makkalot: hi all i'm using the git version of radeon driver ,my card is radeon hd 3650,when i startx with radeon driver i get a black screen and nothing else,startx works well with vesa driver,xorg.xonf : http://paste.pocoo.org/show/197026/ ,xorg.log : http://paste.pocoo.org/show/197025/ ,any ideas ?
mmp: airlied: ping
glisse: mmp: he is likely asleep
mmp: glisse: ahh, ok :-)
mmp: anyway, I was told I can ping also you...
glisse: yeah fire your question
mmp: roughly a month ago I talked to airlied about RS600 support and the still-opened bug
mmp: I offered help; at least I can give remote access to my RS600 hardware to debug
mmp: unfortunately, I was not able to be online too often, as I have to work on my master's thesis
glisse: is the rs600 your working computer ?
mmp: unfortunately yes
mmp: my laptop
mmp: I can't make the setup very comfortable, as that's the only computer I have
mmp: so e.g. webcam, if any, has to be hooked on it
glisse: mmp: what's not working ?
mmp: I'm just a regular student, so I can't offer much :)
glisse: btw what is the model of your laptop
mmp: glisse: basically, http://bugs.freedesktop.org/show_bug.cgi?id=25408
mmp: it's latitude XT
mmp: (argh, looks like I was talking to agd5f, not airlied; I should really read logs before ping-ing someone :)
glisse: he is likely asleep as weel
glisse: mmp: like this one http://cgi.ebay.fr/Dell-Latitude-XT-1-33GHz-Multitouch-Garantie-08-2012_W0QQitemZ230455231153QQcmdZViewItemQQptZDE_Technik_Computer_Peripherieger%C3%A4te_Notebooks?hash=item35a833c2b1 ?
mmp: glisse: yes, +- some variant differences (cpu speed and bluetooth presence)
glisse: i will see if i can buy it
glisse: rs600 is the gpu no dev has access too
Ke: interesting mix of french and german
mmp: glisse: yes, I was told so as well
mmp: glisse: if you are more looking for a laptop, XT2 might be a better alternative :)
mmp: although not as cheap
glisse: mmp: don't worry i will take the cheapest one, i am just looking for the gpu
mmp: ahh :-)
mmp: anyway, as sold from dell, it's quite expensive, so expect the price to grow really high
mmp: here at slovakia, new one used to be sold for ~ 2200 EUR
mmp: but that was 1.5 year ago
mmp: glisse: anyway, if I can help somehow, let me know
mcgregor: just got a drm error I've never seen before:
mcgregor: [63970.561071] [drm:atom_op_jump] *ERROR* atombios stuck in loop for more than 1sec aborting
mcgregor: [63970.561086] [drm:atom_execute_table_locked] *ERROR* atombios stuck executing EA90 (len 2065, WS 0, PS 4) @ 0xEF1B
airlied: mcgregor: what gpu?
mcgregor: rv730 (4650 mobility)
airlied: glisse: probablt need to print out more info ^^
airlied: zzz &
glisse: well need the bios to check the table
glisse: mcgregor: can you dump your bios and attach it to a bug
Ke: sleeping as background process sounds convenient
glisse: mcgregor: by any chance did you experienced a lockup prior to that message ?
mcgregor: glisse, I will ... erm, so fill a bug against what?
mcgregor: no, ntohing psecial or obvious happened at all
glisse: mcgregor: you were under X ?
glisse: sounds like a lockup from which we didn't recover
glisse: is the X still functional after that message ?
mcgregor: yeah, everything seems to be just normal
glisse: let me know once you have a bug
glisse: and paste bug url
mcgregor: hmm sorry I've never reported a bug against dri/drm, could you point me a link please?
mcgregor: https://bugs.freedesktop.org/ there?
mcgregor: glisse, airlied https://bugs.freedesktop.org/show_bug.cgi?id=27438 if you have any more question, feel free to ask
dileX_: mcgregor: summary is bad
mcgregor: hmm could you please give me a tip? what should I add? the problem is, I simply saw this in dmesg, nothing happened really
dileX_: "Error in dmesg kernel log with no obvious effect " is too general
dileX_: sth like "drm: Error atombios stuck with rv730 and 2.6.34-rc3-git1"
dileX_: dunno if you can change summary line
mcgregor: "drm: Error atombios stuck with rv730 and 2.6.34-rc3-git1" sounds better indeed, thank you. I'll try to change it
dileX_: mcgregor: your video-bios is only 21.42 KB?
mcgregor: https://bugs.freedesktop.org/show_bug.cgi?id=27438 changed it.
mcgregor: I did "dmidecode" as root, was that wrong?
dileX: mcgregor: http://paste.debian.net/67283/ (adapt pci-id if necessary)
mcgregor: ok, thanks for h script. the vbios.rom is 61,5kb I'll attach that file too
mcgregor: ok I hve added the vbios
mcgregor: thank you for your help dileX
dileX: mcgregor: n.p.
Wizzup: wow recent changes made xgetimage a lot faster on kms
Wizzup: (r600 driver)
nirbheek: wow, that's what made metacity with compositing slow for me
nirbheek: checks the changes
Hideki: hi, having something of an issue with a Radeon 9600 card...
Hideki: works in X normally, have it set to put the same display out on the VGA and DVI ports
Hideki: however, most things which switch to full screen cause teh DVI output to disappear...
Hideki: meaning I have to restart X to get it back
Hideki: is this a known issue and is there anything I can do about it?
adamk: Hideki: Try changing the primary output device with the --primary option, however it's still my experience that you will have to use 'xrandr' to adjust both monitors back the way they were when you exit fullscreen games.
adamk: There is no need to restart X, though.
Hideki: well, I could have xrandr run on exit if that'd do the trick... to what do I pass the --primary option though, X is started by gdm
BioTube: Hideki: xrandr
Hideki: oh right, lol
adamk: Ahhh, yeah, sorry, I didn't actually say :-)
Hideki: like xrandr --output DVI-0 --primary ?
Hideki: oh, that works
Hideki: spot on, thanks
Hideki: wonder if the CRT is missing its display now
adamk: Unfortunately there is no xorg.conf option for that with radeon, so you'll have to set that each time with xrandr or add that to a start up script.
Hideki: ahh, hrm, VGA display missing now but I can live with that
Hideki: as it's only ever used for testing
Hideki: hrm, seems there's an easier solution still, turn off the VGA display and put it on onyl when needed, lol
Hideki: they were just mirrored while i was getting the projector setup
pazof: I've to thank you all for this great work (M72M "Mobility radeon HD 2400")
meoblast001: my Radeon HD 2000 makes my kernel crash, i was instructed to netconsole
meoblast001: where can i find a tutorial on how to do this?
RIotingPacifist: lame, i used to have an old mobility card which sucked, so i upgraded using dell assuming i woundn't get anything too new, now i have an evergreen chipset, is there an ETA on support for them?
mcgregor: RIotingPacifist, what exactly? 2d acceleration? 3d acceleratioon? non-accelerated modesetting works already
chithead: note that some mobility 5000 chipsets are actually re-branded r700
RIotingPacifist: http://www.x.org/wiki/RadeonFeature looks pretty barren, erm modesettings is working here but glxgears is saying software rendering anyway. How long until suspend/resume and support for kwin4 desktop effects (or compiz i'm not fussed) work?
mcgregor: sorry, I have no idea about that.
RIotingPacifist: I also noticed that passing incorrect parameters to the modules prevents it from loading, is this the desired behavior or should it just ignore them and load normally
krisfremen: hey there, i just upgraded to lenny squeeze, and xvideo doesn't work anymore
airlied: RIotingPacifist: thats standard module behaviour
krisfremen: i have ati radeon X1300
krisfremen: and dmesg is full of complaints about radeon_cp: Failed to load firmware "radeon/R520_cp.bin"
RIotingPacifist: krisfremen: do you have the firmware installed? it should be in /lib/firmware/radeon i think
krisfremen: hmm.. i guess not
krisfremen: where can i get it?
stringfellow_: I think that's in firmware-linux-nonfree on squeeze
RIotingPacifist: krisfremen: i dunno what the properway to install firmware is, but as i g2g i'll post a link to where i found it http://www.megaupload.com/?d=SIYFYOQ0 , then i ran update-initramfs -uk and it fixed firmeware problems
RIotingPacifist: ignore that listen to stringfellow_ he probably wont break your computer
krisfremen: ok, got the package stringfellow_ recommended
krisfremen: yay! works!
krisfremen: thanks stringfellow_!!!
krisfremen: i've been killing myself with this problem for a day now
krisfremen: thanks! :)
makkalot: hi i'm trying to get radeon driver working but didnt have luck with it,vesa works ok. My xorg.conf : http://paste.pocoo.org/show/197026/ xorg.log : http://paste.pocoo.org/show/197025/ any ideas ?
dileX: (II) [KMS] drm report modesetting isn't supported.
dileX: makkalot: ensure you load radeon kernel-module with KMS enabled before X starts
dileX: seems archlinux has also a 2.6.33 kernel
makkalot: dileX: i'm using 2.32
makkalot: dileX: i'm using 2.6.32
dileX: I see it in the Xorg.log
makkalot: when i do lsmod | grep radeon what i have is :
makkalot: radeon 565089 0
makkalot: ttm 37153 1 radeon
makkalot: drm_kms_helper 22299 1 radeon
makkalot: drm 153435 3 radeon,ttm,drm_kms_helper
makkalot: i2c_algo_bit 4935 1 radeon
makkalot: i2c_core 18137 5 videodev,i2c_i801,radeon,drm,i2c_algo_bit
adamk: makomk: Please don't paste more than a few lines here.
makkalot: adamk: ops sorry
adamk: Is KMS enabled in the DRM?
dileX: you have KMS related infos in dmesg?
makkalot: dileX: dmesg http://paste.pocoo.org/show/197307/
adamk: Defaulting to userspace modesetting.
adamk: So you don't have KMS enabled by default and didn't enable it on the kernel boot line
adamk: pass radeon.modeset=1 to the kernel
dileX: yupp, UMS running
adamk: Or recompile the kernel with KMS enabled by default.
Wizzup: Isn't it enabled by default on recent kernels?
Wizzup: hm, probably not.
adamk: Well 2.6.32 isn't really recent :-)
adamk: Not sure about 2.6.33.
dileX: makkalot: check for KMS in yor current/running kernel: grep KMS /boot/config-$(uname -r)
makkalot: dileX: ok trying
dileX: makkalot: build howto (see topic) has also a lot of infos & hints
makkalot: dileX: will try with radeon.modeset=1
makkalot: when enabled radeon.modeset=1 kernel didnt boot it freezes.
makkalot: i'm using radeon hd 3650 should is it r600 type card ?
Wizzup: makkalot: Try a more recent kernel if possible
makkalot: Wizzup: good idea will try 2.6.33 hope to get it working that time,thanks
Magnade: (WW) RADEON(0): Direct rendering is not supported when VGA arb is necessary for the device
Magnade: what causes that error?
Magnade: did a upgrade to lucid and now i see that and i have no acceleration
Wizzup: Silly question: did you use radeon before the update?
Magnade: Wizzup: yes
Wizzup: so no fglrx
Magnade: less the update pulled it in
Magnade: nope it didnt
AstralStorm: Magnade: what's the card?
AstralStorm: 5k series?
Magnade: 3470 iirc
AstralStorm: maybe you're missing the firmware?
Magnade: there is firmware on the computer
AstralStorm: I haven't ever seen that message though
Magnade: could it be kms related?
AstralStorm: that's r500
AstralStorm: shouldn't be, unless you don't have it enabled
AstralStorm: even then it shouldn't
Magnade: well kernel doesnt have it enabled
Magnade: and for whatever reason it complains about loading firmware in the initramfs time of boot
AstralStorm: you can't "not enable KMS"
AstralStorm: the option only twiddled the default
BioTube: Magnade: loading firmware in intramfs?
AstralStorm: it sounds as if some other framebuffer driver took over
BioTube: what exactly is it saying
AstralStorm: and messes up stuff
Magnade: the initramfs message you want?
AstralStorm: I bet the antique radeonfb
BioTube: Magnade: yes
Magnade: pretty sure no fb devices are on
Magnade: im quiet ok with seeing a 80x25 screen scroll by
Magnade: message wise let me reboot
Magnade: ok it does boot its just reallllly slow for some reason
Magnade: is it ok to paste 6 lines?
BioTube: probably better to pastebin
Magnade: looks like it loaded my mobo vid card firmware and then had issues with loading the 3470
BioTube: 2.6.33 and later require an extra frimware for r600/r700 KMS, which wasn't added to the kernel before Linus forbade adding new blobs
AstralStorm: oooh, it's r600?
AstralStorm: I thought 34xx were r500
AstralStorm: somebody is lying
MostAwesomeDude: All HD 2xxx are r600.
BioTube: AstralStorm: a few HD 2xxxs were
Magnade: 24xx/34xx iirc are r600 i thought?
MostAwesomeDude: HD 3xxx are r600, HD 4xxx are r700, HD 5xxx are r800.
AstralStorm: Magnade: so yeah, you're missing the firmware
AstralStorm: install radeon_ucode or radeon-firmware or whatever it's called
Magnade: so im missing half the firmware basicly?
AstralStorm: and rebuild the initramfs
AstralStorm: yeah, the IRQ controller part
Magnade: nothing important... *sigh*
Magnade: kicks ubuntu
AstralStorm: real important
AstralStorm: w/o that you can't really do fast 3D accel (lying)
AstralStorm: (just the devs dropped the previous chunk of code that did it)
AstralStorm: (for some reason unknown, except that it was slower and more messy)
BioTube: AstralStorm: probably because maintaining the code necessary for accel with and without IRQs would be too much of a headache
AstralStorm: or because I still don't have GLSL 1.2 despite less headaches :-(
AstralStorm: no really, that *should* be a small improvement over the 1.1 that it does right now...
AstralStorm: just kidding
AstralStorm: it's some work
AstralStorm: I see the work's being done on power saving now
DanaG: My take on firmware: if he's disallowing blobs from being added, why not go all out and yank all blobs?
DanaG: It seems highly arbitrary to me.
Magnade: hmm these should be in lucid according to a bug report i found
AstralStorm: DanaG: who?
AstralStorm: Magnade: they're non-free
AstralStorm: so they're not in lucid
BioTube: DanaG: that's the plan, but going whole hog without warning would be kind of rude
BioTube: Magnade: see if there's a linux-firmware-nonfree package
AstralStorm: or you're missing them in your initramfs
DanaG: Ah, well then at least they're not being too arbitrary. (Otherwise, it would be "grandfathering" in stuff.
Magnade: that sounds like it got included to me
DanaG: dang, I'm obsessive-compulsive about my parentheses.
AstralStorm: Magnade: so your initramfs sucks
BioTube: Magnade: do you have linux-firmware installed?
AstralStorm: and yeah, install that package if you don't have it
Magnade: it is installed
Magnade: and yes apparently the initramfs sucks
Magnade: i put the firmware by hand on the initramfs and it still didnt load now im curious if thats a error loading rather than error not found
rxt0: is it possible that my R500's tv out doesn't output correctly due to old xserver?
sayao: do i need to enable anything for HDMI audio of my radeon HD2600 to work?
naquad: i've got radeon hd 4870, xorg 22.214.171.1242-1 and radeon 6.12.4-3 driver, kernel is 2.6.32, my xorg: http://dpaste.com/179201/ . the problem: 3d totally sux, glx doesn't work, none of composite managers are working as exptected (artifacts, blinking, almost everything, but not compositing). the question is: what am i doing wrong?
adamk: naquad: The log file is much more important than the config file.
naquad: adamk: http://pastebin.com/SxFU9f8Z
adamk: All that is correct.
adamk: Pastebin the output of 'LIBGL_DEBUG=verbose glxinfo'
adamk: Check to see what group owns /dev/dri/card0, that they have rw access, and that you are in ghat group.
naquad: i'm not in it
naquad: thank you
naquad: did restart
naquad: 3769.548 fps on glxgears =\
naquad: but that helped with glx :)
adamk: glxgears is not a benchmark.
adamk: Under any circumstances, ever.
naquad: xcompgmgr still blinks and shows artifacts :(
naquad: no error messages, no warnings
naquad: what could that be?
adamk: General crappiness of xcompmgr.
unic0rn: try compiz, with --always-indirect
naquad: what else can i try?
unic0rn: will work without it, but only with KMS & DRI2, which you don't have.
adamk: --always-indirect is not a compiz argument.
adamk: xfce4, kde4, and gnome all have their own compositing managers.
unic0rn: blah. --indirect-rendering
unic0rn: at least for 0.8.4 that's an option
naquad: installs cairo-compmgr
unic0rn: cairo-compmgr is just a stinking piece of amateur crap.
unic0rn: tried it. trust me, you don't want it.
unic0rn: it shouldn't be allowed to use 'cairo' in it's name.
naquad: then which one should i use? i've got openbox
unic0rn: not sure what will work with openbox. i would try compiz.
unic0rn: it has some history and heavy testing behind it
adamk: compiz will definitely not work with openbox.
unic0rn: then no standalone window manager will.
unic0rn: that leaves only cairo-compmgr and xcompmgr as far as i know
adamk: I've never used cairo-compmgr, but my experiences with xcompmgr were pretty awful under fluxbox.
naquad: one of them suxx (c) unic0rn, another one doesn't work. nice.
unic0rn: naquad: try something more modular. dunno.
unic0rn: or, actually, wait.
unic0rn: you can't use 2 window managers, at once. and the thing with compositing is, compositing managers are usually integrated into window managers
unic0rn: openbox is a window manager. whatever your environment is (i suspect lxde somehow), screw openbox and launch compiz instead. perhaps it'll work.
unic0rn: that is, if compiz itself will work, then it has to work
naquad: ok, i want to try one more thing, if it won't work i'll install compiz and try it
bridgman: weren't naquad's problems with compositors just the usual "GL programs fighting the compositor" issues that you get without DRI2 and RDR ?
bridgman: ie needs KMS and maybe new kernel ?
unic0rn: nah, i don't think so.
unic0rn: after all, when you're running usual desktop apps and compositor, what else is using opengl, apart from compositor/
unic0rn: nothing, perhaps media player in case of textured video
unic0rn: also, compiz was working fine for me without kms and dri2, just opengl wasn't redirected, but even then glxgears worked fine, just without compositing
bridgman: reads the logs again...
bridgman: naquad's complaints were about 3D programs, at least that's how I interpreted "3D totally sucks"
bridgman: or maybe that base before 3d was working...
bridgman: does xcompmgr use gl or render for compositing ?
bridgman: s/base/was/ wtf ?
MostAwesomeDude: xcompmgr is a proof-of-concept Xrender app. It apparently still blows keithp's mind that people use it seriously.
BioTube: it wasn't all that glitchy when I used it
BioTube: just limited
naquad: MostAwesomeDude: there are no other options for simple window managers like *box (openbox, fluxbox, blackbox, ...)
MostAwesomeDude: naquad: I'm well aware, but on the other hand, those window managers are meant to be sans eye candy.
naquad: MostAwesomeDude: basically, all i need from a composite window manager is to provide transparency, none of the light wms provide this
naquad: so things like xcompmgr and co will live
MostAwesomeDude: naquad: Sure, just pointing out the humor of a utilitarian WM being augmented with pretty effects.
bridgman: naquad, are your remaining problems with all apps or just 3d apps ?
naquad: bridgman: i'm remainding with problem that both drivers have the same issue: no composite managers run on them
bridgman: any error messages ?
bridgman: did you get a chance to try compiz ?
naquad: didn't do it yet
naquad: but i don't have much hope
bridgman: are you pretty sure that 2D and 3D acceleration is working ?
naquad: supertux works :)
naquad: glxinfo says direct rendering: Yes
bridgman: that doesn't mean much these days - for the last year the software renderer supports direct rendering
bridgman: what does the renderer line in glxinfo say ?
bridgman: 'course it's probably ok if supertux is working, I remember that being pretty slow w sw render
naquad: crap... OpenGL renderer string: Mesa DRI R600 (RV770 9440) 20090101 TCL
unic0rn: naquad: that's ok.
unic0rn: that means DRI works.
bridgman: it's fairly old though, isn't it ? recent mesa versions added the mesa version to the render string iirc
unic0rn: not really
unic0rn: i have something around 7.7.1 and still there's 20090101 in renderer string
MostAwesomeDude: 20090101 is the date in one of the GL header files; it doesn't change often.
bridgman: understood; just thought I remembered seeing the Mesa version # in the renderer string for the last couple of mesa releases
BioTube: bridgman: it's in the version string
BioTube: OpenGL version string: 2.0 Mesa 7.9-devel
bridgman: yep, just noticed that, nvm
unic0rn: naquad: bottom line, it seems like your 3D acceleration is all fine. try compiz or perhaps cairo-compmgr, if openbox is really what you want
unic0rn: because as far as i'm considering it crap, cairo-compmgr is possibly the best standalone compositor not bundled with window manager
naquad: unic0rn: ok, will try it.
naquad: thank you for help people
EruditeHermit: is there a status page for r800?
BioTube: EruditeHermit: presumably it's with the rest
EruditeHermit: BioTube, and where is that?
bridgman: look under Evergreen
bridgman: there is no r800 ;)
EruditeHermit: why'd'ya have to go an' use names instead of numbers
bridgman: although the developers want numbers back
EruditeHermit: silly marketing
bridgman: it's kind hard to do something like ">= 770" ;)
EruditeHermit: I would think numbers are easier in comparisons
bridgman: yep, but most of the groups prefer names
bridgman: so names it is
EruditeHermit: it makes things harder
bridgman: not much really; most of the code is driven by attributes more than numbers anyways
EruditeHermit: 800 is newer than 700
BioTube: so what's up next? Deciduous?
EruditeHermit: right but its more confusing for the consumer
bridgman: yep, but 880 is older than 770
EruditeHermit: I wish they would stop doing that
EruditeHermit: stop calling mobile GPUs of lower quality the same as desktop GPUs etc
EruditeHermit: mobility 5870 is not a 58xx GPU
bridgman: agreed; I wish people woudl stop buying numbers
bridgman: then our marketing folks would stop selling them ;)
EruditeHermit: well its really confusing
EruditeHermit: I found a nice website that ranks GPUs
EruditeHermit: but i'd prefer not having to look up stuff if possible =)
EruditeHermit: just knowing by the number
bridgman: just assume the first digit is a function of the year the chip was sold in and everything will make more sense ;)
EruditeHermit: wait that makes less sense
EruditeHermit: you mean the rv numbers
EruditeHermit: or the HD5xxx numbers?
EruditeHermit: neither of which make sense
bridgman: hd numbers
EruditeHermit: its 2010
bridgman: here we use the rv numbers or chip names and they still make sense
bridgman: yeah, so subtract 5-ish
bridgman: don't worry about it too much, just trying to explain how the marketing folks see it
EruditeHermit: those tricky people
EruditeHermit: trying to sell me 3 year old cards as new!
bridgman: well, you say you only want to buy HD5xxx parts ;)
EruditeHermit: i never said that!
bridgman: yeah, well, someone did ;)
bridgman: anyways, as long as engineering can come up with new chips as fast as the marketing folks can come up with new numbers it all works out
Vash63: EruditeHermit: Nvidia's the one selling 3 year old cards new >_>
Vash63: 8800 GT == 9800 GTX == 9800 GTX+ == GTS 250
Vash63: Same chip, different clocks and a few board revisions over like 4 years.
EruditeHermit: yeah Nvidia is worse
EruditeHermit: but I think everyone is doing it
Vash63: AFAIK every HD5000 chip is 100% new.
Vash63: That said, I'm considering going back to Nvidia after my 4870 X2 because of ATi driver issues. I like what they're doing on the Linux side but Crossfire is a bit of a mess on the Windows side.
Vash63: And fglrx... shouldn't ever be mentioned anywhere.
bridgman: fglrx fglrx fglrx ;)
EruditeHermit: bridgman, btw ever get a chance to forward those bugs to anyone on the fglrx team?
Vash63: fglrx wouldn't be that bad if it could actually like, release faster. The OGL performance is more than adequate.
EruditeHermit: 2D is slow
Vash63: Problem is I consider 3 distros on my system: Gentoo, Exherbo and Arch. Rolling releases only... and fglrx is consistantly 6 months behind all of those distros.
Vash63: Making it impossible.
Vash63: Somehow Nvidia manages to keep up with xorg no problem though on their binaries...
Vash63: That said, xf86-video-ati is growing up nicely. Would really like better power management but I'm sure it's coming.
Vash63: Firefox seems a lot faster with the latest updates on git.
MostAwesomeDude: Ever considered a distro with sane release policies? Sounds like you're one for three.
Vash63: Still takes a while when dragging tabs to make the little transparent image of the tab though, but switching and scrolling is awesome.
Vash63: No, I haven't MostAwesomeDude.