[Bug 92765] Linux 4.1.12+, 4.2.5+ -> drm:radeon_pm_late_init [radeon]] *ERROR* + WARNING: CPU: 0 PID: 686 at fs/sysfs/dir.c:31 sysfs_warn_dup+0x68/0x80() +

Holger Hoffstätte holger.hoffstaette at googlemail.com
Mon Nov 2 07:28:46 PST 2015


On Mon, 02 Nov 2015 10:11:45 -0500, Alex Deucher wrote:

>>> Fixed in:
>>> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=49abb26651167c892393cd9f2ad23df429645ed9
>>
>> Alex, I don't know why there are two different bugzillas for this, but
>> please take a look at https://bugzilla.kernel.org/show_bug.cgi?id=106851
>> because the issue does not seem to be fixed by that patch at all.
>>
> 
> You should figure out what the files are failing to registers in the
> first place on your system.  Setting the flag unconditionally just
> fixes a symptom.  I'll change the code to set the flag unconditionally
> to avoid the error spew on resume, but it's really just papering over
> the real problem on your board.

Normally I'd agree (I understand what the code does), but I don't have any
errors - and all sysfs entries! - after boot, so as far as I can tell the
flag isn't set properly even after a successful boot, and then leads to
spews on resume. Maybe the entire process is racy somehow?

My card is an admittedly old, but perfectly working fanless (!) HD 5500,
1GB RAM, driver is r600 and everything - including dpms, suspend/wakeup etc.
works fine. I tried to disable dpm on boot and that of course stopped the
warnings, but then the card was constantly at ~55° instead of the usual
~45°, for no noticeably win in performance.

If you have any other suggestions or patches for me to try I'd be glad
to test/verify. In the meantime I think setting the flag is the easiest
fix; it certainly works for me.

Thanks for looking into it!

cheers
Holger



More information about the dri-devel mailing list