sedzimir: which kernel branch would be the best to use with git X11?
sedzimir: I use drm-radeon-testing now but I see there's a lot of commits to drm-fixes lately
sedzimir: and new xorg-server doesn't seem to work
Nightwulf|work: hi all
TuomasT: Radeon power management stops working after suspend/resume (R670, radeon.dynpm=1). Can I do something to fix that?
TuomasT: I know this because after a boot it draws about 86W idle and after resume it draws 110W idle
TuomasT: Should I file a bug about dynpm=1 ceasing to work after resume (R670 HD3850)?
volter: Some weeks ago I stated here, I wanted to switch from HD2600 to a different model, because I had blank screens out of nowhere.
volter: Well, what shall I say? I did so and it helped.
volter: If the HD2600 is not defective, it might be something about the driver.
agd5f: Jonimus: I didn't pick any number. all you do is register with hwmon
GNUtoo|laptop: hi, I saw a lot of WIP in http://xorg.freedesktop.org/wiki/RadeonFeature for evergreen,is it testable yet? specially for 3d
GNUtoo|laptop: ok thanks a lot
verdelyi: is OGRE supposed to work with the open-source ati driver? I'm getting "assertion failed"s with every second sample program
verdelyi: ^^ on an RV670
verdelyi: radeon_texture.c:94: radeonFreeTexImageData: Assertion `!image->base.Data' failed. (seems to be a popular error, but no solutions afaik)
TuomasT: Should I file a bug report that power management gets disabled after suspend/resume with HD 3850 R670?
Wizzup: I understood more power management was added to 2.6.35, is there a way to enable this, or is it enabled by default? (It seems to no longer recognize dympn)
TuomasT: Wizzup: Yeah, I just booted 2.6.35-rc2 and PM was disabled
TuomasT: Worked with 2.6.34
Wizzup: I guess it may be a different param, but I really don't know where I would be able to find it
TuomasT: Wizzup: If you have it as a module then "modinfo radeon"
TuomasT: Otherwise, I guess looking at the source..
Wizzup: thx,will check
TuomasT: Wizzup: Do tell me if you find the parameter
TuomasT: MODULE_PARM_DESC(dynclks, "Disable/Enable dynamic clocks");
TuomasT: I think it is this one. 2.6.34 had both dynclks AND dynpm
Wizzup: I have that enabled already
agd5f: TuomasT, Wizzup: it's all done dynamically with sysfs now: http://lists.freedesktop.org/archives/dri-devel/2010-May/000492.html
TuomasT: agd5f: Sounds like this could solve my suspend/resume PM issue
TuomasT: will try it right away
Wizzup: Thanks. Won't try right away as in the transition from 34 to 35-rc2 I seem to have lost the ability to use KMS, and opengl version suddely reports 2.1 which means I think it is using cpu rendering (r600 card)
TuomasT: Nice idea, but doesn't work in 2.6.35-rc2
Wizzup: TuomasT: what card you have?
TuomasT: HD 3850
Wizzup: Is that r600/r700 ?
Wizzup: Last q: do you have kms?
TuomasT: Although 2.6.35-rc2 seems to have introduced couple of bugs to that
Wizzup: I see. I currently have no kms or dri, I think. Had it with 34. I'm using vanilla 35-rc2 though.
Wizzup: will investigate
TuomasT: dynpm makes quite a difference. From 108W idle to 85W idle (total system power consumption)
Wizzup: heh, it didn't load the module becaus of unknown param that I was passing...
TuomasT: Wizzup: Hmm. I am doing that also
TuomasT: Maybe that is what messes things up
verdelyi: dunno if it's related but I've just come across this: http://www.phoronix.com/scan.php?page=article&item=linux_2635_fail&num=1
verdelyi: after reading that, I'm no way installing that
Wizzup: It has been resolved already. This kind of things happen in development
Wizzup: Woohoo. It seems like vsync is working.
TuomasT: It looks like it picks up the change only after change of resolution. Now dynpm with 2.6.35 works
TuomasT: And it also works after suspend/resume :)
Wizzup: should enable his gentoo to suspend/resume
Wizzup: Is there a way to retreive the gpu temp? (I googled around but I couldn't find it)
agd5f: Wizzup: I posted a patch. It will be in 2.6.36
Wizzup: great :)
agd5f: Wizzup: http://people.freedesktop.org/~agd5f/0001-drm-radeon-kms-add-support-for-internal-thermal-sens.patch
Wizzup: Can I try to apply this to 35?
mancha: agd5f do older boards say 9100igp have thermal sensors?
agd5f: mancha: no
agd5f: mancha: pre-r6xx cards used external i2c thermal chips if they used one
agd5f: the code for that is already in place
mancha: ah ok, thanks!
TuomasT: agd5f: How does one read those temps then?
soreau: probably lmsensors
verdelyi: I got a backtrace for my problem I mentioned earlier: https://bugzilla.redhat.com/attachment.cgi?id=423024 can someone take an expert look at it (user/config error or bug?)
zhasha: "ParticleFX: radeon_texture.c:94: radeonFreeTexImageData: Assertion `!image->base.Data' failed.\n"
zhasha: verdelyi: it's a kernel bug
zhasha: err, I might have misread that
verdelyi: zhasha: I've read it's something in the mesa code
zhasha: yeah, I saw the __kernel* stuff and jumped to conclusiond
verdelyi: basically that's why I'm asking it here
zhasha: somewhere in the heap of code that is the driver, a radeon_texture_image is created with an invalid base.Data member, OR there's a double free the driver isn't handling correctly
zhasha: either way, I'm going to delegate this one onto a person more well versed in mesa/r600
verdelyi: if I wasn't on Fedora I'd say I test it with fglrx and see if it throws something as well
zhasha: fglrx doesn't support Xserver 1.8 and/or kernel 2.6.33 or what?
verdelyi: I think xorg 1.8 is the main issue
zhasha: gotta love AMD for their speed
zhasha: anyway, this is a driver bug. running fglrx wouldn't do any good
verdelyi: I'm expecting xorg 1.8 support in ~6 months from them
verdelyi: depends on whether opensuse will include it in their 11.3
zhasha: Xserver 1.8 support will be included when Fedora pulls Xserver 1.9 or greater
verdelyi: good point
verdelyi: anyway, let's hope some driver dev notices the issue
verdelyi: dunno if they're grepping through IRC logs though
verdelyi: can someone recommend a 3D engine that the current driver can handle perfectly?
verdelyi: (on RV670)
verdelyi: that doesn't mean there isn't any, does it?
zhasha: verdelyi: it means it's late in europe, close to the weekend and people generally don't seem to be on/know/care
verdelyi: I thought most people are American here
verdelyi: but you're right so I don't wanna jump to conclusions