Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

RadeonHD IRC Logs For 2009-12-15

Search This Log:

Nightwulf|work: hi all
rindolf: Hi all. Can anyone help with this bug - https://bugs.freedesktop.org/show_bug.cgi?id=24172 - "x11-driver-video-radeonhd-1.2.5-1mdv2010.0 hangs / freezes after switching to virtual console and back." (only with DRI enabled).
adamk: Probably the only people who can help are the developers that chimed in on that bug report.
adamk: Have you tried the radeon driver instead?
rindolf: adamk: no, not yet.
rindolf: adamk: should I?
adamk: Well it might work where radeonhd failed.
rindolf: adamk: ah, OK.
udovdh: rindolf,
udovdh: I also have a 2600pro
udovdh: and run radeonhd from git on Fedora 12
udovdh: no text console switch issues here
rindolf: udovdh: ah.
udovdh: I have 2d and even 3d working...
udovdh: (more or less)
rindolf: udovdh: ah.
rindolf: udovdh: OK.
rindolf: udovdh: where can I find instructions on how to build it?
udovdh: dunno what could cause it if you run git as a try
udovdh: in the weiki, moment...
udovdh: wiki
rindolf: udovdh: I'm running the Mandriva driver.
udovdh: http://wiki.x.org/wiki/radeonhd#head-f79351b4e2b19fad40529ce297ac2d2a1e90354c
udovdh: I know
rindolf: udovdh: is it here? http://www.x.org/wiki/radeonhd
udovdh: but trying git is easy
udovdh: radeonhd had the text console issue long ago
udovdh: I run it since 1.0.0
rindolf: Hi all.
rindolf: udovdh: I'm now using radeonhd, and still have the text console issue.
rindolf: udovdh: but I've 3-D (sort-of)
udovdh: mesa 7.6?
rindolf: Extreme Tux Racer kinda works , but its
udovdh: kernel 2.6.32?
udovdh: foobilliard?
rindolf: Extreme Tux Racer kinda works , but it's slower than on my GeForce 4 MX.
rindolf: Well, what I used to have on it.
udovdh: neverball?neverputt?
rindolf: udovdh: let me see.
udovdh: it is all early support
udovdh: can improve
yangman: rindolf: what radeonhd version?
udovdh: maybe try mesa-demos ?
udovdh: if these work then 3d is ok
udovdh: dmesg is not showing stuff you don't need?
rindolf: kernel 2.6.32-desktop-1mnb
udovdh: 2.632 is ok
rindolf: mesa-7.6.1-0.rc3.2mdv2010.1
rindolf: Turns out I was using the "ati" driver all along.
rindolf: Back when I had DRI disabled.
udovdh: 7.6.1 is ok
rindolf: gears yields 412 FPS.
udovdh: kinda slow?
udovdh: 1500 here on a loaded system
yangman: rindolf: well, if it's a radeon issue, you should fix up the description and reassign the bug
rindolf: yangman: git master.
udovdh: phenom 9550
yangman: rindolf: in any case, we can't diagnose this unless you provide us with a log like I asked, since nobody else can reproduce it
rindolf: yangman: log of what?
yangman: rindolf: Xorg.0.log from a failing session
rindolf: yangman: ah.
rindolf: yangman: would it be OK when X is still running?
rindolf: The bug in question is this one - https://bugs.freedesktop.org/show_bug.cgi?id=24172
rindolf: yangman: ping
rindolf: yangman: would it be OK when X is still running?
yangman: rindolf: that wouldn't be as useful as one from where you reproduce the bug
rindolf: yangman: sorry - got disconnected.
rindolf: And then the server was split.
rindolf: yangman: ah, so I should switch to a virtual console and back and then reboot and copy the /var/log/Xorg.0.log ?
yangman: you said you can still ssh in, so grab it that way
yangman: and the output of `dmesg` as well
rindolf: yangman: at which point?
rindolf: yangman: ah, OK.
rindolf: yangman: OK, let me try.
rindolf: Hi.
rindolf: yangman: I've attached my Xorg.0.log and my dmesg file - https://bugs.freedesktop.org/show_bug.cgi?id=24172
yangman: rindolf: thanks. try using the radeon driver as well
yangman: rindolf: sometimes that's broken in one works in the other
rindolf: yangman: the radeon driver from git?
yangman: yeah
yangman: could just be AGP misbehaving, though
rindolf: yangman: OK, let me try.
rindolf: Hi all.
rindolf: yangman: with the "video-ati"/"radeon" driver I can see the wireframes of some of the windows (instead of a big black screen) but X and the keyboard are still hanged.
rindolf: yangman: ping
rindolf: OK, and now highlighting stuff in Firefox gives a wrong preview.
rindolf: At least here - http://en.wikipedia.org/wiki/Doom_WAD
rindolf: How come enabling DRI makes KDE 4.4-beta on Mandriva Cooker slower?
adamk: rindolf, Does /var/log/Xorg.0.log shed any light on the issue?
rindolf: adamk: I attached it to the bug report.
rindolf: adamk: https://bugs.freedesktop.org/show_bug.cgi?id=24172
rindolf: For radeonhd at least.
adamk: rindolf, I really would recommend the radeon driver, frankly. All the developers that work on the 3D Mesa driver use radeon, and most of them also work directly oh the radeon driver.
rindolf: adamk: but I'm getting the same problem with it.
rindolf: adamk: or a similar problem.
adamk: And I don't see anything in that log file to indicate a problem with KDE. You also didn't say if you are using compositing or not.
rindolf: And I'm tired of being ping ponged between video-ati and radeonhd
rindolf: adamk: this log file is not with KDE.
adamk: rindolf, The freezing when switching between console and back? Then open up a bug report for the radeon driver for that bug.
rindolf: It's a different problem.
rindolf: Hi all.
rindolf: According to the Gentoo Wiki radeonhd is dead - http://en.gentoo-wiki.com/wiki/RadeonHD
rindolf: Is it true?
adamk: Why do you think people keep tell you to try radeon :-) My "dead" is going a bit too far, but development on radeonhd certainly isn't as active as it is on radeon.
adamk: And the 3D drivers are really only tested by the developers with the radeon driver.
rindolf: adamk: people are telling me a lot of things.
rindolf: I don't know what to think.
adamk: Well you should think for yourself :-) Do a quick comparison here... This is the commit log for the master branch of radeonhd: http://cgit.freedesktop.org/xorg/driver/xf86-video-radeonhd/log/ And the commit log for the master branch of the radeon driver: http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/log/
adamk: rindolf, As you can see, xf86-video-ati (the radeon driver) sees a lot more development.
janj: Hi, I have a R700 series graphics card which should be identified correctly
janj: Except it isn't
janj: It's a mobility radeon though
janj: Any help on that appreciated
yangman: janj: GPUs aren't identified exactly. as long as the chipset is correct, it's fine, and it should be correct for r7xx
janj: yangman: my xorg log says the radeonhd couln't find a matching id in its device list
janj: the device lists the HD4670
yangman: what radeonhd version?
janj: 1.3.0
yangman: odd
yangman: ping the mailling list
janj: Ok.
janj: btw, in the drivers gpu id list it has separate entries for the mobility radeons
janj: except the list ends in the HD3xxx series for the mobility ones
adamk: You checked the source?
janj: It printed all of them in the xorg log
adamk: Well you could always try adding the PCI IDs to the driver yourself. Or, you could try the radeon driver.
agd5f: janj: what's the id?
janj: agd5f: what should the id look like?
agd5f: janj: lspci -vnn
agd5f: for your video card
janj: 1002:9488
janj: ?
agd5f: janj: that's it and both drivers have that id
janj: agd5f: from the log: (II) RADEONHD(0): Card not in database: 0x9488:0x106B:0x00B6
agd5f: janj: it's there. that's just confusing driver output
janj: agd5f: ok, is there anything I should try?
agd5f: janj: does the driver not work?
janj: agd5f: it doesn't
agd5f: janj: does radeon work?
janj: agd5f: haven't tried. but I skipped the obvious when looking at the log
janj: (EE) RADEONHD(0): Failed to detect a connected monitor
agd5f: janj: try radeon
janj: roger.
yangman: *sigh* people don't seem to want to read past that line
janj: the radeon driver doesn't work either
janj: I'll check the bug reports
janj: Looks like the driver doens't get well with displayport
agd5f: janj: radeonhd doesn't support displayport
agd5f: janj: and for radeon, you need git master as displayport support was only recently added
janj: agd5f: ok that's it then
janj: thanks. and hope to see displayport support soon :)
wltjr: agd5f: no go on patch for xpress200m :( gunna comment on bug
agd5f: wltjr: ok
agd5f: wltjr: btw, I already applied the patch to master and 6.12-branch so you don't need to apply the patch
wltjr: agd5f: anything else to try? happy to try out any patches or what ever
wltjr: agd5f: cool, I just did it to 6.12.4, but good to know it's in there now ):
wltjr: s/did it/applied it
agd5f: wltjr: ok. you might try git master or the 6.12-branch
agd5f: as there was a previous patch that might also help
wltjr: agd5f: ok I think there is a live ebuild for that package on gentoo I was using that, have to see where it's pulling from
wltjr: agd5f: ty to you and anyone else addressing the ancient x200m :)
agd5f: wltjr: for reference: http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?h=6.12-branch&id=3d59746808bc5f335104d27a8dce0fe94ab3cb78
wltjr: agd5f: I might try that stuff as well with the sources I just patched, till I can pull from git
wltjr: agd5f: negative, no diff with those changes either :( still fails at [drm] writeback test failed and 99% cpu blank/black screen
wltjr: agd5f: wait correction :) cpu is idle, so progress there :)
agd5f: wltjr: ok. update the bug
wltjr: agd5f: nah it's 99% so no diff
wltjr: agd5f: will do ty
scgtrp: i'm getting these errors and no acceleration trying to use radeonhd on my laptop: http://pastebin.ca/1716562
yangman: scgtrp: pastebin the entire log
scgtrp: yangman, http://pastebin.ca/1716569
yangman: anything usual in dmesg?
scgtrp: yangman, hm, this appeared upon starting X: http://pastebin.ca/1716573
yangman: hm. that's not right
yangman: try running radeon instead. maybe it's a bug in rhd that we haven't caught
scgtrp: things starting with *ERROR* generally aren't ;)
scgtrp: radeon doesn't even let me start any opengl-based apps
yangman: 3D is handled by mesa. it's a separate issue
scgtrp: drmRadeonCmdBuffer: -22. Kernel failed to parse or rejected command stream. See dmesg for more info.
yangman: but radeonhd is currently failing to initialize drm at all for you
scgtrp: that's what i get on radeon
yangman: what kernel are you running anyway?
scgtrp: 2.6.31
yangman: vanilla?
scgtrp: with one experimental acpi patch to make it work on my laptop
yangman: do you also get the *ERROR* messages in dmesg with radeon?
scgtrp: nope, it shuts up after "[drm] writeback test succeeded in 1 usecs"
airlied: you need a newer kernel for 3D
airlied: 2.6.32
scgtrp: oh, fun
scgtrp: conveniently, 2.6.32 refuses to find my disk when booting :/
scgtrp: "ATI R600/R700 3D and KMS support and other graphic improvements" << so i do.
scgtrp: well, thanks
scgtrp: yangman, airlied: still around? i got 2.6.32 to boot but still have the same errors and no acceleration
airlied: scgtrp: got the latest mesa?
scgtrp: yeah, 7.6
airlied: from git or from package? you need later version for r600
scgtrp: arch's "7.6-2". let me try git then
scgtrp: airlied, still software rendering :/
scgtrp: oh, why thank you google. i got the pastebin i gave you guys while ago as the first result for the error in dmesg :/
airlied: oh you need the irq firmware most likely
scgtrp: where would that be?
scgtrp: http://pastebin.ca/1716671 << not one of those?
airlied: http://people.freedesktop.org/~agd5f/radeon_ucode/
scgtrp: got it, do i need to reboot or just restart X?
airlied: oh you aren't using KMS so that shouldn't matter
airlied: is half awake
scgtrp: airlied, rebooted anyway and still nothing :/
airlied: scgtrp: are you using radeon or radeonhd btw?
airlied: and what version
scgtrp: airlied, radeonhd 1.3.0
airlied: does it work with radeon? it seems like rhd isn't seting up the dri properly or something
scgtrp: radeon gives me 2d acceleration but when i try anything gl-based i get "drmRadeonCmdBuffer: -22. Kernel failed to parse or rejected command stream. See dmesg for more info."
scgtrp: (and dmesg doesn't even mention it)
airlied: scgtrp: did you retry radeon with new kernel/mesa?
scgtrp: ooh, no i didn't. one minute
scgtrp: airlied, btw, do radeon and radeonhd have separate kernel modules? i just noticed the kernel module is called "radeon"
airlied: scgtrp: no same kernel module
scgtrp: ah, ok
scgtrp: airlied, http://pastebin.ca/1716681 << with radeon+mesa-git
airlied: yup thats working 3D
airlied: oh wierd
airlied: that is with 2.6.32?
airlied: and mesa master?
scgtrp: ... that's not working 3D, glxgears didn't start
scgtrp: yeah
airlied: thats still mesa 7.6
airlied: mesa master reports as 7.8
scgtrp: airlied, hm, do i need libgl-git too?
scgtrp: mesa-git only seems to provide libGLU and libGLw
scgtrp: airlied, installed libgl-git, it reports 7.8-devel now but dropped back to software rendering again
airlied: scgtrp: LIBGL_DEBUG=verbose glxinfo
scgtrp: libGL error: dlopen /usr/lib/xorg/modules/dri/r600_dri.so failed (/usr/lib/xorg/modules/dri/r600_dri.so: undefined symbol: _mesa_texformat_luminance_float16) << i think i'll just build all of mesa now instead of trying to build parts of it ;)