[ubuntu-x] Resolution detection issues (timings?)

[Bryce Harrington]

Heya Steve,

At UDS one of the issues we discussed related to timing discovery for
modesetting, and this is in follow up to that.

We've gotten an unusually high number of bug reports about X not being
able to determine proper resolutions in Karmic.  These bugs seem to not
be due to the usual causes (broken monitor EDIDs).  Clues I've gathered
so far (some of which may be irrelevant):

  * Symptom is reduced selection of resolutions available.  Sometimes
    reduced to just 800x600 + 640x480.  Xorg.0.log often just shows that
    fuzzy aspect match is being used.
    E.g.: #471180 #470224 #459547

  * When checkbox was enabled and accessible via System Testing, the
    vast majority of bug reports we got from that was about bad
    resolution availability.
    E.g.: #458625 #466037 #465378 #465306 ...

  * In some cases the EDID is captured properly, just that the
    resolutions don't match up.  In other cases the EDID cannot be
    retrieved in Karmic, but was retrievable in Jaunty.
    E.g.: #473848 #$79323

  * Most reports indicate it is a regression for them since jaunty

  * Reports filed for both fresh installs and upgrades

  * Seeing increased resolution detection bug reports filed against
    pretty much all video drivers:
    -intel   #476481 #470224
    -ati     #484990
    -nv      #455224
    -nvidia  #479323 #473848 #481599
    -fglrx   

  * Sometimes exibits as reduced resolutions available, but sometimes
    resolution is available just doesn't work right.
    E.g.: #472960

I have no evidence that these have a common root cause.  The symptoms
and behaviors vary quite a bit so it may just be coincidence, and the
timing simply be due to the release.

But you had mentioned that there was a kernel timing patch that went in
shortly prior to release, so I wanted to follow up with the data I've
noted just in case that change might be a root cause for some of these
issues.

If you would be able to prepare a kernel with the fix you mentioned, or
else dropping the patch you were suspicious of, I can try to solicit
testing from some of the above bug reporters.

Thanks,
Bryce