happycube: darn... wish i knew about the jtag/smc thing sooner
happycube: i wonder if anyone @ retail's still got a pre-august 09 360 after black friday...
agd5f: airlied: sweet!
ZeZu: get one used for cheap ;p
airlied: agd5f: its a horror show how to support non-kms/kms ;)
airlied: just trying to clean it up sanely now
happycube: zezu - then the trick's finding one that hasn't been updated yet... and isn't too old to rrod
happycube: *old enough to have rrod-issues
airlied: agd5f: I might see if it makes a speed diff on r300->r500 to use a vbo as well
agd5f: airlied: yeah, I wanted to look at that back when I started r6xx accel
ZeZu: happycube, I like to find them already RROD and buy them dirt cheap and fix them
ZeZu: I have 4 salvaged boxes here, all xenon w/ no hdmi though
happycube: cool :)
happycube: bga reflow?
ZeZu: I soldered vga onto one and its pretty nice hooked up to tft though
ZeZu: nah just lame xclamp fix
airlied: oops broke non-kms, will fix that tomorrow and push it out
hifi: speed diff as in "moar fps" or as in "make a diff fast" :o
agd5f: airlied: did you see ossman's patch on bug 24611?
Nightwulf|work: hi all
JohnDoe_71Rus: morning
ZeZu: hi
ossman: airlied, ping
hyc: agd5f: well, the crtc sync polarity fix is progress, I guess
Silas_: hi guys
Silas_: can you help me little
soreau: Not if you don't ask your question a lot
Silas_: what?
Silas_: My question is how to update my ati drivers
Silas_: i'am using Mandriva 2010 Free
hifi: any particular reason why you would want to update them
Silas_: hifi when i try to run any game even the chess 3d in mandriva it runs in slow motion
Silas_: and the compiz effect water is not working correctly
soreau: What do you mean by water not working correctly? Either it works or it doesn't
Silas_: in my laptop when i press Shift & F9 the water drop and stay like it is wet
soreau: and it doesn't dissipate?
Silas_: in my laptop yes
soreau: like it doesn't animate?
Silas_: exactly
soreau: huh
Silas_: on the my laptop everythings is just fine
Silas_: but on my PC
Silas_: no
soreau: You probably should update mesa
Silas_: hmm
Silas_: can you tell me
Silas_: how ?
markinfo: Hi. I have compiled new kernel 2.6.32, newest ATI driver from git and libdrm 2.4.15 My graphic card runs now quite good but there seems to be no 3D acceleration. In Xorg.log is:
markinfo: (EE) RADEON(0): [dri] RADEONDRIGetVersion failed because of a version mismatch.
markinfo: [dri] radeon kernel module version is 2.0.0 but version 1.17.0 or newer is needed.
markinfo: [dri] Disabling DRI.
markinfo: who "needs" version 1.17.0 ?
soreau: markinfo: See the link in the topic. You likely need to make sure to compile libdrm with kms support and compile mesa, ddx and X against that libdrm
Silas_: Package mesa-7.5.2-2mdv2010.0.x86_64 is already installed
soreau: Silas_: mesa is already up to version 7.8 in git
soreau: You should be using 7.6 at the very least
Silas_: hm
Silas_: how to update the mesa?
Silas_: please tell me dude
soreau: You have the option of compiling it yourself or you can ask the mandriva folks if they have any packaged solutions
Silas_: what recommend you?
soreau: If it were me, I'd try to compile latest mesa and see if that helps. but if they're using mesa 7.5, they're likely using other older components too
soreau: I guess you could grab a copy of mesa 7.6 branch
Silas_: and you think that with mesa version 7.6 at least the problem will solve
Silas_: ?
soreau: I have no idea, but it is very possible a mesa upgrade will fix it
markinfo: what means from glxinfo: OpenGL renderer string: Software Rasterizer is it O.K. or not?
amarks: or create another package management nightmare lol
soreau: markinfo: Not ok
soreau: markinfo: You don't have 3D hardware acceleration working
markinfo: mesa and libdrm are two different things?
soreau: yes, see the link in the /topic here
Silas_: soreau i'am not dying so bad for the water effect :) i will let it how it is
Silas_: :D
soreau: Im more surprised new mandriva is shipping old mesa
soreau: Silas_: If you get a wild hair, try Ubuntu 9.10 live cd. Water should work OOTB
eosie: airlied: could you possibly review this: http://old.nabble.com/-PATCH-0-6--r300g%3A-vertex-shader-outputs-mapping-and-rasterizer-setup-rework-ts26550488.html ?
eosie: airlied: Corbin told me he'd like to know from you that it doesn't break anything
soreau: eosie: Is that the same one I have installed?
eosie: yes, it is
soreau: Because I can at least say it doesnt crash anything
Silas_: soreau I already have tried it :) and not just tried.. I Installed it...
Silas_: i uninstalled Mandriva
Silas_: and installed Ubunto 9.10
Silas_: :)
Silas_: butttt no efect dude
soreau: here on my rv350, r300g with eosie patches 2-6 works for compiz, games and xscreensavers with obvious graphical glitches were certain things are not supported yet
soreau: Silas_: What do you mean no effect?
Silas_: I mean that the water effect is the same as in the Mandriva
soreau: hmm
soreau: Which chip do you have again?
Silas_: RV530LE - (Ati Radeon X1650PRo Gamery Edititon)
soreau: Is it 32 or 64bit OS you use?
Silas_: 64 bit
Silas_: but the ubunto was 32 bit
Silas_: and no change
soreau: huh
hifi: I have everything working UMS and KMS with RV515
soreau: Did you try with radeon.modeset=1 for kernel parameter on 9.10?
hifi: on ubuntu 9.10
hifi: desktop effects and everything
Silas_: soreau: no
soreau: and hifi reporting ums working too..
Silas_: hifi: all effects work for me except water
soreau: I wonder what would cause water to show but not animate
hifi: I haven't tried water, one moment
Silas_: soreau: render?
hifi: water works for me
hifi: at least rain
hifi: though it's annoying as hell :p
Silas_: hifi:
Silas_: rain work for me to
Silas_: rain work for me too
hifi: whats the effect that doesn't work
Silas_: water drops but the screen is dosn't get wet...
Silas_: and wiper is not working
hifi: that might be true
soreau: hifi: Best I can tell, he means the water drops render but do not animate
hifi: I can't see anything when I try "Toggle wiper" or "Initiate" for water
hifi: I saw a small wave or something
Silas_: yessss
Silas_: me to
Silas_: me too
soreau: hifi: Try setting the initiate binding to modifier+key instead of modifier+modifier (ie. Super+x)
Silas_: enter youtube and type compiz water effect and you will
Silas_: see how it should work
hifi: soreau: yeah, that works better
hifi: it's a compiz bug, not radeon
Silas_: sec
Silas_: i will give it a tyry
Silas_: i will give it a try
Silas_: nope
Silas_: ...
hifi: try binding it like ctrl+alt+w
hifi: and press them down and move your cursor
hifi: that works for me
Silas_: dudeee
Silas_: this is true
Silas_: but still i don't have the wet effect
soreau: Silas_: I guess you will have to create a video to show us exactly what you mean
hifi: those are probably compiz bugs you are seeing rather than radeon bugs
soreau: If it is what I think, its very strange
soreau: hifi: Not likely
soreau: I have a feeling either it is a mesa bug or something is physically wrong with his card
Silas_: hifi
Silas_: watch and tell me
Silas_: http://www.youtube.com/watch?v=2H-oma_0Y6w
hifi: ok, wiper doesn't work
hifi: but rain and initiate does
Silas_: it should rain and stays wet
Silas_: in our case it only rains and creates a raindrop
hifi: ok, so thats why wiper doesn't do anything
hifi: so, then we can confirm that the "wet" effect of compiz water doesn't work on RV515 and RV530
soreau: What the hell does that mean?
Silas_: :)
hifi: see the youtube video
hifi: the water should "stay" on the desktop after it rains
hifi: I didn't notice that first
soreau: The youtube video shows the effect working, right?
Silas_: on my laptop it is like in the clip
Silas_: yess
Silas_: but in our case it isn't
soreau: We need to see a video of it when it is broken
soreau: I already know what it looks like when it works
soreau: It works fine here, wiper and all
soreau: but I have r3xx
hifi: at 0:30 in that video the water "stays" on the desktop
hifi: I don't know exactly what that even means
Silas_: may be a compiz bug with RV530 i RV515 ?
hifi: does the waves just stay longer visible and clears off just with the wiper
hifi: probably just the wiper is broken
Silas_: after 3 preinstalls ?
Silas_: and with the same distro works on my laptop?
hifi: it rains and the waves slowly dissipate
hifi: but wiper does not clear the waves
hifi: clearly in the video it wipes off the screen, but with RV515 it does nothing
Silas_: yes
Silas_: and the water is not staying...
Silas_: how it should be
Silas_: like in the clip
Silas_: ...\
soreau: make a video already
soreau: This is obviously not something easily explainable with words
hifi: true
Silas_: sec
Silas_: tell me how to make you a video
hifi: I can do it
hifi: uploading
hifi: in the clip I initiated a moment, then I enabled the rain and disabled
hifi: also tried to wipe while it rained
hifi: http://hifi.iki.fi/r5xx-rain.ogg 1.5M
Silas_: hifi
Silas_: same as like me...
hifi: maybe someone will understand now, even I don't know what it should "stay wet"
soreau: So you mean the rippling aftermath effect then?
hifi: s/what/so/
hifi: how
hifi: probably, I have never seen the water effect before
hifi: finally I had something to do with this 3GHz dual core system, record live desktop effects!
Silas_: guys i am from bulgaria... my english is not so well keep that in mind :)
soreau: afaic, that qualifies as working. I have rippling effect after and wiper working on compiz 0.9 and gallium driver fine
hifi: but the wiper didn't work
soreau: But if it is any bug, it is in the r5xx fragment program in mesa
hifi: I tried to wipe during that video
Silas_: the only question is how to fix it
Silas_: when i run starcraft in wine it flickers permanently
soreau: File a bug with those videos and cross your fingers?
Silas_: this is the other problem
Silas_: ...
Silas_: and not just StarCraft
Silas_: World of Warcraft dosn't run at all
soreau: well that is another issue completely
soreau: hifi: Was that on kms or ums?
soreau: Silas_: What is the output of glxinfo|grep renderer ?
hifi: actually I was on kms
soreau: ok
Silas_: direct rendering: Yes
Silas_: OpenGL renderer string: Mesa DRI R300 20060815 TCL
soreau: Silas_: The flickering is fixed by KMS/DRI2/loading the radeon module with modesetting
Silas_: due
Silas_: dude
soreau: Silas_: You should try booting with radeon.modeset=1 which should enable it
Silas_: that will fix and the water effect as wll
Silas_: i will give it a try
soreau: no, it wont fix water
soreau: water is mesa issue
soreau: on r5xx apparently
Silas_: then
Silas_: ..
Silas_: i will not try it :)
Silas_: i don't play the games in Linux
Silas_: i am not installed linux to play games on ti
Silas_: i am not installed linux to play games on it
Silas_: i have windows vista for that purpose
ossman: ajax, ping
cire: how can I determine whether modesetting is enabled?
soreau: cire: If glxinfo shows DRI2 or your X log should tell you
soreau: grep modeset /var/log/Xorg.0.log
cire: thank you soreau
gnuman: Hello, I am using Arch Linux and after latest update (xorg-server 1.6.3.901 -> 1.7.1.902 and mesa 7.5.1 -> 7.6) I have some problems with my RS690M card in laptop. I am using xf86-video-ati 6.12.99.git20091014 and kernel 2.6.31 with UMS (I had lot of artifacts with KMS before, but I didn't tried KMS again after update yet).
hifi: and?
gnuman: The problem is that after some time of running KDE 4.3 with OpenGL composite effects enabled, X server is immediately terminated (I am back at KDM login screen). Without composite effect, this doesn't happen.
gnuman: It did work without problems before update. I have tried to look at Xorg.0.log, but didn't find any error.
gnuman: Is there some known problem with OpenGL composite effects and X server 1.7 + Mesa 7.6 + xf86-video-ati 6.12.99 od Radeon RS690M? Is it fixed in some development Git versions?
hifi: gnuman: you probably have old libdrm
hifi: (that also means the arch devs failed)
gnuman: libdrm 2.4.15
gnuman: hifi: Is that old version?
hifi: in this case, yes
hifi: a new version is not available yet I think which has a fix for the crash that probably happened
hifi: http://cgit.freedesktop.org/mesa/drm/commit/?id=b4312b639d56a6cad78953af0fd4f863182007e3 you need a libdrm with this
gnuman: Thanks, I will try to build libdrm from Git when I get home...
hifi: also you need to build mesa against it
hifi: thats a bit tricky
gnuman: Ah, ok...
gnuman: Btw. should I try KMS again after this update? Should those artifacts with KMS on RS690M be fixed now?
hifi: even better, if it's a issue with missing symbol radeon_bo_is_referenced_by_cs as I think you could ping the arch devs to build libdrm with the patch
hifi: if their mesa already includes the change that requires that function
hifi: I'm not 100% sure, you should see a missing symbol message when X crashes
gnuman: hifi: If it will work, I will definitely fill a bugreport
hifi: first I suggest you find out if it's really that
gnuman: hifi: What about my KMS question? Do you have any information about it (really bad graphics artifacts on RS690M with KMS enabled).
hifi: I have RS690 on an integrated motherboard and it works fine, no artifacts UMS or KMS
glisse: gnuman: can you give the exact reference of your laptop or motherboard ?
gnuman: glisse: MSI PR210X. I am not at home right now, so I can't give you more precise informations...
gnuman: glisse: But like I said last time I tried KMS has been with X server 1.6.3.901 and Mesa 7.5.1, I didn't tried it after update yet...
Ghworg: Argh, 45 minutes of building wasted 'cos I forgot the ChangeLog and thus the packaging failed
glisse: agd5f: when can this happen nb_tom != what lspci says
pendlaren: Hi folks. Running Ubuntu 9.04 on my IBM T60 with X1400 was perfect, but on 9.10 the special effects cause a complete lock; have to cold boot machine. Any suggestions?
Ronis_BR: pendlaren: what driver are you using?
pendlaren: Romis_BR: the /var/log/Xorg.0.log mentions RADEON(0) a lot of times. I've been told that is the driver?
pendlaren: Romis_BR: no Driver line in xorg.conf anymore, so I'm not sure how to change it :-/
pendlaren: Romis_BR: Neither do I remember if I did something special to get it working in 9.04
Ronis_BR: pendlaren: try cat /var/log/Xorg0.log |grep radeon
Ronis_BR: (lowercase)
adamk: pendlaren, That is the correct driver.
Ronis_BR: adamk: radeonhd gives RADEON(0) things also, doesn't it?
adamk: No. It gives RADEONHD :-)
Ronis_BR: ops
Ronis_BR: :0
pendlaren: Romis_BR: (II) LoadModule: "radeon"
pendlaren: (II) Loading /usr/lib/xorg/modules/drivers//radeon_drv.so
pendlaren: (II) Module radeon: vendor="X.Org Foundation"
Ronis_BR: :)
Ronis_BR: yes it is radeon
Ronis_BR: and my nick is RoNis_BR :)
Ronis_BR: pendlaren: what do you mean special effects? compositing?
pendlaren: oops... sorry - bold font, hard to read. Will try to behave!
Ronis_BR: :D
Ronis_BR: pendlaren: you mean you can start X but when you turn compositing on that it locks?
pendlaren: Romis_BR: basically compiz, e.g. a feature like Mac OS X where you move pointer to one corner and get a view of all open windows
pendlaren: Ronis_BR: ARgh...I misspelled your nick AGAIN!
Ronis_BR: :D
Ronis_BR: np
Ronis_BR: pendlaren: can you check if opengl is working? like, try to start glxgears
pendlaren: Ronis_BR: Wheels are turning, but system became very slow - glxgears runs smooth, though
Ronis_BR: pendlaren: try glxinfo
Ronis_BR: glxinfo |grep direct
pendlaren: Ronis_BR: direct rendering: No (LIBGL_ALWAYS_INDIRECT set)
Ronis_BR: I think that is the problem
Ronis_BR: right adamk? :)
adamk: No, that's normal if the terminal was spawned by compiz since compiz is being run with that variable.
pendlaren: Ronis_BR/adamk: Actually compiz IS running, but it leads to trouble if I try to do too much
Ronis_BR: ah
Ronis_BR: so it is normal
Ronis_BR: I though compiz was turned off
pendlaren: Ronis_BR: Yes, sorry, that was a very reasonable interpretation...
adamk: pendlaren, Does your entire machine lockup? Have you tried remotely accessing that machine via ssh?
Ronis_BR: pendlaren: so, 3D accel is working bt when you use too much desktop effects it hangs?
pendlaren: adamk: No, didn't try ssh. Could do that.
pendlaren: Ronis_BR: Actually, sometimes I am able to enter Console and restart gdm service
Ronis_BR: so, maybe only X is hanging
Ronis_BR: man, I have no idea, have you tried on #ubuntu to see if anyone else got this problem?
pendlaren: someone at #ubuntu suggested that I tried booting with kernel param radeon.modeset=1 but that was no success - colors faded and machine locked when I entered the "Visual effects" tab
hifi: (WW) RADEON(0): Option "BusType" is not used
hifi: eh
hifi: I wanted to force my unstable AGP card to PCI
pendlaren: I also got some suggestion on changing to the xorg-edgers driver, but not sure how I do that (installed package, but didn't notice any driver chage)
adamk: pendlaren, Is this AGP or PCIe?
pendlaren: adamk: PCI, I assume, since it is listed in lspci?
adamk: AGP devices show up there, too.
hifi: AGP and PCIe cards are also listed with lspci
adamk: pendlaren, Can you pastebin your /var/log/Xorg.0.log file and the output of 'glxinfo' ?
pendlaren: adamk: Sure
hifi: why was BusType ignored or was just my xorg.conf not read?
adamk: hifi, No idea. Are you sure it's in the correct section?
hifi: it's ubuntu's default bare xorg.conf, I just added it to the "Device" section, it's probably not reading it then
jcristau: hifi: using ums?
pendlaren: adamk: http://www.pastebin.org/58724 [sorry for the delay, but Firefox crashed]
hifi: jcristau: kms
jcristau: hifi: BusType in xorg.conf is only for ums
hifi: that explains
hifi: can I force bustype to PCI in kms somehow?
jcristau: kms needs radeon.agpmode=-1
hifi: ah, thanks
jcristau: in kernel command line
adamk: pendlaren, Well it's PCIe. I'm not sure what's going on, but if you have been able to switch to a console and restart X, it's quite possible that just compiz is crashing for some reason.
pendlaren: adamk: Right. I've tried so many things that I'm not able to remember what caused machine lock and what caused X lock :-/
hifi: btw. libdrm needs a release
pendlaren: adamk: But you think I should keep the current driver, not go for radeonhd?
adamk: You can try, but since the 3D code is provided by Mesa, and is the same whether you use radeon or radeonhd, I'm not convinced it will make much difference.
pendlaren: adamk: OK, I see. I just wonder why it worked well under 9.04.. Perhaps I did install some proprietary driver, then..
pendlaren: adamk: I've got a very bad memory, unfortunately
adamk: pendlaren, Nah, even in 9.04 it would have only worked with the open source drivers.
adamk: pendlaren, But both compiz and the open source drivers have changed between 9.04 and 9.10.
pendlaren: adamk: Right. They shouldn't fix what is working ;)
hifi: wait, did the ubuntu people change the agpmode option too?
hifi: I enable modeset with radeon.modeset=1
adamk: pendlaren, I suggest that next time this happens, drop to the console (if you can) and copy your /var/log/Xorg.0.log file somewhere. Also, grab the output of 'ps ax | grep compiz' to see if compiz is still running.
pendlaren: adamk: Thanks, and than you very much for your time!
pendlaren: Ronis_BR: That goese for you too :)
Ronis_BR: pendlaren: :)
hifi: or how do I know PCI mode was enabled for kms?
hifi: ah, drm stated it forced AGP to PCI
hifi: maybe the lockups will be gone now
hifi: though I'm quite sure it's the motherboard itself
Neo_The_User: Hi. I need help. KDE with desktop effects now completley sucks as of just recently.
hifi: sucks as in X server crashes?
Neo_The_User: I did this: http://www.phoronix.com/forums/showthread.php?t=17603 and movies play good and stuff but.... i end up with draw backs similar as to the crappy via code.
hifi: oh, no
Neo_The_User: like when i move a window, i see like 500 of the same window
Neo_The_User: but if i turn on wobby windows, it goes away but the inside of the window is messed up... so... i dunno
Neo_The_User: *wobbly
Neo_The_User: oh this dude has the same problem as i do! http://www.phoronix.com/forums/showthread.php?t=20532
Neo_The_User: i just asked in #kde and they said they didnt break it
Neo_The_User: 'We ain't did touch nothin at all"
Neo_The_User: is anybody here>
Ghworg: Neo_The_User: Revert 286bf89e5a1fc931dbf523ded861b809859485e2 in mesa and it will work
Neo_The_User: i cant find the git web for that one
BioTube: Neo_The_User: it's two weeks old
BioTube: http://cgit.freedesktop.org/mesa/mesa/commit/?id=286bf89e5a1fc931dbf523ded861b809859485e2
Neo_The_User: Thanks buddy!
stikonas: why nobody fixes this regression?
stikonas: after it is bisected, it should not be too hard to fix it
Neo_The_User: i no rie!
Ghworg: wine no longer crashes when using fbo with high-res textures, yay
Ghworg: It gets 0 fps, but at least it doesn't crash
Ghworg: The car demo in 3dmark2001 shows amusing behaviour now.
Ghworg: The textures on the ground ahead of the car are fine, but behind it they are black. It's as if the car is bringing eternel darkness to the land by driving over it :-)
Neo_The_User: wine is too slow for me. thats why i have a dual boot on win 7 64 RTM
Ghworg: I just like running 3dmark to get a nice visual representation of progress in the drivers
Neo_The_User: oh..
Neo_The_User: i thought you meant for like hardcore intense online gaming
Neo_The_User: and no you didn't imply it. i mis interpretate and miss read things. sry
Neo_The_User: do? ati chips do everything with an ALU with free shift
Neo_The_User: not DDX related. just a question
MostAwesomeDude: The shader core's a bit more complex than just an ALU.
MostAwesomeDude: There's usually dedicated reciprocal and exponent circuits, and of course the texture fetcher is an entire thing to itself.
MostAwesomeDude: But I don't know too much about the deep innards of the implementation, just how it's programmed.
tmoertel_work: After upgrading to Fedora 12 (from F10), my video is so slow that I can't really use my system productively. I've tried both radeon and radeonhd drivers and have the same problem: "(WW) RADEON(0): Direct rendering is not supported when VGA arb is necessary for the device"
tmoertel_work: I've tried using KMS, but when I do, I get a kernel oops: https://bugzilla.redhat.com/show_bug.cgi?id=540593
tmoertel_work: When I turn off KMS (nomodeset), I get the VGA-arb warning and no DRI, and my system renders with Molasses Vision 3000.
tmoertel_work: Is there any way I can get around the VGA-arb/DRI issue?
tmoertel_work: Is there any way I can get faster rendering w/out using DRI?
logari81: tmoertel_work: I suppose you should pastebin your /var/log/Xorg.0.log to get help from the people here :)
MostAwesomeDude: Goddamn ES1000.
MostAwesomeDude: tmoertel_work: Can you turn off the onboard?
tmoertel_work: MostAwesomeDude: No, it's a Dell server. The onboard video cannot be turned off, neither by BIOS nor jumper.
MostAwesomeDude: tmoertel_work: Hm. Well, it *is* correct in that you can't DRI as long as the VGA arb is involved.
MostAwesomeDude: And it has to be involved, because of the ES1000. Hm.
tmoertel_work: The funny thing is, back when I was using Fedora 10, I didn't have this problem. Somehow, radeonhd worked like a champ. (Again, no KMS.)
MostAwesomeDude: This is a question for airlied. He'll probably comment directly on your bug.
tmoertel_work: That's what I'm hoping. I cc'd him on the bug report. ;-)
MostAwesomeDude: Yeah, the whole KMS thing is good in that multicard magically started working for a lot of people.
MostAwesomeDude: But, on the other hand, rn50 (ES1000) bites hard.
tmoertel_work: My /var/log/Xorg.0.log: http://gist.github.com/245579
MostAwesomeDude: tmoertel_work: Re-reading the report, I get the feeling that you'd be okay with just magically skipping over the rn50, right?
tmoertel_work: MostAwesomeDude: yeah, if I could get the onboard video to be ignored, that would be great.
MostAwesomeDude: tmoertel_work: Alright. I don't think there's actually an option for that ATM, but there should be.
tmoertel_work: I wonder how the older driver managed to avoid the VGA-arb/DRI problem.
MostAwesomeDude: Oh, the VGA arb didn't exist.
MostAwesomeDude: So devices would kind of just stab each other repeatedly until one of them gave up. :3
MostAwesomeDude: Actually, the ES1000 probably just sat there and never got flicked on.
tmoertel_work: Any way to just disable the VGA arb?
MostAwesomeDude: I had thought that it was its own module (vgaarb), but I'm not seeing it in my list. Maybe it's part of the drm module?
tmoertel_work: I saw a Git patch float by not too long ago that seem to enable one to compile it out.
tmoertel_work: I don't know whether that patch made it into the Fedora source, however.
MostAwesomeDude: Whoa.
MostAwesomeDude: I just realized my other video card is enabled.
MostAwesomeDude: That's unexpectedly awesome.
MostAwesomeDude: Anyway, Fedora wouldn't enable that; for the vast majority of cases, the VGA arb makes life better, not worse.
MostAwesomeDude: You're just stuck with this stupid little onboard chip that can't be flicked off.
MostAwesomeDude: Anyway, airlied would know this. I'm so incredibly not involved with this stuff usually.
tmoertel_work: MostAwesomeDude: I mean, did they include the patch in their upstream sources so that I could rebuild the RPMs using my own flags to disable VGA arb.
tmoertel_work: MostAwesomeDude: yeah, maybe I should just ping airlied.
tmoertel_work: airlied: you around?
MostAwesomeDude: I think .au is still asleep.
tmoertel_work: MostAwesomeDude: right.
tmoertel_work: MostAwesomeDude: I'll try back later, then. Thanks for your help! :-)
MostAwesomeDude: tmoertel_work: NP
agd5f: tmoertel_work: using kms? the problem is the drm module doesn't deal with chips with nothing connected properly at the moment
tmoertel_work: agd5f: I can't use KMS: https://bugzilla.redhat.com/show_bug.cgi?id=540593
agd5f: tmoertel_work: right. you can't use kms because the drm module doesn't deal with chips with nothing connected properly at the moment
tmoertel_work: agd5f: Ah! You're saying I should just hook up a monitor and see if it works?
agd5f: tmoertel_work: it might
tmoertel_work: agd5f: It's worth a try. I'll see if I can find a monitor...
tmoertel_work: agd5f: maybe I can use the VGA port on my existing monitor.
MostAwesomeDude: tmoertel_work: Wait, when you said DRI wasn't working, did you mean on the ES1000, or on the nice card?
MostAwesomeDude: 'Cause ES1000 doesn't even have a 3D engine.
tmoertel_work: MostAwesomeDude: even on the nice card.
tmoertel_work: MostAwesomeDude: there's nothing hooked up to the ES1000.
agd5f: the drm modules will oops when it loads on the card with no monitors connected
MostAwesomeDude: That's unfortunate, since I know that that oops is just because nobody thought of a better course of action. :3
agd5f: MostAwesomeDude: yeah
tmoertel_work: agd5f: I've hooked the ES1000's VGA out into the VGA port of my LCD monitor. Now I'm going to reboot, hoping that KMS will now work.
tmoertel_work: here goes nothing . . .
tmoertel_work: agd5f, MostAwesomeDude: thanks for your help!
MostAwesomeDude: agd5f: Would it just be better maybe to turn off the card in that case?
MostAwesomeDude: Or is it possible to hotplug connectors still?
agd5f: MostAwesomeDude: need to decide on some default when nothing is connected
MostAwesomeDude: I like airlied's plan of "if (rn50) light_up_everything(); else turn_off_card();
agd5f: MostAwesomeDude: probably should just load then not attempt to set a mode
MostAwesomeDude: agd5f: Would hotplugging outputs still work in that case?
agd5f: MostAwesomeDude: sure. the drm should be fine. it's how to deal with console and X that gets problematic
tmoertel_work: When I enabled KMS and rebooted, I got the same oops.
tmoertel_work: Maybe I need to plug the ES1000's output into a separate monitor?
agd5f: tmoertel_work: you could blacklist the es1000
tmoertel_work: agd5f: how could I blacklist it?
agd5f: so that the drm won't try and load on it
agd5f: tmoertel_work: not srue if you can blacklist a pci id. might have to remove the id from the list claimed by the radeon drm
marvin24: this one keeps crashing my machine (rs780/kms/2.6.32-rc8+drm-linus) every few hours: http://pastebin.com/m30bf4685
tmoertel_work: agd5f: so I would need to patch the radeon drm source and rebuild?
agd5f: tmoertel_work: yes
marvin24: forgot the first line: "BUG: unable to handle kernel paging request at ffff880048c13930"
tmoertel_work: agd5f: ok. well, I'll need to find a machine with good video to do that kind of hacking. right now, I'm running in text mode.
hnsr: should I worry about mtrr errors on dmesg? or are they harmless
agd5f: tmoertel_work: it's a one liner. just remove the line for the es1000 in include/drm/drm_pciids.h
hnsr: i seem to get these when radeon module is loaded
hifi: hey, I'd like the light_up_everything() option
hifi: sometimes I have no monitor attached to my HTPC when I boot it
tmoertel_work: agd5f: I'm running Fedora, so I need to download the package sources, add a patch, and then rebuild. (I must replicate this solution on a number of machines.)
osiris: MostAwesomeDude: texture size calculation in cs checker is currently broken for compressed textures
ossman: agd5f, irqs? :)
agd5f: ossman: just about to send the patch
ossman: agd5f, woot
glisse: 0fps+nobugs->software
ossman: agd5f, against what tree?
agd5f: ossman: airlied's
ossman: drm-next?
agd5f: radeon_dp, but it should apply cleanly to any of them
glisse: agd5f: btw do you know why rs400 report an aperture space 2 time bigger than stolen vram ?
agd5f: glisse: pci aperture sizes are often wrong
agd5f: some 128 MB cards report 64, etc.
agd5f: some 8 MB cards report 16, etc.
glisse: yeah, this leads me to wrong track while hunting down my rs400 issue
agd5f: ossman: patches sent
deus: If i understood this rite
deus: R300 cards will be supported in 2.6.32
adamk: R300 cards have been supported for years now.
deus: oh
deus: 9600 is a R300 card no?
adamk: Yes, it is from the r3xx family.
deus: huh
deus: well
deus: specificly TV-out
Zajec: agd5f: !!!!!
Zajec: :D
Zajec: agd5f: thanks :)
adamk: deus: I do not know if tv-out is supported.
deus: but..weird. i heard the new 2.6.32 kernel would be adding more support
adamk: deus: It adds support for KMS.
deus: Kernel mode setting?
hifi: yes, but it's been there for r3xx family for some time now
hifi: I think?
deus: huh
hifi: 2.6.31?
deus: well
deus: im upgrading to it now
hifi: or even 2.6.30, I can't remember
deus: now i got 2.6.28
deus: hmm
deus: sweet
deus: this might go easier then i excpected
deus: RV350 ok?
deus: Is that still in the area of R300
happycube: why'd they even bother to make a rn50/es1000 - couldn't the rv100 been made small enough w/a die shrink?
adamk: deus: Yes.
ajax: happycube: also remember that rn50 may or may not have a working 3d engine.
deus: Do i need the DRI drivers after i get the new kernel?
ajax: you get better yield if you don't have to make sure that works.
MostAwesomeDude: May or may not? I thought they never had them.
spreeuw: deus: libdri, xf86-video-ati, mesa
spreeuw: all from git
spreeuw: built in this order
MostAwesomeDude: spreeuw: Uh, no.
spreeuw: dont worry its quite stable
MostAwesomeDude: He needs a KMS kernel, too.
ajax: MostAwesomeDude: there's a 3d engine _there_. it just might not work.
MostAwesomeDude: ajax: Oh. Weird.
spreeuw: 2.6.32-rc8 is ok
ajax: i've had gears running on one, but apparently i got lucky
deus: MostAwesomeDude: A module? or is it a specific kernel?
MostAwesomeDude: ajax: Oh, BTW, there's some paperwork funkiness, so I don't get paid during Winter Break for work. Guess which oft-postponed thing I get to work on during break? :3
spreeuw: deus: get 2.6.32-rc8 or better
spreeuw: its easier
MostAwesomeDude: deus: Please *please* build a full kernel if you're trying out KMS. Out-of-tree modules probably won't work.
deus: well, cant hurt to try before im bothering to compile an entire kernel
spreeuw: hmm 3d clouds seem to have zbufer issues in flightgear
spreeuw: or its broken bumpmapping
MostAwesomeDude: deus: Sure it can. :3
MostAwesomeDude: Well, it won't hurt, but it'll be a waste of everyone's time.
hifi: deus: do you really need to compile, doesn't your distribution offer dev packages?
Zajec: what is "case 181: /* CP EOP event */" ?
deus: hifi: probably its ubuntu
hifi: deus: then you could use packages from karmic or upgrade to karmic from jaunty
hifi: karmic has KMS support for R3xx family out of box
deus: Thats what im currently upgrading to now
agd5f: Zajec: End of Pipe interrupts
hifi: deus: full upgrade to karmic recommended
hifi: then you just need to boot with "radeon.modeset=1" kernel option and you should have working KMS as I do
agd5f: deus: for kms tv-out you need 2.6.32. ums has supported tv-out for ages
hifi: OpenGL renderer string: Mesa DRI R300 (RV350 4150) 20090101 x86/MMX+/3DNow!+/SSE TCL DRI2
hifi: agd5f: ah, sorry, I didn't know that
Zajec: agd5f: thanks
hifi: deus: then, upgrade to karmic and get a kernel from lucid :))
Zajec: what is VLINE interrupt? is that end of vblank?
Zajec: and start of displaying new frame?
deus: hifi: so im back to my original plan then :)
deus: but UMS?
hifi: deus: UMS what?
deus: " ums has supported tv-out for ages"
hifi: yes, you don't have to upgrade for that
deus: whats UMS?
agd5f: Zajec: you can get an interrupt at a particular line in the vtrace
Luzipher: deus: UMS = User Mode Setting ... the old (and still usual) way of doing things before modesetting went into the kernel
deus: ah
spreeuw: deus if you will build yourself check this http://www.phoronix.com/forums/showthread.php?t=20484
Zajec: how can I sync with VBLANK? should I use drm_wait_vblank?
deus: To think i was going to reformat to debian, so i could get xorg 7.4 so i could get an unsupported binary drivers :)
Zajec: drm_wait_vblank it hard to understand, it's documented with other arguments that I actually takes
muep_: can the current free graphics stack do antialiasing?
Ronis_BR: why I can't use KMS if uvesafb is built-in on kernel (but not loaded at startup)
Zajec: :q
spreeuw: Ronis_BR: dunno but KMS replaces it
yangman: Ronis_BR: having it built-in means it's being loeaded
Ronis_BR: spreeuw: when uvesafb is built-in, if I remove nomodeset on grub line than the screen was black :(
yangman: Ronis_BR: uvesafb, that is. you can't be running both
Ronis_BR: yangman: but I did not but video=uvesafb
Ronis_BR: yangman: so it isn't loaded right
ossman: ajax, airlied, would it be possible to get a koji kernel build with agd5f's patch and my patch? :)
ajax: if i knew which patches you were talking about, maybe.
ossman: ajax, I'll get some links
glisse: irq ones i guess
adamk: Wow... Moving a Folder View plasma widget around on the right hand screen in a dualhead setup is painfully slow on my x850. If I move it to the left screen, it's just fine.
ossman: ajax, here's one: http://bugs.freedesktop.org/attachment.cgi?id=31552
adamk: wonders what it would be like on an HD3450....
ossman: ajax, and then agd5fs stuff: http://marc.info/?l=dri-devel&m=125960757404659&w=2 and http://marc.info/?l=dri-devel&m=125960785305050&q=p3
agd5f: adamk: if your desktop is larger than the max texture size you are likely falling back to software
adamk: Even without any composite manager running?
deus: You are sure i need the 2.6.32 kernel for tv-out?
agd5f: deus: for kms yes
agd5f: adamk_: yes
adamk: Well that sucks :-(
ossman: ajax, think that's doable?
deus: agd5f: any alternatives?
agd5f: deus: don't use kms
agd5f: deus: non-kms has supported it for years
deus: UMS, that is fidgeting with the xorg.conf?
deus: or am i way of
deus: f
Zajec: agd5f: do you have who I can ask about syncing with vblank?
agd5f: deus: just means modesetting is done inthe ddx rather than the drm. it can still be configured on teh fly with xrandr
airlied: eosie: my review says push it and I'll fix it later if it braks anythig ;-)
agd5f: Zajec: what do you want to know?
Zajec: agd5f i thoguht of drm_wait_vblank but it needs drm_file as argument
agd5f: Zajec: what are you trying to do?
Zajec: agd5f: i want to wait for VBLANK with reclocking engine
agd5f: Zajec: follow mjg59
agd5f: 's old patch as a guide
Zajec: agd5f: mjg59 used internal radeon way for that
Zajec: agd5f: i wanted to use drm infrastructue
Zajec: agd5f: drm's vblank handling
agd5f: Zajec: in the interrupt handler, schedule the reclock when processing the interrupt
Zajec: agd5f: yeah, i know i can do this that way
Zajec: agd5f: i thought it would be nicer to use drm's "framework" if I can call it that way
agd5f: Zajec: I don't think that will work
Zajec: agd5f: that's probably why we call drm_vblank_hanldle
airlied: glisse: there are two apertures insdie the aperture that the gpu can make do different things
airlied: ossman: I'll do builds later
airlied: ossman: generally for Fedora I want the patches in my upsteram tree first
masa-: was that some shader compiler rewrite work a while back just for intel or does that affect radeons too?
airlied: ossman: can you send your patch to the mailing list?
airlied: I don't pick patchse out of bugzilla, make sure it has signed off by in it
ossman: airlied, dri-devel? or lkml?
ossman: patch already has a sob though
ossman: and an ack
agd5f: ossman: dri-devel
ossman: agd5f, airlied, patch sent
deus: agd5f: do you have any litterature i can look at on howto go about doing this?
agd5f: deus: http://wiki.debian.org/XStrikeForce/HowToRandR12
deus: agd5f: is any of this depented on the 2.6.32 or 2.6.31 kernel?
agd5f: deus: no. you only need 2.6.32 for tv-out with kms.
ossman: agd5f, was there no code for it up until then? or some bug?
agd5f: ossman: ?
ossman: the kms tv-out
agd5f: tv-out was ported to kms after 2.6.31 was released
ossman: I thought that was in with more or less the first versions of the atombios code
agd5f: ossman: ums has had tv-out support for ages
ossman: ah
deus: Even for R300
taiu1: übüü
deus: "the ATI driver on r200/r300/r400/r500/r600 boards (since 6.7.191) where RandR 1.2 replaces MergedFB, " they are talking about the free drivers rite?
adamk: rnoland: I take it that r6xx is working again?
adamk: deus: Yes.
agd5f: deus: yes
ossman: agd5f, did my mail get anywhere? archives seem to be slow at updating
agd5f: ossman: I got it
ossman: then it should be out of my hands at least
deus: the xorg.conf on my machine (ubuntu karmic) seems to have some sort of placeholder for its configuration, is that then under KMS or such?
agd5f: deus: don't need to worry about your xorg.conf
agd5f: you can configure it all with xrandr
deus: you mean --addmode and that jass
rnoland: adamk: um, i think so...
rnoland: libdrm is fixed. and mesa is building for me...
agd5f: deus: you can try enabling load detection
rnoland: don't think that i've installed today
agd5f: deus: xrandr --output S-video --set load_detection 1
adamk: Cool.
agd5f: xrandr --auto
rnoland: adamk: i was getting a page fault last week, which i have a patch for....
rnoland: adamk: not sure what caused it to start faulting in radeon_cp_texture() for me when i started any game, but....
adamk: agd5f: FYI, it doesn't appear to be MAX_TEXTURE_SIZE related. On an HD3450, I'm getting the same problem, but in reverse. Moving the widget is slow on the left monitor, but fast on the right.
glisse: airlied: yeah i know that
glisse: but i don't have doc to tell me if it's true on rs400 too
glisse: rs690 doesn't seems to have that
airlied: glisse: all gpus have it
airlied: glisse: before r600
airlied: it jsut not always enabled
glisse: well as i didn't see the reg on rs690 i assumed it was removed for ig[
airlied: I though rs690 had HDP_APER_CNTL
deus: agd5f: what am i suppose to look for?
glisse: not in the doc i have
agd5f: deus: whether S-video is listed as connected
rehabdoll: does anybody know of any benchmarks comparing 4670 and newer cards?
rehabdoll: every damn benchmark i can find only compares cards of the same gen.
otaylor: rehabdoll: it's not relevant to the current open source drivers
rehabdoll: yes i know, but i figured theres lot of knowledge here
rehabdoll: im inquiring regarding wintendo :)
rehabdoll: i have recently gotten addicted to modern warfare 2 :(
otaylor: rehabdoll: I think the iniital reviews on tomshardware had r7xx vs. r8xx benchmarking - tomshardware also has massive charts across different cards
rehabdoll: cheers, im not really into the hardware sites these days. ill check it out. thanks!
otaylor: tomshardware is pretty old-skool for hardware sites - they've been around at least a decade
adamk: agd5f: Hah... It's the wallpaper I had selected (which was a rotating globe showing a satellite image of earth). If I change it to a static image, the widgets work fine.
deus: adamk: xrand say its not connected
adamk: deus: Sorry, I really don't know anything about tv-out. I've never used it before.
chithead: rehabdoll: there exist benchmarks comparing 4670 against nvidia gt220/240. but the midrange r8xx are not released yet, so no benchmarks against newer radeons
agd5f: deus: then you can force it on with xrandr
deus: agd5f: how do you do that?=
rhodan: Can I still have tear-free video with QXGA and a Mobility X1400?
deus: agd5f: nevermind im findit it ou
agd5f: deus: xrandr --newmode 800x600 38.25 800 832 912 1024 600 603 607 624 -hsync +vsync; xrandr --addmode S-video 800x600; xrandr --output S-video --mode 800x600
adamk: rnoland: Alright, I just updated and can confirm that everything is working here fine.
rnoland: cool
airlied: glisse: the patch seems to check crtc enabled for turning them back on
glisse: oh i missed the if (enabled)
mroconnor: im having some wierd kms issue with dp support
glisse: need to get some sleep
mroconnor: i get a hi-res screen only in the upper left hand corner of my laptop
mroconnor: well it isnt really hi-res compared to what it was a few days ago ;)
airlied: I think there are some fixse I need to apply still
mroconnor: i see alot of this in dmesg: [drm] got edid (null) from DP
mroconnor: airlied: no worries
mroconnor: I just wanted to let you know
airlied: mroconnor: there are two patchse on dri-devel from agd5f
airlied: I think they fix that
mroconnor: airlied: thanks I'll have the guys merge it in
deus: agd5f: X Error of failed request: BadName (named color or font does not exist)
agd5f: deus: what did you type?
Ronis_BR: airlied: it seems that tuxonice just doesn't work with KMS+Radeon, it in fact is working with KMS+intel
deus: agd5f: xrandr --newmode 800x600 38.25 800 832 912 1024 600 603 607 624 -hsync +vsync; xrandr --addmode S-video 800x600
agd5f: deus: can you pastebin the output of xrandr
deus: http://pastebin.com/m9de26c8
tmoertel_work: airlied: would it be sane of me to rebuild the kernel, patched to remove support for the ATI ES1000, to get around a boot-time oops when the kernel detects the chip? (see my earlier bugzilla link for details)
tmoertel_work: (I ask because I'm rebuilding such a kernel in the background, and if it's a bad idea, I'd rather not wait it out ;-)
agd5f: deus: just the output of 'xrandr'
agd5f: tmoertel_work: yes, it's fine
deus: agd5f: hang on im into the middle of something here, i dug up a s-video cord and connected that dirrectly to the tv through its own s-video input, what i was using before that was svideo to scart
deus: now i got something
agd5f: deus: ok
tmoertel_work: agd5f: thanks. (I already rebuilt the kernel once, but forgot the "--with firmware" option for rpmbuild, so rebuild again :-(
icewaterman: hm, my radeon hd2600 is much slower on linux than my radeon 1600
icewaterman: approximately by factor 10, even though both have dri enabled and working
deus: agd5f: http://pastebin.com/m359b2170 thats how it looks now
adamk: icewaterman: On the HD2600, what's the output of 'glxinfo | grep -i render' ?
deus: but the screen is flickering
deus: you can see part of it
icewaterman: adamk: dunno, switched back to the 1600 to have more speed
agd5f: deus: what tv standard do you need? pal or ntsc?
agd5f: deus: you may need to change it
icewaterman: adamk: this is ubuntu karmic
adamk: icewaterman: Well we can't exactly debug the problem, or even tell you if the drivers for the HD2600 were installed properly, if you aren't using the card :-)
icewaterman: adamk: same driver setup than with the 1600
deus: agd5f: if i set it to pal i loose the signal
adamk: icewaterman: But there are different drivers for the two cards, so it can't be the same driver setup :-)
icewaterman: 1:6.12.99+git20090929.7968e1fb-0ubuntu1 <-- seems to be from 20090929
agd5f: deus: try turnign it off then on after changing it
deus: sorry if i set to ntsc i loose the signal
agd5f: deus: what standard is your tv?
deus: well xrandr says it is connected
deus: agd5f: i...dont....know
icewaterman: anyways, i'll stick with the radeon 1600 for now.
deus: cant remember what the standard my country used
deus: is PAL used in USA?
agd5f: deus: if you live in teh us or japan it's ntsc, most other places is pal
deus: or was it vica versa
icewaterman: deus: nope, they have NTSC
adamk: icewaterman: OK...
deus: what about europe
deus: noray
deus: well
icewaterman: deus: most countries have pal
deus: it is a advanced flatscreen tv
icewaterman: in europe
deus: pal it is
adamk: Does Ubuntu 9.10 even ship with the r6xx driver?
icewaterman: adamk: direct rendering was set to yes
adamk: icewaterman: That doesn't mean anything.
adamk: icewaterman: The software rasterizer will give you direct rendering.
icewaterman: adamk: oh, well, i'll continue to use my 1600 and switch over to my 2600 together with the next ubuntu release
icewaterman: maybe its gonna work then
icewaterman: originally i switched to the 2600 to gain performance, not the reverse.
deus: well
deus: this is moot anyway
icewaterman: anyways, gtg
icewaterman: bye
deus: The s-video cable is to short
deus: i had to drag my tv across the room and even then it the cable was stretched in ht emiddle of the air
deus: i need to focus on the svideo-scart cable
Vash63: Sometime in the last week or so from the git driver, my 4870 X2 stopped booting into Xorg when KMS is enabled.
deus: agd5f: http://pastebin.com/m4741af59 anyway here is the current xrander detailed xrandr output if you are interested
Vash63: xorg.log doesn't seem very useful, but hopefully tere's something here... http://dpaste.com/127197/
Vash63: It seems to just hang and never produce an error.
agd5f: Vash63: probably a missing symbol. you need a newer libdrm
Vash63: Guess I should try with a vanilla kernel first, just weird that it happened after a xf86-video-ati update and not a kernel update.
Vash63: One from the git repo, or is there newer than 2.4.15?
agd5f: Vash63: you need libdrm from git
Vash63: I'll try that, thanks.
deus: agd5f: might it be time to consider the KMS solution?
agd5f: deus: you can try it, but the code is pretty much the same
deus: it must be pal
deus: ^ignore that
Ronis_BR: deus: :D
deus: uh strange
deus: i just replaced the scart cables
deus: and
deus: it works
deus: well
deus: again
deus: i have the flickering
deus: but i get sound
Zajec: err "[PATCH] radeon: Disable both CRTCs during mode switch"
Zajec: will i get blinking on mode switch?
Zajec: for example turning on my second monitor will make first blink?
deus: i miss ctrl alt backspace
AndyWas: zajec: You see Christian's HDMI audio patch yesterday?
AndyWas: I know this should be in #radeonhd but you are not there...
AndyWas: I tried it but get some git errors
AndyWas: If I pastepin what I am getting maybe you can see where I am going wrong?
Zajec: AndyWas: i saw, but didn't test, I don't have RV730/770
Zajec: AndyWas: i'm going sleep
Zajec: AndyWas: please,post on ML
Zajec: AndyWas: it's second time you catch me right when I'm done for day :)
Zajec: night :)
AndyWas: OK...I let you sleep now!
deus: well
deus: I DID IT
deus: sorta
deus: but yes
deus: mplayer is getting a clear picture
airlied: tmoertel_work_: I should have a look at fixinfg the oops
airlied: iit doesn't look too hard
airlied: whats the bug num? and can you attach a full dmesg if not already
Oipo: So I tried adding the r6xx/r7xx interrupt handler to the kernel via the patch, placing the firmware in /lib/firmware/radeon/R700_rlc.bin. And when booting the kernel, it hangs on requesting the firmware.
deus: the colors are not rite
airlied: tmoertel_work_: oh wierd we can't find the ROM for the es1000 by the looks of that
airlied: tmoertel_work_: can you get lspci -vv and dmesg from nomodeset boot?
t0dd: I have a Radeon HD 3450 video card, and I'm running Fedora 12. Whenever I set the device driver in xorg.conf to "radeonhd" or "radeon", X gets hung up. I'm using a generic vesa driver right now, but I'm not sure what the problem is.
spreeuw: should work
spreeuw: try disabling compiss
soreau: spreeuw: Hey hey now
airlied: t0dd: remove xorg.conf
airlied: you shouldn't need one
stikonas: Oipo: put firmware in kernel source firmware/radeon and add those files to firmware/Makefile nad compile firmware into kernel, this works fine for me
t0dd: airlied: how will X know which drivers to use then?
jcristau: t0dd: it knows what hardware you have
jcristau: so it can do the maths
t0dd: so the mere fact that the xorg.conf exists is screwing things up?
airlied: t0dd: its all autodetected
airlied: t0dd: no I'm just guessing its a good place to start
airlied: since F12 works on my hds
t0dd: alright, I'll give that a try, thanks.
airlied: if it still doesn't work then try adding nomodeset on the kernel command line
t0dd: okay, thanks again.
t0dd: we'll see what happens.
Oipo: stikonas, Alright, I'll try that.
Oipo: Thanks.
airlied: ossman: something trashed all the whitspace in that patch
airlied: I've fixed it up locally
t0dd: Whenever I boot Fedora, I have to manually start X from the terminal. But when I do that, it always freezes. Even after I deleted the X config. I'm using a generic vesa driver again. Unfortunately, I think the problem relates to some proprietary ATI drivers I installed this morning. Maybe that messed up some settings?
t0dd: Oh, and again, I have an ATI Radeon HD 3450
airlied: if you isntalled fgglrx then I'm not sure what we can do for you
t0dd: Would a clean install be necessary if I already did that?
airlied: did X work before you did it?
gimzo: do you have kms running when you start x ?
t0dd: well, it worked when I used Fedora 11. I just upgraded
t0dd: gizmo: yeah, I tried that too. Same problem. It crashes
gimzo: it always crashes?
t0dd: not when I use the vesa driver, but otherwise, yeah
gimzo: oh, I got it wrong then, I thought it crashes with vesa driver
gimzo: time for me to get some sleep :)
airlied: t0dd: can you pastebin dmesg?
t0dd: http://pastebin.org/58946
airlied: t0dd: looks fine, got an x log /var/log/Xorg.0.log
t0dd: Do you want me to pastebin that too?
airlied: yup
t0dd: one sec
t0dd: could I just email it to you? It's huge
t0dd: nvm
t0dd: http://pastebin.org/58947
airlied: can you get one without vesa?
airlied: one with no xorg.conf
t0dd: ok
MostAwesomeDude: Alright, so apparently radeon cold-boots my secondary card without me doing anything. This is fairly awesome.
MostAwesomeDude: List of things I wanna try:
MostAwesomeDude: - fbcon on the other card (/dev/fb1)
MostAwesomeDude: - mplayer on the other card
MostAwesomeDude: - Dual-card X
t0dd: oh god
MostAwesomeDude: Now, Google's not really helping me out here.
t0dd: sorry.
t0dd: http://pastebin.org/58949
airlied: MostAwesomeDude: system-config-display --reconfigure
airlied: "might" work on fedora
MostAwesomeDude: airlied: How about for fbcon and mplayer?
MostAwesomeDude: I get "Can't put VSCREENINFO: Invalid argument" with mplayer.
t0dd: I use the radeonhd driver in that one. You can see the segfault towards the end
airlied: t0dd: I said no xorg.conf
airlied: we don't support radeonhd on fedora
airlied: MostAwesomeDude: patches to fix that are in one of my trees
t0dd: okay, every log says xorg.conf is being used (even though I deleted it the last time)
MostAwesomeDude: airlied: Oh, is this the thingy with the pixel clocks?
airlied: MostAwesomeDude: yup
t0dd: let me delete all these logs and try again with no xorg.conf and see exactly which log it generates?
airlied: soudns good
t0dd: I deleted the X config file and rebooted. X failed to start, and when I tried to manually start it from the terminal, my system crashed, and it didn't create a log in /var/log
airlied: if X faisl to start it leaves a log file
t0dd: It created one, but it was empty
t0dd: it had 0 lines
airlied: agd5f, glisse : so if we have no BIOS and the card isn't posted, should we just ignore it?
airlied: t0dd: ouch
t0dd: I should probably just reinstall Fedora
airlied: so when X fails to start when you boot it doesn't log anything?
airlied: before it crashes when youy start it yourself
t0dd: the old.log files magically reappeared, even though I removed them before rebooting, but they had the same radeonhd info as before
t0dd: and no, it doesn't
airlied: you don't have an xorg.conf in /root orsomwher else do you?
agd5f: airlied: sounds like a plan
t0dd: Let me look. brb
t0dd: it generated a log this time: http://pastebin.org/58954
t0dd: (w/o any mention of xorg.conf)
airlied: and it crashes there?
airlied: wierd
t0dd: that's nice. epic crashes when I open a new tab
t0dd: anyway
tavl_: anyone using openSuSe 11.2 + KMS + Radeon?
airlied: okay multi-op r600 pushed
airlied: agd5f: hopefully it won't show regressions this time ;)
amarks_: airlied: that colour corruption issue we talked about around the time of your tokyo trip, any chance you can take a look at it?
amarks: airlied: can't remember what you called it
airlied: amarks: you'll have to remind me a lot more about it ;)
amarks: airlied: let me screenshot it, easier
amarks: airlied: http://imagebin.org/73564 and http://imagebin.org/73565 - a lot of the time the colour is black so makes task bar tabs unreadable - it flickers in and out depending on what you hover or click on
amarks: the taskbar tabs should not be blue
Hiryu: "/usr/lib/xorg/modules/drivers//radeon_drv.so: undefined symbol: DRIOpenDRMMaster", anyone know what's causing this?
airlied: amarks: what gpu?
amarks: rv790
airlied: latest -ati?
amarks: yes, well 2 days ago or so
amarks: has been there for a while
airlied: KMS or non-kms?
amarks: kms
airlied: wierd not sure what is causing that, compositing on?
amarks: yes kwin opengl enabled
airlied: does it happen without that?
airlied: or with xrender compositing?
amarks: let me try xrender
amarks: i've applied xrender, do you need to restart kde?
airlied: not sure I don't use it
amarks: airlied: still see similar corruption with xrender
airlied: compositing off ?
amarks: airlied: worse, disabled all effects, http://imagebin.org/73568
amarks: these are after an x restart just to be certain
airlied: amarks: might need to file a bug for agd5f to look at
airlied: we might be doing something wrong on rv790
amarks: what component
airlied: freedesktop xorg Drivers/radeon
amarks: ok will do
amarks: i assume u r not seeing this on r600?
zhasha: airlied: did you have a look at the false svideo detection on rv515 or should I open a bug?
airlied: I don't run KDE but I haven't had it reported from what I can see
airlied: zhasha: my rv515 is working, so you should probably file, attach the bios if you can
zhasha: how would I go about doing that?
zhasha: it working in F11 by the way
amarks: airlied: ok thanks for your advice
zhasha: and with rawhide 1 week before release or so
airlied: zhasha: cd /sys/bus/pci/devices/
airlied: echo 1 > rom; cat rom > /tmp/rom; echo 0> rom
airlied: zhasha: works with radfeon.tv=0?
zhasha: airlied: should I add that to my kernel line a check it out?
airlied: zhasha: yup
airlied: that assume I implemented the check right ;-)
zhasha: airlied: well, that fucked it up good
zhasha: displayed garbage on my lvds
airlied: zhasha: oh nice
zhasha: and no, I didn't write radfeon :P
eosie: MostAwesomeDude: ping
spstarr: interrupt handling eh
spstarr: agd5f: your interrupt handler + firmware addons going into drm-next-testing or so?
airlied: spstarr: no he wrote them to leave on the mailing list
spstarr: airlied :P
airlied: though the firmware isn't going to go in the kernel
spstarr: well i donno the process
spstarr: patches to ml first then how long for a review before merge into git?
airlied: as long as it takes
spstarr: its not an issue for me, i can just take drm-next-testing + patch and test it with the firmware addons
airlied: it should be there later in the week I'd guess
airlied: I have to do some testing on it first
spstarr: ok, you need some additional testing?
spstarr: i can test on the w500
airlied: give it a go,
spstarr: im about do do my weekly libdrm + ddx + mesa master + git kernel updates
spstarr: ok