I've come across a weird problem in the Radeon drivers in 4.2 -
depending upon the horizontal scan rates I specify for my monitor, I can
cause a divide by zero within the Radeon init code.
Here's the relevant sections of my XF86Config file:
Section "Monitor"
Identifier "PX-982"
HorizSync 30-80
#HorizSync 30-85
VertRefresh 50-100
Modeline "1600x1200" 175.50 1600 1664 1856 2120 1200 1201 1203
1230
Modeline "1600x1200@60" 156.20 1600 1664 1856 2120 1200 1201
1203 1230
Modeline "1600x1200@72" 187.75 1600 1664 1856 2120 1200 1201
1203 1230
Option "DPMS"
DisplaySize 362 273
EndSection
Section "Screen"
Identifier "Screen 1"
#Device "3dfx"
Device "radeon"
Monitor "PX-982"
DefaultDepth 24
Subsection "Display"
Depth 24
Modes "1600x1200" "1600x1200@60" "1600x1200@72" "1280x1024" "1024x768"
"800x600" "640x480"
ViewPort 0 0
EndSubsection
Now, if I change the horizontal scan rate of the monitor from 30-80 to
30-85, then X does a divide by zero at startup. I've traced it down to
file radeon_driver.c, routine RADEONInitDDARegisters, save->post_div is
zero. Now, WHY save->post_div is zero is beyond the few minutes I've put
into it.
My questions are:
1) How can I get symbols loaded on the drivers, so that I can debug this
in a civilized fashion? Do I need to build the Radeon driver into X
(i.e. is the module load process stripping the symbol data)?
2) Does anybody have any pointers on this? For example, what routine
fills in the save structure, since clearly that is where the barf is
occurring?
3) Am I the only one who's ever seen this?
Additional information:
System is a PIII-800 running RH7.2 with 2.4.18 with pre-empt and XFS.
Card is a Radeon 7500
I've seen this both with the binaries on XFree and with my own builds
from today's CVS from DRI.
-------------------------------------------------------
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
