Bug 27738 - [915] 1920x1200 resolution fails, probable FIFO issue.
Summary: [915] 1920x1200 resolution fails, probable FIFO issue.
Status: CLOSED INVALID
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: x86 (IA32) Linux (All)
: medium normal
Assignee: Daniel Vetter
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-04-19 07:37 UTC by Julien Cohen
Modified: 2017-07-24 23:08 UTC (History)
4 users (show)

See Also:
i915 platform:
i915 features:


Attachments
xorg log file (27.28 KB, text/plain)
2010-04-19 07:37 UTC, Julien Cohen
no flags Details
dmesg output (46.37 KB, text/plain)
2010-04-19 07:39 UTC, Julien Cohen
no flags Details
xrandr output (5.29 KB, text/plain)
2010-04-19 07:40 UTC, Julien Cohen
no flags Details
intel_reg_dumper output (gzipped) (111.14 KB, application/gzip)
2010-04-19 07:47 UTC, Julien Cohen
no flags Details
Fix wm calculation. (2.32 KB, patch)
2010-08-20 09:27 UTC, Chris Wilson
no flags Details | Splinter Review

Description Julien Cohen 2010-04-19 07:37:07 UTC
Created attachment 35161 [details]
xorg log file

System environment:
-- chipset: 915GM
-- system architecture: 32-bit i686
-- xf86-video-intel: 2.11.0
-- xserver: X.Org X Server 1.7.6.901 (1.7.7 RC 1)
-- mesa:
-- libdrm:
-- kernel: 2.6.32-3-686-bigmem
-- Linux distribution: debian sid
-- Machine or mobo model: dell latitude X1
-- Display connector: VGA

Reproducing steps:
When I try to use 1920x1200 or 1600x1200 resolution, the screen gets "out of sync" (there is no message error on the screen, the image is unreadable and moving from left to right, the screen info says : analog, 74.0 kHz, 60Hz PN 1920x1200). I use xrandr or grandr to switch to these resolutions. The problem is the same if I boot directly within this resolution. The problem is the same is the laptop screen is on or off (well at least while testing with older versions, I have not tried with this version). 


Additional info:

The problem is the same with older versions of the driver.

It works well under windows with this screen (samsung syncmaster T240).
I have the same problem with an other laptop (ASUS eeepc) with the same graphic card (915GM). I tried several distributions (ubuntu, debian, mandriva).

I manage to get 1600x1200 with the vesa driver.

With older versions, I tried many modelines following various hints on internet (I don't remember exactly which ones).
Comment 1 Julien Cohen 2010-04-19 07:39:15 UTC
Created attachment 35162 [details]
dmesg output
Comment 2 Julien Cohen 2010-04-19 07:40:09 UTC
Created attachment 35163 [details]
xrandr output
Comment 3 Julien Cohen 2010-04-19 07:47:38 UTC
Created attachment 35164 [details]
intel_reg_dumper output (gzipped)
Comment 4 Julien Cristau 2010-04-19 10:44:32 UTC
On Mon, Apr 19, 2010 at 07:37:07 -0700, bugzilla-daemon@freedesktop.org wrote:

> -- kernel: 2.6.32-3-686-bigmem

can you upgrade to 2.6.32-4-686-bigmem?  that has newer drm code from
2.6.33, so could behave differently.
Comment 5 Julien Cohen 2010-04-20 00:20:39 UTC
I upgraded to 2.6.32-4-686 and the result is the same (at least on the screen).
Comment 6 Julien Cohen 2010-04-20 03:37:15 UTC
Well, now I perceive that the problem occurs as soon as the console switches to frame buffer at boot time, before the X server is launched. 

I think this occurs only since I use that new kernel (but I am not sure).

So it might not be related to the X server...
Comment 7 Chris Wilson 2010-07-14 04:33:56 UTC
As always there have been a few KMS fixes over the last couple of months, and in particular I think this one may be relevant for VGA connections:

commit 9bc354998f8aa8a83b0cd430e8fcf2cbc3be7367
Author: Adam Jackson <ajax@redhat.com>
Date:   Fri May 28 17:17:37 2010 -0400

    drm/i915: Honor sync polarity from VBT panel timing descriptors
    
    I'm actually kind of shocked that it works at all otherwise.
    
    Signed-off-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Eric Anholt <eric@anholt.net>
Comment 8 Chris Wilson 2010-07-20 06:48:28 UTC
Another cause might be FIFO underruns. Can you attach a dmesg with drm.debug=0x4?

If it is, one workaround is to try:

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 9ddb7b5..edaae6c 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1434,6 +1434,13 @@ static int i915_load_modeset_init(struct drm_device
*dev,

        I915_WRITE(INSTPM, (1 << 5) | (1 << 21));

+       /* XXX hack for bug 22996, preset the FIFO to accommodate a 2048
external display */
+       {
+               uint32_t size = I915_READ(DSPARB) & 0x7f;
+               size = (95 - size) << DSPARB_CSTART_SHIFT;
+               I915_WRITE(DSPARB, size);
+       }
+
        ret = intel_fbdev_init(dev);
        if (ret)
                goto cleanup_irq;
Comment 9 Chris Wilson 2010-08-20 09:27:03 UTC
Created attachment 38014 [details] [review]
Fix wm calculation.

This fixes the flicker on my 915GM with a 1920x1080 panel.
Comment 10 Julien Cohen 2010-08-26 02:53:58 UTC
> This fixes the flicker on my 915GM with a 1920x1080 panel.

Great, I have to try this. Is there a way to get a .deb package or do I have to compile it ? (I am not used to do this.)



> Can you attach a dmesg with drm.debug=0x4?

Can you precise what I have to do? Is it a parameter I have to pass to the kernel with grub? How should I do it?

Julien.
Comment 11 Eugeni Dodonov 2011-08-22 09:38:46 UTC
> > Can you attach a dmesg with drm.debug=0x4?
> 
> Can you precise what I have to do? Is it a parameter I have to pass to the
> kernel with grub? How should I do it?

Just add this parameter to the grub boot line (when booting grub and selecting a choice of what kernel to boot, there should be an option for you to edit the command line); or append it into the kernel boot parameters in /boot/grub/menu.lst.
Comment 12 Eugeni Dodonov 2011-08-22 09:40:22 UTC
(However, by judging from the age of this bug report, I wonder if it is still accurate).
Comment 13 Daniel Vetter 2012-02-10 12:18:39 UTC
Please retest with latest kernels (at least 3.2) and report whether you still have this issue.
Comment 14 Chris Wilson 2012-10-21 14:30:01 UTC
Timeout. Please do reopen if you can still reproduce the issue and help us diagnose the problem, thanks.


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.