I thought I would put these matters in a new thread. I largely just need a directive where to report them, if they are solid enough to report. I am running on standard slackware-13.1 w/robbie workman's config for 2.6.35.4, and this patch for the vram table which restores agp compatibility
1. First boot into X in the morning goes wrong. Yesterday it was a kernel oops. Today, I was kicked into some test cycle on the monitor, which was showing as switched off but was cycling White, Red, Green, Blue, Off, every 5 seconds or so. I tried Ctrl_Alt_Backspace, and tried switching the monitor on (or mebbe off again & on); I had the console there, so I hit up arrow and return, and X loaded :-/. Every kernel oops switches off the monitor, but the monitor is a pretty robust thing, doing 62khz at least. Xorg.0.log for that session finishes as follows
...
(II) RADEON(0): Modeline "1440x900"x0.0 136.75 1440 1536 1688 1936 900 903 909 942 -hsync +vsync (70.6 kHz)
(II) RADEON(0): Modeline "1280x960"x0.0 108.00 1280 1376 1488 1800 960 961 964 1000 +hsync +vsync (60.0 kHz)
(II) RADEON(0): Modeline "1152x921"x75.0 111.98 1152 1224 1352 1552 921 922 925 962 -hsync +vsync (72.2 kHz)
(II) USB Keyboard: Close
(II) UnloadModule: "evdev"
(II) USB Keyboard: Close
(II) UnloadModule: "evdev"
(II) HID 1241:1166: Close
(II) UnloadModule: "evdev"
(II) Power Button: Close
(II) UnloadModule: "evdev"
(II) Power Button: Close
(II) UnloadModule: "evdev" EOF
2. It runs the 'atlantis' screensaver now with no issues. I got a kernel oops from Ctrl_alt_backspace, although this only happens after a few idle hours.
I also got weirdness by jumping to console terminals and back. Last night & this morning, jumping to a console loses me the Alt_F7 vt. I started this session with X on Alt_F7. I jumped to a console, read a file, and jumped back. Now Alt_F7 is blank, and X is on Alt_F8 :-/. I can't trick it to move on to vt9 though.
3. Starting X without the radeon module loaded is also likely to produce a kernel oops. From my ssh terminal I gather this crash was in the radeon module. I have circumvented that with a 'modprobe radeon' in rc.local.
I have participated in solving something elusive like this before ('overcurrent change' log spam from ehci_hcd up to 2.6.19). What was done there was that a patch was made which logged values of certain registers to syslog. I patched the kernel and ran tests on it, and the problem became clear (My Via hardware violated the specs). Via had 'solved' this quietly by internally disabling 2 usb ports, but of course, I had them.
If you guys are interested, you may want to do something similar, but don't tell me to fire up gdb!
1. First boot into X in the morning goes wrong. Yesterday it was a kernel oops. Today, I was kicked into some test cycle on the monitor, which was showing as switched off but was cycling White, Red, Green, Blue, Off, every 5 seconds or so. I tried Ctrl_Alt_Backspace, and tried switching the monitor on (or mebbe off again & on); I had the console there, so I hit up arrow and return, and X loaded :-/. Every kernel oops switches off the monitor, but the monitor is a pretty robust thing, doing 62khz at least. Xorg.0.log for that session finishes as follows
...
(II) RADEON(0): Modeline "1440x900"x0.0 136.75 1440 1536 1688 1936 900 903 909 942 -hsync +vsync (70.6 kHz)
(II) RADEON(0): Modeline "1280x960"x0.0 108.00 1280 1376 1488 1800 960 961 964 1000 +hsync +vsync (60.0 kHz)
(II) RADEON(0): Modeline "1152x921"x75.0 111.98 1152 1224 1352 1552 921 922 925 962 -hsync +vsync (72.2 kHz)
(II) USB Keyboard: Close
(II) UnloadModule: "evdev"
(II) USB Keyboard: Close
(II) UnloadModule: "evdev"
(II) HID 1241:1166: Close
(II) UnloadModule: "evdev"
(II) Power Button: Close
(II) UnloadModule: "evdev"
(II) Power Button: Close
(II) UnloadModule: "evdev" EOF
2. It runs the 'atlantis' screensaver now with no issues. I got a kernel oops from Ctrl_alt_backspace, although this only happens after a few idle hours.
I also got weirdness by jumping to console terminals and back. Last night & this morning, jumping to a console loses me the Alt_F7 vt. I started this session with X on Alt_F7. I jumped to a console, read a file, and jumped back. Now Alt_F7 is blank, and X is on Alt_F8 :-/. I can't trick it to move on to vt9 though.
3. Starting X without the radeon module loaded is also likely to produce a kernel oops. From my ssh terminal I gather this crash was in the radeon module. I have circumvented that with a 'modprobe radeon' in rc.local.
I have participated in solving something elusive like this before ('overcurrent change' log spam from ehci_hcd up to 2.6.19). What was done there was that a patch was made which logged values of certain registers to syslog. I patched the kernel and ran tests on it, and the problem became clear (My Via hardware violated the specs). Via had 'solved' this quietly by internally disabling 2 usb ports, but of course, I had them.
If you guys are interested, you may want to do something similar, but don't tell me to fire up gdb!