mcgrof: hm, so TMDSA and LVTMA may HDMI code for video
airlied: mcgrof: later chips just use UNIPHYs
mcgrof: airlied: what' sthat
airlied: just another digital encoder
airlied: same as TMDSA or LVTMA
mcgrof: I was looking if to add a new hdmi struct for, and then one for audio, but so far I see the audio stuff just needs the hdmi specific offset which I can just store into the hdmi audio struct. The only use I see for the hdmi struct is for startin the hdmi engine and it seems this is required for TMDSA and LVTMA
mcgrof: not even sure what those are
mcgrof: I see
mcgrof: do many cards have those encoders? Or should I not bother right now
airlied: yup all the r500s
airlied: I'm not sure radeonhd ever got r600 HDMI audio going
agd5f: I think only rs6xx and r6xx/r7xx have hdmi audio
mcgrof: I see, well I'll find out once I test this, I do see a check for >= CHIP_FAMILY_RS600
agd5f: r5xx don't have audio I don't think
airlied: oh yes r5xx hasn't got HDMI :)
agd5f: mcgrof: that works
mcgrof: ok, yeah that's the port name
mcgrof: should I bother with the porting the non-audio hdmi engine stuff?
airlied: mcgrof: thats all dealt with by atombios, you only need the register writes that deal wiuth audio
mcgrof: airlied: so TMDSA and LVTMA encoders are also handled by atombios?
airlied: mcgrof: yes
agd5f: mcgrof: all encoders are
mcgrof: interesting
mcgrof: I was asking because of this comment
agd5f: mcgrof: the whoel display pipeline is
mcgrof: /* some version of atombios ignore the enable HDMI flag
mcgrof: * so enabling/disabling HDMI was moved here for TMDSA and LVTMA */
mcgrof: In RHDHdmiEnable
mcgrof: whatever that means
agd5f: mcgrof: I guess that's probably the audio stuff
mcgrof: wall that I see audio gets from that is the hdmi ofset
mcgrof: I can't type
mcgrof: s/wall/all
airlied: mcgrof: you probably just want to add the offset somewhere to the encoder info
agd5f: just like is done for cretc regsiter offset (crtc_offset)
agd5f: crtc even
mcgrof: airlied: ok, I was adding it to an radeon_hdmi_audio_struct which I currently stashed into radeon.h RADEONInfoPtr, do we prefer the radeon_hdmi_audio struct in the encoder ?
agd5f: mcgrof: yeah since they are tied to the encoders
mcgrof: ok
twoerner: airlied: ping
airlied: twoerner: pong
twoerner: airlied: is the latest driver sane?
twoerner: airlied: id did not have had much time to test the older one
airlied: twoerner: should be, what in rawhide now should be fairly sane.
twoerner: airlied: ok, so you found the corrutor?
airlied: there were a few things that could have caused it
twoerner: airlied: ok
fpoibaf: anyone managed to run google earth (4.3) on a R500 at decent speed?
fpoibaf: openarena, sauerbraten end other games run fine, but GE is really slow, like it's using software rendering
fpoibaf: although the about box tell me the R300 driver is being used
airlied: does it print out any sw fallbacks?
fpoibaf: it only prints:
fpoibaf: do_wait: drmWaitVBlank returned -1, IRQs don't seem to be working correctly.
fpoibaf: Try adjusting the vblank_mode configuration parameter.
fpoibaf: (I am using the mesa 7.2 branch)
MrCooper: have you tried running it with vblank_mode=0?
taiu1: that version probably did some fallbacks
taiu1: was smooth line or line stipple
fpoibaf: great, thanks! Running with vblank_mode=0 (drirc generated with dri conf) make it runs at decent speed!
Kano: hi, how to patch radeon driver to use exa by default when drm module is available?
Kano: i have got current drm modules but i want that exa works without xorg.conf too
jcristau: should take all of 2 minutes looking at the code to figure out..
Kano: jcristau: i am sure you find it in 30s
Kano: jcristau: i would prefer a configure option
MrCooper: configure option shouldn't be necessary, but I think it should probably also take the server version into account, only enabling EXA by default as of 1.5 or so
Kano: MrCooper: but the target xserver here is 1.4 ;)
MrCooper: and maybe only on >= r300, as older cards tend to have too little VRAM to reserve enough EXA offscreen memory
peterz: should Radeon Xpress 200G thingies work with the current driver -- or are they know borken?
MrCooper: OTOH maybe just enable it by default and see what happens...
Kano: peterz: if you do that for cards without xv hardware support no problem for me
airlied: peterz: should allwork
peterz: airlied: then I have a bug report for you :-)
peterz: airlied: mouse pointer renders good, the rest of the screen is utterly scrambled
airlied: peterz: rawhide?
Kano: airlied: xv without exa works too?
peterz: airlied: kubuntu, xserver-1.6 radeon-git
Kano: i got at least no kaffeine error
airlied: peterz: wierd got a logfile?
airlied: peterz: does KDE enable ayn desktop effects?
peterz: airlied: yeah, let me enable mode debug for you though
peterz: airlied: it would, but I doubt kdm does
airlied: true.
MrCooper: peterz: is it only kdm or the session as well?
peterz: MrCooper: dunno, can't really log in with all the screen garbled
MrCooper: you can't type blindly? :)
peterz: dunno what it looks like, I've just installed this box
MrCooper: airlied: should I push the radeon-rewrite fix I posted?
airlied: MrCooper: yeah please Ive jsut been distracted by non-mesa
MrCooper: peterz: I'd expect 'username -> tab -> password' to work
MrCooper: peterz: I ask because of http://bugs.freedesktop.org/show_bug.cgi?id=19579
peterz: this is definately more than a few seconds
MrCooper: well, apparently with KDE 4.2.1 they're only seeing corruption in kdm anymore
peterz: tries startx
peterz: still hosed
MrCooper: k, not the same problem then, thanks for trying
peterz: http://programming.kicks-ass.net/sekrit/x200-xorg.conf
peterz: http://programming.kicks-ass.net/sekrit/x200-Xorg.0.log
peterz: airlied: ^
airlied: peterz: try Option "accelmethod" "EXA"
peterz: airlied: same thing
peterz: updated the Xorg.0.log file
airlied: peterz: can you test VGA?
peterz: vesa driver?
airlied: no VGA plug :)
peterz: ah, sure thing
peterz: similar, but more unstable screwage
airlied: Option "Noaccel" "true" then
peterz: now it very slowly drew crap :-)
b0le: airlied: just tested radeon-rewrite again (with a R430 (X800)), and glxgears works without kms, but starting compiz causes a lockup. With kms: glxgears locksup and trivial/clear locksup after resizing it (and I haven't even tried compiz)
MrCooper: b0le: did you have the fix I just pushed for compiz without KMS?
b0le: MrCooper: no, will try nowl.
b0le: s/nowl/now/
airlied: b0le: wierd, I played with my x700 today and it seems fine
airlied: need to play some more then :)
airlied: peterz: can you edit src/pcidb/ati_pciids.csv go to the 5954 line
airlied: and add an 1 in the fourth field
airlied: like the 5955 case
airlied: then run ./pcidb/parse_pci_ids.pl ./pcidb/ati_pciids.csv
peterz: "0x5954","RS480_5954","RS480",1,1,,,1,"ATI Radeon XPRESS 200 5954 (PCIE)"
peterz: "0x5955","RS480_5955","RS480",1,1,,,1,"ATI Radeon XPRESS 200M 5955 (PCIE)"
airlied: yup
airlied: I'm grasping at straws now
peterz: goes install a fwe extra perl modules
peterz: airlied: ok, should I rebuild after that perl stuff ran?
airlied: yes
peterz: sorry, wrong straw :/
airlied: peterz: any ideas if it ever worked? :)
peterz: airlied: no, just got the machine from a friend..
airlied: peterz: run memtest on it yet? :)
peterz: nope, haven't :-)
peterz: but windows did boot before I stuffed in the linux cd
airlied: so probably not that.
airlied: noaccel drawing garbage is just wierd
peterz: looks like the mode is just set wrong
peterz: lemme get my digicam
MrCooper: airlied: think http://bugs.freedesktop.org/show_bug.cgi?id=20672 is more likely DRM or DDX issue?
peterz: airlied: http://programming.kicks-ass.net/sekrit/P1010002.JPG
b0le: MrCooper: now the screen just goes blank (except for cursor, which still moves). I loose keyboard input, but can still ssh in and kill compiz, but my screen remains blank. After killing X (over ssh), moving the mouse no longer moves the cursor (and I still have no keyboard input).
airlied: MrCooper: probably kernel.
airlied: peterz: same on VGA and DVI then?
peterz: airlied: yep
MrCooper: b0le: anything interesting in the X server log file or stderr output?
airlied: hmm try adding a virtual 1600 1200 line in SubSection "Display" for 24 bits
peterz: airlied: same
airlied: runs out of ideas that is just wierd
airlied: it looks like a stride issue alrigh
MrCooper: yeah, that or tiling
b0le: MrCooper: from dmesg: [drm:drm_lock_take] *ERROR* 3 holds heavyweight lock
b0le: and from X output: CS section size missmatch start at (r300_cmdbuf.c,emit_tex_offsets,186) 4 vs 2. CS section end at (r300_cmdbuf.c,emit_tex_offsets,206)
MrCooper: I reported the latter in my post about the fix I pushed, but the former looks like it might be caused by the fix...
twoerner: airlied: ping
airlied: b0le: what you running for the CS section mismatches?
airlied: twoerner: pong
b0le: airlied: I started compiz
MrCooper: airlied: seems to happen with pretty much everything here
twoerner: airlied: i am toturing the driver and it is working for now, but Xv needs way more cpou than before
twoerner: s/cpou/cpu
airlied: MrCooper: oh wierd.
airlied: twoerner: probably the tearing avoidance code
twoerner: s/toturing/torturing
twoerner: airlied: is this enabled by default, now?
twoerner: airlied: i have no xorg.conf with the new installation
airlied: yes should be on Fedora.
b0le: MrCooper: even starting trivial/clear is causing the same problem (same output in dmesg) and I loose keyboard input.
MostAwesomeDude: b0le: I did already ask you about dmesg output with r300 Gallium, right?
MostAwesomeDude: And the conclusion was that there wasn't any dmesg garbage, so the kernel was okay with it?
b0le: MostAwesomeDude: yes. Last time I checked, there was nothing in dmesg (when running gallium)
MostAwesomeDude: 'k. Thanks. BTW, I know almost exactly where the r300 problem is; I'll be able to handle it when I do the r300 fragment shaders. :3
twoerner: airlied: exavsync is not meantioned in Xorg.0.log
twoerner: s/meantioned/mentioned
MostAwesomeDude: Just been busy with finals.
airlied: twoerner: its on by default, its not exavsync
airlied: its xv stuff
airlied: there is an xvattr to play with it
twoerner: airlied: i used it before with the GIT driver with about half the cpu usage (~ 5 days ago)
twoerner: airlied: cpu usage is now 80-99% for HD stuff
twoerner: for Xorg
MrCooper: b0le: hmm weird, maybe the lock is already taken outside of those functions for you, but I can't see where...
b0le: MostAwesomeDude: sounds good :) can you ping me if you have something I can test (though I will probably try every now and again anyways, depending on how busy I am)
airlied: MrCooper: I thought ScheduleSwap was taking it
airlied: I wondered why it worked for me at all
airlied: ah if the cliprects are 0 its failing with the lock unheld
airlied: so I think I need to make ScheduleSwap a bit better
MostAwesomeDude: b0le: You got it. It'll probably be a few days, but hopefully soon.
MrCooper: airlied: ah, but only from SwapBuffers, not from CopySubBuffer
Kano: airlied: i had also display corruption when i switched from fglrx to ati (fglrx module was unloaded of course)
airlied: MrCooper: oops
Kano: airlied: 6.12.0
MrCooper: airlied: fixing...
airlied: Kano: what gpu
airlied: ?
Kano: 3450
MrCooper: b0le: does http://people.freedesktop.org/~daenzer/radeon-rewrite-locking.diff work better?
airlied: MrCooper: care to fix the return if the cliprects test in ScheduleSwap fails?
MrCooper: airlied: I'm not sure exactly what your intention is for that, so I'll leave it to you :)
airlied: hehe no worries :)
glisse: sometimes changing mc fb loc helps finding bug :)
peterz: airlied: what's the base of drm-next? git diff drm/master drm/drm-next gives me a huge patch far outside of the drm stuff
kepstin: peterz, the base of drm-next is linus's tree. Last I checked, at v2.6.29-rc8
airlied: peterz: yup 2.6.29-rc8
peterz: thanks! git diff v2.6.29-rc8..drm/drm-next seems to generate useful stuff
ech0s7: enabling xfce compoistor X freeze, and i have to restart X with ctrl+alt+backspace, i have rv620 and radeon and drm compiled from git, this is my xorg.log: http://rafb.net/p/yBniG525.html
twoerner: airlied: ping
twoerner: airlied: using actual GIT of xf86-video-ati results in 40-99% cpu usage with the same HD tests and it looks to be a lot smoother
twoerner: airlied: is this a result of the KMS stuff?
linmax: I'm trying to get acceleration working on my r600 card
linmax: but all I get is this: [drm] wait idle failed status : 0xA0003030 0x00000003
twoerner: linmax: are you using Fedora rawhide?
linmax: No, I'm using airlied's drm-next branch
linmax: on a debian system
twoerner: linmax: i used upstream drm (r6xx... branch) with latest kernels
twoerner: now i am using fedora rawhide kernels and drm is workign there
twoerner: linmax: which ati driver version do you have?
linmax: I'm using a self compiled 6.12.0
Kano: linmax: do you want to test my script for drm r6xx branch + dkms?
twoerner: linmax: do you have other drm messages?
twoerner: linmax: failed writeback?
linmax: [drm] Setting GART location based on new memory map
linmax: [drm] Loading RV620 CP Microcode
linmax: [drm] Loading RV620 PFP Microcode
linmax: [drm] Resetting GPU
linmax: [drm] writeback test succeeded in 1 usecs
linmax: [drm] wait idle failed status : 0xA0003030 0x00000003
linmax: Kano: what does you script do?
Kano: installs latest drm
Kano: you need dkms
twoerner: linmax: i have had this some time ago, but it is fixed for me in latest GIT versions
airlied: twoerner: are you using the 6.12.0 from koji?
twoerner: airlied: xorg-x11-drv-ati-6.12.0-2.fc11.i586
airlied: twoerner: wierd the modeset patch must be causing another conflict maybe
twoerner: airlied: jup, seems like this
glisse: waooo got fbcon :)
twoerner: airlied: most hd stuff is at ~50% with GIT version, but at minimum 80% with this rawhide drivr
airlied: twoerner: that is lateset git from today?
twoerner: airlied: but i have one stream that is at 99%+ with boith
linmax: twoerner: by latest GIT you mean mesa/drm from freedesktop.org?
twoerner: airlied: a55ced5ee20c07e743c7c0978803fd10589c1531
twoerner: linmax: jup
twoerner: airlied: Date: Sun Mar 15 12:08:04 2009 -0400
airlied: twoerner: nothing jumping out at me, will have to look tomorrow.
twoerner: airlied: if there is anything i can do...
twoerner: airlied: oha... kernel oops
twoerner: only got one line
twoerner: airlied: fiesystem seems to be still ok ;-)
peterz: airlied: agd5f: got this r600 exa stuff working, brilliant stuff, only got the occasional display corruption when I drag windows between screens
linmax: twoerner: I tried latest git now, but it's still the same problem
twoerner: linmax: drm and/or drv-ati ?
linmax: drm
twoerner: hmm
linmax: should I try ati, too?
twoerner: this should work
twoerner: it does for me here with a RV630
linmax: this is a HD3470
linmax: lspci shows it as ATI Technologies Inc Mobility Radeon HD 3470
linmax: but it's a PCI-express card
twoerner: mine is also PCIe
nanonyme: Aren't most cards nowadays PCIe? ;)
linmax: I just found the 'Mobility' a bit strange
nanonyme: Why so?
linmax: because it sounds like it's a laptop card
nanonyme: linmax: Why wouldn't you have PCIe in a laptop?
nanonyme: ;)
linmax: of course you can have PCIe in a laptop
linmax: but this card is in a PCIe slot in my desktop
nanonyme: Ahh.
nanonyme: That's an interesting choice, yes.
nanonyme: Might be actually a good one sometimes though.
Kano: airlied: what chip has 1002:949C?
linmax: hm, still the same with latest ati
ajax: src/radeon_chipset_gen.h: { PCI_CHIP_RV730_949C, "ATI FirePro V7750 (FireGL)" },
Kano: ok, RV730
Kano: there are some errors in the pci.ids, why does nobody update it
boghog: oh bummer, turning on compositing makes my video tear again :(
spstarr_work: hmm
spstarr_work: looks at git
Kano: 949F is V3750 from win inf file, not V5700...
mharris: Kano: pciids.sf.net - go wild ;)
spstarr_work: MrCooper: I will test your patch tonight
MrCooper: boghog: enable 'unredirect fullscreen windows' in the compositing manager
damidalla: hi everybody, I am back asking silly things :D
damidalla: I did manage having DRI on my RS600, but opengl windows stay on top of everything else...
agd5f: damidalla: need dri2 support for that, which isn't done yet
agd5f: or don't run compiz
spstarr_work: MrCooper: have any idea why this might be happening
damidalla: ok, thanks - well, at least is not a bug of mine, and I can forget how to install fglrx
spstarr_work: MrCooper: http://www.sh0n.net/spstarr/corrupt-composite.ogv
nanonyme: Wonder which one manages to get the back&forth scrolling fixed first, fglrx or open drivers. :)
damidalla: nanonyme: well, for a lot of people fglrx is not an option anymore... they just retired from the "competition" :)
nanonyme: No, they didn't. :)
nanonyme: Fglrx is about as featureful as the open drivers for my card.
nanonyme: Except a little more.
nanonyme: (Actually realistically a lot more)
nanonyme: Just too bad they don't yet support X server 1.6. :/
damidalla: nanonyme: well, if the news are true, I will not get a 1.6 X.org - compatible fglrx driver
damidalla: as fglrx 9.4 will not support many graphic cards
nanonyme: damidalla: Then again, you have 3D with open drivers anyway.
nanonyme: damidalla: I wouldn't *want* fglrx if I had opensource Mesa drivers and opensource power management. :P
damidalla: nanonyme: yes, I did suffer a little to make it work, but I am happy of not depending anymore on the slow development of new X.org-compatible fglrx
nanonyme: damidalla: New versions of fglrx now happily only supports the people who don't have an alternative. :)
nanonyme: s/supports/support/
damidalla: nanonyme: and I really hope the alternative is not far away! :D
nanonyme: damidalla: Heard it mostly depends on how many people think open drivers for new cards are interesting enough to invest work hours in. :)
damidalla: nanonyme: my only problem is the (almost) complete lack of knowledge...
damidalla: "only" problem :D
nanonyme: damidalla: If you have interest, that only slows you down, it doesn't stop you from doing anything. ;)
damidalla: nanonyme: well, I still have to learn how to write decent code :D
nanonyme: damidalla: That can also be dealt with by practice given enough interest. ;)
damidalla: nanonyme: I do not know if it's "my road" to become a programmer - however, my studies will help me to find it out, I hope! :)
Zajec: hello, does xf86-video-ati compile for X Server 1.6.0?
bgoglin: Zajec: why not ?
Zajec: will pastebin
Zajec: http://pastebin.com/m79bbc69c
Zajec: I've X installed from packages for openSUSE... can that be a problem?
bgoglin: no idea about suse. no error/warning before these?
Zajec: no, don't see any warning/error before
Zajec: and ./autogen.sh was clean
nanonyme: Could check config.log for not found messages.
Zajec: ok
Zajec: should I worry about
Zajec: conftest.c:10:28: error: ac_nonexistent.h: No such file or directory
Zajec: i've libtool installed
nanonyme: Probably not.
tlp: c
tlp: woops
Zajec: ok, i updated packages from XOrg openSUSE repo and now it works fine :) should have done that before making noise :)
ech0s7: enabling xfce compoistor X freeze, and i have to restart X with ctrl+alt+backspace, i have rv620 and radeon and drm compiled from git, this is my xorg.log: http://rafb.net/p/yBniG525.html
ech0s7: anyone have some idea?
adamk_: airlied: FYI, you called it. After upgrading, KMS seems to be working properly.
adamk_: Well, KMS is fine. glxgears just hard locked the machine.
airlied: adamk_: what hw?
adamk_: x850.
adamk_: I can try with an x1950 on the same machine.
adamk_: Dual core Xeon, if it matters.
adamk_: All updates as of this morning.
nanonyme: Blah, another merge conflict in drm-next. :/
adamk_: Oh, PCIe
adamk_: I'll have to check the other card later tonight or tomorrow as I'm about to leave for dinnerw ith family.
airlied: adamk_: no worries
oyvind: Hi, I currently use snaphots of the radeon-driver and stay up to date with the master branch at http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/. Should I also be using latest snapshot of DRM (http://cgit.freedesktop.org/mesa/drm/) ? I usually compile the stuff under linux-core/ and install drm.ko and radeon.ko manually. Is that the correct way to do things ? I'm on Ubuntu Intrepid using the default kernel 2.6.27 ..
nanonyme: oyvind: You very likely don't need a newer libdrm so I'd suspect the answer is no. :)
adamk_: Alright, the family is running late. I swapped cards quickly and checked. The x1950 locks up solid too.
adamk_: The network is dead.
adamk_: I should setup the serial console on this box.
adamk_: Though it does seem unlikely that would actually help in this case. In any case, I'll be back later.
oyvind: nanonyme: OK, thanks. But just to avoid confusion, I am talking about the kernel modules, not the userspace libdrm.so library (the one that comes with Ubuntu, libdrm.so.2.3.1 seems to work just fine with the newest versions of drm.ko and radeon.ko).
yangman: oyvind: if you're running r6xx/r7xx, you need drm from the r6xx-r7xx-support branch
nanonyme: oyvind: Those snapshots on the page were strictly libdrm snapshots.
nanonyme: yangman: Assuming he has such a card, yes.
oyvind: yangman: k, thanks. I'm on R500 (ATI X1400/M54)
nanonyme: Yups yups.
oyvind: nanonyme: well, they contain a subdir with kernel modules called 'linux-core'
nanonyme: Hmm, right.
oyvind: Point is, though, I'd like to stay on the edge of developments with the radeon and radeonhd-driver, and I just wanna do things correctly, without having to update huge parts of my distro :)
nanonyme: I'd probably rather use the drm-next kernel tree git branch if I wanted to stay on the edge. (I'd expect it's getting updates for other stuff than just r6xx/r7xx too considering it's getting into kernel 2.6.30)
oyvind: nanonyme: Jepp, that what's triggered my original question, because I read this post: http://www.phoronix.com/forums/showpost.php?p=66754&postcount=29, from the discussion of Phoronix-article about the 6.12.0-release.
nanonyme: oyvind: I would strongly suspect you can keep on compiling the kernel modules from drm master since you don't need the r600/r700 support.
nanonyme: If you want to.
oyvind: nanonyme: Jepp, it seems to work very well, at least :) (stable and all).
AStorm: will test latest driver again :)
nanonyme: I wish Linux 2.6.29 would get out soon.
glisse: nanonyme: i wish the other way :)
nanonyme: glisse: How's that?
nanonyme: Oh. :)
nanonyme: glisse: You meant as in "hopefully it will take a while still so there'll be more time to finalize drm-next"?
glisse: well i don't think we will be ready for 2.6.30 but it gives us more times to have more testing on that
arekm: even finalized drm-next will contain tons of bugs ;>
nanonyme: glisse: What would "not ready" be at worst? ;) Not being allowed to merge? :o
glisse: of course we won't merge if we are not ready
nanonyme: Right.
MostAwesomeDude: What.
MostAwesomeDude: Why does RadeonProgram have R600 and R700 entries?
MostAwesomeDude: And how is Compiz platinum on the R700?
glisse: MostAwesomeDude: fglrx ?
MostAwesomeDude: glisse: Probably. Or swrast.
MostAwesomeDude: No, wait, can't do compiz on swrast. So it's either fglrx or somebody screwing around.
ajax: compiz platinum?
ajax: the shark, ladies and gentlemen, has been jumped.
kkuno: lol
kkuno: lol nexuiz is gold on r700
kkuno: ahhh
kkuno: 2fps
biker_rat: Anyone have sauerbraten running with r5xx on 64-bit?
AndrewR: Hi all! This is possible that ""Move vblank_init to driver load time." commit in drm-git causes kernel panic on rv280 + kernel 2.6.27.19?
biker_rat: It's not working for me.
soreau: I need help getting my radeon driver working. Here is glxinfo and X log
soreau: http://pastebin.com/m5bbd5a2e
biker_rat: What distribution are you using?
biker_rat: soreau, what distro?
AStorm: AndrewR, uh, old kernel and old version of drm branch?
soreau: biker_rat: It's all in the X log
soreau: Using gentoo atm
AStorm: soreau, VIDEO_CARDS=radeon is set?
AStorm: if so, you need recent drm
AStorm: that is, x11-drm package
AndrewR: AStorm, old kernel, but drm was updated ... and kernel panic just seconds after i kill X with my usual ctrl-alt-bs was bad surprise. right now i use drm-git with two commits reverted - after some startx/kill X - no panics
AStorm: if VIDEO_CARDS=radeon wasn't set, set it and rebuild mesa
da1l6: dlopen: /usr/lib/xorg/modules/extensions//libdri.so: cannot open shared object file: No such file or directory
da1l6: dlopen: /usr/lib/xorg/modules/extensions//libglx.so: cannot open shared object file: No such file or directory
AStorm: interesting
da1l6: that sounds suspecious
AStorm: soreau, also, USE=dri opengl should be set
soreau: AStorm: It was but fglrx is as well
AStorm: soreau, uh
soreau: AStorm: So I have VIDEO_CARDS="radeon fglrx vesa"
AStorm: eselect opengl set xorg-x11
chithead: looks like some drivers need rebuilding after you upgraded xorg-server
AStorm: and retry
soreau: AStorm: Already have that set
soreau: AStorm: Even though ati is still an option
AStorm: chithead, not drivers, dri and glx
soreau: AStorm: I'm remerging mesa atm then I'll try again
AStorm: why are these broken I don't know
chithead: AStorm: evdev is also broken
soreau: and remerging and xf86-video-ati
soreau: evdev and udev I seem to always have some related problem to them
chithead: and consider starting acpid before x
biker_rat: So, anyone know how to get sauerbraten to run on rv515: -f0 option gives me one frame rendered then a frozen system. Without -fo I get one frame per minute with software fallback enabled or a greyed out screen with software fallback disabled. Has anyone solved this conundrum and care to share? I am exausted due to boredom from rebooting my computer and trying variations in driconf.
PuckPoltergeist: hello to the ati community
Enrico[ITA]: love ati community :D
PuckPoltergeist: I hope someone help me with my problems with the AGP r600
PuckPoltergeist: on vt-switch, the X server consumes 100% of CPU
PuckPoltergeist: with: [drm] wait idle failed status : 0xA0003030 0x00000003"
spstarr: hello
soreau: I'm still having a problem, seems fglrx is still loading http://pastebin.com/m167a3671
soreau: eselect opengl list still shows ati (even though xorg-x11 is selected) and fglrx is in use by agpgart
soreau: Is it the libGL.so files from fglrx are still present?
chithead: soreau: I think your problem is gentoo specific, you need to eselect opengl set xorg-x11 and remove the fglrx module from /lib/modules and with rmmod
soreau: What would the module in /lib/modules be called? I see nothing that says fglrx
chithead: soreau: find /lib/modules -name "*fglrx*"
soreau: Ok, that's the kernel module still installed huh
soreau: Thanks chithead, brb
soreau: chithead: I still can't get it right: http://pastebin.com/m2ede5588
chithead: soreau: you unloaded the fglrx module, yes? and eselect opengl set xorg-x11 finished successfully?
soreau: sudo eselect opengl set xorg-x11
soreau: Switching to xorg-x11 OpenGL interface... done
chithead: and the user is in the video group or dri permissions are 0666
soreau: user's not in the video group
soreau: but dri perms are 0666
soreau: But last time I think when I removed fglrx it only showed xorg-x11 and not ati. I think somehow fglrx is still partially installed
soreau: iirc, I removed the libGL.so files and remerged mesa and stuff
soreau: (last time I did this)
soreau: Does having radeonfb modules loaded cause any problem?
airlied: yeah don't do that.
soreau: radeonfb causes problems?
soreau: Or don't rm libGL.so.1.2
jcristau: mode setting in two places causes problems..
jcristau: or, well, two pieces of software thinking they own the hardware
soreau: I see
crdlb: is there a way to avoid thinkpad suspend power drain without radeonfb?
yangman: crdlb: for an older thinkpad?
crdlb: a T42 with an RV200
yangman: that was fixed in kernel years ago, iirc
yangman: unless there were more than one suspend power issues
DanaG: Oh yeah, so with this EXA acceleration, what exactly does that cover? Video rendering, or what?
DanaG: I'm not quite clear on that. =รพ
yangman: common 2D operations: copying regions, colour fills, drawing basic shapes, etc
yangman: video rendering is part of Xv, not EXA
DanaG: ah.
crdlb: it was added to radeonfb in 2.6.18
crdlb: I guess I can try that script on thinkwiki to see if works or not
spstarr: hmm
soreau: So I got the driver going by adding intel agp in the kernel and compiz runs but is extremely slow with either xaa or exa. Can anyone take a look here's glxinfo and X log http://pastebin.com/m24930e02
soreau: Would evdev not loading have anything to do with it?
soreau: Looks like having Alpha Blur enabled was causing things to go slow
spstarr: has never seen blur work ever
soreau: Focus blur is working great here
spstarr: which rxxx?
soreau: r350
soreau: switches back to exa
spstarr: soreau?!
spstarr: soreau: DRI2?
spstarr: AGP or PCIe?
soreau: No, just regular radeon
soreau: no dri2 yet
soreau: agp
soreau: 9600 radeon
spstarr: oh, yeah that works, but i never saw blur work
spstarr: i have a rv350 on laptop
spstarr: the mobile edition
soreau: It was the blur plugin making other things very 'choky' too
soreau: So even having focus blur enabled I guess is a problem
soreau: Now everything's running very smoothly
spstarr: don't VT switch with composite on :)
soreau: Why not?
soreau: I do all the time, works fine
soreau: spstarr: Again, I'm not using all the latest and greatest git stuff
soreau: Just whatever's in portage
soreau: But, I'm getting geared up to try DRI2 here soon enough
soreau: Very excited about it
spstarr: hmm
spstarr: kde or gnome
soreau: neither