New monitor, pink vertical line and crazy screen resolution with Evergreen + KMS

[Dave Witbrodt]

Hi,

I've posted some successful results on this list lately when testing my 
Evergreen (HD 5750) card with drm-radeon-testing + KMS.

I bought a new monitor, which arrived this week, and found that I am 
getting a pink vertical line on the left side of the screen from top to 
bottom.  It is only a few pixels wide, probably 2 pixels (see below). 
The line is not present during POST/BIOS, GRUB, and early kernel boot; 
it appears when radeondrmfb kicks in, and is present in virtual 
terminals and X alike.  The line does not appear in screenshots, so it 
is not actually screen corruption, but likely a problem setting the mode 
correctly for the new monitor.

The monitor only supports HDMI and VGA/Dsub connections.  I do not have 
an HDMI cable and none was provided with the monitor.  The manufacturer 
did provide a DVI-to-HDMI converter cable, so that is what I am 
currently using.

I have been troubleshooting and collecting information for several days, 
and have a few more items on my list that I will be trying tomorrow. 
Here is what I know so far:

1) booting Windows Vista and setting the preferred 1920x1200 resolution 
results in a perfect display, no pink vertical line on left side of 
screen.  Conclusions:  monitor is not defective; it is at least possible 
to drive the monitor correctly using the DVI-to-HDMI converter.

2) I have been using custom kernels:  2.6.32.9, 2.6.33-rc8, and 2.6.33 
... all with the drm-radeon-testing branch merged in.  Suspicion:  I may 
have made errors in merging; however, these custom kernels worked fine 
with the previous monitor at the same 1920x1200 resolution.

3) Building a kernel directly from the drm-radeon-testing branch (no 
messing around with merging) gives the same pink vertical line. 
Conclusion:  problem exists even if I goofed up the merging in the other 
kernels.

4) The monitor allows on-screen display of information, and it reports 
an insane 1922x1200 resolution when using Linux with DRM + KMS.  That 
corresponds with an error of 2 pixels horizontally, which looks about 
right for the width of the pink line judging by eye.  (The monitor 
reports a correct 1920x1200 resolution when used with Vista.) 
Hypothesis:  DRM is somehow programming the wrong resolution for the 
monitor -- but is getting it nearly right, so that I get a bizarre mode 
with 2 extra pixels but no other corruption or artifacts on the display.

5) xdpyinfo reports this (notice particularly the size in millimeters):

screen #0:
   dimensions:    1920x1200 pixels (508x317 millimeters)
   resolution:    96x96 dots per inch

while Xorg.0.log includes this:

(II) RADEON(0): Supported standard timings:
(II) RADEON(0): #0: hsize: 1920  vsize 1200  refresh: 60  vid: 209
(II) RADEON(0): #1: hsize: 1680  vsize 1050  refresh: 60  vid: 179
(II) RADEON(0): #2: hsize: 1600  vsize 1200  refresh: 60  vid: 16553
(II) RADEON(0): #3: hsize: 1440  vsize 900  refresh: 60  vid: 149
(II) RADEON(0): #4: hsize: 1400  vsize 1050  refresh: 60  vid: 16528
(II) RADEON(0): #5: hsize: 1280  vsize 1024  refresh: 60  vid: 32897
(II) RADEON(0): #6: hsize: 1280  vsize 960  refresh: 60  vid: 16513
(II) RADEON(0): #7: hsize: 1152  vsize 864  refresh: 75  vid: 20337
(II) RADEON(0): Supported detailed timing:
(II) RADEON(0): clock: 154.0 MHz   Image Size:  593 x 371 mm
(II) RADEON(0): h_active: 1920  h_sync: 1968  h_sync_end 2000 
h_blank_end 2080 h_border: 0
(II) RADEON(0): v_active: 1200  v_sync: 1203  v_sync_end 1209 
v_blanking: 1235 v_border: 0
(II) RADEON(0): Serial No: 001LK37A00557
(II) RADEON(0): Ranges: V min: 56 V max: 75 Hz, H min: 24 H max: 80 kHz, 
PixClock max 154 MHz
(II) RADEON(0): Monitor name: HH281

This is what I have in xorg.conf for the monitor:

Section  "Monitor"
         Identifier      "HH281HPB"
         VendorName      "Hanns.G"
         ModelName       "HH281HPB"
         HorizSync       24-80
         VertRefresh     56-75
         DisplaySize     593     371
         Option          "DPMS"
EndSection

Apparently 'xdpyinfo' has a mind of its own!  ;)

6) There is a warning about fixing the timing clock in Xorg.0.log:

(II) [drm] DRM interface version 1.3
(II) [drm] DRM open master succeeded.
(II) RADEON(0): Output DP using monitor section HH281HPB
(II) RADEON(0): Output HDMI-0 has no monitor section
(II) RADEON(0): Output DVI-0 has no monitor section
(II) RADEON(0): Output DVI-1 has no monitor section
(**) RADEON(0): Option "ModeDebug" "true"
(II) RADEON(0): EDID for output DP
(II) RADEON(0): EDID for output HDMI-0
(II) RADEON(0): EDID for output DVI-0
(WW) EDID preferred timing clock 154.00MHz exceeds claimed max 150MHz, 
fixing
(II) RADEON(0): EDID for output DVI-1
(II) RADEON(0): Manufacturer: HSD  Model: 20e5  Serial#: 557
(II) RADEON(0): Year: 2010  Week: 1
(II) RADEON(0): EDID Version: 1.3
(II) RADEON(0): Digital Display Input
(II) RADEON(0): Max Image Size [cm]: horiz.: 59  vert.: 37
(II) RADEON(0): Gamma: 2.20
(II) RADEON(0): DPMS capabilities: StandBy Suspend Off


I'm not finished exhausting my list of steps to try and info to collect, 
but I have stopped for tonight and will finish tomorrow.  (Need sleep.) 
  I just thought I would post the info I have so far to get a headstart 
on finding out if anyone here can help me fix this.

The next items on my list are:

A)  Try booting with "radeon.modeset=0".  [Simple idea, but it didn't 
occur to me until just now.]

B)  Pull the HD 5750 and install my HD 4850.  The older card was working 
perfectly with KMS (even with a KVM + Dsub connection!) before I bought 
the new card to try out the Evergreen support.  I can try VGA/Dsub, DVI, 
and DVI-to-HDMI cables with the old card and see which permutations (if 
any) cause the pink vertical line to disappear.  If no pink line occurs 
at all, then the problem would seem to be Evergreen related.

C)  Try a real HDMI cable instead.  I mentioned that I don't own one, 
but one is in the mail and should be here early next week.  :)

The monitor is actually working beautifully.  It was a poor design 
decision for the manufacturer to include an HDMI port but no DVI port. 
Anyone owning a video card with DVI but no HDMI would be _forced_ to use 
the DVI-to-HDMI converter if they wanted the highest quality connection 
possible.  I just wonder if a workaround is possible to allow the 
DVI-to-HDMI converter to work (for those forced to use one).


Thanks,
Dave W.