obviously can't place those devices on the second vPHB > any more, hence the failure. That's not quite what happens: as you can see (step 3 in the description) the PHBs get re-added automatically, but for some reason they end up after the devices rather than before them, and QEMU can't handle having device and controller specified in that order. I'll look into making it so the PHBs get re-added before the devices using them. Fix posted upstream. https://www.redhat.com/archives/libvir-list/2017-September/msg00084.html v2 patches posted upstream. https://www.redhat.com/archives/libvir-list/2017-September/msg00168.html Andrea, any update on getting this upstream and downstream? (In reply to David Gibson from comment #8) > Andrea, any update on getting this upstream and downstream? Sorry for taking so long to reply. It turns out that reordering controllers can break migration[1], so given that the problem scenario described above was kinda convoluted to being with, I think it's better to avoid further breakage and close the bug as CANTFIX. Doing so now. [1] https://www.redhat.com/archives/libvir-list/2017-September/msg00734.html |