arnal: hi
arnal: lately, I've got entire lock-up of the X server (I can only move the mouse and I have to reset my computer): http://www.mini-dweeb.org/~arnau/tmp/Xorg.0.log.1
arnal: any idea?
arnal: another Xorg log files: http://www.mini-dweeb.org/~arnau/tmp/Xorg.0.log.2 http://www.mini-dweeb.org/~arnau/tmp/Xorg.0.log.3
MrCooper: glisse: btw the new GPU reset is also killing the box when trying to fall back from AGP to PCI, this was working before
Ke: is there any support for HD4290?
BioTube: yes
Ke: I get only KMS with no accel
Ke: using 6.13-rc*
Ke: and linux-2.6.33
BioTube: with 2.6.33 and later, you need an extra firmware
adamk: Pastebin your /var/log/Xorg.0.log file.
adamk: And the output of 'dmesg | grep drm'
Ke: hmm need to boot then
adamk: Ops: Are there any plans on removing the restriction making it so that only registered nicks can talk?
amarsh04: adamk: no ops in channel
adamk: You mean to tell me that airlied, agd5f, and MrCooper aren't ops?
MrCooper: I can't really say anything about that as I wasn't involved in enabling it
amarsh04: I just checked some other channels and some have ops, some don't
adamk: I believe it was because of the spammer issues, but it's been like this for a while and I'm getting PMed a few times a day from folks asking why they can't talk here.
adamk: Probably because towards the top of the nick list :-)
adamk: amarsh04: Freenode discourages people from sitting around with ops in the channel.
amarsh04: ah, ok adamk
adamk: amarsh04: So typically you don't know they're ops till nickserv gives them +o
MaDMaLKaV: open sources commies
MaDMaLKaV: they don't like authority
amarsh04: but Nickserv isn't here either
MaDMaLKaV: :)
adamk: Hmmm.
amarsh04: Chanserv I meant
adamk: Ahhhh, chanserv, that's right.
adamk: Well, maybe he can give ops if you PM him. It's been a while since I've had to do it on #compiz
adamk: Anyway.
Ke: wonder which one if these kernels was the best
Ke: radeonfb is bad?
adamk: I don't mind getting these PMs, but clearly there are a number of people who come here for help who aren't getting it.
adamk: ke: When using KMS, yes.
Ke: http://users.tkk.fi/~jkarlson/files/Xorg.0.log http://pastebin.com/vugVtu6E
Ke: hmm failed to load firmware
Ke: iirc I had one that didn't
Ke: what could cause this for radeon built as a module
Ke: If I use make install modules_install for the build
Ke: /lib/firmware/radeon/ lists quite a bit of binaries
adamk: But does it list that one?
suokko: Ke: see build how to in topic if you are missing rlc firmware
TheBrayn: howdy folks
adamk: Oh, the error doesn't mention which one is missing.
adamk: Heh.
MaDMaLKaV: MostAwesomeDude, about what you said yesterday about about the possibilty of the XBox 360 working with a radeon-ish gpu + a framebuffer/modesetter , it is that way
MaDMaLKaV: you see were can I see info / code about other similar systems? I want to compare the info we have extracted from the modesetter with other similar systems to see if there is a similar implementation already coded
edwin: adamk: also you can only talk if you connect via ssl
suokko: agd5f: Looks like drm kernel code assumes at least in a few places max 2 crtcs. Might have to fix that for evergreen support
edwin: adamk: which is silly since this channel is publicly logged afaik
adamk: edwin: I'm not connecting via ssl, that I know of.
adamk: Yet here I am talking.
edwin: then it must have changed, let me try
edwin: appears to work
Ke: suokko: missing though r780 only
edwin: (w/o ssl now)
TheBrayn: I installed xf86-video-ati with portage
TheBrayn: and still I get this: http://codepad.org/b3Q6xI25
Ke: this card is fairly new which could lead me to believe it's unsupported for now
glisse: MrCooper: it kills the box at module load time ?
Ke: but let's try again
suokko: Ke: Do you have RS780_me.bin and RS780_pfp.bin?
adamk: ke: You need to resolve that firmware problem.
adamk: Your card is certainly supported.
MrCooper: glisse: at initialization time, yeah, if I try AGP 2x or 4x (which have never worked reliably with KMS)
MrCooper: previously it would fall back to PCI, now the GPU reset hardlocks
suokko: TheBrayn: video-ati doesn't handle 3D
TheBrayn: ok
adamk: TheBrayn: What video card?
suokko: TheBrayn: But there is 3d support in mesa-dri
TheBrayn: ok
TheBrayn: hd4870
Ke: suokko: nope
adamk: TheBrayn: Yeah, you definitely need a newer version of mesa.
MrCooper: glisse: at least I assume that's what happens, given the GPU reset hardlocks in other circumstances
TheBrayn: 7.5.2 atm
TheBrayn: which version should I install?
MrCooper: afk for a bit, bbl
glisse: MrCooper: yeah likely will plug back an agp computer to play with tomorrow, on pcie the reset stuff seems harmless
adamk: TheBrayn: At least 7.6.1, though if portage lets you install a git version, that's probably better.
Ke: still not getting any r780 firmware
suokko: Ke: Which version of kernel are you trying to compile?
Ke: http://wiki.x.org/wiki/radeonBuildHowTo#TroubleshootingExtraFirmwareforR600.2BAC8-R700 not even with these
Ke: 2.6.33.1
suokko: Do yo uhave the RS780_me.bin and RS780pfp.bin in /lib/firmware/radeon/. Did you regenerate initrd after adding them there?
Ke: no initrd and no
Ke: oh there was an error
Ke: No rule to make target `/lib/firmware/radeon/R600_rlc.bin'
Ke: http://pastebin.com/D2NSL9c4 from .config
suokko: ls /lib/firmware/radeon/R600_rlc.bin ?
Ke: no such thing
Ke: blind
suokko: Ke: You would need to copy R600_rlc.bin and R700_rlc.bin to there
Ke: yes
Ke: emerge radeon-ucode
okias: why it can compile for 64bit but for 32bit can't? I use nouveau and it's fine, only radeon have problems
okias: please help
okias: http://pastebin.ca/1854609
okias: sorry, I wasn't logged, so problem is compilation 32bit libdrm with --enable-radeon :-(
okias: in attached pastebin is compilation log
suokko: okias: make clean?
okias: suokko: lol, thanks a lot! :-)
Ke: OpenGL renderer string: Mesa DRI R600 (RS880 9714) 20090101 TCL DRI2
Ke: yay
okias: it's enought build mesa for rs690 with dri=radeon,r300 ?
suokko: okias: You don't need radeon but swrast is good idea
okias: suokko: why? when iit's already accelerated?
okias: hmm, compilation fails... and now I did make realclean - http://pastebin.ca/1854624
suokko: okias: just in case there is some fallback driver
suokko: okias: That linker is missing -m32
okias: hmm I hoped it is covered by --enable-32bit
suokko: It should be
okias: not helping :-(
Tecan: :)
Tecan: how would i check my radeon version ?
Tecan: it seems to be working well, glsl not working but thats semi-expected
jcristau: it's in the X log
Tecan: hmmm thanks
suokko: Tecan: mesa version is in glxinfo
Tecan: 1.2
Tecan: glx version 1.2 is that the latest ?
suokko: Tecan: loo farther down
suokko: look
jcristau: that's the protocol version. not the software version.
Tecan: OpenGL renderer string: Mesa DRI R300 (RS400 5955) 20090101 NO-TCL
Tecan: OpenGL version string: 1.5 Mesa 7.6
Tecan: linux mint is very well done
Tecan: its ubuntu with an upgrade
adamk: Tecan: So it supports opengl 1.5 with Mesa version 7.6
suokko: Tecan: Your card doesn't support GLSL well. No idea if broken implementation will be made or not
adamk: On your GPU, you could get GLSL if you used gallium3d, but that's still heavily under development and, I believe, slower than the regular Mesa driver.
adamk: Oh... RS400.
adamk: Heh, yeah, avoid gallium3d then.
Tecan: no worries, i suck at shader development anyway haha
okias: suokko: I tried add CFLAGS="-m32" even for gmake, but it doesn't help
Tecan: so does ati unofficially support this project ?
Tecan: it seems like something they should be investing in
suokko: okias: LDFLAGS="-m32"
okias: where should I add? to ./autogen or gmake?
adamk: Tecan: They support this project with documentation and by paying developers.
Tecan: yes,a nd the government should be supporting linux development too
Tecan: the world's just one big capitalist greasy oil orgy
adamk: Well that's a discussion for another channel.
okias: suokko: with ./autogen it doesn't help
suokko: okias: I don't understand how the build system works so no idea how to fix
okias: suokko: aah, this problem appeared on my laptop too :-(
okias: build worked on my laptop fine before few days :-(
Tecan: checking out the git repo... doubt i'll be able to help but i'll try. looks like fun :)
Tecan: opengl 4 will be a tough mark to hit. hopefully there's some developers from siggraph willing to chip in too
Tecan: nvidia drivers arnt even opensource either so radeon should get alot of attention
LordVan: hi
LordVan: got a quick question: how is Radeon HD 5750 support atm ? - can'T use fglrx due to Xorg downgrade pains .. ;)
okias: http://pastebin.ca/1854642
glisse: LordVan: no acceleration
okias: with CC="gcc -m32"
LordVan: glisse, u mean 3d or 2d as well?
glisse: no 3d no 2d because today 2d -> 3d
Tecan: lordvan probably no dri
suokko: okias: http://sprunge.us/hAYT ?
LordVan: hmm sorta sux .. u happen to know about xvideo support? cuz watching vids without that is a pita ..
[Enrico]: LordVan: no xvideo support atm
adamk: LordVan: DRM support is needed for Xv support on modern radeon GPUs.
[Enrico]: LordVan: xvideo is 2d acceleratio. so no 2d no xv
glisse: no xvideo neither, xvideo|2d needs 3d
adamk: LordVan: There is a 10.4 catalyst prerelease that supports X server 1.7.*
LordVan: meh .. sucks .. guess i'll keep my geforce 6600 in there for now ^^
LordVan: adamk, i read about it yes but i duno if i wanna try catalyst prerelease .. ;)
adamk: Meh, I've used it in Ubuntu without issues.
LordVan: got a radeon hd 5750 laying around on top of the pc for a while
LordVan: 'a while' -> months
adamk: And I've spoken with folks who have used it in Slackware -current.
LordVan: hmm maybe i'll give it a go .. i wonder if there's an ebuild for gentoo floating around somewhere ..
Tecan: lol @ the wiki page change log >> 10 Sep 2009: 6.12.4: brown paper bag release for X.org 7.5.
LordVan: cuz u know to be honest for now i couldn't care less if the driver is opensource or not but i bought this card ages ago .. if the closed driver lets me use it then so be it -- doesn't mean i won'T try the OSS driver as soon as it supports what i want/need ;)
[Enrico]: LordVan: there is, i've done it :)
[Enrico]: LordVan: it is the 8.721 ati-driver
[Enrico]: +s
LordVan: [Enrico], ;) but not in the official tree yet or is it
[Enrico]: LordVan: you can find it in the main portage tree (hard masked of course)
LordVan: hmm ok gotta sync then ;)
LordVan: hmm .. /usr/portage/x11-drivers/ati-drivers/ChangeLog got a lot 'Enrico' in ;)
LordVan: so then i guess i'll try it when i get some time to put my pc off and stick the card in ;)
[Enrico]: LordVan: i help the mantainer since he doesn't have the hardware to test with (he is going to buy one) so i do the tests for him. but i've not tested the beta yet since i keep the test system stable, but i uploaded the ebuild in the bugzilla and there are reports of user using it with success
okias: hmm, look like that patch didn't work :-( or I hitting more problems
[Enrico]: [the current stable xorg-server is 1.6 and the 10.4 beta works only with 1.7]
Tecan: all all the radeon cards very simular in api ?
glisse: Tecan: what you mean by api ?
okias: or maybe, with CC="gcc -m32" + that patch it work. On my Nouveau laptop it's enought CC="gcc -m32", on radeon computer is needed patch...
Tecan: the hardware commands
okias: please fix that build problem, it take a lot time to figure out what workaround use
Tecan: instruction sets
glisse: Tecan: different from one generation to the other
Tecan: figures
glisse: thought shaders isa has a lot of similarities btw r6xx,r7xx
suokko: okias: I don't have multilib. THat was just blind guess what was missing
dk_mag: okias: have u installed libdrm 32-bit compability libs?
okias: dk_mag: udev-151 32, libdrm 32, mesa 32
dk_mag: xf86-video-ati doesn't need multilib just mesa and libdrm. I'm running radeon kms on RV635 and it works fine
dk_mag: maybe u need to recompile libdrm again for multilib, i'm not sure how multilib works for your distro, which one are u running?
suokko: dk_mag: that is not problem
okias: dk_mag: libdrm is fine, recompiled into /usr/lib32...
okias: problem is build system
suokko: okias: What is error when you build with that patch and (no special CC or LDFLAGS set?)
okias: compiling
dk_mag: well if it doesn't work now maybe passing this to autogen could work, since it's trying to link the wrong architecture of libs. PKG_CONFIG_PATH="/usr/lib32/pkgconfig:$PKG_CONFIG_PATH", sorry if im totally off here, just trying to help from what i know of your issue.
suokko: dk_mag: Problem is simple that build system has bug and it has to be fixed
dk_mag: ok i understand now, then i guess i'm not helping, sorry 'bout butting in.
okias: r300_dri.so.tmo: could not read symbols: File in wrong format
okias: without CC="gcc -m32"
mcencora: MrCooper: does the xorg state tracker require specific kernel drm version? because it doesn't work for me (the X crashes)
mcencora: nha: I've been playing with fragment shaders recently and discovered a possible field of improvement
mcencora: nha: writing to result.color in one instruction is way more faster than in a few
mcencora: nha: while simple fp gave me ~400FPS with one instruction writing to result.color.xyzw, 4 instruction writing to every channels seperately gave ma ~200FPS
mcencora: nha: the hw is rv535
suokko: Yes. FS is "hot" code in gpu side :)
suokko: okias: crtc = i830_covering_crtc(pScrn, &box, NULL, &crtcbox); | return id;
suokko: okias: http://sprunge.us/CLVB
nha: mcencora: that's quite interesting
nha: so it would make sense to try to merge those instructions, and if it's not possible, even writing to a temporary register first might be beneficial
mcencora: nha: yeah
nha: that could be done in an additional compiler pass
okias: suokko: if it work, will be patch merged?
suokko: okias: I will send it to ML for review. I just on't know if that is correct fix
okias: suokko: you forget in first line s/-32/-m32... (i fixed it locally)
okias: right now compiling
okias: suokko: fine, works!! thanks
suokko: okias: ok. I will end the fixed version
okias: suokko: ok :-)
okias: so one problem fixed, next to go - every wine app fail with lastest git
suokko: okias: What is error there?
okias: suokko: wait a second, uploading log
okias: http://paste.ubuntu.com/403923/
suokko: I don't know what is the real error there
suokko: okias: Do yo uhave comparison from olderworking version?
okias: suokko: uuuh, it worked with default 32bit emulation packages
okias: suokko: but terribly slow (0,5fps ...)
okias: http://paste.ubuntu.com/403931/ - WoW.exe fail, looks similiar to PlantVsZoombies.exe
suokko: okias: Can you post a working log for comparision?
suokko: okias: and if you build 32bit mesa demoes do they work?
okias: suokko: I can build it, wait a sec :-)
okias: suokko: what should I test?
okias: how enable mesa debug? LD_DEBUG?
okias: suokko: which test should I run?
suokko: okias: try glxgears and glxinfo
okias: suokko: works
suokko: and info reports that it is hw rendering instead of sw
okias: suokko: works as hw
okias: suokko: maybe recompile wine against new mesa?
suokko: okias: Shouldn't be required
suokko: okias: Does wine work if you do LIBGL_ALWAYS_SOFTWARE=1 wine ... or LIBGL_ALWAYS_INDIRECT=1 wine ...?
themgdn: .
themgdn: aw, it works
themgdn: hello, i have some troubles when compiling driver from git
themgdn: here is the log http://dpaste.com/177047/
themgdn: with old driver and libdrm kms doesn't work - very slow 2d, e.g. scrolling in firefox, but 3d is ok
themgdn: kernel headers - 2.6.33. i don't know, where i cat get newest drm.h...
suokko: themgdn: libdrm
themgdn: from git, latest
suokko: themgdn: Then that fail in include is some old stail fail I think
themgdn: you think, i need to copy drm.h from libdrm's /include dir?
okias: suokko: hmm, I recompiled wine and still see: http://paste.ubuntu.com/403971 :-(
suokko: okias: That log doesn't help me because I can guess what part of that wine spam is relevant and wht is not
okias: suokko: I try found some demo with that
suokko: and that first crash in begin happens inside wine code
themgdn: wow, compling with headers from libdrm dir
themgdn: nice, thanks
themgdn: no, kms is still buggy :)
themgdn: [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation !
themgdn: in dmesg
themgdn: and sloooow 2d
Wizzup: can confirm
agd5f: themgdn: see bug 27284
Tecan: holy wow, 3 tb harddrives only 80 dollars at future shop
themgdn: agd5f, thanks
themgdn: but where is patches? :)
suokko: themgdn: dowgrade to working version ;)
themgdn: ..or wait for merging patches to master?
themgdn: thanks for help and patches
themgdn: bb
Digital_Pioneer: OK, my 3D is gone again.
Digital_Pioneer: RV770, 2.6.33, recent git of libdrm, ddx, mesa...
Digital_Pioneer: Direct rendering is on, says glxinfo. If I turn on Kwin's compositing, my screen goes black. BZFlag falls back to software rasterizer instead of the normal 3D.
Digital_Pioneer: r600_cp: Failed to load firmware "radeon/R700_rlc.bin"
Digital_Pioneer: Hm!
Digital_Pioneer: -rw-r--r-- 1 root root 4096 Feb 27 19:50 /lib/firmware/radeon/R700_rlc.bin
Digital_Pioneer: Hm!
[Enrico]: Digital_Pioneer: do you use radeon builtin in the kernel by any chance?
Digital_Pioneer: [Enrico]: Possible.
Digital_Pioneer: CONFIG_DRM_RADEON=m
[Enrico]: Digital_Pioneer: then you need to add the firmware to the kernel too or to the initramfs if you use it
Digital_Pioneer: CONFIG_DRM_RADEON_KMS=y
[Enrico]: Digital_Pioneer: ok it is a module not builtin, but if you include it in the initramfs then you must include the firmware too
Digital_Pioneer: [Enrico]: So how do I do that? :D
rehabdoll: CONFIG_FIRMWARE_IN_KERNEL=y
[Enrico]: Digital_Pioneer: no idea sorry i don't use an initramfs i use it builtin in the kernel to get kms early at boot
[Enrico]: well i use CONFIG_EXTRA_FIRMWARE to include only what i need
Digital_Pioneer: rehabdoll: That one is set here.
[Enrico]: Digital_Pioneer: i think that is not enough since the radeon/R700_rlc.bin is not included by default in the linux firmware
rehabdoll: you need to specify what firmwares to include
rehabdoll: ie: CONFIG_EXTRA_FIRMWARE="radeon/R700_rlc.bin"
[Enrico]: Digital_Pioneer: but you can use CONFIG_EXTRA_FIRMWARE=radeon/R700_rlc.bin to force it
[Enrico]: yeah with double quote is better
Digital_Pioneer: Indeed. Sucks when you don't build your own kernels...
Digital_Pioneer: Cursed regressions, this _was_ just working.
Digital_Pioneer: ducks out to restart X and the like.
Digital_Pioneer: THAT's better.
[Enrico]: :)
Digital_Pioneer: Yeah, it can't find that firmware at boot.
Digital_Pioneer: I just built it into my initrd, don't know if it worked or not -- I just reloaded radeon and it works for now. :P
Digital_Pioneer: Too lazy to reboot and test right now.
Ke: btw is there some direction with gallium3d/classic mesa
Ke: is it just duplicated work, is there some reason to develop mesa classic further
BioTube: Ke: Gallium is the future, at least for cards that it supports(r300 and later in the radeon family)
Ke: BioTube: that's what I hear, why do we have the classic mesa for radeon r600+
BioTube: Ke: because r600g's pretty new
BioTube: and the winsys hadn't been stabilized yet
BioTube: Gallium itself has changed a lot since r600c was started as well
dougmencken: anybody online?
Pallokala: work on r600g has started?
BioTube: Pallokala: check glisse's fd.o repo
Pallokala: can I expect it to work?
dougmencken: I have Radeon; and those black (just black quares) makes me sad; even Transmission's ones and Firefox's
Pallokala: ie. if something isn
dougmencken: and even VLC icon in Gnome panel
BioTube: Pallokala: last I checked, glxgears locked up my machine
Pallokala: t implemented yet, does it use software path or crash?
Pallokala: ok
glisse: Pallokala: won't work
glisse: it's just ground stuff
Ke: doesn't any lockup implicate a bug in drm rather than userspace
glisse: more is coming soon hopefully
spstarr: glisse: hey
rhodan: Still not enough developers?
MostAwesomeDude: More important things to work on, also some of us think that some infrastructure work needs to be done.
unic0rn: hi. i have few questions and i'm pretty sure few will be kinda silly, but after recent fighting with stuff i'm just not sure anymore how this thing works.
unic0rn: first, there's radeon_drv and r200_dri. i have libdrm and dri updated, mesa 7.9 devel, and those are fine
unic0rn: glxinfo displays mesa 7.9-devel
unic0rn: but the radeon_drv in xorg drivers folder is old. where is it? looks like either i've screwed up something, or it isn't part of mesa
unic0rn: second thing, and that's the main issue, is: why DRI2 for r200 (radeon 9200, 128mb, agp x8) isn't working?
adamk: unic0rn: radeon_drv.so isn't part of mesa.
adamk: unic0rn: It's from xf86-video-ati, it's an Xorg DDX.
unic0rn: i'm pretty sure it should, yet doesn't here. xorg log says screen 0 isn't dri2 capable, and that's it. result is pretty obvious, glx isn't being redirected, so simple glxgears screws up screen with compiz and aiglx
adamk: unic0rn: Are you using KMS?
unic0rn: no. kms disabled.
adamk: Well that's the problem, then.
adamk: Or at least one of them.
unic0rn: :/
unic0rn: so kms has to be enabled
adamk: Yes.
unic0rn: and what are the other problems?
adamk: No idea. But till you use KMS, there's no point trying to figure out why DRI2 isn't enabled.
unic0rn: i've tried yesterday. white screen with compiz, obviously DRI failed again. have to check the logs what went wrong
adamk: If DRI is failing with KMS, chances are libdrm and/or xf86-video-ati isn't built with KMS support.
unic0rn: yeah, i guess that's the case. as far as i know. libdrm and mesa/dri are built with kms now, so that leaves xf86-video-ati
adamk: And since you said that radeon_drv.so is old,chances are it wasn't built with KMS support.
unic0rn: well, it's debian sid. what's kinda crazy, imho, is that with most recent radeon driver, most recent xorg and libdrm and mesa, it fails. that's why i've started compiling it on my own.
unic0rn: fails as recent update of radeon driver enabled kms in modprobe config, while it's not supported in this build.
unic0rn: that screws up basically everything
BioTube: unic0rn: the one in sid right now has KMS support
unic0rn: radeon_drv?
BioTube: xserver-xorg-video-radeon is the package name
unic0rn: as far as i know, it hasn't.
unic0rn: unless it was changed within past 48h
BioTube: it used to be in experimental
jcristau: unic0rn: it changed a week ago
unic0rn: maybe. maybe instead of playing with compiling stuff, i'll just check the experimental repo. the thing is, i was almost sure i have it from experimental already
unic0rn: jcristau: i've installed the most recent package no longer than 3 days ago.
jcristau: there's nothing in experimental. sid has radeon 6.12.192, with kms enabled.
unic0rn: together with recent xorg, drm and mesa
BioTube: unic0rn: make sure you don't have something lying around /usr/local
unic0rn: and radeon driver had kms enabled in /etc/modprobe/ yeah. but it wasn't compiled with kms support
unic0rn: OR something else wasn't.
unic0rn: that or another, it didn't work.
jcristau: how about you provide logs?
unic0rn: would love to. trouble is, i have only 3 recent xorg log files. and since then, half the things has changed. but actually, i have current xorg radeon package installed, i've only compiled mesa and libdrm, so yeah, why not. i'll reboot and check it.
unic0rn: ok. i've overwritten everything compiled manually, by doing update/reinstall of all stuff (xorg, radeon driver/drm modules, mesa)
unic0rn: KMS enabled, radeon added to /etc/modules
unic0rn: tried updating kernel, but looks like there's something wrong with new initrd in sidux, so i'm using 2.6.32 now
unic0rn: BUT... according to the logs, everything works fine.
unic0rn: even DRI2
unic0rn: the problem is, it's out of sync
unic0rn: (II) RADEON(0): Setting screen physical size to 270 x 203
unic0rn: i guess that may be the problem, but why it uses that, no idea. it's set to 1024x768 specifically in xorg.conf, and that worked without kms just fnie
unic0rn: fine*
BioTube: unic0rn: the physical size is in millimeters, not pixels
unic0rn: hm
unic0rn: so what the hell..
unic0rn: i'm getting black screen instead of gdm
BioTube: sounds like a linking error
BioTube: there's probably a 'failed to find symbol' message waiting for you n the VT that launched X
unic0rn: there is not
BioTube: strange then
unic0rn: i can see nothing unusual in .xsession-errors either
unic0rn: i've tried both init 5 and startx
unic0rn: with startx, everything started
unic0rn: including compiz and awn
unic0rn: just... no picture.
unic0rn: i don't get it. :/
BioTube: it's always frustrating when that happens
unic0rn: i saw tons of different shit happen
unic0rn: but this time, Xorg.0.log says everything is just fine
unic0rn: even more, it's perfect. dri works, dri2 works, everything works!
unic0rn: just nothing can be seen on the monitor.
unic0rn: -.-
BioTube: not event he cursor?
unic0rn: i still suspect it has something to do with resolution
unic0rn: not even the cursor.
BioTube: if the modeline was wrong, your monitor should tell you
unic0rn: and when i press a button on my CRT, i don't see anything, so it's obviously out of sync
unic0rn: i have a weird feeling it may have something to do with a kernel
unic0rn: need to check what they've screwed up with that initrd, getting kernel panic with it
unic0rn: and obviously 2.6.32 may be not good enough for kms
thansen: any dev around who'd like to help me get display port output working? :)
Jonimus: thansen: it should work already...
thansen: I know that :).. unfortunately for me it doesn't
airlied: thansen: what hw?
thansen: airlied: https://bugs.freedesktop.org/show_bug.cgi?id=27314
thansen: firepro 7740
thansen: (I can attach any other details on there that will help)
thansen: or give access to the box etc
airlied: can you get a log with drm.debug=15
thansen: airlied: do you want the dp connected at boot?
airlied: thansen: yes
thansen: ok, I'm on the machine now so I'll reboot and brb...anything else you want me to collect on this reboot?
airlied: nope that should be tell me some more
thansen: k, thanks for the help...brb
thansen: airlied: ok, posted https://bugs.freedesktop.org/show_bug.cgi?id=27314
airlied: so the screen is really 2560x1440?
thansen: airlied: yes
thansen: dell u 2711
thansen: http://accessories.us.dell.com/sna/productdetail.aspx?c=us&cs=04&l=en&s=bsd&sku=224-8284&redirect=1
Jonimus: wants that monitor now....
thansen: I have another sitting here :)
thansen: but it's already claimed
thansen: one sec, gotta reconnect to the net..
airlied: thansen: can you try booting with nomodeset and starting X? just to see if the UMS code goes any better
Jonimus: airlied: he just left
airlied: oops
Jonimus: man that is a really nice monitor sadly it costs almost as much as my Dell laptop did.
Jonimus: airlied: hes back ^
thansen: yeah, sorry about that
thansen: damn intel 5300 wireless doesn't play well with dd-wrt
thansen: gotta look into it later and see what's going on :(
Jonimus: hmm mine doesn't have any issues, and its also a dell laptop
thansen: Jonimus: what's the card and what router do you have?
Jonimus: I really wish I had a DP monitor to messwith.
airlied: thansen: can you try booting with nomodeset and starting X? just to see if the UMS code goes any better
Jonimus: thansen: its an intel 5300 same as yours, and the router is a WRT320N running dd-wrt
thansen: Jonimus: yeah, I have old school g router
thansen: airlied: sure, what's the exact syntax for killing kms
thansen: is it 'nomodeset'
thansen: or radeon.modeset=0 or something
Jonimus: thansen: both will work
thansen: ok, brb again...
Jonimus: also the card worked just as well with my WRT54GS before I upgraded
thansen: airlied: still no dice
airlied: thansen: got the Xorg.0.log?
thansen: interesting that xrandr calls the displays 'DisplayPort-X' and this time the connector is 1 instead of 0
thansen: sure, I'm in it right now..sec
airlied: yeah UMS and KMS output naming sorta diverged
thansen: airlied: ok, attached to the bug
airlied: hmm I'm guessing the fact that both connectors are on the one uniphy might be causing issues
airlied: is there two DP connectorS
airlied: ?
thansen: airlied: it has dock but it's not connected
thansen: like, bottom of laptop has the ports for it so I'm guessing that's what it is
airlied: yeah its probably on the dock then
airlied: can you attach a copy of the bios? cd /sys/bus/pci/devices/
airlied: pciid can be found in lspci
thansen: airlied: ok, see attachement on the bug..I hope it's right
thansen: Jonimus: so I have wrt54GL running v24sp1
Jonimus: thansen: well as I was saying my wrt54gs running the micro version of that release.
thansen: Jonimus: I need to reload them anyway cause I can't seem to login anymore
airlied: thansen: you got -ati installed from git?
airlied: I wanna try a few hacks on the ums code to see if its something stupid or not
thansen: airlied: no, 6.12.192...but I can get git if you want
thansen: running gentoo and I know someone has an ebuild out there for it
airlied: well if you can apply patches to 6.12.192 that'll work either
airlied: just need something you can patch cleanly
thansen: airlied: whatever is easier for you
airlied: thansen: grab git then ;-)
thansen: k
unic0rn: small update on my KMS/kernel issue: xorg seems to set 1024x768, reports modesetting in kernel enabled, everything in logs looks fine, yet no picture at all. trying to get some info on #sidux@oftc to fix initrd for 2.6.33, yet no luck for now.
unic0rn: anyone succeeded with r200 & KMS & DRI2 on 2.6.32 kernel?
airlied: thansen: http://fpaste.org/WgQj/ is the first test patch, use nomodeset and post the Xorg.0.log
thansen: airlied: sorry, got a phone call..lemme see if I can get this patch to apply with the ebuild...
thansen: airlied: looks like it applied clean...I'll reboot and brb with info
thansen: airlied: no dice, log attached bug
airlied: thansen: hmm might have to talk to agd5f to get some more ideas
thansen: airlied: he was helping me a day or two ago but got pulled away to something else :( Thanks for the help so far though :)
airlied: btw I assume the bios shows up okay on it?
thansen: airlied: yes, and through boot etc until driver/drm loads
thansen: I still have yet to see the screen at it's full capacity though :(
thansen: I boot for about 3 minutes into windows and their (dell's) tool for manipulating screens sucked and couldn't figure out how to make it not mirror
thansen: *booted