b0le: I was wondering what would cause a "RADEONDRIGetVersion failed to get the DRM version" error in Xorg.0.log (after this it disables DRI) - this is after recompiling DRM modules (I have also recompiled libdrm, and X proto/libs/drivers since, which didn't help)
MrCooper: do you have the full log file and dmesg output somewhere to look at?
b0le: MrCooper: just a minute, and I will
b0le: MrCooper: http://pastebin.org/20546 http://pastebin.org/20547 http://pastebin.org/20548 (xorg.0.log, kern.log, dmesg)
MrCooper: odd, not sure what the problem could be - which version of the DRM kernel modules were you using previously?
b0le: master (and master)
b0le: I didn't even git-pull (I modified drm_drv.c change the err_1 from displaying retcode to make it display a new variable failcode (but the function still returned retcode), which I changed in various places in the function (I thought it was failing in drm_unlocked_ioctl - but most likely I was misunderstanding the code) - as I was trying to figure out why SWCursor is breaking aiglx
MrCooper: and if you build the previous snapshot, it still works?
b0le: MrCooper: (specifically causing a kernel panic)
b0le: MrCooper: I will try git-reset --hard and then recompile DRM and see if that changes anything
b0le: MrCooper: that fixed it... I wonder why my simple changes to drm_drv.c would cause it to fail to get drm version?
MrCooper: well, obviously something's wrong with your changes :)
b0le: evidently... now just to find the cause of the kernel panic
MrCooper: I wonder if it's actually trying to use silken mouse... have you tried disabling it explicitly?
b0le: MrCooper: silken mouse = SWCursor? (if so, I need it to use swcursor, as I am trying to get it to work with mpx (but atm I only have master))
MrCooper: no, if I mean SWCursor I say that...
MrCooper: silken mouse is a feature for smooth (SIGIO driven) cursor motion that can only work with hardware cursors
MrCooper: but it looks like it may be trying to use it regardless
b0le: hmm, it says Silken mouse enabled. Using software cursor
b0le: MrCooper: how would disable it explicity?
MrCooper: using the command line option -nosilk or (IIRC) Option "SilkenMouse" "off" somewhere
b0le: MrCooper: option "SilkenMouse" "off" worked, but it still kernel panics (when moving mouse over glxgears or starting compiz (both with LIBGL_ALWAYS_INDIRECT=1))
MrCooper: oh well, there goes that theory
airlied: b0le: whats the kernel panic?
b0le: MrCooper: here is the kernel log: http://pastebin.org/20554 (how relevent is the last entry, to the cause of kernel panic?, or is it not possible to tell)
b0le: airlied: flashing leds, no output (if I understood you correctly...)
airlied: ah so you can't see it.
b0le: hmm the kernel log didn't seem to copy everything (I'll post whole lot on googlepages, just a minute)
b0le: airlied: unforutanely
airlied: b0le: do you have an xorg.log from a working startup
airlied: I think there might be a problem with the atombios memory buffer
airlied: and the PCIE gart table. but it may be nothing..
b0le: joel.bosveld.googlepages.com/kernNew.log (except that is alot of logs kernel logs mashed into 1, I'll find line number in a min)
b0le: airlied: one sec and I'll paste that (do you want the xorg.log of the settings that causes the kp when I run compiz?)
airlied: yes..
airlied: did it work with ati 6.7.197 driver?
airlied: and break with 6.8.0?/
b0le: this is with all from master
airlied: try tsting out 6.7.197 maybe to see if the same problem happens
b0le: http://joel.bosveld.googlepages.com/Xorg.0.log
airlied: I'm too tired to do much now, but I'll look tomorrow if you leave the log and check if 6.7.197 works.
b0le: airlied: thanks
airlied: hmm maybe it isn't that from reading the log.. but check with 6.7.197 just in case..
airlied: if not netconsole or a serial console are your best bet..
b0le: "Feb 21 20:27:53 JOEL-GENTOO kernel: [drm:radeon_cp_dispatch_indirect] buf=16 s=0x0 e=0x9c" from kernNew.log is the last line before one of the panics
airlied: thats just the X server sending some rendering to the kernel
b0le: airlied: also kernel panic with 6.7.197
airlied: oh bad.. then netconsosle or serial console might be the only thing..
moogle: good morning everyone, I downloaded the tarball containing the 6.8 driver, just want to know ho to compile...
moogle: first i launched configure...
moogle: and then? anyone knows?
airlied: ./configure --prefix=/usr ; make ; make install
b0le: airlied: ok I will try setup a netconsole (just need to figure out how...)
moogle: oh... thank you very much
moogle: i'll try immediately
moogle: after the make install i just need to change the line in xorg from fglrx to radeon right?
airlied: moogle: yes.
moogle: just another noob question... :) the version i'm installing will be recognized by my package manager?
moogle: or it will cause conflicts on future updates?
airlied: moogle: nope.. it'll get nuked by OS updates.
airlied: moogle: sosme distros will package 6.8 soon.\
moogle: uh, ok... thank you very much for your help!
b0le: will the talks at fosdem (from x dev room) be available online for download?
Temujin_: hey airlied, i'm trying to help someone with ubuntu. They have a FireGL V5250 / R520 chip and want to run compiz and multi-monitor. They're currently using fglrx 8.37 w/xgl and it's not working for them. Will your driver work for them?
ajax: none of the open drivers support 3d on r500 yet.
ajax: which means compiz won't work.
Temujin_: gotcha, thanks
maks_: why insist on compiz Temujin_ :)
Temujin_: maks_: idk, i don;t use the damn eye candy myself, but the people insist..
PSYCHO___: Temujin_: and friend have compiz 0.7
PSYCHO___: that is first that suppot more than 1 screen
Temujin_: oh yeah, i did read that, thanks for the pointer
OipOS: What's the current status for GLSL support in the r300 branch?
glisse: OipOS: i think there is no status and no work on it
OipOS: Drat.
damentz: airlied: what does the r100 patch fix?
damentz: like, what are exa transforms used usually for?
airlied: damentz: rotation
damentz: ok, then i wont use git again i suppose
damentz: not worth the time it takes to compile
OipOS: It doesn't take THAT long...
damentz: OipOS: well, i have a 1.2ghz pentium 4-m
damentz: trust me, its pretty slow
damentz: its slower than my 1ghz athlon thunderbird desktop
OipOS: A kernel build takes longer... He he he.
PSYCHO___: Thu Feb 21 20:05:17 2008 >>> x11-drivers/xf86-video-ati-9999
PSYCHO___: merge time: 1 minute and 40 seconds.
PSYCHO___: it does not take e that long
PSYCHO___: :)
damentz: well its also building the new alsa 1.0.16 drivers
damentz: its been about 10 minutes
damentz: still goin
b0le: airlied: I have got a netconsole log - http://joel.bosveld.googlepages.com/netconsole.log (btw, this is for the kernel panic with SWCursor and indirect rendering (moving mouse over glxgears or starting compiz)) - I also ran "mcelog --ascii < netconsole.log > mcelog.log" (which replaced some of the errors down the bottom with more details) at http://joel.bosveld.googlepages.com/mcelog.log
b0le: It says it isn't a softare problem, but a hardware one. Yet it works fine on my other partitions?
airlied: b0le: oh an mce...
b0le: airlied: does this mean that it is hardware and not software problem? (and is it possible to find what is playing up?)
Magnade: mce means there is a hardware fault and it was corrected
Magnade: it could mean bad things depending on the errors it spews
b0le: it kernel panics
b0le: I'm afraid that it means that my harddrive is dying... (I had problems with kernel panics before, and it was because harddrive was dying) though it seems odd that I only get it on gentoo, my other partitions (ubuntu and windows) seem to work fine
cjb: sounds more like a kernel bug.
airlied: b0le: it sounds like an interaction between the driver and your gpu causes bad things..
airlied: we've had issues with mice interactions before.. but I can't t hink why swcursor would cause such problem.
b0le: I may try reinstall (again :D)
b0le: does the kernel version make any difference (and is there a recommended version - does it make any difference that I am using everything from master?)
airlied: nope I can't think of anything that would've broken it.. most likly just not a very well tested path..
Magnade: b0le: it sounds like other issues to me run memtest try setting up sensors to see if your voltage from powersupply is good etc
b0le: Magnade: how long should I run memtest for? (I did run it for 3 passes with no errors). Also, if it is a hardware proplem, wouldn't you expect it to be apparent on ubuntu and windows?
Magnade: b0le: 3 passes is prob good id look into voltages then
Magnade: if you think its the hard drive btw check out smartmon tools
Magnade: windows is so flakey its hard to tell hardware issues from normal running
Magnade: its possable ubuntu isnt pushing it as hard as you are now