[Intel-gfx] [PATCH 0/3] [drm/i915] - LVDS mode setting fixes
isely at isely.net
Thu Mar 17 06:57:43 PDT 2011
Back in 2008, I generated these bug reports:
These were entered due to a project I've been involved in where we pair
up a special purpose display device with unusual video timings, via
LVDS, to a commercial x86 single board computer with an Intel GPU. The
SBC's video BIOS can't have any implicit knowledge of the nature of the
otherwise unrelated display device, and since the xorg driver had been
trying to probe the hardware / use the video BIOS, this put our display
into an unusable state.
I later submitted patches to fix these issues and they were merged into
the userspace xorg Intel driver. These fixes allowed for some
additional LVDS manual configuration and made it possible to directly
set the LVDS output video timings via a normal xorg modeline.
Fast forward to 2011 and this same project, due to the SBC being EOL'ed,
is updating to later hardware with a more modern Intel GPU (GMA-4500),
which then forced an update to a later Linux kernel, later X server,
dri2, and most significantly kernel mode setting. With KMS being used
now for Intel GPUs, the previous mode setting logic in the userspace
xorg Intel driver had since been removed, and along with it the fixes I
had made. The KMS implementation must have forked from the userspace
code before I had submitted those patches because the same problems are
in fact happening there and the corresponding fixes never made it into
the drm/i915 module.
This patch series (3 of them) basically implement the same fixes as was
previously done for the userspace driver back in 2008. The fixes are
not a direct port; I coded the changes and obviously tested again.
This is also why there are 3 patches not 2; the third one is a fix for
another problem uncovered while debugging the fixed mode change.
I had also opened a new bug report for the 24 bit LVDS issue. Patch 1/3
in this series is the same as the patch posted to that bug report, found
Note that these patches all have default behavior that preserves what
the i915 module already does. So there's no impact to users until the
relevant kernel module option is tweaked to enable the new behavior
implemented by the patches. (This is just like what I did in 2008
except now it's a module otion instead of an xorg driver option.)
I'd very much appreciate it if these could please be merged into
drm/i915. The changes are not complex and should be easy to follow.
Signed-off-by: Mike Isely <isely at pobox.com>
isely at pobox.com (or isely at isely.net)
isely @ isely (dot) net
PGP: 03 54 43 4D 75 E5 CC 92 71 16 01 E2 B5 F5 C1 E8
More information about the Intel-gfx