Regressions when upgrading 7.2 -> 7.4, ATI Radeon 9250 PCI

[Alex Deucher]

On Sun, Apr 19, 2009 at 2:33 PM, Jeremy Henty <onepoint at starurchin.org> wrote:
>
> On Sun, Apr 19, 2009 at 12:49:56PM -0400, Alex Deucher wrote:
>
>> In the bad case, the 1600x1200 mode is getting picked.  The mode comes
>> from the EDID in your monitor:
>> (II) RADEON(0): clock: 260.0 MHz   Image Size:  54 x 3688 mm
>> (II) RADEON(0): h_active: 1600  h_sync: 1664  h_sync_end 1856
>> h_blank_end 2160 h_border: 17
>> (II) RADEON(0): v_active: 1200  v_sync: 1201  v_sync_end 1204
>> v_blanking: 1250 v_border: 0
>>
>> The physical size is obviously wrong.
>
> Is  that  a  bug  in  the  driver  or  is  my  monitor  sending  bogus
> information?
>
>> Does your monitor  support 1600x1200?  We'd have to  add a quirk for
>> your monitor if the 1600x1200 mode is indeed bogus.
>
> According  the manual  it supports  1600x1200 at 80Hz, but  I  have never
> pushed it  that far.   (I'm squinting  enough as it  is.  :-)  It's an
> Iiyama Vision Master 450.
>
> Another thing: after fixing things by adding the Modes lines there was
> a fonts problem: all text in the Icewm manus and dialogs, and also GTK
> and  Qt widgets  (but *not*  FLTK widgets)  was rendered  as  a minute
> scribble.  It  looked as though  everything was rendered in  2-3 pixel
> type!  The fix  was to uncomment the line "#DisplaySize  360 270 # mm"
> in the configuration.  I think that forces the server to calculate the
> DPI from the real display size.
>
> Are any of these problems to  do with my card not being recognised, or
> is that a separate issue?

These are all related to the EDID in your monitor.  First the
1600x1200 mode it announces doesn't seem to work, secondly, it has
bogus size data: 54 x 3688 mm.  The bogus size data is what causes the
DPI to be calculated incorrectly which causes your font problems.

Alex