Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2009-12-29

Search This Log:

Nightwulf|work: hi all
Demo1: hi
scarabeus: http://paste.pocoo.org/show/160299/
scarabeus: yesterdays mesa
scarabeus: xserver is
scarabeus: anything else i should specify to help you guys see what is borked? :]
Nightwulf|work: scarabeus: do you use a very recent libdrm and mesa<7.7 and KMS?
Nightwulf|work: very recent means 2.17 to be exact
scarabeus: nope about kms
scarabeus: libdrm is indeed latest
scarabeus: mesa is yesterdays git
Nightwulf|work: ok...not exact the combination i had...this null pointer occured for me when using mesa 7.6.1 together with libdrm 2.17 and KMS
Nightwulf|work: since i updated to mesa 7.7, libdrm 2.17 with latest radeon head git and no KMS, it works like a charm (well, some kwin redraw bugs left besides)
scarabeus: Nightwulf|work: by any chance using raster on your qt? it might cause few drawing glitches
Nightwulf|work: raster?
scarabeus: in that case answer is nope :P
Nightwulf|work: well, if you tell me, what you mean with "raster", i could try ;-)
soreau: scarabeus: Why aren't you using kms? With latest components, compiz chokes here without it
scarabeus: Nightwulf|work: it will just break stuff for you :P
scarabeus: Nightwulf|work: it is another rendering engine for qt
scarabeus: possibly faster
scarabeus: soreau: dunno kwin works peachy
Nightwulf|work: ah, i c
soreau: scarabeus: What kills it for you?
scarabeus: soreau: dunno why it crashed
soreau: It was an isolated incident?
scarabeus: i was minding my own stuff and then
scarabeus: nope i got it for 3rd time in few weeks so it is not entirely isolated
soreau: If it's not reproducible I don't think it will get much attention especially if it's with dri1
soreau: My recommendation would be to enable kms
scarabeus: soreau: there is no gallium for r600 cards anyway, so even with kms it would die in the same mesa code anyway
Nightwulf|work: soreau: kms worked for me until libdrm 2.16...after that i get the same null pointer exception, scarabeus posted above
dileX: hi Nightwulf|work
Nightwulf|work: hi dileX
soreau: scarabeus: You wont know that until you try it. kmks uses different paths
knoppers: Hello
knoppers: using most recent libdrm and ddx and kernel and the screen is flickering a little bit
knoppers: in X
knoppers: RV630
knoppers: after a while you dont see the flickering anymore
knoppers: because your eyes or brain adjust :D
Narigo: can someone in here help with setting up dual head with the ati radeon driver?
Narigo: i can set up my two monitors like this via xrandr: LVDS (laptop monitor) on the left, HDMI-0 (external) on the right. the external one is being seen as the primary monitor. i want the LVDS to be the primary one
MrCooper: Narigo: xrandr --output LVDS --primary
Narigo: that doesn't help :(
Narigo: kde taskbar is still on the HDMI monitor... :S
MrCooper: probably a window manager / client issue then
MrCooper: probably a KDE issue then
Narigo: so i should try to use a different wm?
MrCooper: that probably won't affect the taskbar
Shuren: hello, i provided logs from my situation: mesa 7.6.1 make 3d slow, then 7.7 works well (both compiled with driver r600,swrast and --disable-gallium) with libdrm 2.4.17
Shuren: mesa 7.6.1: http://shuren.pastebin.com/m38f388d4 http://shuren.pastebin.com/m2efa63dc
Shuren: mesa 7.7: http://shuren.pastebin.com/m7909f9e7 http://shuren.pastebin.com/m6c632bb1
Shuren: xf86-video-ati are yesterday git, my hardware is a radeon hd 2400 pro, and i use kms
Narigo: MrCooper, i know it won't help with the taskbar in kde, but maybe in gnome or something different it's better then?
adamk: Shuren: Under 7.6.1, pastebin the full output of 'LIBGL_DEBUG=verbose glxinfo'
Shuren: here: http://shuren.pastebin.com/ddcfa3b6
mmp: hello; any progress with RS690 and DRI2 support; on anything I can help with by testing?
Shuren: and here mesa 7.7: http://shuren.pastebin.com/d2b275a5a
adamk: Shuren: I'm fairly certain that your Mesa and libdrm versions are out of sync.
adamk: Or perhaps it's just that Mesa wasn
Shuren: mmm, what libdrm want mesa 7.6.1? :)
adamk: Err... Wasn't built against the correct libdrm...
adamk: Not sure, honestly. I just use mesa from git.
Shuren: i try
Narigo: MrCooper, just tried: gnome works
MrCooper: right, that's my experience as well
Narigo: do you have a suggestion what DE or WM i could try else? i don't really like the gnome look&feel
Narigo: the strange thing about kde not working is: with the proprietary drivers it used to work - maybe that's related to the labels of the monitors? they are called differently with each driver, right?
Shuren: adamk, thank you, you are right:
Shuren: OpenGL renderer string: Mesa DRI R600 (RV610 94C7) 20090101 x86/MMX/SSE2 TCL DRI2
Shuren: OpenGL version string: 1.2 Mesa 7.6.1
Shuren: i was sure that i compiled first drm... too newbie sometimes, thank's again
moobie: hi
moobie: I have a problem using kernel 2.6.33 with kms
moobie: If anyone have time, plz see my Xorg.0.log
moobie: Xorg.0.log with kernel 2.6.33: http://pastebin.com/m54edb7d0
moobie: With kernel everything was ok
Nightwulf|work: we're at ;-)
moobie: I have tried latest git libdrm,mesa and ddx without luck
jcristau: moobie lives in the future
Nightwulf|work: ah...i c ;-)
moobie: I meant :)
moobie: And the 9, is the Ubuntu version number
moobie: ;)
moobie: It like my libdrm and the drm in kernel doesn't match?
Nightwulf|work: please try libdrm 2.16 and leave all other versions as they are
Nightwulf|work: I'm eager to see wether that works like it was for me
adamk: Shuren: Glad you got it working.
adamk: moobie: Sounds like xf86-video-ati wasn't built with support for KMS.
moobie: Nightwulf|work, I use libdrm git? Is that too recent?
moobie: adamk, well if I boot my 2.6.32 kernel, without touching my ddx, everything works fine
adamk: moobie: But was libdrm built with KMS support?
Nightwulf|work: well, i used 2.16 and it stopped working with an update to 2.17
moobie: adamk, yes sir
Nightwulf|work: when i got back to 2.16 it worked again
moobie: Nightwulf|work, but why does it work with an earlier kernel then?
er0x: r600?
Nightwulf|work: moobie: it throws a null pointer exception for me even on 2.6.31.y (as described in the build howto)
Nightwulf|work: er0x: for me on rv635 and rv790
moobie: er0x, r700
er0x: 32.2 dont like avivo r600+ cards
er0x: with kms
moobie: Nightwulf|work, well I don't see any errors in dmesg
moobie: http://pastebin.com/m7b2992c0 <- kernel 2.6.33 log
moobie: Xorg.0.log with kernel 2.6.32: http://pastebin.com/m5ab92eee
moobie: Xorg.0.log with kernel 2.6.33: http://pastebin.com/m54edb7d0
Nightwulf|work: moobie: k, seems to be a different bug/behaviour then
moobie: Seems others have the same problem:
moobie: http://www.phoronix.com/forums/showthread.php?t=17603&page=10
moobie: See the first post
moobie: Anyone know when one of the developers are on the channel? :)
moobie: hi Zajec
moobie: How's the developing of powermanagement
GoGi: I have installed libdrm, radeon driver and mesa from git
GoGi: do I need anything else to properly test 3d on r700?
GoGi: I currectly get "do_wait: drmWaitVBlank returned -1, IRQs don't seem to be working correctly. Try adjusting the vblank_mode configuration parameter."
GoGi: and glxgears have clearly wrong colors
BioTube: IRQ code wasn't merged into the mainline kernel until the 2.6.33 cycle and requires additional firmware not allowed into the kernel
hifi: that line is irrelevant and can be ignored
GoGi: firmware not allowed into the kernel even in 2.6.33?
GoGi: ok
Badiss: i all !
GoGi: the main gear is violet instead of red
BioTube: kernel policy is no new firmware
Badiss: i'm trying to setup compiz with my ATI RadeonXpress 200M
Badiss: with AIGLX
stikonas: linux-libre maintainers will be bored now...
Badiss: and i've got the error AIGLX : 3D claims not support 0x23
Badiss: till 0x32
GoGi: BioTube: why?
Badiss: any idea ???
BioTube: GoGi: no idea
Badiss: ok thanks
adamk: Badiss: That's not a problem. Just an informational message.
e7th04sh: hi, i foun "(WW) RADEON(0): Direct rendering disabled" in /var/log/Xorg.0.log
e7th04sh: what can i check now, to find the source of the problem
e7th04sh: (btw glxinfo reports "driect rendering: yes"
BioTube: e7th04sh: we need the whole log to make heads or tales of that message
BioTube: e7th04sh: the software renderer can do direct rendering
e7th04sh: so it doesn;t mean i get hardware accelaration?
BioTube: no, it doesn't
e7th04sh: ok, that explains a lot, many people thought it means exactly that
e7th04sh: i shall paste the log somewhere
BioTube: it did at one point
Badiss: adamk : do you think it might raise an issue for compiz ??
e7th04sh: it has 1.5 k lines, but here it is : http://pastebin.ca/1730656
e7th04sh: if you want it in some other form, just tell me
adamk: Badiss: Not at all.
adamk: Badiss: Newer versions of the driver just supress those warrnings.
Badiss: ok thank you adamk : )
e7th04sh: btw, it all worked fine before i tried to upgrade from debian lenny to squeezwe
BioTube: e7th04sh: does dmesg list anything about vgaarb?
adamk: e7th04sh: Pastebin 'dmesg | grep drm'
e7th04sh: BioTube, no
e7th04sh: adamk http://pastebin.ca/1730662
e7th04sh: there is an error present there
BioTube: debian needs the firmware-linux package installed
GoGi: i have recompiled everything and still wrong colors
GoGi: is this a known problem?
e7th04sh: thank you, i am now installing the package
e7th04sh: but it's strange since everything used to work... was that a change between lenny and squeeze or did i get the package uninstalled during upgrade? :/
BioTube: eth04sh: starting with squeeze they finally decided to remove the firmware from the kernel package
e7th04sh: damn, this means that my problem shouldbe easy and obvious for someone who knew that...
e7th04sh: but they couldn';t help me over at #debian
e7th04sh: thank you for help
e7th04sh: i suppose i have to reboot to see the change?
BioTube: closing x and reloading the module should be enough
e7th04sh: thank you then, and hopefully i won't have to bother you anymore :) bye
Eddi|zuHause: is there some info on which wine registry key settings work best with the radeon driver on r300? (http://wiki.winehq.org/UsefulRegistryKeys <- HKCU/Software/Wine/Direct3D)
Ronis_BR: airlied: man, for some reason, radeon driver + KMS becomes much faster with compositing after my last update. I still getting a very bad score at glxgears, but now I can use my system with compositing on without problems. congratulations :)
moobie: Any devs in here today?
hifi: many
moobie: :)
moobie: hifi, are you able to help me debugging my problem?
moobie: http://pastebin.com/m7b2992c0 <- kernel 2.6.33 log
hifi: probably not, but state your problem here and someone else might be able to help
moobie: moobie: Xorg.0.log with kernel 2.6.33: http://pastebin.com/m54edb7d0
moobie: My laptop startup with a black screen with kernel 2.6.33
moobie: Kernel 2.6.32 works fine
BioTube: do you have the new firmware?
moobie: Yeah
moobie: As of dmesg for kernel 2.6.33 there is no errors
moobie: BioTube, in my Xorg.0.log for kernel 2.6.33, there is some errors about libdrm and drm in kernel not the same version
moobie: But the weird part is, that it works with kernel 2.6.32 without changing any drm,radeon or mesa versions
moobie: BioTube, just to be sure. You do mean the new irq firmware?
BioTube: yes
marvin24: moobie: have you tried with KMS disabled?
moobie: marvin24, acutally not. Should I try that?
moobie: Well I actually think kms is enabled by default with kernel 2.6.33?
BioTube: you have to enable on-by-default
marvin24: moobie: yes - you have radeon.modeset=1 in your kernel parameters, change it to 0 and try again
moobie: BioTube, I haven't compiled my kernel myself
moobie: marvin24, if it works. What can we conclude?
marvin24: maybe some kernel bug
moobie: ok
moobie: Well something in the two Xorg.0.log's isn't right. They expose different ddx git versions?
moobie: How is that possible
adamk: They do? They both say module version = 6.12.99
marvin24: adamk: further down, there is some "Built from git commit 9d0f3af7278dc939fd4e6f3ea69d9f488a9fbed7" message
marvin24: (which is the current version)
adamk: INteresting. Only one of them shows that.
moobie: Hmm
adamk: But that doesn't necessarily mean that they are different versions.
adamk: moobie: Is this a generic vanilla kernel?
moobie: It is Ubuntu's mainline kernel
marvin24: ever further down, there is a version mismatch message - I guess that means that libdrm is to old
adamk: Yuck.
adamk: God only knows what Ubuntu does to the kernel.
moobie: marvin24, I got libdrm git
moobie: Git version radeon for 2.6.32: 9d0f3af7278dc939fd4e6f3ea69d9f488a9fbed7
adamk: Before even thinking of filing a bug report, you should probably reproduce this with a vanilla kernel.
marvin24: what does "dpkg -s libdrm2 | grep Version" say
moobie: marvin24, Version: 2.4.17+git20091221.fdb33d56-0ubuntu0sarvatt2
moobie: marvin24, but that doesn't matter. I replaced it with libdrm git from yesterday
moobie: adamk, well you are probably right
marvin24: moobie: how?
moobie: marvin24, sudo make install
moobie: didn't use the package manager
marvin24: did you compiled with --enable-radeon-experimental-api ?
moobie: adamk, you are probably right. I just hate building my own kernel.
moobie: marvin24, yes
moobie: marvin24, it works with kernel 2.6.32
moobie: :-)
moobie: Thats the weird part
marvin24: ok - no idea than - go compile a stock kernel
moobie: and frustrating part
marvin24: I'm also running 2.6.33-rc2 now - no problem
moobie: Could I just use the default .config?
moobie: marvin24, well with ubuntu config?
marvin24: no - my own
moobie: which distro?
marvin24: karmic
moobie: Can I borrow you config file?
marvin24: I'm using a very striped down config, which fits exactly to my system (probably not yours)
marvin24: you can try the ubuntu config
marvin24: but with the master of the kernel source
moobie: marvin24, ok
moobie: btw did you use: INSTALL_MOD_STRIP=1 CONCURRENCY_LEVEL=3 make-kpkg --initrd --append-to-version=-radeonkms kernel_image kernel_headers modules_image
moobie: to make the deb packages
marvin24: no packages here
moobie: Just sudo make install?
marvin24: just use: git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6
marvin24: and copy the /boot/config-2.6.xyz to .config
moobie: marvin24, ok I'll try
marvin24: make -j bzImage modules && make modules_install
moobie: thx. Haven't tried the non ubuntu way yet
chant: With KMS (w/ patch to remove IRQ call in crtc) & DRM/DDX/MESA head as of yesterday on a RS780, I get garbage when I start to use xorg.
chant: My guess is I should update KMS - should I use drm-next head, or the kernel.org head?
marvin24: chant: try + drm-radeon-testing
chant: What's drm-radeon-testing ?
marvin24: (I later patched to by removing the drivers/gpu and include/drm stuff)
chant: from libdrm?
marvin24: no from arlied tree
chant: ok.
Ghworg: As of a few hours ago (last time I fetched) the only different between kernel.org master and drm-next is some intel stuff
marvin24: Ghworg: true, but there is difference to 2.6.32.y
chant: If I just clone drm-radeon-testing, that's the same thing right?
Ghworg: drm-radeon-testing doesn't have any of the stable branch patches, and there is more bleeding edge drm code in there, but pretty much the same
taiu1: agd5f: afaics for vert_array_bgra to work needs smth like http://cgit.freedesktop.org/~andrem/mesa/commit/?h=r600-test&id=71c235a95fc5b260f7455b6ba9b8ddaa39adf968
DanaG: hmm, can HD3200 do two independent LVDS interfaces, for things such as those dual-screen laptops?
zaphire: problem: when starting x, it doesn't start, but gives this output:
zaphire: http://pastebin.com/f14db1237
zaphire: vga is ATI Rage 128 Pro Ultra TF
zaphire: [Enrico]: checked with livecd, card is ok
[Enrico]: zaphire: no clue sorry :'(
zaphire: i'm using "r128" driver, but it's the same with "ati"
BioTube: ati is just a wrapper
zaphire: than that's why
DanaG: hmm, what would be expected behavior for an RV200 with no connected monitors?
DanaG: Right now, Radeon just gives up. I guess that makes sense...
Hackus: hello
Hackus: Isn't /usr/X11R6/lib/modules/dri suppose to contain dri modules for X?
Hackus: My /usr/X11R6/lib/modules/dri directory is empty for some reason.
Hackus: In fact, I don't have a dri directory.
Hackus: I thought mesa was suppose to have the radeon dri modules when you build it from GIT?\
knoppers: my X is installed to somewhere else
knoppers: /usr/lib/xorg/modules/dri
knoppers: and it contains the drivers or whatever it is
knoppers: the .so
knoppers: r300_dri, r600_dri, radeon_dir, ...
Hackus: I cannot locate the X dri modules in the mesa tree from git either.
knoppers: Hackus, maybe you can look for the files to find the .so
knoppers: swrast_dri.so
Hackus: Ok, I found them.
Hackus: Anyoen know if the radeon_dri.so modules are built in the mesa tree or are they built in the current X tree?
Ghworg: mesa
Hackus: I just built the mesa tree from git, and I can't find the modules.
Ghworg: $ dpkg -S /usr/lib/dri/radeon_dri.so
Ghworg: libgl1-mesa-dri: /usr/lib/dri/radeon_dri.so
adamk_: Hackus, If they were built properly, they'd be in lib or lib64 (depending if you are using a 32 bit or 64 bit system) at the base of the source tree.
Hackus: After builing, the only thing I have in the lib64 directory is a gallium directory and a bunch of libGL, libGLU, and libGLw, libOSMesa.
adamk_: How did you build mesa?
Hackus: I pulled the tree down from git, and basically did a make linux-x86-64 in the git tree
adamk_: You should use the autotools to configure the build.
adamk_: And then run 'make'
Hackus: The directions don't tell me to do that though.
adamk_: linux-x86-64 is a target that does not support DRI.
adamk_: What instructions would those be?
Hackus: http://dri.freedesktop.org/wiki/Building
adamk_: There is probably an old style target for x86_64 with dri.
adamk_: Indeed, there is linux-dri-x86-64.
Hackus: It says once you download the git tree, you select the architecture and do a make.
adamk_: But I still recommend autotools.
Hackus: yeah, but that is for amd.
Hackus: I have a intel.
adamk_: And I say not to :-)
NForce25: hi all, is KMS supposed to be working with RS690 card?
Hackus: Well, which architecture should I pick then?
chithead: NForce25: yes. rs690 is igp though, not a discrete card
adamk_: Hackus, x86_64 is x86_64 is x86_64. Unless you are using intel's old 64-bit architecture (which, trust me, you're not) then you are using amd64 :-)
adamk_: Hackus, And, again, i recommend using the autotools.
adamk_: Hackus, I suggest taking a look at the wiki in the topic for instructions on how to configure mesa with autotools.
NForce25: right, i called it wrong. But i get random artefacts and screen corruptions with kms enabled. What could be wrong? Kernel is 2.6.32, mesa 7.7, libdrm 2.4.17, xf86-video-ati from git
chithead: in what application? using opengl compositor?
Hackus: Shouldn't I just be able to do autogen?
NForce25: no application, in random places. yes, using compositor, dont remember if i tried without it
NForce25: for example mouse cursor dissappears and i have a huge rectangle in its place which i can move
NForce25: or all top part of the screen gets garbaged
adamk_: Hackus, You probably want to specify --with-dri-drivers so you don't spend time building more than you need, and you may need --with-dri-driverdir to specify where to install the drivers (which can vary from distribution to distribution),
Hackus: I just usually copy them manually from the build directories.
soreau: Hi guys, I am getting the following error http://pastebin.com/m1dbbadaf when trying to build a program and the developer is telling me it's GL/gl.h's fault because it's defining #define GL_VERSION_1_3 but not defining other things that sdl would define if GL_VERSION_1_3 was not defined
soreau: Does this make any sense and is it an easy bug to fix or something that will be fixed in the future (or not a mesa bug)
soreau: I used to be able to build this program last week, so I don't know if something changed in mesa or the program wrt this
cxo: but gl is under sdl, so that doesnt make sense
cxo: sdl might have to update what it #ifdefs with GL_VERSION_1_3
SnowRaptor: Hey there! I'm using yesterday's git kernel with airlied's patches in an xpress 1150, KMS enabled
SnowRaptor: weird thing is: I randomly get flickering horizontal lines (a screenful of them) sometimes during the boot process, when KMS is enabled
SnowRaptor: the screen then keeps lickering until it goes blank for powersaving
SnowRaptor: then, I press a key and it comes back from blank, with normal display and no more flikering
SnowRaptor: ideas where to search for bugs?
spreeuw: do you think it uses wrong refresh or so?
spreeuw: because there was someone else in here yday who had that
SnowRaptor: there are some "kernel: integrated sync not supported" in dmesg, but those messages also appear when there is no flickering
DanaG: [ 4.140278] WARNING: at /build/buildd/linux-2.6.32/drivers/gpu/drm/drm_crtc_helper.c:1032 drm_helper_initial_config+0x5b/0x60 [drm_kms_helper]()
DanaG: that's also no-monitor-connected condition.
DanaG: Interestingly enough, it never actually says anything to the effect of "no monitors connected".
evil_core: SnowRaptor: what airlied patches?
SnowRaptor: airlied/drm-2.6.git
SnowRaptor: drm-radeon-testing I guess
SnowRaptor: I gotta go now, I'll be back in a few hours
jwm: is mplayer acceleration available on rs780?
stikonas: yeah, it should be, only Evergreen cards are unsupported
jwm: cool
jwm: any docs on how to enable it?
stikonas: mplayer should use XV by default on new distributions
jwm: I use gentoo heh
jwm: maybe I didn't set the xv use flag
jwm: nah it is set
adamk_: jwm, Add 'vo=xv' to ~/.mplayer/config
adamk_: Of course, this all supposes that you have acceleration working.
chithead: mplayer uses xv by default, even on gentoo
chithead: verify with xvinfo that xv works
jwm: yeah it works
jwm: damn I wish radeon supported XvMC
BioTube: i don't think anybody does
jwm: nvidia does right?
chithead: intel does
chithead: but the gain is questionable, mpeg-2 does not place high burden on cpu
jwm: so h264 can't use xvmc?
mentor: I want to hook up mesa/DRI to X logging. However, there doesn't seem to be any generic X11 headers for this. Would it be bad to pull in an inluce/xorg/os.h dependency?
Hackus: Yeah, looking at the code for that stuff logging would have to be done VERY carefully.
mentor: In terms of external dependencies? or?
Hackus: In terms of what operations of MESA and the DRI infrastructure you want to log.
Hackus: Did you have an idea what operations you want to log in MESA/DRI?
Hackus: I have a suggestion. :-)
mentor: Hackus: Well, I was mostly aiming at the fprintf's already there, as I can't seem to get them to output
Hackus: You can't use the stream library to print out DRI operations in kernel space.
Hackus: (Clib).
Hackus: Frankly, what I would do is make a /proc directory and update the stats of DRI system, and MESA.
Hackus: That way user space tools could use the already existing logging API's.
foreste: hi
foreste: i problem on 3d ati radeon x200m is debian
foreste: debian version 6 squeeze
foreste: new kernel 2.6.30-2
foreste: i not instaling fglrx 9.3 in new kernel
mentor: Hackus: I'm looking at the DRI drivers portion of mesa (i.e, userspace only)
foreste: what instaling open driver radeon x200m in debian ?
mentor: Sorry, I keep losing this thread of conversation
Hackus: I will second that. Installing fglrx these days seems to be asking for trouble.
Hackus: I noticed after rebuilding my mesa libs too, that all of my opengl screensavers stopped working.
foreste: please help me ;]
soreau: foreste: Pastebin your X log
jwm: are you guys switching back to fglrx's opengl?
jwm: heh
foreste: soreau: fglx my card not support in 9.3
soreau: foreste: This channel is for the open source drivers. We might be able to help you get them working but first you need to pastebin your /var/log/Xorg.0.log file
foreste: i install opendrive 3d acelared my card ?
soreau: Yes, the open drivers provide 3D for your card
agd5f: taiu: looks good
foreste: u not seach power splly notebook :(
foreste: i
foreste: supmly
foreste: pastebin log 30 minutes or 1 h
nightmorph: do you radeon devs prefer to see the latest mesa (7.8-git) in wiki/RadeonProgram or current release (7.7 stable)?
nightmorph: because afaik it's always supposed to be the latest results, which is why that KenMays guy is pissing me off by overwriting my results with outdated (7.7, 7.6) mesa versions
MostAwesomeDude: We prefer either latest stable, or git, whichever. We don't really read it; it's more for you guys than for us.
MostAwesomeDude: But I can put a note up if you like.
nightmorph: i sometimes wonder if it's worth splitting it out into two wikis -- one that tracks stable (which most distros package) and one for the latest git checkouts
nightmorph: because time and time again i have to revert ken's edits and replace them with something relevant, or what i consider relevant
nightmorph: MostAwesomeDude: sure, sounds good, but there are already so many notes at the top i doubt anyone reads 'em :(
amarsh04: yay, mesa 7.6.1 is in Debian unstable
chithead: "yay"? 7.7 is latest release
MostAwesomeDude: You could probably scrap most of them.
MostAwesomeDude: The notes at the top, I mean.
amarsh04: it's better than what was there before, chithead
nightmorph: MostAwesomeDude: i'll take another look, then
MostAwesomeDude: nightmorph: What might be useful is going through the footnotes. There's a handful that no longer apply.
nightmorph: true; keeping those up-to-date is hard, but most of them are for specific system tests
nightmorph: e.g. not stuff that i've tried
Wizzup: r300g ran battlefield1942 through wine succesfully, with some missing graphics :)
MostAwesomeDude: Heh, awesome.
Wizzup: I cannot test on my r600, since mesa is 64 bit, and wine needs 32 bit ones. Still having a hard time building the 32bit libs
wind-rider: hi
wind-rider: i'm using Kubuntu 10.04 with the latest packaged open source radeon driver
wind-rider: on a Radeon 4530 HD
wind-rider: (laptop)
wind-rider: and i get quite a few artefacts in my screen on gui elements
wind-rider: you can see it on http://launchpadlibrarian.net/37279305/snapshot2.png
wind-rider: (i created a bug report at https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-ati/+bug/500607)
wind-rider: but can anybody see if this is a bug in the radeon driver? or in qt? or in kde?
er0x: i get 'em too sometimes
er0x: hit refresh and all be fine
wind-rider: i can not test with the fglrx driver because that one does not support the x server shipped in Kubuntu 10.04 yet
er0x: so far looks like its 2d problems
wind-rider: er0x: if I mouse over the artefacts in the taskbar, they indeed go away
er0x: with opengl its fine
wind-rider: er0x: should i post the bug-report somewhere else?
er0x: i think its known bug
er0x: atleast you arent first to tell it
wind-rider: er0x: i hope somebody can pick it up, then :)
er0x: for temproraty fix u can use kwin's opengl effects :D
wind-rider: er0x: doesn't that crash everything?
wind-rider: er0x: i'll try it :)
er0x: 3d works fine here with kms
wind-rider: er0x: when enabling desktop effects with opengl and texture from pixmap, it disables itself again
er0x: with any error message?
er0x: if not hit resume in general section
er0x: its litle buggy :d
wind-rider: er0x: strangely enough, after a few times trying, it is enabled now
er0x: yah bugs
wind-rider: er0x: but the artefacts are not away
wind-rider: er0x: still black boxes over text in the taskbar and the system tray is orange
er0x: hmm here it fixed my artefacts
wind-rider: er0x: maybe plasma should be restarted
er0x: u can try that
wind-rider: er0x: pity, still artefacts
er0x: then no ideas
soreau: MostAwesomeDude: I was able to start X with radeong.. and start compiz.. though it was very slow and I had to ssh in to toggle monitor on/off to get it to show something other than black screen - glxinfo was reporting software rasterizer..
wind-rider: er0x: is there a bug reporting site for the radeon driver?
soreau: I have never seen compiz run with swrast and everything not be white until now
er0x: wind-rider: check bugzilla on x.org
Ghworg: https://bugs.freedesktop.org/
wind-rider: thank you :)
wind-rider: er0x: i filed a bug report at https://bugs.freedesktop.org/show_bug.cgi?id=25825
wind-rider: er0x: i must leave now, thx for your help
soreau: Here are some shots I took of enemy territory with the gallium driver http://www.speedyshare.com/settings.php?id=606609711&idh=7f96d7e96bdd8b29f42f4a6b36c3977f this is a big step up from what it used to be, in both performance and graphics (note, it might not be obvious this is not what textures the game uses)
foreste: i moment pastebin log my ;]
ferret_: I have an awesomely stupid question
ferret_: Is this file necessary in order for KMS to work? : /usr/lib64/xorg/modules/drivers/modesetting_drv.so
ferret_: A gentoo developer in his infinite wisdom disabled the building of that file in the latest mesa ebuild version
Ghworg: ferret_: I don't have that file and KMS works fine for me, so apparantly not
ferret_: wonders what it does, then...
ferret_: Ghworg: do you have any mention of modesetting_drv.so in /var/log/Xorg.0.log ? You might have the file in a different place, but when it loads it is mentioned there
Ghworg: ferret_: No, nothing
Ghworg: ferret_: I did a find of /usr for that filename and it didn't find anything and all the radeon stack is installed in /usr
Ghworg: Ah, it looks like it is related to gallium somehow and I don't have that enabled
Ghworg: src-ext/mesa/src/gallium/winsys/drm/radeon/xorg/Makefile:TARGET = modesetting_drv.so
NForce25: hello there. Can anybody take a look at the madness which happens if i enable kms on my rs690? Here are the screenshots:
NForce25: http://ikelk.lt/storage/images/000/015/116/15116/original_img.png
NForce25: http://ikelk.lt/storage/images/000/015/117/15117/original_img.png
NForce25: http://ikelk.lt/storage/images/000/015/118/15118/original_img.png
NForce25: http://ikelk.lt/storage/images/000/015/119/15119/original_img.png
NForce25: Xorg.0.log is here: http://pastebin.com/f7bca0265
NForce25: I have amd64 box, mesa 7.7 (multilib), libdrm 2.4.17, xf86-video-ati from git. Did i configure something wrong or should i post a bug?
bartek: Solefald: log xorg paste ?
bartek: soory
foreste: soreau: http://pastebin.pl/15590 my log xorg
foreste: what actyvation open driver ati radeon x200m ?
foreste: help please ;]
NForce25: everyone seem to be sleeping :/
DanaG: hmm, for one thing, there seem to be some language-barrier issues going on, too.
foreste: ok i using translate ;]
NForce25: foreste can you speak russian?
foreste: i speak polish ;D
NForce25: right, i dont understand that. but we're neighbours
DanaG: Anyway, what's specifically not working with the default drivers for your video card?
foreste: ok moment open translate and continue speak ;]
foreste: Driver beguile with (from) kernel ati and it does not argue my card (kart) already radeon ati
foreste: Show xorg.conf?
foreste: my system debian 6 squeeze
DanaG: Please copy /var/log/Xorg.0.log into a pastebin -- there's a "pastebinit" package that can do it for you.
DanaG: Install "pastebinit" package, and then run: "pastebinit /var/log/Xorg.0.log
DanaG: oops, screwed up my quote marks on that second one.
adamk_: He did.
adamk_: http://pastebin.pl/15590
DanaG: ah.
foreste: http://pastebin.pl/15590
foreste: ;d
adamk_: foreste: There is some problem with your Xorg installation.
foreste: what ?
adamk_: Did you ever try to install fglrx on that machine?
DanaG: hmm, the fact that it never TRIES dri... makes me think perhaps xorg.conf is screwed up.
adamk_: Nah, it's all because of the error from loading the glx module.
adamk_: All the hits on google show people trying to use fglrx.
adamk_: Which, of course, won't work on his GPU.
DanaG: righty-oh. Might be nice to have fglrx itself check GPU during installation pre-config. =þ
foreste: Then, < sweat > I fitted so odinstalowałem because it does not serve this card (kart) new fglrx
foreste: unistall*
adamk_: I don't understand that. But you need to completely remove fglrx and then reinstall xserver-xorg-core, libgl1-mesa-dri and libgl1-mesa-glx
foreste: my notebook benq joybook a52 ;]
foreste: adamk_: ok
foreste: lol
NForce25: adamk_ or DanaG, maybe you can help me with my problem? i mentioned it earlier
adamk_: Sorry, I really don't know.
foreste: 2 modules fglrx dont uninstall
_KAMI_: i
DanaG: hah, adding input device "Broadcom Corp"? So the mouse is the corporation? Nice.
DanaG: =þ
foreste: Now become (stay) unistall
NForce25: DanaG: maybe the name is messed up, but i have no problems with mouse or touchpad :) KMS doesnt work though
foreste: Make mute now fglrx?
adamk_: Make mute? You must completely remove it.
SnowRaptor: And I'm back!
foreste: adamk_: soory stupid my translate ;x
SnowRaptor: I'll repost my issue
adamk_: foreste: If you installed it via synaptic or apt-get, you should remove it that way. If you downloaded it from the AMD website, there should be an uninstall script in /usr/share/ati/
SnowRaptor: I'm using yesterday's git kernel with airlied's patches in an xpress 1150, KMS enabled. Weird thing is: I randomly get flickering horizontal lines (a screenful of them) sometimes during the boot process, when KMS is enabled
foreste: adamk_: i install fglrx in repro debian
SnowRaptor: the screen then keeps f lickering until it goes blank for powersaving. then, I press a key and it comes back from blank, with normal display and no more flickering
SnowRaptor: ideas where to search for bugs?
foreste: i dont using scripts amd site
NForce25: SnowRaptor: you have similar hardware as me, and a bit similar error as me. http://ikelk.lt/storage/images/000/015/117/15117/original_img.png maybe that look like yours?
SnowRaptor: NForce25: no, much worse. the screen is completely covered by alternated black white horizontal stripes, no more than 10px high, flickering
SnowRaptor: The problem is that the problem is random
SnowRaptor: or seems to be handom, I couldn't notice a pattern
SnowRaptor: so I don't know how to reproduce it
NForce25: my problem starts about 1 minute after logging into kde and keeps going all the time :(
SnowRaptor: mine not. Used to happen when I switched to console with KMS off
SnowRaptor: I can get rid of it unless I wait until the screen goes blank and then come back from blanking. Thenm, picture gets fine and stays so.
NForce25: seems like i am the only one with my problem, everyone else got that working and i am still waiting while someone fixes it. Maybe i should post a bug report?
SnowRaptor: That's my intent, but I'm trying to gather useful information. I can't report a bug saying "screen randomly gets psychedelic on boot, comes back from the trip after video blank"
SnowRaptor: or, worse"
SnowRaptor: "sometimes screen psychedelic on boot, comes back from the trip after video blank"
SnowRaptor: the "sometimes" part is tricky
NForce25: maybe your mesa is old? With mesa 7.6.x KMS was even more useless for me
SnowRaptor: I'm using the klatest mesa from git
SnowRaptor: or, at least, sunday's latest
NForce25: havent tried a stable one?
SnowRaptor: no, stable ones don't even work
SnowRaptor: X11 segfaults when I run glxgears
SnowRaptor: xpress 1150 (or 220M) is the buggiest IGP from ATI
spreeuw: hehe
NForce25: strange. mine does work with 7.7. maybe i should try git too
NForce25: x1250 is a lot buggier i think :)
SnowRaptor: I guess it's the same
SnowRaptor: but, same family
SnowRaptor: those three are "flour from the same bag"
chithead: I think rs4xx is worse than rs6xx
spreeuw: two girls from the same cup
SnowRaptor: LOL
NForce25: are they? as far as i know, x1250 is rs690, 1150 is older, isnt it?
spreeuw: the manpage should list it
SnowRaptor: then never mind what I said
NForce25: and i am just curious, how much fps do you get from glxgears when they are working?
chithead: if vblank is working, glxgears fps should be equal to monitor refresh rate
NForce25: well, i think it isnt if i get ~150fps with lcd monitor
SnowRaptor: I get around 400
SnowRaptor: but I use compiz roo
SnowRaptor: *too
foreste: adamk: complete reinstall ;]
NForce25: it doesnt matter for glxgears if am using compiz (kwin) or not...fps is the same
mentor: So, I have an RV620, I think, and I'm running Linux (git drm-linus), libdrm, mesa, xf86-video-ati (git master) and every so often in X the screen blanks and the computer appears to lock up, no keyboard, network, or even ACPI power button events.
mentor: I amm using KMS, but it appears to happen with and without 3D
chithead: you can also tell if vsync is active by resizing the window. if fps stays the same then you probably have vsync
foreste: i rv 410 ;d
SnowRaptor: anyway, I'm more concerned with the flickering issue
GoGi: usb controllers have a standard interface
GoGi: is this impractical for graphic controllers?
mentor: Not even remotely
GoGi: ?
mentor: WEll
mentor: They do have a standard interface, they use PCIE
chithead: they have a standard interface, vesa vbe
mentor: Sorry
GoGi: a standard interface that supports all of the features
chithead: but no acceleration and limited choice of resolution
mentor: Yes, it is impractical; it's not even remorely likely
chithead: that is somewhat contradictory
mentor: ?
GoGi: you mean it's unlikely to ever happen?
mentor: yes
GoGi: why is it impractical?
mentor: Because USB devices do standard, repeated, small things
mentor: Graphics controllers are more like an entire computer, and not a simple one like a x86 PC
mentor: Which is not simple;
GoGi: and we want to draw triangles?
GoGi: ah and it is not possible to put all the software as firmware on the graphics card?
GoGi: if they are already as capable as an entire computer?
SnowRaptor: GoGi: are you looking for a possibility to get rid of drivers?
GoGi: hm well you still need one driver
Droste: isn't that exactly what the gallium drivers are trying? universal software interface? same for all vendors?
GoGi: I meant a universal hardware interface
mentor: As I said, PCIE
jimenez: does my gpu support KMS
jimenez: os[Linux 2.6.32-ck2 i686] distro[Ubuntu "karmic" 9.10] cpu[2 x Genuine Intel(R) CPU T2300 @ 1.66GHz (GenuineIntel) @ 1.66GHz] mem[Physical: 1006.7MB, 67.1% free] disk[Total: 36.7GB, 79.6% free] video[ATI Technologies Inc M56P [Radeon Mobility X1600]] sound[HDA-Intel - HDA Intel]
BioTube: jimenez: KMS should support your chipset
jimenez: but is enabled by default compiled in kernel ?
BioTube: it's always compiled in, but by default you've got to enable it with radeon.modeset=1
jimenez: ohohohoh
jimenez: i was using modeset=1 only
poot: takes a shit
poot: !ops i gotta take a dump.
PinkFreud: uhgh
PinkFreud: not this dumbass again
PinkFreud: poot was just booted from #debian for trolling
poot: feces
mentor: MY powers don't work in here
mentor: It was a bit of a shock to me, actually
PinkFreud: snerks
PinkFreud: now there's the way to do it. :)
Hackus: Good, now that we got that crap out of the way...back to open source!!
PinkFreud: mentor: what, your ability to kick him from #debian, or your inability to kick him from here? :)
Hackus: :-)
mentor: Yes, I thought I was still over there
PinkFreud: lol
Ghworg: Since 2.6.29 I've been using 64-bit kernels because of a really obscure suspend bug that only affects 32-bit. Latest drm code has introduced a different suspend bug that only affects 64-bit
Ghworg: bangs his head against the wall
jimenez: nope
jimenez: is not working
jimenez: no kms
jimenez: no framebuffer
BioTube: the framebuffer is separate
jimenez: the system just hangs
chithead: kms requires framebuffer console enabled in kernel
BioTube: modprobe fbcon
jimenez: all is compiled in
jimenez: the kernel is
jimenez: .32
Ghworg: Missing firmware possibly?
jimenez: i dont need │ │ < > Userspace VESA VGA graphics support │ │
jimenez: or │ │ [ ] VESA VGA graphics support │ │
Wizzup: no vesa,no
jimenez: well all is configuredok
Hackus: Looks like there is significan regressions in in 2.6.31 vs 2.6.29 looking at the software rasterizer.
Hackus: Around 15%.
mentor: are
mentor: Hackus: Is the sw. rasterizer in the kernel?
cxo: i thought its in mesa
mentor: AOL
Hackus: Oh sorry.
Hackus: No, the software rasterizer I think is a Xorg module.
BioTube: Hackus: it's in mesa
Hackus: Oh, part of mesa.
Hackus: We have a LONG way to go me things as far as a decent accelerated display architecture for Linux. The code is depressing, and exilerating at the same time.
Hackus: The regressions for example from 2.6.29 to 2.6.32 is depressing.
BioTube: that's what KMS and gallium are about
Hackus: But, AMD opening its hardware is exilerating and hopeful.
spstarr: looks at git
erilliam: I'm looking at the RadeonFeature matrix, and it looks like if I build from source I should be able to play OpenGL 2.0 games if I build from source? Is that correct? (I'm on R600 and Fedora 12)
cxo: well you're not going to play anything cool on wine yet
cxo: but yeah, the support is sorta there
tavl: anyone here using opensuse + kms + radeon driver?
erilliam: cxo: Thanks, I'll give it a shot! I'm thinking about a native game (Heroes of Newerth)
cxo: you need to build 2.6.32, drm, mesa and x86-video-ati
tavl: cxo, this was to me or erilliam?
erilliam: Sounds like fun. Hopefully won't mess up my system too badly :D
cxo: tavl, well that sort of the requirement to get anything working properly right now
cxo: erilliam, make sure you do it in that order too, and look at the configure --help options, you need to enable experimental api and what not on some of the packages
erilliam: ok, thanks
erilliam: When getting this stuff from git -- is there anyway to only get the latest rev (rather than the whole history) as to say I'm on an insanely terrible internet connection would be an understatement
cxo: yeah, --depth or something
cxo: look at the man git-clone
erilliam: ok, thanks
tavl: cxo, yeah, but i tried rebuilding some of those and broke my system completely... i think the problem was when updating xorg... what is the minimum version?
tavl: my xorg server is v1.6.5, x protocol is v11rev0 and kernel is i already know about 2.6.32, coz of KMS support... how to check DRM, MESA and x86-video-ati versions?