Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2009-6-06

Search This Log:

zhasha: dmb: what do you mean?
DanaG: http://www.gossamer-threads.com/lists/linux/kernel/1084970 -- heh, interesting things going on in wireless stuff, in addition to the graphics stuff.
dmb: zhasha, well, usually programs (like games) are able to switch the video mode
dmb: so full screen will work
dmb: at different resolutions
dmb: that doesn't seem to work with modesetting on
zhasha: ah, well I don't know about KMS
DanaG: fbset also doesn't work with it.
DanaG: Bummer.
DanaG: I forgot... who was it that had AMT on the Thinkpad? Looks like compiling compat-wireless from June 2 (the June 5 is broken) will give the AMT fix!
DanaG: ... I think.
DanaG: ummm.... ah, using new driver with old firmware, causes lockups.
DanaG: Speaking of non-ATI firmware.... how about a question of ATI firmware? =þ
DanaG: How often does ATI update stuff like that? I've never heard of it being announced -- it's usually just silent.
DanaG: s/if you use the old firmware/likes to lock up/
DanaG: whateer.
DanaG: ah, spstarr was the one with the thinkpad.
clotho67: Is rv610 supported by ~glisse/drm-next drm-next-radeon branch?
agd5f: _Groo_: you need to stick around if you want answers
agd5f: http://www.botchco.com/alex/xorg/fix_xv.diff
hifi: can you add options to the display driver without writing the whole xorg.conf
hifi: or generating
nanonyme: Sure.
nanonyme: You only need the section for the display driver then.
nanonyme: I've done that plenty of times.
nanonyme: I usually only have that and a files section with a ModulePath.
dileX: good. Revert "drm: don't associate _DRM_DRIVER maps with a master" is now in Linus-tree.
nanonyme: Commit message pretty much says nothing to me. :)
dileX: if you use glisse's drm-next-radeon with Linus-tree on top, it will say a lot (no starting Xorg) :-)
dileX: to be more precise: Linus-tree and radeon is b0rked with 2.6.30-rc8-git{1,2}. fix is in rc8-git3.
osiris_: if anyone is running radeon-rewrite branch of mesa, please test newest version of my patches in "More radeon-rewrite patches" thread and report.
amarsh04: osiris, what's the mailing list again?
osiris_: mesa3d-dev
amarsh04: ah ok
tormod: osiris_: can you please push them to a git repo?
osiris_: I don't have one
osiris_: guess it's high time to request for account on fdo
tormod: absolutely :)
tormod: you don't have a web account somewhere? git works over http as well
osiris_: unfortunately no
tormod: osiris_: www.freeshell.org
dileX: osiris_: are you speaking of ?
osiris_: dileX: yes
dileX: osiris_: reading this thread, one request (seen this on other MLs): can you add a "versioning" (ChangeLog) within your patches and in the subject-field? I was irritated while folowing this thread. an excellent example is . this will surely attract more testers.
dileX: osiris_: ...or as tormod mentioned a git-web would be fine.
osiris_: dileX: I've just requested for an account on fdo, so next time I will push the patches to my git tree
osiris_: also if you are testing those patches, you could test the one in "Few shaders related patches against radeon-rewrite" thread - these actually can fix some games
elena09: My card is ATI Mobility Radeon HD 3450, is it ok for your forum?
elena09: hello?
elena09: My card is ATI Mobility Radeon HD 3450, is it ok for your forum?
chithead: if you have a question, just ask
elena09: http://dl.getdropbox.com/u/1052075/Ubuntu%209.doc
elena09: There is my qouestion
elena09: help with dark screen
elena09: http://dl.getdropbox.com/u/1052075/Ubuntu%209.doc
chithead: how about summarizing your issue in a few words, what your configuration is and what you have tried, and a pastebin of the logs and all relevant information. instead of a .doc file on an obscure file hosting site
elena09: ok
elena09: a second please
elena09: http://paste.ubuntu.com/189595/ here it is, that is not obscure
elena09: I wrote it once with all details and I don't want to write it anymore
elena09: system log viewer doesn t appear. instead it it envyng, so i can't find logs chithead
elena09: and dropbox is not obscure, the bugsquad helped me to get access there, it is very useful
legend2440: where are the "radeon" drivers loaded in jaunty? i look in /etc/modules and only lp is listed there and no mention of it in xorg.conf
MrCooper: legend2440: the X driver can load it on demand
legend2440: MrCooper: so how would i make it so it loads ati instead of radeon
chithead: /etc/modules is only for kernel modules. there exists no ati kernel module
legend2440: chithead: when i type lsmod it lists "radeon" any way to get it to use "ati" instead?
chithead: there exists no ati kernel module
legend2440: chithead: ok thanks. the reason i asked is i upgraded to jaunty with xorg-xserver 1.6 and tv out stopped working. i get a signal to the tv but its either scrambled or flipping like vertical hold problem. in intrepid with xorg-xseerver 1.5 it worked fine with the fglrx drivers
elena09: can't start X with the default drivers in ubuntu 9.04, and am using the mobility hd3450
bigtrev: http://lists.opensuse.org/radeonhd/2009-04/msg00217.html heres the same laptop
elena09: I know but there is not any solution bigtrev
elena09: funny, ati peope sent me here
kernelpanic: Hello everyone! I'm wondering whether I can get *smooth* fullscreen video using adobe's flash-player. I'm using version 10 on my gentoo amd64 system with a week-old radeon-git driver and a rv530. When switching to full-screen (2560x1600), the video is really jerky. Anything I can do?
kernelpanic: I would lower the desktop resolution, but since switching from fglrx (thank god), using the randr-tray-icon to go to e.g. 800x600 results in a black screen
nanonyme: kernelpanic: Got OpenGL with hardware rendering on your system?
nanonyme: The only acceleration method proprietary Flash provides is via OpenGL.
chithead: the adobe flash player is not the fastest, 2560x1600 is probably too much for it. try mplayer `youtube-dl -g $URL` or similar
nanonyme: Yeah, that's a point. If you use mplayer, you probably get to use Xv which is preferrable since 2D rendering is nice and fast in the open drivers...
kernelpanic: http://www.nopaste.org/p/a5m3sXLB7 To be honest, I'm not sure. googleearth seems to say no, glxinfo looks good to me
nanonyme: Looks good to me.
kernelpanic: mmh. that only works for youtube, right?
nanonyme: kernelpanic: It does, yes. And only for videos.
nanonyme: Not for any fancy Flash content.
chithead: there exist similar tools for other video sites too
MrCooper: I think you need to tweak some configuration file to make Adobe Flash use free GL drivers
nanonyme: Sounds odd. :)
kernelpanic: From what I read, it uses rgb video instead of yuv, so thats why it doesn't use xv. But I still wonder why it doesn't use a texture of the video's size and then scale it up to my screen's size. Or maybe it does, but then I'm surprised thats so slow
gentooer: here's some details of Flash hardware acceleration: http://blogs.adobe.com/penguin.swf/2008/05/flash_uses_the_gpu.html
nanonyme: kernelpanic: Adobe Flash doesn't do Xv at all.
gentooer: it requires several extensions that you may not have
kernelpanic: nanonyme: thats what I'm saying
nanonyme: Right. :)
gentooer: like you don't have "GL_ARB_shader_objects"
osiris_: glisse: ping
kernelpanic: gentooer: mmh, seems you're right. From the list, I can only find GL_ARB_multitexture in glxinfo's output
nanonyme: Gallium will get shipped with some other acceleration architecture for Flash too iirc. OpenVG?
kernelpanic: So, I'll just have to wait until the free radeon-driver supports these?
nanonyme: Well, waiting or joining in development are probably the only choices, yes. :3
phoenix64: "the Flash Player requires that the client glx vendor string be something besides "SGI"."?
nanonyme: Egh.
phoenix64: @client glx vendor string: SGI
MrCooper: that's just silly of them
kernelpanic: perfect conditions for a seconds-per-frame video experience, then :)
nanonyme: phoenix64: Meaning they require that it's a licensed OpenGL vendor.
kernelpanic: "We have this logic in place to detect whether software rendering is in place and fall back on our own software fullscreen in that case."
nanonyme: Or?
nanonyme: Oh, right.
kernelpanic: Well, that might be a bad idea, but the free drivers pretending to be from SGI seems a little odd, too
nanonyme: puts a donkey hat on
nanonyme: Might in fact even be illegal. :p
nanonyme: Unless it actually is from SGI.
MrCooper: it's based on the SGI SI
Netzpython: hi people!
nanonyme: MrCooper: Yeah but based on doesn't mean it's their product.
nanonyme: It might be illegal use of a trademark.
glisse: osiris_: pong
MrCooper: neither does the the vendor string mean that
MrCooper: it doesn't mean anything really, a point which unfortunately still seems lost on Adobe
osiris_: glisse: cs_write_reloc refs the bo, which function should unref it?
nanonyme: Hmm, SGI actually seems to be the primary author of several Mesa components...
Netzpython: maybe its the wrong place for my question, but i expirience a strange bug within xorg-server+radeon while trying to start compiz: get a white screen only and a mousepointer and such msgs in the log: EXA bug: Calling FinishAccess on pixmap 0x9004c88 with index 1 while it should have been (nil).
nanonyme: Including GLU.
nanonyme: (Or Mesa can use SGI's GLU anyway and Mesa page claims it's more advanced than Mesa GLU *shrug*)
nanonyme: Might be out of date. :)
MrCooper: Netzpython: the white screen is probably due to the X server failing to load/initialize the hardware *_dri.so for some reason
Netzpython: i cant spot any EE or WW for .so files here :/
glisse: osiris_: it's unref at bo submission
glisse: cs
Netzpython: all code is from master-git
glisse: so we know that bo doesn't disapear until submited
osiris_: glisse: somewhere in the texture code the bo isn't unrefed and the results after some time in OOM and GPU hang
MrCooper: Netzpython: grep AIGLX /var/log/Xorg.0.log
Netzpython: oh, that shows strange output
Netzpython: will paste 4 lines ok?
Netzpython: may i?
kernelpanic: you've already sent 3 :)
Netzpython: (**) AIGLX enabled
Netzpython: (II) AIGLX: Screen 0 is not DRI2 capable
Netzpython: (II) AIGLX: Screen 0 is not DRI capable
Netzpython: (II) AIGLX: Loaded and initialized /usr/lib/dri/swrast_dri.so
MrCooper: Netzpython: so first of all you need to figure out why the X driver fails to enable the DRI; is the radeon kernel module loaded? What does 'dmesg | grep -e agp -e drm' say?
glisse: osiris_: sounds like there is too many ref taken somewhere
glisse: osiris_: might happen if bo is referenced more than once
glisse: osiris_: what is the gpu hang ?
glisse: is computer still sshable ?
glisse: osiris_: by more than once i mean if in the cs the same buffer is rellocated more than once i think the cs code only unref once while it should unref as many time as bo is validated other solution is that on validation if buffer is already validated in the current cs we don't take another ref
osiris_: glisse: I don't have second machine, but I use SAK sequence to kill X and the app, then ctrl-alt-del to safely reboot
glisse: cleaning up radeon for 2.6.31 was sucking my time, hopefully i should be done tomorrow with patch testing
osiris_: glisse: I think this hang is caused in the end by out of gart memory access
glisse: osiris_: that's bad to not have another computer when doing gpu stuff :)
glisse: got to go try to see if my guess is the right one
glisse: bbl
legend2440: i am trying to build these drivers xf86-video-ati-6.12.2. ./autogen.sh --prefix=/usr works fine but when i make i get error atimodule.c:39: error: ‘PACKAGE_VERSION_MAJOR’ undeclared here (not in a function) anyone know why?
kernelpanic: Is the non-working rand-resolution-changing on rv530 a known bug?
bigtrev: question, is there an X lockup in this log? http://paste.ubuntu.com/189656/
Netzpython: ah, got it, thanx to mrcooper, even if hes gone ;)
linmax: hello, I'm trying to get 2d acceleration working with my r600 card
linmax: I'm running kernel 2.6.30-rc8-git3 and xf86-video-ati git from today
linmax: but when I start the X server I only get a black screen and the cursor is a horizontal bar with lot's of colored pixels
linmax: here is my Xorg.0.log: http://pastebin.com/d1d5d2891
linmax: my dmesg: http://pastebin.com/d2910bd19
linmax: with Option "DRI" "off" everything is fine except I have no HW acceleration
linmax: any ideas?
Netzpython: #
Netzpython: (EE) AIGLX error: dlopen of /usr/lib/dri/r600_dri.so failed (/usr/lib/dri/r600_dri.so: cannot open shared object file: No such file or directory)
Netzpython: #
Netzpython: that sounds bad
Netzpython: (EE) AIGLX: reverting to software rendering
Netzpython: und even worse, your xorg-server is "a little bit" OLD :D
Netzpython: X.Org X Server 1.4.2, so try at least 1.5.x or git
linmax: Netzpython: yeah, it's from Debian testing
linmax: I'll try a newer Xserver
nanonyme: Netzpython: No, it's not bad at all.
nanonyme: That's supposed to happen.
nanonyme: There is no 3D accel yet for r6xx/r7xx.
nanonyme: Or at least nothing released.
Netzpython: yeah that right
Netzpython: but the system should not say: /usr/lib/dri/r600_dri.so: cannot open shared object -> No such file or directory
nanonyme: Sure it should.
nanonyme: That's the name of the r600 Mesa driver.
Netzpython: and it does not exist yet?
nanonyme: 18:50 < nanonyme> There is no 3D accel yet for r6xx/r7xx
Netzpython: no 3d accel, no _dri.so at all?
nanonyme: It only exists if you install either of the r6xx/r7xx development Mesa branches with r600 support compiled in.
Netzpython: ah!
nanonyme: It's strictly a Mesa thing, nothing more.
nanonyme: So it shouldn't cause that.
nanonyme: I'm all for getting a newer X server though. ;)
Netzpython: hehe
Netzpython: xorg-server from git-master is not compileable here at the moment :/
nanonyme: Well, not that much newer. :D
nanonyme: Call me a conservatist but I prefer to stick closer to the released versions.
nanonyme: Like, say, atm branch server-1.6-branch sounds good to me.
nanonyme: Apparently 12 commits since 1.6.1 in 1.6 branch. :)
nanonyme: wonders what the heck is trying to be.
nanonyme: 1.6.2 RC1?
mikkoc: X.Org X Server (1.6.2 RC 1)
nanonyme: Right.
nanonyme: I wonder if that was a release on request. I haven't seen RC's for minor versions before.
Netzpython: fails compiling here...
Netzpython: git from 3 days ago works fine though -.-
nanonyme: How about newest from server-1.6-branch?
nanonyme: It's a tad newer than
Netzpython: but i can see impressive performace improvements on my M22 ati card + compiz :)
nanonyme: Heh.
nanonyme: Hmm, looks like I'll get to play with KMS too, after all...
nanonyme: Laptop just started refusing to start OSX so might install a Linux on it. It has an older ATi GPU. :)
Netzpython: well OSX is always a subject to change...
nanonyme: It's fine when it works.
erjc: [drm:drm_mode_getfb] *ERROR* invalid framebuffer id
nanonyme: It's mostly that when it doesn't, it's a black box and all you can do is curse or take it to maintenance.
linmax: ok, I now have X Server but it's still the same problem
linmax: anything else I can try?
erjc: http://erjc.pastebin.ca/1449836
Netzpython: linmax: which version of mesa lib?
nanonyme: bridgman: Hey.
bridgman: hi nanonyme
bridgman: I was just starting to look at all of elena09's docs but fortunately they just got happy on #ati ;)
bridgman: so I can get a bit further into Saturday without having to think
nanonyme: Say, what? o.O
nanonyme: needs an inbuilt Google
nanonyme: Right, some user then. :)
bridgman: check backlog on #radeon, I think -- nice organized page of "here's what I tried, here's what happened, here are my logs, here are my bios screens, here is the hardware..."
_Groo_: hi/2 all
_Groo_: any dev awake?
nanonyme: Could ask in any case and wait for them to wake up. ;)
groo_: ping all? oO
groo_: summons the great ones, rise agd5f, airlied, glisse and all the demons
bridgman: groo; throw yourself on the miracle of IRC logging
groo_: hu, didnt work
groo_: hi bridgman
bridgman: hi; AFAIK the summoning thing only works with me
bridgman: maybe I'm the only one who reads backlog ;)
groo_: so i summoned bridgman... need to get to lvl 80 to summon the rest
bridgman: pretty much
groo_: bridgman: can you take a look at xv code?
bridgman: sure, if you'll accept the usual disclaimers about hogs and digital watches
groo_: bridgman: btw is ok to start pouring bugs in bugs.freedesktop for dri2/kms? or to early?
groo_: bridgman: i can live with that if you fix the )(*@#)(@*# clipping again
nanonyme: :3
bridgman: groo, I feel your chances of immediate happiness slipping away
bridgman: you're on the kms/gem/ttm code, right ?
groo_: bridgman: i tried changing the code myself but it wasnt pretty ¬¬
groo_: bridgman: yep, latest from glisse with xorg backport from master.. cutting edge.. bleeding all over :D
groo_: bridgman: actually its pretty good. the only thing bugging me is the stretched xv video
bridgman: ahh... (bridgman looks for a way to gracefully get out of this ;))
groo_: bridgman: it was working with a patch from.. hmm i believe agd5f, but the code changed again and the patch doesnt apply anymore.. maybe you could take a look at the patch and apply it acordignly to the new code
dileX: groo_: Tried ?
groo_: dileX: its exactly that the one that doesnt apply anymroe
dileX: groo_: are you sure? take a deeper look
groo_: dileX: hmm you are right.. its newer code.. let me apply.. thanks dilex, let me see if that will make me a more happy clam
groo_: restarting x, brb and thanks again dilex
Emme_NK: Hi! I'm back at playing with KMS... Just grabbed the latest git, but it doesn't work (anymore)
Emme_NK: Display is out of sync
Emme_NK: serial console still working: http://www.nopaste.de/p/a3jAx1oxpb
groo_: it worked!!!
groo_: is a very happy clam..
dileX: groo_: thank agd5f (and if you like ppl reading backlogs)
groo_: dileX: what kind of donation you prefer? dead kitties or dead puppies?
groo_: dileX: i cant read backlogs since i dont stay online :P
dileX: what tld is dot-br?
groo_: dileX: brasil
groo_: btw with latest code, opengl is faster... i can do a lot of fun stuff with latest opengl games, like world of goo
groo_: unfortunatelly cairo-dock is still broken
dileX: groo_: last what I associate with brasil was the cinema film "city of god"
groo_: dileX: lol i wonder why :D
_Groo_: dileX: but im portuguese, not brazilian, i just live here for now
dileX: _Groo_: cool film (one of my favourites)
_Groo_: dileX: theres a very good one, "tropa de elite" in portuguese
_Groo_: dileX: see imdb.. i recomend it, since you aparently like that type of film
_Groo_: bridgman: can you explain me something?
_Groo_: bridgman: with latest xorg from master, the rgba color codes are correct for gtk finally
_Groo_: bridgman: BUT i still see some color errors in cairo render (mainly in firefox)
_Groo_: bridgman: is that a gtk error or a radeon one? i believe the latter since it was working before dri2
_Groo_: freenode is lagging a lot
bridgman: have to admit I haven't been following that problem; is there a bug report or ticket I could skim ?
bridgman: groo; I think that's me lagging today ;)
_Groo_: bridgman: no, its freenode alright
_Groo_: bridgman: for ex, if i run latest cairo-dock 2.0 (which was rebuilt with opengl, it gives me)
_Groo_: CS section size missmatch start at (r300_cmdbuf.c,emit_tex_offsets,185) 4 vs 2
_Groo_: CS section end at (r300_cmdbuf.c,emit_tex_offsets,189)
_Groo_: drmRadeonCmdBuffer: -22
_Groo_: and dmes
_Groo_: dmesg
MostAwesomeDude: _Groo_: There should be an error message in your dmesg.
_Groo_: MostAwesomeDude: http://pastebin.ca/1449927
_Groo_: MostAwesomeDude: i was dropping it to pastebin kde plasmoid :)
MostAwesomeDude: Ah.
MostAwesomeDude: I see it.
Emme_NK: how can I see the modeline radeon-kms is currently using?
Emme_NK: something in /sys ?
_Groo_: MostAwesomeDude: did you see the pastebin? my provider crashed
nanonyme: 21:23 < MostAwesomeDude> I see it.
_Groo_: nanonyme: the back logger king :)
dileX: Emme_NK: what says 'cat /sys/module/ttm/holders/radeon/parameters/modeset' ?
_Groo_: MostAwesomeDude: should i start opening bugs around dri2? or is it too soon?
Emme_NK: dileX: just "1"
dileX: Emme_NK: with nomodeset you will see 0
dileX: cheat-code*
Emme_NK: fbset says "timings 4294967295 0 0 0 0 0 0"
Emme_NK: but I don't assume these values are reliable...
_Groo_: MostAwesomeDude: ping?
osiris_: agd5f or bridgman: is it allowed for two instructions to output data to the same location (e.g. MOV vertex.result.color.xy, temp1; MOV vertex.result.color.zw, temp2;)
MostAwesomeDude: _Groo_: Pong?
MostAwesomeDude: osiris_: I don't see why not.
bridgman: osiris; this is two sequential instructions ?
bridgman: hold on, this is two different component subsets of the same register
_Groo_: MostAwesomeDude: :) did you see the pastebin? what do you think? fixable or not yet?
osiris_: bridgman: not necessarily
osiris_: and what in case of MOV vertex.result.color.xyzw, temp1; MOV vertex.result.color.xyz, temp2; (xyzw vs xyz writemasks)
bridgman: in the same instruction I think you would run into write port limitations
bridgman: are we talking 3xx-5xx or 6xx-7xx here ?
osiris_: r300
bridgman: in different instructions probably fine
osiris_: bridgman: what about the case where all source operands come from different memory types (e.g. MAD OUTPUT[0], INPUT[0].wwww, STATE[3], TEMP[0];)
MostAwesomeDude: _Groo_: Yeah, not exactly sure how to fix though.
_Groo_: MostAwesomeDude: do you have cairo-dock in order to reproduce the bug?
MostAwesomeDude: _Groo_: I think there's already a bug opened for it.
bridgman: osiris; head exploded, looking in docs ;)
dmb: glisse, is there a way for modeswitching with KMS?
dmb: so for example, a game can switch resolutions, or xrandr can?
MostAwesomeDude: dmb: xrandr works as usual.
MostAwesomeDude: So does xxf86vm.
dmb: tries xrandr
dmb: MostAwesomeDude, whats the proper syntax of the xrandr command to change resolutions?
chithead: dmb: recommended read http://wiki.debian.org/XStrikeForce/HowToRandR12
dmb: xrando is only showing 1920x1200 as a possible resolution
dmb: tries something
dmb: is it possible my monitor isn't reporting EDID information correctly?
chithead: maybe you disabled the scaler?
dmb: how did i do that?
dmb: this only seems to happen when using kms
dmb: dmb@pi:~$ xrandr --addmode LVDS 1024x768
dmb: xrandr: cannot find mode "1024x768"
adamk: dmb, You have to create the mode with the newmode option first.
chithead: xrandr --output foo --set scaler bar
dmb: adamk, heh, it didn't like that command very much
dmb: X Error of failed request: BadName (named color or font does not exist
dmb: erm
dmb: chithead,
dmb: the set command
chithead: of course replace "foo" and "bar" with the appropriate values
dmb: chithead, so would xrandr --output LVDS --set scaler 1024x768 be correct?
chithead: no
chithead: scaler takes a scaler type as argument, eg. off or depending on hardware and driver support center, aspect or full
dmb: chithead, they all seem to give the X error
chithead: off should not give an error
dmb: even off gives the error
dmb: dmb@pi:~$ xrandr --output LVDS --set scaler off
dmb: X Error of failed request: BadName (named color or font does not exist)
chithead: maybe it is a limitation of kms
dmb: without kms, it actually lists all the proper modes in xrandr
dmb: looks at MostAwesomeDude :P
MostAwesomeDude: dmb: Don't look at me; it should magically work.
dmb: maybe KMS doesn't read EDID information yet
MostAwesomeDude: I doubt that. More likely, it's a bug in KMS quirks.
cbmuser: airlied: how is the status with kms-code ? can you estimate when the code will be in the main kernel-tree ?
airlied: cbmuser: planning on merging it in the merge window hopefully
airlied: just initial code, for experimental use
_Groo_: hi airlied
_Groo_: guys any news on bug 16537? i cant think of anything i can do to help you guys debug it more
airlied: _Groo_: no idea I' am just havck from holdiays
airlied: back even
_Groo_: airlied: its the same bug i filled months ago, garbled screen on rs485 when returning from suspend
_Groo_: airlied: some stupid register isnt getting properly activated
_Groo_: airlied: i used to be able to workaround it with vbetool quirks.. but now that im testin dri2/kms.. i cant do a thing...
_Groo_: airlied: btw, very good job with dri2/kms indeed.. im happy as a clam, using it for almost 2 months now... and always compiling and sending bugs.. every dev in here hates my guts :D
nanonyme: _Groo_: Or then they love you for helping them in their work of making good drivers. ;)
nanonyme: Helping them notice things they otherwise would have missed. :3
_Groo_: nanonyme: naah, probably they wish i might drop dead :D
cbmuser: airlied: thanks for the feedback
cbmuser: tried kms on my Thinkpad T43 with Radeon without success
cbmuser: Radeon X300 Mobility, I meant
_Groo_: cbmuser: which chipset?
cbmuser: 01:00.0 VGA compatible controller: ATI Technologies Inc M22 [Mobility Radeon X300]
cbmuser: I'm on Ubuntu 9.10/unstable
_Groo_: cbmuser: is that a 3xx-5xx or a 6xx-7xx?
cbmuser: rv370
cbmuser: so r300, right ?
nanonyme: Yes.
_Groo_: cbmuser: should work then.. did you followed glisses blog?
cbmuser: I followed Dave's blog
cbmuser: can you point me to glisses blog ?
_Groo_: try this http://jglisse.livejournal.com/1822.html
nanonyme: cbmuser: Could you also link Dave's blog?
_Groo_: cbmuser: youll have to do some pretty big changes, but make , make install should do the trick
nanonyme: Could compare the two.
agd5f: dmb: kms code doesn't add scaled modes for lvds yet. you have to add them manually with --newmode and --addmode
cbmuser: nanonyme: http://airlied.livejournal.com/62269.html
nanonyme: cbmuser: Yeah, definitely glisse's instructions then. ;)
cbmuser: ok
cbmuser: so far I managed kms to work on an intel chipset
nanonyme: If you take a closer look, differences between the two blog posts are in kernel branches, ddx branches and Mesa branches. (which is already pretty major) :)
nanonyme: They use the same libdrm though, I guess.
cbmuser: still cloning the first git-repo ;)
cbmuser: the drm-next stuff is very large
airlied: nanonyme: wierd it should be the same mesa branch
airlied: nanonyme: and the DDX is only one patch different
airlied: maybe two
nanonyme: airlied: Yours has r300-bufmgr Mesa branch, glisse's has radeon-rewrite Mesa branch.
airlied: nanonyme: my blgo post is very old
nanonyme: I noticed. :)
nanonyme: I was just going through the differences.
cbmuser: isn't there a single patch I can apply against 2.6.30 to get radeon-kms into the kernel ?
airlied: cbmuser: no.
nanonyme: cbmuser: If you have a git repo for main kernel tree, you might be able to use it as a reference
nanonyme: Local repo, that is.
cbmuser: usually never pull kernel-sources via git
nanonyme: You might want to take the habit of keeping Linus' tree as git, then using it as reference.
nanonyme: Saves loads of time when cloning other trees based on it.
cbmuser: hehe
cbmuser: actually I have stopped building my own kernels, the kernels provided by the distributions are very good
nanonyme: Yeah, true.
cbmuser: I hope the radeon stuff will be in Linus' tree soon
nanonyme: airlied might know about schedules if anyone does...
osiris_: airlied: could you take a look at the mipmap_limits crashing bug? here's the bt http://pastebin.com/m75839930
osiris_: airlied: t->mt is null
glisse: airlied: i got the patch ready for kms i need more test on agp/pcie so i am sure i didn't break anythings while cleaning up checkpatch.pl complains
glisse: anyway way too late over here i will post it tomorrow and then i think we should be ready to face any complains on lkml ;)
Netzpython: is the dynamic-clocks-option in xorg.conf now deprecated?
chithead: Netzpython: it has been renamed, and other functions added. see the radeon manpage
Netzpython: thx
Enrico|Chicco: mhm i'm trying to use radeon driver for my r600 kernel but Xorg freeze. i'm on .29 kernel.... do i need some particular version of some software? i'm on gentoo
Netzpython: Enrico|Chicco: when does it freeze?
airlied: though you might convince Intel with a 965 driver but you might not/win 21
airlied: oops
Enrico|Chicco: Netzpython: at startup
Enrico|Chicco: Netzpython: the screen is black with some red artifrat in the upper border
Enrico|Chicco: artifact*
Netzpython: Enrico|Chicco: ok, which versions of video-ati, xorg-server, mesa drm etc is installed?
Netzpython: all from stable or ~arch or even git?
Enrico|Chicco: video-ati 6.12 xorg-server 1.5.x mesa 7.4
Enrico|Chicco: lib drm .11
Netzpython: k
Enrico|Chicco: Netzpython: i run stable but xorg stuff is all almost in ~arch
Netzpython: had some probs like lockups with these versions too, ati m22 (r300 ased) chip
Netzpython: upgrading xorg+drivers+libs to git could be worth a try
Enrico|Chicco: Netzpython: eheheh i prefer to wait then
Enrico|Chicco: the worst thing is that Xorg.0.log is not generated!
FallenWizard: afaik you need an updated drm kernel module
FallenWizard: for r6xx-r7xx and 2.6.29
FallenWizard: 2.6.30 has it
Enrico|Chicco: oh that's why
Enrico|Chicco: FallenWizard: i've changed my xorg.conf a bit and now i see that xorg fails drmOpenDevice
Enrico|Chicco: with errore -1 No such device
FallenWizard: Try 2.6.30
FallenWizard: -rc8 runs fine here
Enrico|Chicco: FallenWizard: well i can wait its release (some days :D )
FallenWizard: ok
Enrico|Chicco: FallenWizard: but in the mean time i can try to update drm modules pulling them from git
Enrico|Chicco: and build them against .29
FallenWizard: ok
Enrico|Chicco: gentoo has an ebuild for that
zc00gii: [20:23] I tried the catalyst drivers with arch, but I failed configuring it
zc00gii: [20:23] I then switched back to the open source drivers
zc00gii: [20:23] but then for some reason compositing broke
zc00gii: I'm hotplugging btw
zc00gii: I may have had an idea that fixed it
zc00gii: brb
zc00gii: rebooting
zc00gii: nope
zc00gii: didn't do crap for it
_Groo_: hi/2 all
_Groo_: agd5f: ping
_Groo_: airlied: ping
_Groo_: MostAwesomeDude: ping
zhasha: good morning _Groo_
_Groo_: zhasha: good night zhasha:)
_Groo_: anyone tested google earth yet? im having complete screwed up rendering when zooming
_Groo_: and a lot of
_Groo_: [35449.654012] [drm:r300_cs_track_check] *ERROR* [drm] No buffer for color buffer 0 !
_Groo_: [35449.654016] [drm:radeon_cs_ioctl] *ERROR* Invalid command stream !