Nightwulf|work: hi all
Paulie889: Hi when testing r300g driver I noticed this warning in terminal "r300: Warning: cs_count off by -5", in some games terminal output is spammed with it. This warning appeared with some last mesa changes, however I didn't noticed any visible problems. So I'm wandering if this is just a harmless warning or if this is sign of a bug and should be reported.
glisse: sounds like a bug
mcgreg: can you use 2 gfx cacrds with the radeon driver, both at the same time (one onboard) like: gfxcard1 -> monitor 1 gfxcard2 -> monitor2 ?
glisse: mcgreg: not as one desktop
mcgreg: so, you mean, it would be two totally seperate x servers?
glisse: yes
mcgreg: thank you
Zaba: hrm, strange
Zaba: using a few hours' old version of mesa/libdrm/radeon driver from git, I get a strange issue in X: For instance, I open a terminal, and it's visible on the screen even if I close it. Likewise for everything else: It just doesn't go away, not even when I restart X.
Zaba: I don't get anything in xorg's log or dmesg, though
Zaba: my kernel is 2.6.34
mcgreg: I am getting corruptions after every fullscreen 3d application now - but I dont have anything in dmesg as well
Zaba: I get them with windowed 3d stuff, too
mcgreg: Zaba: http://i.imagehost.org/0109/screenshot1.jpg have a look at that - got that yesterday after starting wine Wow (world of warcraft)
Zaba: ah, well, I don't get anything like that
mcgreg: even after restarting Xorg the screen was corrupted
Zaba: but yes, my screen also was still full of trash when I restarted X
Zaba: mcgreg, what I get is simply that things don't get erased. Contents of closed windows are visible until something draws over them, and such.
mcgreg: for me it's mostly trashed fonts
Zaba: mcgreg, I wonder whether it's been bug-reported.
mcgreg: I dont know. I wouldnt even know to what report it against. it could be "everything" (from my point of view)
mcgreg: ouch , sorry sounds like terrible english
Zaba: mcgreg, I'm not sure what the problem is in, either.
Zaba: Could be mesa, could be the radeon driver, I guess.
Tommeh: agd5f: could you remind me what the kernel option was to disable vsync on my R600? I was having flickering problems onboot/on load of wine programs, and you suggested that might be the case.
Tommeh: *that might be the fix
Zaba: hrm
adamk: Morning all. The r600 driver has some issues with NeverwinterNights. This is what it should look like: http://thorn.visualtech.com/fglrx-nwn2.png and this is what it looks like with the r600 driver: http://thorn.visualtech.com/r600-nwn2.png
adamk: I disabled s3tc support in the game when using fglrx to get an accurate comparison.
adamk: Basically, the texture of the cloak is all screwed up, and it remains standing when I'm resting :-)
adamk: Any thoughts before I open up a bug report?
crocket: After I resume from sleep in slackware, the monitor gliches a lot when X session updates.
crocket: I heard it was related to radeon drvier.
crocket: or graphic drivers in general.
WintermeW: hi ppl...got an issue with radeon OSS driver and radeon pro 9200 (R280). when i set the Device driver to "radeon" and that i restart X, i get awful colors and horiz/vert offsets ...my screen and monitor section are correct i think, since when i use vesa, it's perfet..any clue about how to fix that ?
chithead: you don't need any xorg.conf, what happens if you rename or remove it
WintermeW: same thing
WintermeW: 01:00.1 Display controller: ATI Technologies Inc RV280 [Radeon 9200 PRO] (Secondary) (rev 01)
adamk: When using radeon in the xorg.conf file, does it work properly if you force it to use XAA?
WintermeW: for my previous problem, i tried to force the AccelMethod but still no luck..screen with offsets and really fucked-up colors
dileX: $ sudo modinfo radeon | grep -i dyn
dileX: parm: dynclks:Disable/Enable dynamic clocks (int)
dileX: is that module-option correct?
mcgreg: Zaba: just played warzone2100 for 1 hour ... my konsole is totally trashed. could you test that perhaps too?
Zaba: mcgreg, which part of that should I test?
mcgreg: well, I simply played it , fullscreen , normal single player game against computer (AI) - I guess it doesnt matter what you do
Zaba: any kind of opengl activity trashes X for me, in the way I've described above.
rah: my X server (from git) crashes on startup with the followingL
rah: :
rah: (II) [KMS] Kernel modesetting enabled.
rah: X: pixmap.c:118: AllocatePixmap: Assertion `pScreen->totalPixmapSize > 0' failed.
rah: Aborted
mcgreg: yeah well, I believe it also depends of how long or how intensive the usage was - like, if I played warzone for 2 minutes I had just few corruptions, if I played for about a hour, I have massive corruptions
rah: anyone have any idea what my be causing this?
rah: s/my/may/
mcgreg: and those corruptions slowly go away with time
Zaba: it was enough for me to start quake2 in developer mode (which just opens the console, doesn't even render anything 3dish), and then after quitting it, it wouldn't get erased, and the same would happen to any other X app.
Zaba: they don't go away for me at all
rah: answer me.. ANSWER ME!!!
rah: pokes agd5f
adamk: Wow... A development version of the X server crashes... Now there's a shocker.
rah: adamk: you find it shocking that a development version of the X server crashes?
adamk: No, I do not. That was sarcasm.
rah: I see
rah: have you any idea what may be causing this assertion to fail?
adamk: Something in the pixmap code. I'd guess it's unrelated to the radeon driver.
adamk: But it is code under development, so things change rapidly. Wait a bit and try again. If it continues, open up a bug report.
rmrfslash: Fellas, I'm looking into a laptop that I want to develop GPGPU stuff
rmrfslash: OpenCL or CUDA
rmrfslash: Now, I'm looking at a couple of laptops with video cards from NVIDIA and ATI
rmrfslash: I know my laptop now, which has an ATI Radeon Mobility HD 3670 is god awful
rmrfslash: in terms of working under Linux (though I'm using Kubuntu)
rmrfslash: But the only I'm looking at has an ATI Radeon MObility 5145
rmrfslash: How well supported is this in the new xorg radeon driver
philv: Well, I'd check ATI's OpenCL compatibility list first, because the 5145 isn't on it iirc.
agd5f: 51xx are actually r700 class hardware
rmrfslash: that was my next question
rmrfslash: what class are these so I can look up the radeon featuers list
rmrfslash: I see OpenGL in the radeon driver features page as yellow and it says 2.0/3.3
rmrfslash: what is this?
rmrfslash: if it says that except for "MOSTLY"
rmrfslash: s/except/instead
rmrfslash: oh, also... what class is the Radeon Mobility HD 3670?
thansen: agd5f: ever find out any more info on this.. http://bugs.freedesktop.org/show_bug.cgi?id=27314
agd5f: rmrfslash: 3670 is r6xx class
rmrfslash: so I'd look under r600
agd5f: thansen: no. that dell 27 seems problematic
thansen: agd5f: got others with the same issue on that monitor?
agd5f: yeah
agd5f: the link doesn't want to train for some reason
thansen: have any bug numbers handy for me to look at/cc myself?
agd5f: thansen: it's already marked as a dupe of your bug
thansen: oh
thansen: usually sees those emails
agd5f: thansen: does 2.6.35rc3 work any better?
thansen: I tried one of the .35s but it's been a little while
thansen: unfortunately my monitor is in storage unit right now too :(
thansen: agd5f: I'll see if I can find a way to hook it up to an old dell 30"
thansen: is there such thing as displayport to dual link dvi cable?
agd5f: thansen: that will probably work fine. I have a dell 30" and it works fine with dp
thansen: the m6500?
agd5f: thansen: dp to single link for passive adapters
thansen: wants full res
thansen: guess it doesn't matter for testing
thansen: I couldn't get the 27" to work at any res
glisse: dp can handle higher res that dvi
thansen: right
thansen: I have 30" that doesn't have dp
glisse: and dp->dual dvi would need active adapter which might not even exist
thansen: that's what I was wondering
agd5f: thansen: dp to dvi adapters don't use dp signaling so it's tmds pass through
agd5f: they exist, but they may have the same issue with the problematic monitor
thansen: oh
thansen: ok
agd5f: thansen: the active adapters talk DP to the card, then convert dp to tmds for the monitor
thansen: you're well over my head at this point :) I trust you
agd5f: thansen: if you get a chance, try dumping the regs with the problematic monitor as noted on the bug in my latest comment
thansen: yeah, I will
thansen: so I can still have kms enabled right (or should)
agd5f: thansen: you want to dump the regs while the monitor is still on, so don't load radeon or X
thansen: isn't sure where the lines are for all this stuff so I don't know when 'drm' kicks in
agd5f: thansen: blacklist radeon and boot into runlevel 3 or whatever runlevel doesn't start x on your distro
thansen: runs gentoo so I'm good with that
thansen: ok, so kill kernel stuff completely
agd5f: thansen: just add radeon to the module blacklist while testing, just to make sure it doesn't load
thansen: yeah
agd5f: /etc/modprobe.d/blacklist.conf
thansen: nods
agd5f: might be slightly different per-distro
thansen: it's already in there right now while I'm using fglrx
agd5f: thansen: you can dump the regs using fglrx if that's easier for you
thansen: I just don't have the monitor right now :(
thansen: is there some other tool I should use in the fglrx case
agd5f: thansen: avivotool is fine
agd5f: same instructions
taiu: agd5f: + if (track->cb_color_bo_offset[i] & (track->group_size - 1)) {
taiu: agd5f: is this ok? bo_offset seems to be upper 32 bits already
thansen: agd5f: with fglrx I *can* get into X though? (just out of curiosity)
agd5f: thansen: yes. we just need a dump of the working state
agd5f: I just suggested the bios since I didn't know you were using fglrx
thansen: agd5f: got it, I'll be on it as soon as I get my monitor out (or before since I have another)
agd5f: taiu: good catch
thansen: agd5f: thanks again for the responsive help..I really appreciate it!
agd5f: thansen: np
Tommeh: agd5f: could you remind me what the kernel option was to disable vsync on my R600? I was having flickering problems onboot/on load of wine programs, and you suggested that might be the fix for me to try.
Tommeh: promises to write it down this time :)
agd5f: taiu: it's already shifted so we should probably shift back to bytes or shift group size
agd5f: Tommeh: no option that I know of
agd5f: Tommeh: you'd have to mess with the dri2 stuff
agd5f: Tommeh: the drm gets told what to do by userspace
Tommeh: agd5f: I see. I wonder if there's any log you'd like to see?
Tommeh: It's not like the displays flicker constantly - it's only the one display that does it, and only one boot or when loading wine programs (like Spotify)
Tommeh: And *only* with KMS.
Tommeh: *on boot
Tommeh: I've been waiting to see if it just gets fixed by new releases, but so far (2.6.35-rc3 + Sarvatt's edgers repo with all updates) it's not stopped doing it. :)
agd5f: Tommeh: there are several dri2 vsync related bugs and patches floating around on the xorg mailing list
Tommeh: I see!
Tommeh: I'm a development newbie, I don't really know how to go about reporting things. Sorry :)
agd5f: Tommeh: no worries
Tommeh: Hopefully with it being r600, it is still under lots of active development and testing - so hopefully it may be fixed yet! :)
Tommeh: I can't find any recent bugs in the last year via google search of the xorg mailing list, however.
Tommeh: Are there really outstanding bugs that old for r600 dev?
agd5f: Tommeh: maybe this bug? https://bugs.freedesktop.org/show_bug.cgi?id=28410
agd5f: Tommeh: the kms vsync stuff is only very recent, so if it's been an issue for a while, it's probably something else.
Tommeh: Sinnnce.. 2.6.32-rcX, to my knowledge.
Tommeh: That was when I finially got KMS working *and* DRI at the same time.
agd5f: Tommeh: so it's probably not vsync related
Tommeh: Jun 03 16:11:03
Tommeh: realised xchat is logging stuff :)
Tommeh: I'm daft and forgot what was up.
Tommeh: I'll go test radeon.disp_priority=2 now.
Tommeh: brb
Tommeh: agd5f: Yep, sadly that doesn't help
Tommeh: The 2nd screen still flickers on boot, after login and probably when I load Spotify :)
Tommeh: i've always thought it was a DRI/RandR mix-up but I don't really know that much about it all :)
agd5f: Tommeh: if it flickers constantly then it's probably a mode problem, likely either pll or watermark related
agd5f: Tommeh: does radeon.new_pll=0 help?
Terrax121: Wauu. Decoding h.264 through shaders seems really hard!
Sarvatt: Tommeh: its not related to your problem but to disable vsync with xserver 1.8+ you need mesa from git and this in your ~/.drirc - http://paste.ubuntu.com/450867/
Terrax121: agd5f, based in your experience. Would it be possible decoding h.264 proberly though shaders. I mean, is there anything to gain compared to using the cpu only?
Sarvatt: (note the driver name, using your actual driver name doesn't work)
MostAwesomeDude: Terrax121: Some of the steps make a lot of sense to accelerate.
Thunderbird: idct for instance is not that hard
Thunderbird: take a look at the whole decoding chain
Thunderbird: and start at the right of it
Terrax121: I am just wondering if transfering data from the gpu to the cpu creates a bottleneck
Jonimus: Terrax121: that would be the bottleneck on many systems yes, unless you can do as much as possible on one chip or the other
Daekdroom: Or avoid any data transference by doing as many steps in a row as you can..
Terrax121: Another thing I am wondering about
Terrax121: If I can complete this project within 3 month :)
Jonimus: lol Terrax121
Jonimus: and would flash actually use it :P
MostAwesomeDude: Terrax121: You should be in #dri-devel too.
MostAwesomeDude: My original proposal for this kind of stuff was that you had to start on the right and accelerate all the stages in a row.
MostAwesomeDude: So it wouldn't ping-pong.
Terrax121: MostAwesomeDude, yeah. Well my plan is to implement a workable vdpau backend
Terrax121: Then slowly moving all the different decoding stages to galium one at a time
MostAwesomeDude: Terrax121: You mean, all in software?
MostAwesomeDude: Might want to look into libavcodec.
MostAwesomeDude: Or I suppose that's backwards, isn't it? HM.
Terrax121: MostAwesomeDude, I want to get a hang on all the decoding steps before I move it to galium
Terrax121: Does anybody know what the vdpau-wrapper is for?
agd5f: Terrax121: I've never implemented it, but MC parallelizes well, so I'd suggest doing MC and higher on the GPU as a start
MostAwesomeDude: Terrax121: Gallium's API pretty much guarantees that a lot of features will be available.
MostAwesomeDude: MC and higher already exist in pipe_video_context. Reuse it.
Terrax121: agd5f, okay, i'll try
Terrax121: MostAwesomeDude, I will take a look on it. But don't you think it would be better implementing it on the cpu at first, and then moving the different function over to galium?
MostAwesomeDude: Terrax121: Not particularly. But that's my opinion.
Terrax121: Ok :)
Terrax121: MostAwesomeDude, I'll give it a try. But I will need you to help me with the galium specific stuff
Thunderbird: initially I would just do it using GL + GLSL using a stable driver
Thunderbird: just to get a feeling on how it should work
Thunderbird: and later on port it to gallium3d
Terrax121: Thunderbird, yeah I have had those thoughs too
Terrax121: I could even implement it with OpenCL
Daekdroom: OpenCL looks like a good idea.
Daekdroom: I mean, it could even get ported to other drivers.
mwc: You could even test an OpenCL implementation today as you develop it, fglrx + stream sdk has an OCL 1.0 implementation
mwc: I was looking into this as well, once of the nicest features seems to be the ability to share buffers between OpenGL and OpenCL
mwc: so you can render your final frame into an OpenCL buffer and then immediately render it via Glx to a drawable
Thunderbird: vdpau allows cooperation with gl these days as well (there were some changes)
Thunderbird: so perhaps that's how you should do it for now as well
Thunderbird: and it would not be limited to gallium3d
Daekdroom: Too many possibilities o.o
Thunderbird: but personally I would wish to use the hw decoder
Thunderbird: people will start complain to you why their laptop battery drains so quickly
mwc: vdpau is a proprietary Nvidia API. Are there any plans to implement it on top of gallium?
Ke: radeon and laptop is nono anyways
mwc: Yeah, nobody's really working on power management are they
leio: they are
Terrax121: mwc, yes
Thunderbird: there are multiple implementations of vdpau already
Thunderbird: for instance good old S3 even has one
leio: said work was just recently rejected by Linus into 2.6.35-rc3 :P
leio: if I am to believe what I read from phoronix
Thunderbird: s3 has no open code
Terrax121: Yeah well implementing vdpau though OpenCL may be the way to go. But I would also like it to be implemented with galium.
Reilithion: Jonimus: You lurkin'?
mwc: Hmm, I thought va-api was the way forward for open source video accel
Daekdroom: Wouldn't it end up being the same? Gallium will end up having OpenCL sooner or later.
Reilithion: I've found the specific commit that broke radeon+kms for me. It's 6a9ee8a, made by Dave Airlie. Does he hang around in here?
mwc: Reilithion: ask airlied ;)
Reilithion: airlied: Hi. I think I may have found a bug with a commit you made to the Linux kernel.
Terrax121: Daekdroom, it might
mwc: Daekdroom: supposedly, the folks at VMware are working on one http://zrusin.blogspot.com/2009/02/opencl.html
agd5f: leio: the temp sensor patch was rejected but the other pm stuff is already in there
Ronis_BR: hi all
Ronis_BR: when I trying to play a game with wine sometimes I got the message
Ronis_BR: radeon 0000:01:05.0: vbo resource seems too big for the bo
Ronis_BR: and X crashes
Ronis_BR: can anyone help me?
Ronis_BR: anyone??
Ronis_BR: I'll try to update video driver
airlied: Reilithion: breaks how?
Reilithion: airlied: When I upgraded to 2.6.34, I was suddenly unable to boot. I found out that if I had kms enabled, loading the radeon driver caused the system to completely freeze. Even Magic SysRq wouldn't work.
Reilithion: I tracked it down to commit 6a9ee8a -- vga switcheroo
Reilithion: Excuse me. By "radeon driver" I of course mean "radeon kernel module"
airlied: Reilithion: are you on a laptop with two GPUs?
Reilithion: airlied: No. Desktop with 2 separate graphics cards. One a Radeon X850 in a PCIe 16x slot, the other an older Radeon in a PCI slot.
airlied: Reilithion: that commit doesn't do anything on machines that aren't laptops
airlied: did you enable CONFIG_VGA_SWITCHEROO?
Reilithion: Let me look.
Terrax121: Okay I got an implementation specific question
Terrax121: http://http.download.nvidia.com/XFree86/vdpau/doxygen/html/index.html
Terrax121: Why would nvidia typedef all their function calls?
Reilithion: airlied: It is not set.
Reilithion: airlied: That commit is the first commit in which I can reproduce the problem.
Terrax121: All their functions, which is needed to be implemented is typedef'ed :/
airlied: Reilithion: try setting CONFIG_VGA_SWITCHEROO thouigh it shouldn't make a differne
airlied: Reilithion: you have ACPI support?
soreau: Does compiz alpha blur work with r600 mesa driver?
airlied: Reilithion: try commenting out the call to radeon_register_atpx_handler in radeon_drv.c
mokoloko: blur works with kwin effects pretty well
Reilithion: airlied: I do have acpi support. I'll try that commenting out first.
adamk: soreau: Admittedly it's been a while since I've tried, but when I did it was definitely not working.
adamk: (A while > 3 weeks)
soreau: oh wow
soreau: adamk: I thought it was working with r600 already
adamk: Hmm...
adamk: Blur doesn't work on r300-r500 if you have two monitors (even if you have an r500 with a MTS of 4096 and are well within that limit). Perhaps the same is true with r600 and I just never tried with only one monitor.
soreau: that sounds strange
soreau: adamk: If you ever get a chance to test, could you let me know if it does or not?
soreau: (blur work on r600)
Reilithion: airlied: That does not seem to have had an effect. Loading the radeon module still causes a freeze.
Reilithion: airlied: I'll try turning on CONFIG_VGA_SWITCHEROO
airlied: Reilithion: yes that would be interesting, though none of that code runs in theory if the config is off
airlied: and you comment out that line
adamk: soreau: Yeah, just remind me in a day or two if you haven't heard from me.
soreau: adamk: Ok
MostAwesomeDude: Terrax121: Have you looked at libvdpau yet? It's hosted on FDO.
Terrax121: MostAwesomeDude, I don't get why nvidia would define the functions as typedefs?
airlied: Terrax121: function pointers?
Terrax121: airlied, yes
airlied: thats pretty normal
airlied: it makes it a lot cleanr to pass function pointers as typedefs
Terrax121: airlied, for an API?
airlied: yes, GL does it
Terrax121: Okay didn't know that
airlied: granted GL does it ugly as hell
airlied: look for typedef in usr/include/GL/glext.h
airlied: then recoil in horror at the PFNGLOMGIMSHOUTINGCANYOUHEARME typedef names
Terrax121: airlied, haha
Terrax121: :)
Terrax121: airlied, do you know where the corrensponding .c file for glext.h is placed?
airlied: they just end up inside mesa
Terrax121: yeah of course
Terrax121: I'll take a look
Terrax121: OMG: typedef Bool ( * PFNGLXQUERYMAXSWAPBARRIERSSGIXPROC) (Display *dpy, int screen, int *max);
Terrax121: Well thats ugly
MostAwesomeDude: PFNMMMSANDWICHESOHGODILOVESANDWICHESSOMUCH
Daekdroom: What the?
Terrax121: =)
vital: GL oh teh beautiful...
Daekdroom: Well..
Daekdroom: Atleast it tells what it does
vital: loves sandwiches
MostAwesomeDude: It's completely reasonable in context. The functions pointed to might not even exist, so you obviously can't have them at compile time.
MostAwesomeDude: But I wish they wouldn't have gone with shouting names.
wirry: cat-like-typing detected
Terrax121: airlied, I don't see the point in declaring typedef'ed functions, instead of actual functions
Terrax121: Like this:
Terrax121: typedef VdpStatus VdpDeviceCreateX11
Terrax121: you would have to do this afterwards:
Terrax121: VdpDeviceCreateX11 vdp_device_create_x11;
Terrax121: What is the purpose of that?
airlied: Terrax121: because you want to pass function pointers
airlied: its not for defining actual function
Terrax121: airlied, okay maybe I will see the point in it some time :)
Terrax121: or the pointer in it ;)
airlied: when you pass function pointers it gets really ugly
airlied: since the function paramters needs to be complete function pointer specification
airlied: so if you change the function prototype, you have to change *everywhere* its passed
airlied: whereas if you typedef it you don't
Terrax121: Ahh I see
Ronis_BR: guys, I have updated everything and I still getting the message: radeon 0000:01:05.0: vbo resource seems too big for the bo
Ronis_BR: it happens while playing an opengl game with wine
Ronis_BR: when the problem occurs X crashes
Reilithion: airlied: Neither change you suggested made any difference. The radeon driver freezes whether CONFIG_VGA_SWITCHEROO is enabled or disabled, and whether or not I comment out that function call.
airlied: Reilithion: ah I have an idea
airlied: Reilithion: in radeon_get_bios, try avoiding the call to igp_read_bios_from_vram
Reilithion: same file?
airlied: in radeon_bios.c
Reilithion: airlied: Ok. And how shall I avoid it? Comment it out?
airlied: yeah
Reilithion: I'll try that.
Ronis_BR: airlied: can you help me?
Reilithion: airlied: That seems to have done the trick.
Reilithion: airlied: The radeon module did not freeze this time.
airlied: Reilithion: could you file a bug on bugs.freedesktop.org? against xorg-x11-drv-ati
airlied: I'm not working today so I'll forget over the weekend, I'll write a patch for Linus next week
Reilithion: airlied: Sure. Mind explaining what's going on?
Reilithion: I don't understand any of this.
airlied: Reilithion: its trying to read the BIOS from VRAM, but the GPU probably isn't posted yet, so the VRAM isn't setup
airlied: we used to check for that only on IGP chips, but I've see the BIOS in VRAM on the switcheroo machines as well so I dropped the check for IGPs
airlied: but broke unposted GPUs
Ronis_BR: well, I found another guy with the problem: https://bugs.freedesktop.org/show_bug.cgi?id=28574
Ronis_BR: does anyone have any idea what commit should have made this? It seems to be a very new problem
Reilithion: airlied: I found Product: xorg , Component: Driver/Radeon , is that where I should be filing?
airlied: yes
Terrax121: airlied, but would using typedefs instead of normal prototypes, not cause the library to be useless as an implementation?
Jonimus: I just built a fesh mesa form git and now when I try to play UT2k4 there is major corruption al over the screen
Terrax121: airlied, I mean. If I have to do this: VdpDeviceCreateX11 vdp_device_create_x11
Terrax121: The one who might want to use my implementation, would have to use my specific headers and not the one made by nvidia
airlied: Terrax121: you shouldn't be using them for prototypes gernally, just for pointers
airlied: and if the API is compatible they both should be the same
Jonimus: I'm rebuilding libdrm,mesa, and ddx but I can't seem to find what the issue is/was.
Terrax121: I do use them for pointers only, but how would one call the VdpDeviceCreateX11 function if its only typedefs? They would have to call vdp_device_create_x11 and therefor need my specific headers
Reilithion: airlied: Bug 28592 - When KMS is enabled, radeon module freezes system on load.
airlied: Reilithion: thanks
Reilithion: airlied: No, thank you.
airlied: Terrax121: the whole point is you use their library though
airlied: and justprovide a backend for it
Terrax121: airlied, you got a point
adamk: soreau: No go. http://thorn.visualtech.com/r600-compiz-blur.jpg
adamk: soreau: If I use OBS to make thunderbird transparent, I get that awful red, orange, green rainbow thing.
soreau: Nice
adamk: soreau: As you can see, though, the gnome-terminals that have their own transparency do blur.
soreau: that's strange
adamk: soreau: But I can't actually see the content of the blurred terminal :-) They have text in them that's not at all visible.
soreau: hmm.. I see that now
soreau: I recall similar behavior before blur was working with r300x drivers
soreau: adamk: Thanks, I might put a note in the compiz wiki
adamk: This is with 0.8.6 and mesa from git from 10 minutes ago.
soreau: That is if you don't want to do it ;)
adamk: If I get a chance tomorrow I'll test with compiz++
soreau: adamk: I doubt it will make a difference
adamk: But for now I'm off.
adamk: Yeah, I'm guessing this is a driver issue not a compiz one.
adamk: ttyl.
soreau: it's obviously a driver issue and the ++ implementation is quite similar
meoblast001: where is the Mesa IRC channel, i'm having a few problems after Ubuntu nuked my drivers
soreau: meoblast001: Do you have a radeon graphics card?
meoblast001: yes
meoblast001: oh, i don't even want to think about the fun i'm going to have when i restart my system
soreau: meoblast001: What makes you thing ubuntu nuked your drivers?
meoblast001: it updated
soreau: from/to what?
meoblast001: from my custom ones to its DEBs
meoblast001: so now i went from GL 2.1 to GL 1.5
soreau: mesa breakage is typically not a showstopper
meoblast001: well now drm won't compile to a non /usr/local directory if i compile with Intel
meoblast001: and Mesa won't let me compile without Intel
soreau: I'm still not clear why you don't reload your system and start from there
meoblast001: reload my system?
soreau: As you put it, 'restart my system'
RAOF: Or use xorg-edgers, which will happily provide you with mesa crack.
meoblast001: ok... brb
meoblast001: RAOF: what?
soreau: xorg-edgers is a repo that provides latest user space bits
meoblast001: ok, i'll use that
invalidrecord: hi guys is the mobility radeon hd 5470 an r600/r700 need to know if i need indirect rendering for compiz
RAOF: No.
soreau: invalidrecord: It's an r800 and no 3D yet with the radeon driver
RAOF: Actually, EPARSE. What's your question?
invalidrecord: thanks guys, im on fglrx :( keep up the good work it is much appriciated
invalidrecord: ohh also when did amd aquire ati was it recent?
invalidrecord: gimf
soreau: nearly two years ago now?
invalidrecord: 2006 omg
meoblast001: RAOF: the guy who compiled these seems to have stripped the libraries
meoblast001: :/
RAOF: Yes, of course.
meoblast001: #0 0x0000000000000000 in ?? ()
RAOF: Install the -dbg packages if you want the symbols.
meoblast001: RAOF: i don't know which one that is, libgl1-mesa-glx-dbg didn't work
RAOF: libgl1-mesa-dri-dbg is probably what you're after. That's got the symbols for the dri drivers.
meoblast001: k
meoblast001: RAOF: nothing
meoblast001: i'll just use good old printf debugging
meoblast001: hm, dies in glCreateShader
meoblast001: i blame Mesa
meoblast001: i get no GL errors
meoblast001: nvm
hifi: bzflag flickers on RV740 and KMS
hifi: RV730*
hifi: if I force indirect rendering it doesn't flicker but is a bit slower (it does AIGLX?)