[Bug 17902] [830M missing dvo driver] need support for DVO-LVDS via na2501

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=17902

--- Comment #81 from thor at math.tu-berlin.de 2012-06-01 06:10:32 UTC ---
Well, I've got a bit futher:

1) One can get i2c access mostly reliable when the drm.debug option is set
high. Thus, whether or not the i2c interface works is likely a matter of
timing. Could it be that the bit-banging interface is too critical to reach the
chip in all cases? If so, where could one set the timing?

2) I can get a stable picture by using the video bios to write the chip
configuration for me. For this, use

$ vbetool post
$ vbetool vbemode set 280

I also performed an i2c dump with and without the mode set through the video
bios:

Properly initialized through the video bios:

     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f    0123456789abcdef
00: 05 13 26 67 01 a5 19 a2 35 95 81 ff 03 04 00 00    ??&g????5??.??..
10: 00 a0 02 02 02 02 02 02 07 00 00 11 54 03 02 40    .????????..?T??@
20: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
30: 00 00 00 00 03 ff 00 44 88 88 00 00 00 00 00 00    ....?..D??......
40: 80 00 00 80 00 00 00 00 00 00 00 00 00 00 00 00    ?..?............
50: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
60: 11 00 00 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0    ?..?????????????
70: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 18 00    ..............?.
80: ff 07 3d 05 00 00 00 00 00 00 00 00 10 02 10 00    .?=?........???.
90: ff 07 a0 02 00 00 05 00 00 00 88 00 24 00 25 03    .???..?...?.$.%?
a0: 28 01 28 05 84 00 00 00 00 04 70 4f 00 00 03 03    (?(??....?pO..??
b0: 00 00 00 00 00 00 09 03 00 a0 00 20 33 33 33 33    ......??.?. 3333
c0: 01 90 00 0f 03 16 00 02 02 00 00 00 00 00 00 00    ??.???.??.......
d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
e0: 00 00 1a 00 80 00 00 00 80 00 00 00 00 00 00 00    ..?.?...?.......
f0: 86 eb ca 90 00 00 00 88 0a 00 0a 0a 0a 0a 0a 0a    ????...??.??????

Improperly initialized (initially):
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f    0123456789abcdef
00: 05 13 26 67 01 a5 19 a2 31 95 81 ff 03 04 00 00    ??&g????1??.??..
10: 00 a0 02 02 02 02 02 02 07 00 00 11 54 03 02 40    .????????..?T??@
20: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
30: 00 00 00 00 03 ff 00 28 88 88 00 00 00 00 00 00    ....?..(??......
40: 80 00 00 80 00 00 00 00 00 00 00 00 00 00 00 00    ?..?............
50: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
60: 11 00 00 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0 f0    ?..?????????????
70: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 18 00    ..............?.
80: ff 07 3d 05 00 00 00 00 00 00 00 00 10 02 10 00    .?=?........???.
90: ff 07 a0 02 00 00 05 00 00 00 88 00 24 00 25 03    .???..?...?.$.%?
a0: 28 01 28 05 84 00 00 00 00 04 70 4f 00 00 03 03    (?(??....?pO..??
b0: 00 00 00 00 00 00 09 03 00 a0 00 20 33 33 33 33    ......??.?. 3333
c0: 05 90 00 0f 03 16 00 02 02 00 00 00 00 00 00 00    ??.???.??.......
d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00    ................
e0: 00 00 1a 00 80 00 00 00 80 00 00 00 00 00 00 00    ..?.?...?.......
f0: 8f 07 58 90 00 00 00 88 0a 00 0a 0a 0a 0a 0a 0a    ??X?...??.??????

Differences are in registers 0x44 and 0xc0 which must contain 0x37 and 0x01,
respectively.

Just using i2cset to define these registers alone does not seem to help, though
the "screen ripping" looks different. Register 0xc0 seems to be related to the
screen scaler, at least bits 7,2 and 0 do have a function, though I do not know
which.

With a couple of vbemode calls issued from a remote console, I could play a
round of tuxracer. Given that this is a 1Ghz Pentium III, 10 fps are quite
acceptable and I believe that this means that this is accelerated video.

So basically, the dvo chip kernel module should "just" call the video bios to
set the screen to the proper dimensions, and should then the chip untouched -
this might properly work. Of course, you wouldn't have xrandr support with all
that as the resolution would stay fixed, but there would be hardware
acceleration.

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