Bug 57811

Summary:	Can not detect the resolution of the external monitor after upgrade to 3.9.1
Product:	Drivers	Reporter:	wei sun (wsun07)
Component:	Video(DRI - Intel)	Assignee:	Imre Deak (imre.deak)
Status:	RESOLVED INSUFFICIENT_DATA
Severity:	high	CC:	daniel, intel-gfx-bugs
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	3.9.1	Subsystem:
Regression:	Yes	Bisected commit-id:
Attachments:	dmesg with drm.debug=0xf in kernel command line dmesg with drm.debug=0xf in kernel command line with the two patches

Description wei sun 2013-05-08 21:10:23 UTC

Hi there,

I have a DELL XPS 13 laptop and it supported external monitor without any problem before when I was using kernel 3.8. Two days ago, I upgraded the kernel to 3.9 and I could not set up the resolution for the external monitor correctly. Today, I tried 3.9.1 and got the result. I checked the syslog and found the following error message:
"May  8 15:11:24 debian kernel: [ 5772.815695] [drm:intel_dp_i2c_aux_ch] *ERROR* too many retries, giving up".

Please let me know if you need more information.

Thanks,
Wei

Comment 1 Daniel Vetter 2013-05-09 09:13:22 UTC

One more for Imre's wait_for_event fixes.

Comment 2 wei sun 2013-05-14 17:44:05 UTC

I tried 3.9.2 and the issue was still there.

Comment 3 Daniel Vetter 2013-05-20 20:17:00 UTC

Ping for Imre to paste his patches for testing ...

Comment 4 Imre Deak 2013-05-20 20:30:38 UTC

(In reply to comment #3)
> Ping for Imre to paste his patches for testing ...

The two fixes attached to Bug 57461 could be tried, but judging from the error message the root cause can be something unrelated. Could you Wei also attach a dmesg log with drm.debug=0xf in the kernel command line?

Comment 5 wei sun 2013-05-22 04:29:49 UTC

Hi there,

I tried the two patches and they worked partially. The resolution of the external monitor is still wrong when I plugged it first time. However, the correct resolution can be detected with the gnome display configuration tools and I can manually set it up. The error message is still there. 

I will send you the log file tomorrow.

thanks,
Wei

Comment 6 wei sun 2013-05-24 13:36:49 UTC

Created attachment 102411 [details]
dmesg with drm.debug=0xf in kernel command line

Comment 7 wei sun 2013-05-24 13:37:48 UTC

Created attachment 102421 [details]
dmesg with drm.debug=0xf in kernel command line with the two patches

Comment 8 Imre Deak 2013-05-24 13:43:03 UTC

(In reply to comment #7)
> Created an attachment (id=102421) [details]
> dmesg with drm.debug=0xf in kernel command line with the two patches

This doesn't actually have drm.debug=0xf set .. If it was you would see it in the dmesg at 'Command line' and a lot more messages from DRM.

Comment 9 wei sun 2013-05-26 03:25:50 UTC

The set up for the kernel command without patch and with patch are exactly same. I do not know why the messages are not logged with the two patches. Anyway, you may start with the log file without the two patches and see if you can find any clue.

Comment 10 Jani Nikula 2013-08-08 09:44:50 UTC

Wei Sun, please provide the requested dmesg with drm.debug=0xe module parameter set to make progress on the bug. Thank you.

Comment 11 Jani Nikula 2013-08-09 06:29:08 UTC

Wei Sun wrote:
> This issue was already gone after I upgraded to 3.10.3 kernel in Debian.

Apparently we broke something in 3.9 but fixed it again in 3.10. We don't know what it was, and I don't expect us to find out. (Unless you're willing to bisect this for completeness...) 3.9 is EOL anyway. 

Just to make sure, is your 3.10.3 a vanilla upstream kernel, or a Debian kernel? I'd like to know we have the fix upstream, and not only in Debian.

Comment 12 Jani Nikula 2013-10-09 12:32:19 UTC

Apparently reporter is happy this is working in 3.10, closing. Please reopen if the problem reappears in newer kernels.