Bug 101081

Summary: NULL pointer dereference at radeon initializing at boot
Product: Drivers Reporter: Christoph Haag (haagch.christoph)
Component: Video(Other)Assignee: drivers_video-other
Status: RESOLVED CODE_FIX    
Severity: normal CC: wshuman3
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 4.2-rc1 Subsystem:
Regression: No Bisected commit-id:
Attachments: dmesg
dmesg from linux 4.1 for comparison

Description Christoph Haag 2015-07-06 14:30:16 UTC 
Created attachment 182011 [details]
dmesg

$ lspci | grep VGA
00:02.0 VGA compatible controller: Intel Corporation 3rd Gen Core processor Graphics Controller (rev 09)
01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Wimbledon XT [Radeon HD 7970M] (rev ff)

See attached dmesg for the complete error.

First thing that I find strange is

   Direct firmware load for radeon/TAHITI_vce.bin failed with error -2

1. This file is not in upstream linux-firmware git repository yet. I've seen it here https://secure.freedesktop.org/~agd5f/radeon_ucode/ and put it into /usr/lib/firmware/radeon, but the message and null pointer dereference doesn't change at all.
2. I don't have a TAHITI GPU, I have a PITCAIRN GPU.

There is __pm_runtime_disable in the stack trace so I tried booting with radeon.runpm=0 radeon.dpm=0 (as in the attached dmesg), but the exact same thing happens with and without power management.

Symptoms: Boot is extremely slow and X.org completely hangs on startup. It should so some interaction with the radeon GPU on startup, i.e. waking it up from runpm for a short time, perhaps there is the cause in its hang.
Comment 1 Christoph Haag 2015-07-06 14:31:50 UTC 
Created attachment 182021 [details]
dmesg from linux 4.1 for comparison
Comment 2 Christoph Haag 2015-07-09 14:39:48 UTC 
I built agd5's drm-fixes-4.2-wip and the issue is gone.

I haven't tested which commit fixed it but I guess this could have had something to do with it:
http://cgit.freedesktop.org/~agd5f/linux/commit/?h=drm-fixes-4.2-wip&id=05995ea6b002dffa1831212435689b7802de5be1

I will close this when rc2 is out and it's still gone.
Comment 3 Christoph Haag 2015-07-13 13:38:23 UTC 
Okay, fixed in rc2. Thanks.