[Bug 31943] [drm:radeon_dvi_detect] *ERROR* HDMI Type A-1: probed a monitor but no|invalid EDID

[bugzilla-daemon@freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=31943

Alban Browaeys <prahal@yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #40595|0                           |1
        is obsolete|                            |

--- Comment #5 from Alban Browaeys <prahal@yahoo.com> 2010-12-03 20:06:47 PST ---
Created an attachment (id=40792)
 View: https://bugs.freedesktop.org/attachment.cgi?id=40792
 Review: https://bugs.freedesktop.org/review?bug=31943&attachment=40792

Fix the version in the edid (v2.3 makes no sense) checksum is correct with v1.3

This patch is not good for upstream , ie if we had this fix why not add  the
hundreds of other one for each broken edid kind.
Still it showcase the issue of this bug report.
First discarding the checksum would not help as here the reported incorrect
version is 2 and we only support 1 (thus the valid check would still fail).
The checksum is in fact correct if one replace this v2.3 by v1.3. It looks like
the version was mangled a posteriori.

One option is to completely discard the validity check via an option for the
drm module . Still it would move the issue to other layers if those expect a
valid edid or do a check of validity (they would also require a bypass option).

Another option is to load valid edid (from disk or initramfs,) like we do for
firmwares. One of the main issue of this approach is how to identify if and
which edid to load : phase out an id for the monitor or one for the invalid
edid (the latter via sum).
There are gaps to fill for this option though : will distro need to ship
hundreds of fixed edid in initramfs and such.

-- 
Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.