Bug 1281580 - Boot fails in DRM init with bochs ('vga') and cirrus drivers since 2015-11-12
Boot fails in DRM init with bochs ('vga') and cirrus drivers since 2015-11-12
Status: NEW
Product: Fedora
Classification: Fedora
Component: kernel (Show other bugs)
rawhide
All Linux
unspecified Severity urgent
: ---
: ---
Assigned To: Kernel Maintainer List
Fedora Extras Quality Assurance
:
Depends On:
Blocks:
  Show dependency treegraph
 
Reported: 2015-11-12 15:45 EST by Adam Williamson
Modified: 2015-11-12 15:51 EST (History)
7 users (show)

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Environment:
Last Closed:
Type: Bug
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)

  None (edit)
Description Adam Williamson 2015-11-12 15:45:36 EST
Rawhide 2015-11-12 nightlies fail to boot in virtual machines using the 'vga' or 'cirrus' virtual adapters. (This affects our openQA deployment, as openQA instances use 'vga', so all openQA tests currently fail).

With both adapters, boot appears to be failing in DRM init. Using a serial console you can see that the boot proceeds to that point, then hangs.

https://www.happyassassin.net/temp/20151112-rawhide-vga.png is a screenshot of the serial console after a boot attempt with the 'cirrus' driver has hung. The 'bochs' driver is, I guess, a bit less chatty - the last message with that one is '[TTM] Initializing DMA pool allocator', but it's clearly failing in more or less the same place.

This seems very likely to be caused by the new kernel that landed on that date, kernel-4.4.0-0.rc0.git7.2.fc24 . There is one downstream change that has something to do with DRM:

- Drop CONFIG_DRM_DW_HDMI_AHB_AUDIO for now

though it *seems* unlikely to be causing this. It seems more likely this is caused by an upstream change, since that kernel also bumped to a newer upstream state. The affected kernel is at commit c5a3788:

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=c5a37883f42be712a989e54d5d6c0159b0e56599

The previous kernel was at bd4f203:

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=bd4f203e433387d39be404b67ad02acf6f76b7bc

and I've verified that that kernel works OK, so something between those two commits may have broken this.
Comment 1 Adam Williamson 2015-11-12 15:51:03 EST
So guess what, airlied's big drm-next merge landed between those two:

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit?id=3e82806b97398d542a5e03bd94861f79ce10ecee

which makes it a pretty likely candidate for this, I guess.

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