the great sony digital lcd ordeal

[dep]

hey, chilluns . . .

having a week ago posted here my dismay over my failure to get the 
digital input of my gorgeous and nifty 23-inch wideview sony sdm-p232w 
lcd panel to work, i thought i'd drop a note describing how i 
ultimately made it into the incredible -- really! -- 1920x1200 fully 
digital display it is today (and has been for the last 24 hours). i 
figure that this will maybe help somebody sometime.

first let us consider the things that didn't work. these include, of 
course, sax2, which in my estimation should never be allowed anywhere 
near an XF86Config file. still, i tried -- sax2 must be good for 
optimally configuring some monitor somewhere, and i hoped i'd gotten 
lucky. i hadn't.

searching all over the web, i saw a post (from years ago) which said 
that the digital registers are not covered by the radeon driver (i have 
a radeon 9000 pro, a/k/a rv250), and that the ati drivers must be 
installed. i got them and installed them, breaking so many things that 
a complete inventory is still to be made. (i've uninstalled them, 
reinstalled the kernel, modules, and xfree, and still some things that 
used to work don't anymore.)

i found an XF86Config that somebody had cobbled together which he 
claimed worked at achieving 1920x1200 on a dell notebook that had a 
radeon chip in it. the remarkable thing about this file is that it was 
assembled with no sense of order that i could find and that it 
contained page after page of options the meaning of which were 
mysteries to me. it did not work as listed, and close to 20 hours of 
trying to tune it resulted in no improvement.

i then simply began to guess. the radeon manpage helped. as i added 
things that didn't work and restarted x, i read the xfree log the way a 
roman seer read chicken guts. usually there were no errors -- but there 
was no image theough the digital input (though i got text, just fine 
and very sharp, when x wasn't running).

then, last night, the xfree logfile divulged its secret. the edid 
information it recorded from the monitor contained numbers that looked 
strangely familiar:

(II) RADEON(0): Supported additional Video Mode:
(II) RADEON(0): clock: 154.0 MHz   Image Size:  495 x 309 mm
(II) RADEON(0): h_active: 1920  h_sync: 1984  h_sync_end 2016 
h_blank_end 2120 h_border: 0
(II) RADEON(0): v_active: 1200  v_sync: 1201  v_sync_end 1204 
v_blanking: 1250 v_border: 0
(II) RADEON(0): Supported additional Video Mode:
(II) RADEON(0): clock: 162.0 MHz   Image Size:  495 x 309 mm
(II) RADEON(0): h_active: 1600  h_sync: 1664  h_sync_end 1856 
h_blank_end 2160 h_border: 0
(II) RADEON(0): v_active: 1200  v_sync: 1201  v_sync_end 1204 
v_blanking: 1250 v_border: 0
(II) RADEON(0): Ranges: V min: 55  V max: 65 Hz, H min: 30  H max: 92 
kHz, PixClock max 170 MHz

so i strung them together thusly:

"1920x1200" 154 1920 1984 2016 2120 1200 1201 1204 1250

and i was *almost* there. it was doing something goofy, though, giving 
me a screen and a half. i went back to the logfile and found the 
listing for the maximum pixel clock, which was 170, not 154. so:

"1920x1200" 170 1920 1984 2016 2120 1200 1201 1204 1250

bingo! screen worked through the digital input, and the difference was 
the difference in wiping the vaseline off the lens. just incredible.

now, if i can get the frigging trackball to behave, and glx to work 
again, i'll be a happy man.
-- 
dep

The secularists have not wrecked divine things; but the secularists
have wrecked secular things, if that is any comfort to them.
				    -- G.K. Chesterton, "Orthodoxy"