Bug 85441
Summary: | [vfio] [lockdep] Deadlock when attempting to unbind device from a running VM | ||
---|---|---|---|
Product: | Drivers | Reporter: | Marti Raudsepp (marti) |
Component: | PCI | Assignee: | drivers_pci (drivers_pci) |
Status: | RESOLVED PATCH_ALREADY_AVAILABLE | ||
Severity: | normal | CC: | alex.williamson |
Priority: | P1 | ||
Hardware: | x86-64 | ||
OS: | Linux | ||
Kernel Version: | 3.17.0-rc7 and Ubuntu 3.13.0-36-generic | Subsystem: | |
Regression: | No | Bisected commit-id: |
Description
Marti Raudsepp
2014-10-02 10:11:25 UTC
The 3.17 deadlock is fixed by this: https://lkml.org/lkml/2014/9/29/745 I've already tagged it for stable, so it will hopefully show up in 3.17.1, after the 3.18 merge window opens. Only 3.17 is affected by that bug. I believe the older locking issues have already been fixed in more recent kernels. (In reply to Alex Williamson from comment #1) > I believe the older locking issues have already been fixed in more recent > kernels. Oh, are you referring to this one? https://bugzilla.redhat.com/show_bug.cgi?id=1025700 Was there a stable tree patch with the fix? (In reply to Marti Raudsepp from comment #2) > (In reply to Alex Williamson from comment #1) > > I believe the older locking issues have already been fixed in more recent > > kernels. > > Oh, are you referring to this one? > https://bugzilla.redhat.com/show_bug.cgi?id=1025700 > Was there a stable tree patch with the fix? Ping, was there a patch for this? Or is that a Red Hat trade secret? Uh, no trade secrets, vfio is upstream, but I'm also not planning to figure out which patches fixed your specific issue since it's already fixed on current upstream. There was a try_reset_ set of patches for pci and vfio, that's a likely target. (In reply to Alex Williamson from comment #4) > Uh, no trade secrets, vfio is upstream Sorry, that was a bad joke. :) > I'm also not planning to figure > out which patches fixed your specific issue since it's already fixed on > current upstream. There was a try_reset_ set of patches for pci and vfio, > that's a likely target. Fair enough, I'll see if I can find it, thanks. I think this bug can be closed. |