i915, framebuffer: wrong xoffset?

i915, framebuffer: wrong xoffset?

[Lucas De Marchi]

Since 3.0-rc1 when booting my computer the text in the fb console is
displayed with a wrong xoffset. I took a picture (sorry for the poor
quality):

http://www.politreco.com/files/IMG_20110610_090045.jpg

It appears that there's a negative xoffset applied when booting the
computer. After starting X and returning to console everything is back
to normal, though.

I started  to bisect this, but there are too many iterations missing:

$ git bisect log
git bisect start
# good: [61c4f2c81c61f73549928dfd9f3e8f26aa36a8cf] Linux 2.6.39
git bisect good 61c4f2c81c61f73549928dfd9f3e8f26aa36a8cf
# bad: [55922c9d1b84b89cb946c777fddccb3247e7df2c] Linux 3.0-rc1
git bisect bad 55922c9d1b84b89cb946c777fddccb3247e7df2c
# good: [c44dead70a841d90ddc01968012f323c33217c9e] Merge branch
'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
git bisect good c44dead70a841d90ddc01968012f323c33217c9e
# bad: [a09ed5e00084448453c8bada4dcd31e5fbfc2f21] vmscan: change
shrink_slab() interfaces by passing shrink_control
git bisect bad a09ed5e00084448453c8bada4dcd31e5fbfc2f21
# good: [9461702d2a54cd4d9da09b7755c96815791a9d07] m68k: let Makefile
sort out compiling mmu and non-mmu lib/checksum.c
git bisect good 9461702d2a54cd4d9da09b7755c96815791a9d07


My card:

00:02.0 VGA compatible controller: Intel Corporation Mobile 4 Series
Chipset Integrated Graphics Controller (rev 07)
00:02.1 Display controller: Intel Corporation Mobile 4 Series Chipset
Integrated Graphics Controller (rev 07)

dmesg:

[    0.619044] i915 0000:00:02.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[    0.619049] i915 0000:00:02.0: setting latency timer to 64
[    0.712209] i915 0000:00:02.0: irq 45 for MSI/MSI-X
[    0.712215] [drm] Supports vblank timestamp caching Rev 1 (10.10.2010).
[    0.712216] [drm] Driver supports precise vblank timestamp query.
[    0.764605] composite sync not supported
[    0.765713] fixme: max PWM is zero.
[    0.813774] composite sync not supported
[    0.827331] fbcon: inteldrmfb (fb0) is primary device
[    0.863748] Console: switching to colour frame buffer device 160x50
[    0.866442] fb0: inteldrmfb frame buffer device
[    0.866443] drm: registered panic notifier
[    0.904054] acpi device:06: registered as cooling_device0
[    0.904172] input: Video Bus as
/devices/LNXSYSTM:00/device:00/PNP0A08:00/LNXVIDEO:01/input/input4
[    0.904179] ACPI: Video Device [GFX0] (multi-head: yes  rom: no  post: no)
[    0.904208] [drm] Initialized i915 1.6.0 20080730 for 0000:00:02.0 on minor 0


It says "multi-head: yes", but there's no other monitor connected whatsoever.




Any ideas what could be happening?

-- 

Lucas De Marchi

Re: i915, framebuffer: wrong xoffset?

[Keith Packard]

[-- Attachment #1: Type: text/plain, Size: 594 bytes --]

On Sat, 18 Jun 2011 13:29:47 -0300, Lucas De Marchi <lucas.de.marchi@gmail.com> wrote:

> Since 3.0-rc1 when booting my computer the text in the fb console is
> displayed with a wrong xoffset. I took a picture (sorry for the poor
> quality):

Can you capture register dumps in the working and broken configuration?

git clone git://anongit.freedesktop.org/git/xorg/app/intel-gpu-tools

Build that and run tools/intel_reg_dumper and send along the output of
that in working and non-working configurations. That should show us
where the problem lies.

-- 
keith.packard@intel.com

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: i915, framebuffer: wrong xoffset?

[Lucas De Marchi]

Hi Keith,

On Sun, Jun 26, 2011 at 19:55, Keith Packard <keithp@keithp.com> wrote:
> On Sat, 18 Jun 2011 13:29:47 -0300, Lucas De Marchi <lucas.de.marchi@gmail.com> wrote:
>
>> Since 3.0-rc1 when booting my computer the text in the fb console is
>> displayed with a wrong xoffset. I took a picture (sorry for the poor
>> quality):
>
> Can you capture register dumps in the working and broken configuration?
>
> git clone git://anongit.freedesktop.org/git/xorg/app/intel-gpu-tools
>
> Build that and run tools/intel_reg_dumper and send along the output of
> that in working and non-working configurations. That should show us
> where the problem lies.

I've just build a new kernel and noticed it's fixed in current Linux
tree. Since -rc4 still had this bug, it was very quick to bisect this,
and the fix was this commit:


commit efc2924e733631a64c0afed8dbba2741d186998c
Author: Keith Packard <keithp@keithp.com>
Date:   Mon Jun 6 17:12:49 2011 -0700

    drm/i915: Call intel_enable_plane from i9xx_crtc_mode_set (again)

    This change got placed in the ironlake path instead of the 9xx path
    during a recent code shuffle.

    Signed-off-by: Keith Packard <keithp@keithp.com>



Does it make sense? Do you still want the register dumps or they
aren't needed anymore?


Lucas De Marchi

Re: i915, framebuffer: wrong xoffset?

[Keith Packard]

[-- Attachment #1: Type: text/plain, Size: 502 bytes --]

On Mon, 27 Jun 2011 19:17:48 -0300, Lucas De Marchi <lucas.de.marchi@gmail.com> wrote:

> I've just build a new kernel and noticed it's fixed in current Linux
> tree. Since -rc4 still had this bug, it was very quick to bisect this,
> and the fix was this commit:

It doesn't entirely make sense, but then we don't know why this call is
needed for any hardware, so it doesn't surprise me that this fixes your
problem too. Good to know, and thanks for retesting.

-- 
keith.packard@intel.com

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]