Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2009-6-05

Search This Log:

adamk_: I'm trying to get dualhead working on an x1050 in rawhide.
adamk_: This xorg.conf file only ends up in the monitors cloning each other: http://pastebin.com/m7ca5d6e3
adamk_: Here's the log file: http://pastebin.com/m7d17a2a1
adamk_: Without an xorg.conf file, no combination of xrandr commands that I've tried seems to work, either. I only get cloned images. Using the --right-of or --left-of option either doesn't do anything, or simply switches which part of the combined screen is displayed on both monitors.
adamk_: I have modesetting enabled, and haven't tried it with kms disabled yet.
glisse: adamk_: do you have virtual in your xorg.conf ?
adamk_: Yep.
adamk_: #
adamk_: Virtual 2880 2048
glisse: adamk_: sounds like a bug
adamk: Yeah, as I suspected, if I boot with KMS disabled, it works fine.
airlied: adamk: you need to specify crtc on the xrandr command line
airlied: kms works fine for me dual-head on my rv370 if you are using my tree
adamk: Oh wait... I don't have the rawhide repo enabled on this machine.
nanonyme: Well, F11 will probably come out soonish in any case. :)
erjc: kernel: [drm] Initialized radeon 2.0.0 20080528 for 0000:01:00.0 on minor 0
erjc: kernel: [drm:drm_mode_getfb] *ERROR* invalid framebuffer id
glisse: erjc: it's ok
hifi: was about to say
erjc: libGL error: dlopen /usr/lib/dri/swrast_dri.so failed (/usr/lib/dri/swrast_dri.so: undefined symbol: _glapi_tls_Context)
hifi: erjc: have you installed radeon-rewrite
glisse: erjc: mesa badly installed
hifi: glisse: I can handle this my sensei!
erjc: like playing whack-a-mole
hifi: just try to do thing literally as the blog entry says
erjc: was better with tormod PPA packages
hifi: I tried to be nice to my system and not install anything to /usr with make install
erjc: Buffer too small for color buffer http://erjc.pastebin.ca/1448820
dileX: erjc: this is with (commit a3fc6259ac683c92c25d9ea17d8e34d15acd0984) "radeon: improved command checking for r3xx hw"?
erjc: updated this morning
erjc: git whatchanged
erjc: commit 4c034441a8f12e4bf60b8c5f58202d5ae00ff31d Author: Michal Krol Date: Fri Jun 5 15:00:15 2009 +0200
erjc: http://erjc.pastebin.ca/1448828 dmesg looks so pretty
mjt: what's needed to open/use drm device by more than X display?
mjt: now second attempt to open it fails
mjr: I believe ol' dri not supporting it at all, don't know about all this newfangled dri2 development stuff
mjt: hmm.
mjt: fglrx used its own drm or the in-kernel one?
mjt: (or something entirely different)
MostAwesomeDude: fglrx uses its own kernel module.
mjt: i know about the module. radeon.ko is also "its own" module
MostAwesomeDude: No, I mean, radeon is built as part of DRM. fglrx is completely separate.
mjt: ok
mjt: interesting. so intel drivers also disallows multiple X sessions?
mjt: i think i used that stuff even with simpler drivers
MostAwesomeDude: If I remember correctly, the multiple master code (for multiple X) was broken for a while, but is currently unbroken. I also think it depends on whether you're using DRI1 or DRI2.
mjt: how to check?
MostAwesomeDude: Xorg.0.log would be the place. I don't know the specifics offhand, sorry.
mjt: the kernel stuff is from r6xx-r7xx-support branch
MostAwesomeDude: Ah. In that case, DRI1 only.
mjt: interesting... ;)
bridgman: I thought airlied said that drm support for multiple masters (ie multiple X driver instances) was only added relatively recently
mjt: but where it's added to?
bridgman: one of the branches where KMS/MM work was being done; probably added a branch or two ago but propagated through to the latest
bridgman: this is all guess & vague recollection though
mjt: heh ok
MostAwesomeDude: If that's the case, it almost certainly isn't in the r6xx DRM branch.
mjt: that branch last updated on Apr-6
bridgman: absolutely; I look at all the commits going in there and I'm sure even I would have noticed that ;)
spstarr: hello bridgman
bridgman: hi spstarr
spstarr: is going to be around Leslie @ 7 3pm
spstarr: recruiter interview
bridgman: good to hear
bridgman: is playing hookey today and working from home
bridgman: where I have big screens and fast PCs ;)
dileX: damn. read the MLs before bisecting .
dileX: agd5f: ^ was the problem for my yesterdays asking. so Revert "drm: don't associate _DRM_DRIVER maps with a master" solves it.
agd5f: dileX: yeah, I saw that
dileX: agd5f: me too late, 1st I thought it was glisse's d-n-r kernel (started there to discover the problem). next time I will strive the MLs first.
_Groo_: hi/2 all..
mcgreg: hi
dileX: hi mcgreg & _Groo_
_Groo_: hi dileX
_Groo_: anyone knows if the latest code drop fixes xv with dri2/kms? compiling it now
koolfy: wow
koolfy: I just tried out some PSX emulators for linux
mcgreg: is there one good?
koolfy: more than one
koolfy: works great with x1650 on ubuntu with OS driver
koolfy: xorg hangs when the emulator is launched with Xpress200M
koolfy: (same driver, -9999 up to date)
mcgreg: koolfy: ok, what one?
koolfy: same behaviour with epsxe, pcsx ans pSX
koolfy: and*
koolfy: (those are really good emulators)
koolfy: but it looks like an ugly bug in the radeon driver
koolfy: that's weird, never had that hang with any game
koolfy: and why do two different PSX emulators share the same problem ??
agd5f: _Groo_: http://www.botchco.com/alex/xorg/fix_xv.diff
zhasha: agd5f: why isn't this in the various branches? :P
agd5f: zhasha: it'll all work once the kms bits hit master
zhasha: when will that happen?
nanonyme: wonders if the answer is Soon (tm)
zhasha: Soon™ is the property of ©Microsoft™ Corporation® and you will be sued for using it. See you in court!
nanonyme: zhasha: I kinda wish.
nanonyme: It's an overly abused word. :)
nanonyme: Btw, am I the only one who's starting to get a little tired to those continuous benchmarks on Phoronix? :)
zhasha: no, the kernel benchmarks? ridiculous
nanonyme: Benchmarks on pretty much everything imaginable. ^^
zhasha: I'm guessing it means there's very little to report on
nanonyme: Probably...
koolfy: Soon : a benchmark of the average benchmarks per month
koolfy: recursive benchmark :)
koolfy: Everything is funnier with recursivity
koolfy: except sex.
zhasha: yo dawg I heard you like cars so we put a car in your car so you can drive while you drive!
nanonyme: koolfy: But iteration has less overhead!
nanonyme: (Unless we're using tail recursion)
koolfy: haha
nanonyme: (And I'm too tired)
zhasha: bring me a pepsi!
koolfy: ok those emulators work fine with fglrx
nanonyme: One of those "kids, you learn this when you grow up" things, btw. That is, when you try to write your recursive code into iterative and when it's not worth it. Not sure if I still have grown up properly. :)
koolfy: you can't even imagine how fucked up id fglrx on 2.6.29…
koolfy: is*
nanonyme: Erm, when you should try to write even.
koolfy: ok I found an app that hang only with radeon driver, only with Xpress200m chipsets, and those two emulators are the only ones that reproduce the bug
nanonyme: Hmm, emulators?
koolfy: how in hell can I find out what they use that is so special ?
koolfy: epsxe ans pSX-bin
nanonyme: I *seriously* hope it's opensource.
koolfy: haha
koolfy: the only one opensource doesn't even compile on my amd64
nanonyme: Heh. :)
nanonyme: Who forces you to use amd64? ;)
nanonyme: is still kinda sad we had to leave 16bit programs behind with amd64 :(
koolfy: nonono, I won't recompile my whole system :D
nanonyme: :D
nanonyme: debootstrap a Debian on it? ;)
koolfy: uarkhh
koolfy: *pukes*
koolfy: nobody has a similar chipset ?
koolfy: see if you can reroduce it :(
nanonyme: *sigh* Seriously. It seems evil OS people are telling me to either upgrade my CPU or use 32bit OS's if I want to keep running all the programs I want. :/
nanonyme: (Not that the new CPU wouldn't solve most of my problems but... they cost money)
koolfy: oh, my bad, that fglrx patch for .29 is not that bad
koolfy: but I would like to find what is responsible of the hang on the opensource driver…
koolfy: C'mon I won't be using fglrx my whole life just to play some FF7 :(
otaylor: koolfy: either debug or report a bug, it's that simple
koolfy: of course, but I'd like to give some more information on the bug report
koolfy: not just «it doesn't works.»
otaylor: well, provide details about your hardware, your software, and how to reproduce the problem, and that's a pretty good start at a good bug report
koolfy: as it only affects my old xpress200m with those little-used PSX emulators, little people will be able to reproduce it
koolfy: that's why I try to give some help :p
koolfy: most people could go on without any fix and never notice it
nanonyme: thinks nearly every time someone mentions a problem, they also mention xpress200m
osiris__: koolfy: what mesa version do you have?
otaylor: koolfy: if someone is responding to a bug report, they'll be able to give you instructions for anything else they consider useful.
koolfy: nanonyme: can't change the laptop's integrated GPU :p
koolfy: osiris__: I tried with 7.4.2 and 7.5_rc2
osiris__: koolfy: try radeon-rewrite branch
koolfy: on my gentoo (x1650 pro, same driver, mesa 7.4.2 it works)
koolfy: ubuntu*
koolfy: is it a branch of mesa ?
osiris__: yes
koolfy: if the same mesa works with another chipset, isn't it more likely a driver problem ?
koolfy: (just asking
nanonyme: Time to poke at a quirk table?
osiris__: koolfy: x200 is non tcl chip, and the rendering paths in r300 driver for swtcl and hwtcl are pretty different
koolfy: ok' thanks :)
dwery: 'morning. I would like to know if OpenGL 2.0 support for the R300 is still under development
MostAwesomeDude: Yep.
dwery: nice. anything that I can test?
MostAwesomeDude: No.
MostAwesomeDude: Well, kind of.
MostAwesomeDude: You can test radeon-rewrite, but it doesn't actually have GLSL support.
dwery: I guess I made a bad decision when I bought a laptop with an r300 ;)
MostAwesomeDude: I wouldn't say that.
MostAwesomeDude: What part of OGL 2.x do you need? FBOs work fine.
dwery: I was trying th Google O3D plugin for Firefox and it seems it doesn't work on 1.4 but I still don't know which features are indeed required
onox: MostAwesomeDude: does normal radeon driver have support for GLSL?
onox: I tried in nexuiz but it made the game extremely slow
MostAwesomeDude: onox: No.
_Groo_: hi/2 all
_Groo_: any dev awake? agd5f, glisse, airlied, anyone?
MostAwesomeDude: I'm up.
_Groo_: hi MostAwesomeDude
MostAwesomeDude: You probably don't want me, though. :3
_Groo_: MostAwesomeDude: could it take a look at this?
_Groo_: rats.. now that i need a error, it doesnt give me, just a sec
_Groo_: MostAwesomeDude: well anyway, with latest code, xv is broken again, could you pleeeease fix the clipping changes? now it stretches my videos instead of cuting them
dwery: MostAwesomeDude: I guess Google will fi it some day. thanks anyway
_Groo_: MostAwesomeDude: are you there?
_Groo_: wonders where everyone went
dwery: MostAwesomeDude: I'm verifying the code.. it seems it requires ARB_vertex_buffer_object and GLEW_EXT_framebuffer_object. My current driver has the first and guess the second one is th FBO you were talking about. So I guess I will try radeon-rewrite tomorrow
_Groo_: dwery: radeon-rewrite doesnt havent the second one
dwery: _Groo_: isn't FBO?
_Groo_: dwery: it only has GL_EXT_framebuffer_object
dwery: ouch
_Groo_: dwery: just checked.. using latest dri2/kms, drm, xf86 from glisses branche
dwery: _Groo_: thanks!
_Groo_: dwery: for what you need it anyway?
dwery: _Groo_: Google's O3D plugin
_Groo_: dwery: ah.
dwery: _Groo_: I'll try removing the test and see if it works with simple objects
_Groo_: dwery: good luck :)
dwery: The other option would be the proprietary driver.. but I don't like it and probably the package has not been compiled for Xorg 1.5
_Groo_: agd5f: ping
_Groo_: glisse: ping
_Groo_: airlied: ping
_Groo_: MostAwesomeDude: ping
rhodan: is it just me or did there really happen almost nothing in the last few weeks?
dwery: _Groo_: it crashed badly :) at least, I've tried!
rhodan: where does the gallium3d development happen? any gitwebs i can bookmark to keep up with development'
_Groo_: dwery: figured :) if it needs fbo, it needs fbo.. but its in the makings i believe
dwery: time to go to bed... thanks for your help!
_Groo_: dwery: thank you for your effort :) i didnt do anything
rhodan: yould somebody please give me a short roundup on the state of gallium3d and dri2 on r500?
rhodan: *could
rhodan: glisse and airlied didn't update theis blogs in a while
nbz: rhodan: dri2 works in fedora.
rhodan: would be cool to have opengl2.0 already
nbz: rhodan: airlied has been on holidays for a few weeks, so some less development. glisse got distracted in the effort to get radeon-rewrite upto scratch
rhodan: ok
nbz: rhodan: someone would need to write it. Osiris was working on adding support for that in the classic mesa driver, but I think he mentionedabandoning that effort.
nbz: MoseAwesomeDude has been working on galium, but I don't think its ready for anyone but developers yet.
rhodan: hm kay
nbz: currently glisse and airlied are *probably* working on getting the radeon drm ready to merge the KMS into the linux kernel within the next couple of weeks
rhodan: but opengl 2.1 is in gallium, so when gallium support in xf86-video-ati ist ready, it will automatically have gl 2.0, right?
rhodan: *is ready
nbz: rhodan: I think so, but I am not sure if the driver MAD is wroking on is just for r300 or r3xx tor5xx
rhodan: ok, i'll try dri2
rhodan: should i get the rawhide or the rawhide-testing branch of the kernel?
nbz: no idea. I would presume rawhide (rawhide-testing? do you mean linus-testing?)
nbz: oh, seen that branch now - seems ooold.
DanaG: heh, seems like the git "tree" needs some gardening care. i.e. cleaning / pruning.
rhodan: would that make the download smaller?
DanaG: I don't know how git works, actually... but it would make it more sane to navigate.
DanaG: For example, what's the current state of R600, overall? There are so many branches, it's hard to tell.
rhodan: btw. how do i get the vt-scrollback-buffer to work properly? it only saves around 2 1/2 pages for me
nbz: DanaG: I think there is only one branch that matters for that. for us people following on at home it would be useful if the developers appended each commit with "Doesn't work yet". til it does. :)
spstarr: _Groo_: Friday/Sat are usually quiet for development
rhodan: "fuck that gallium, tgif!"
nbz: DanaG: but I assume that when that branch is ready, they will blog it/phoronix mention it.
_Groo_: spstarr: unfortunatelly i can only help on weekends, altough i compile during the week
DanaG: which branch is it, by the way?
_Groo_: spstarr: acutally i just need xv to work..
nbz: r6xx-rewrite
spstarr: _Groo_: which gpu
_Groo_: spstarr: rs485
_Groo_: spstarr: aka the evil 200m one
spstarr: so the r3xx code
_Groo_: spstarr: yep
_Groo_: spstarr: its stretching my videos.. i tried looking at the code but to no avail, i dont have enough info
_Groo_: spstarr: it was working with two patches glisse sent me, but the code changed and i cant apply them anymore..and i cant use the old code anymore because with latest drm/kernel/mesa it gives me strange rendering bugs.. so im stuck
_Groo_: spstarr: i can send you the diffs if you want
spstarr: im on a r6xx :)
_Groo_: spstarr: ah, so no 3d for you
spstarr: well, 2D is using the 3D engine but no GL
_Groo_: spstarr: i hope soon :)
spstarr: will stay in his Intel GMA mode until there's some work merged into fedora
spstarr: then I can start using the r6xx
spstarr: nice having that ability ;p
_Groo_: spstarr: ¬¬
DanaG: what will be the holy grail: compiz on top of KMS.
DanaG: yeah, I know there's a bunch of stuff in the middle I'm ignoring. =þ
_Groo_: DanaG: using it already
DanaG: on R600, I mean.
_Groo_: DanaG: ah :D
_Groo_: DanaG: use fglrx.. it has dri2
rhodan: 2D is frickin' fast no
rhodan: *now with dri2
_Groo_: hi/2 all
_Groo_: great news
_Groo_: anyone awake to listen to them?
DanaG: dude, if you want to say something, just say it. =þ
_Groo_: i just backported to jaunty the latest and greatest xorg from master... along dri2/kms that i already did backported, im happy to report that gtk doesnt have rgba problems anymore :)
amarsh04: wonders what is needed to get the best results out of: Display controller: ATI Technologies Inc RV280 [Radeon 9200 SE] (Secondary) (rev 01)
_Groo_: amarsh04: dri2/kms
_Groo_: so now i just need a new patch for the xv stretch bug and ill be happy as a clam
_Groo_: wonders if all clams are happy
amarsh04: has disk space, will compile from source if needed (-:
_Groo_: amarsh04: follow glisses blog post.. should be easy at that point? which distro? or just use fedora 11
amarsh04: I'm on Debian unstable... do you have a URL handy?
_Groo_: sure just a sec
_Groo_: amarsh04: http://jglisse.livejournal.com/1822.html
amarsh04: thanks _Groo_
_Groo_: amarsh04: good luck :)
_Groo_: amarsh04: im playing world of goo with dri2, using compiz to rotate the game just for fun, and everyhting is smooth as silk
amarsh04: I might restart X without an xorg.conf to see what settings are really necessary
dmb: so... any idea on when programs/x will be able to change the resolution for full screen stuff?