Bug 28479 - Black screen after using 1024x768 that manually added by xrandr.
Summary: Black screen after using 1024x768 that manually added by xrandr.
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: 7.5 (2009.10)
Hardware: x86 (IA32) Linux (All)
: medium major
Assignee: Jesse Barnes
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-06-10 03:05 UTC by CSJ
Modified: 2010-08-03 01:11 UTC (History)
4 users (show)

See Also:
i915 platform:
i915 features:


Attachments
/var/log/Xorg.0.log (20.59 KB, text/plain)
2010-06-10 03:07 UTC, CSJ
no flags Details
xrandr -q --verbose (3.08 KB, text/plain)
2010-06-10 03:07 UTC, CSJ
no flags Details
VBIOS dump (64.00 KB, application/octet-stream)
2010-06-10 03:08 UTC, CSJ
no flags Details
intel_reg_dumper (7.14 KB, text/plain)
2010-06-10 03:09 UTC, CSJ
no flags Details
Be more permissive with pll frequencies. (2.13 KB, patch)
2010-07-02 14:41 UTC, Chris Wilson
no flags Details | Splinter Review
dmesg with kernel 2.6.35-rc4 + patch (15.39 KB, text/plain)
2010-07-06 04:07 UTC, CSJ
no flags Details
[Patch 1/4]: Add fixed panel mode for eDP without fixed mode in VBT (1.29 KB, patch)
2010-07-16 00:26 UTC, ykzhao
no flags Details | Splinter Review
[Patch 2/4]: Always enable panel fitting for eDP (1.08 KB, patch)
2010-07-16 00:26 UTC, ykzhao
no flags Details | Splinter Review
[Patch 3/4]: Use fixed panel timing to drive eDP (2.29 KB, patch)
2010-07-16 00:27 UTC, ykzhao
no flags Details | Splinter Review
[Patch 4/4]: validate modes for eDP by using fixe panel size (1.51 KB, patch)
2010-07-16 00:28 UTC, ykzhao
no flags Details | Splinter Review
dmesg (6.40 KB, text/plain)
2010-07-20 01:46 UTC, CSJ
no flags Details
intel_reg_dumper (10.33 KB, application/octet-stream)
2010-07-20 01:48 UTC, CSJ
no flags Details
Fix panel fitting regression. (1.11 KB, patch)
2010-07-20 01:58 UTC, Chris Wilson
no flags Details | Splinter Review
dmesg with 2.6.35-rc4 + 4 patches + Fix panel fitting regression since 734b4157 (6.40 KB, text/plain)
2010-07-20 03:25 UTC, CSJ
no flags Details
[Patch]: add the support of setting other modes for eDP (4.26 KB, patch)
2010-07-20 19:03 UTC, ykzhao
no flags Details | Splinter Review

Description CSJ 2010-06-10 03:05:54 UTC
I use xrandr to get resolution list, it only has 1366x768
there is no 1024x768.

Then, I use result of `gtf 1024 768 60` 
for xrandr to addmode as below steps:

xrandr --newmode "1024x768_60.00"  64.11  1024 1080 1184 1344  768 769 772 795  -HSync +Vsync
xrandr --addmode DP1 1024x768_60.00
xrandr --output DP1 --mode 1024x768_60.00

I got black screen.
but system is still alive.

Machine: Dell E6510
CPU: Intel Core i5 M 540 @ 2.53GHz
OS: Ubuntu 10.04

PS, On Windows,
there is 1024x768x60 and can be used normally.
Comment 1 CSJ 2010-06-10 03:07:33 UTC
Created attachment 36190 [details]
/var/log/Xorg.0.log
Comment 2 CSJ 2010-06-10 03:07:59 UTC
Created attachment 36191 [details]
xrandr -q --verbose
Comment 3 CSJ 2010-06-10 03:08:34 UTC
Created attachment 36192 [details]
VBIOS dump
Comment 4 CSJ 2010-06-10 03:09:04 UTC
Created attachment 36193 [details]
intel_reg_dumper
Comment 5 Chris Wilson 2010-06-29 06:29:10 UTC
Modesetting, including EDID probing, is handled in the kernel. I don't know how many of the Arrandale/Ironlake timing fixes from post-2.6.32 Ubuntu have applied to their kernel - it would be worth checking against the current mainline as well. In the meantime, can you please upload the dmesg in case something is reported whilst probing - and adding drm.debug=6 to your kernel commandline would increase the amount of debugging information whilst modesetting.

Was the xrandr output after adding the new mode?
Comment 6 Chris Wilson 2010-07-02 14:41:30 UTC
Created attachment 36699 [details] [review]
Be more permissive with pll frequencies.

This is the most likely candidate to fix missing modes, although there are a few other timing fixes for Ironlake/eDP.
Comment 7 CSJ 2010-07-06 04:07:49 UTC
Created attachment 36787 [details]
dmesg with kernel 2.6.35-rc4 + patch

This is dmesg using 2.6.35-rc4 + below patch
https://bugs.freedesktop.org/attachment.cgi?id=36699
Comment 8 CSJ 2010-07-06 04:10:26 UTC
(In reply to comment #6)
> Created an attachment (id=36699) [details]
> Be more permissive with pll frequencies.
> 
> This is the most likely candidate to fix missing modes, although there are a
> few other timing fixes for Ironlake/eDP.

Hi, I use (In reply to comment #5)
> Modesetting, including EDID probing, is handled in the kernel. I don't know how
> many of the Arrandale/Ironlake timing fixes from post-2.6.32 Ubuntu have
> applied to their kernel - it would be worth checking against the current
> mainline as well. In the meantime, can you please upload the dmesg in case
> something is reported whilst probing - and adding drm.debug=6 to your kernel
> commandline would increase the amount of debugging information whilst
> modesetting.
> 
> Was the xrandr output after adding the new mode?

After adding the new mode, 
I can see it listed in `xrandr` output.

But screen black when I switch resolution to 1024x768.
but I still can type command `xrandr -s 0` to go back from black screen.
and resolution come back to 1366x768
Comment 9 ykzhao 2010-07-16 00:26:05 UTC
Created attachment 37105 [details] [review]
[Patch 1/4]: Add fixed panel mode for eDP without fixed mode in VBT
Comment 10 ykzhao 2010-07-16 00:26:45 UTC
Created attachment 37106 [details] [review]
[Patch 2/4]: Always enable panel fitting for eDP
Comment 11 ykzhao 2010-07-16 00:27:23 UTC
Created attachment 37107 [details] [review]
[Patch 3/4]: Use fixed panel timing to drive eDP
Comment 12 ykzhao 2010-07-16 00:28:10 UTC
Created attachment 37108 [details] [review]
[Patch 4/4]: validate modes for eDP by using fixe panel size
Comment 13 ykzhao 2010-07-16 00:33:05 UTC
Hi, CSJ
    Will you please try the attached four patches in comment #9-12 on the latest Eric's drm-intel-next tree and see whether the issue can fixed?
    
    Please add the boot option of "drm.debug=0x04" and attach the output of dmesg, intel_reg_dumper after trying the other display mode created by yourself.

thanks.
Comment 14 CSJ 2010-07-20 01:46:42 UTC
Created attachment 37230 [details]
dmesg

dmesg after `xrandr --output DP1 --mode 1024x768`
with insmod drm.ko debug=4
Comment 15 CSJ 2010-07-20 01:48:26 UTC
Created attachment 37231 [details]
intel_reg_dumper

the result of intel_reg_dumper after `xrandr --output DP1 --mode 1024x768`
Comment 16 CSJ 2010-07-20 01:52:20 UTC
(In reply to comment #13)
> Hi, CSJ
>     Will you please try the attached four patches in comment #9-12 on the
> latest Eric's drm-intel-next tree and see whether the issue can fixed?
> 
>     Please add the boot option of "drm.debug=0x04" and attach the output of
> dmesg, intel_reg_dumper after trying the other display mode created by
> yourself.
> 
> thanks.

Hi ykzhao,
I applied 4 patches(#9 ~ #12) on kernel 2.6.35-rc4
(but I removed some PCH_eDP for compile succesffuly)

and add 1024x768 resolution got from gtf,
the screen not black now,
but become fuzzy(like many small screen overlaps likes a garbage screen)

dmesg and intel_reg_dumper are attached.
Comment 17 CSJ 2010-07-20 01:54:00 UTC
(In reply to comment #16)
> (In reply to comment #13)
> > Hi, CSJ
> >     Will you please try the attached four patches in comment #9-12 on the
> > latest Eric's drm-intel-next tree and see whether the issue can fixed?
> > 
> >     Please add the boot option of "drm.debug=0x04" and attach the output of
> > dmesg, intel_reg_dumper after trying the other display mode created by
> > yourself.
> > 
> > thanks.
> 
> Hi ykzhao,
> I applied 4 patches(#9 ~ #12) on kernel 2.6.35-rc4
> (but I removed some PCH_eDP for compile succesffuly)
> 
> and add 1024x768 resolution got from gtf,
> the screen not black now,
> but become fuzzy(like many small screen overlaps likes a garbage screen)
> 
> dmesg and intel_reg_dumper are attached.

P.S:
I got fuzzy screen after `xrandr --output DP1 --mode 1024x768`
Comment 18 Chris Wilson 2010-07-20 01:58:40 UTC
Created attachment 37232 [details] [review]
Fix panel fitting regression.

Sounds like you then hit the panel fitting regression in 2.6.35-rc1.
Comment 19 CSJ 2010-07-20 03:25:59 UTC
Created attachment 37236 [details]
dmesg with 2.6.35-rc4 + 4 patches + Fix panel fitting regression since 734b4157

dmesg with debug=4,
remove drm_mode_set_crtcinfo(adjusted_mode, 0);

https://bugs.freedesktop.org/attachment.cgi?id=37232
Comment 20 CSJ 2010-07-20 03:27:08 UTC
(In reply to comment #18)
> Created an attachment (id=37232) [details]
> Fix panel fitting regression.
> 
> Sounds like you then hit the panel fitting regression in 2.6.35-rc1.

Hi Chris,

I applied [PATCH] drm/i915: Fix panel fitting regression since 734b4157
But looks like the same, got a garbage screen.

dmesg attached.

thanks
Comment 21 ykzhao 2010-07-20 19:03:52 UTC
Created attachment 37256 [details] [review]
[Patch]: add the support of setting other modes for eDP

Hi, CSJ
    thanks for the testing.  Very sorry that the patches in comment #9-12 is not based on 2.6.35-rc4.
    From the output of intel_reg_dumper in comment #15 it seems that the panel fitting is not set. Not sure whether the patch in comment #10 is correctly applied. 
   > PFA_CTL_1: 0x00800000 (disable, 
    PFB_CTL_1: 0x00000000 (disable, 
    PFA_WIN_POS: 0x00000000 (0, 0)
    PFB_WIN_POS: 0x00000000 (0, 0)
    PFA_WIN_SIZE: 0x00000000 (0, 0)
    PFB_WIN_SIZE: 0x00000000 (0, 0)

   Will you please double check whether the patch in comment #10 is correctly applied?  
   Of course I merge the patches in comment #9-12 into one patch for your convenience. And it is refreshed based on 2.6.35-rc4 kernel(The IS_PCH_eDP is also removed).
    Will you please try it and see whether the screen is still garbled?

thanks.

thanks.
Comment 22 CSJ 2010-07-20 19:48:13 UTC
(In reply to comment #21)
> Created an attachment (id=37256) [details]
> [Patch]: add the support of setting other modes for eDP
> 
> Hi, CSJ
>     thanks for the testing.  Very sorry that the patches in comment #9-12 is
> not based on 2.6.35-rc4.
>     From the output of intel_reg_dumper in comment #15 it seems that the panel
> fitting is not set. Not sure whether the patch in comment #10 is correctly
> applied. 
>    > PFA_CTL_1: 0x00800000 (disable, 
>     PFB_CTL_1: 0x00000000 (disable, 
>     PFA_WIN_POS: 0x00000000 (0, 0)
>     PFB_WIN_POS: 0x00000000 (0, 0)
>     PFA_WIN_SIZE: 0x00000000 (0, 0)
>     PFB_WIN_SIZE: 0x00000000 (0, 0)
> 
>    Will you please double check whether the patch in comment #10 is correctly
> applied?  
>    Of course I merge the patches in comment #9-12 into one patch for your
> convenience. And it is refreshed based on 2.6.35-rc4 kernel(The IS_PCH_eDP is
> also removed).
>     Will you please try it and see whether the screen is still garbled?
> 
> thanks.
> 
> thanks.

Hi ykzhao,

After appling #
Comment 23 CSJ 2010-07-20 19:49:45 UTC
(In reply to comment #21)
> Created an attachment (id=37256) [details]
> [Patch]: add the support of setting other modes for eDP
> 
> Hi, CSJ
>     thanks for the testing.  Very sorry that the patches in comment #9-12 is
> not based on 2.6.35-rc4.
>     From the output of intel_reg_dumper in comment #15 it seems that the panel
> fitting is not set. Not sure whether the patch in comment #10 is correctly
> applied. 
>    > PFA_CTL_1: 0x00800000 (disable, 
>     PFB_CTL_1: 0x00000000 (disable, 
>     PFA_WIN_POS: 0x00000000 (0, 0)
>     PFB_WIN_POS: 0x00000000 (0, 0)
>     PFA_WIN_SIZE: 0x00000000 (0, 0)
>     PFB_WIN_SIZE: 0x00000000 (0, 0)
> 
>    Will you please double check whether the patch in comment #10 is correctly
> applied?  
>    Of course I merge the patches in comment #9-12 into one patch for your
> convenience. And it is refreshed based on 2.6.35-rc4 kernel(The IS_PCH_eDP is
> also removed).
>     Will you please try it and see whether the screen is still garbled?
> 
> thanks.
> 
> thanks.

Hi ykzhao,

After appling #12, it works fine on the new added mode 1024x768


Thanks a lot
:)
Comment 24 ykzhao 2010-07-20 20:17:34 UTC
Hi, CSJ
    Thanks for so quick response. It is good news that it can work as expected after applying the patch in comment #21. 

thanks.
Comment 25 CSJ 2010-07-20 20:54:59 UTC
(In reply to comment #24)
> Hi, CSJ
>     Thanks for so quick response. It is good news that it can work as expected
> after applying the patch in comment #21. 
> 
> thanks.

sorry for typo, 
I applied comment #21. but not #12.

thanks
Comment 26 Chris Wilson 2010-08-03 01:11:51 UTC
The patches are now in Eric's drm-intel-next branch and will be included in 2.6.36-rc1.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.