DanaG: a hardware question: http://www.tomshardware.com/reviews/radeon-hd-5450,2549-3.html -- can a 5450 with HDMI, DVI, and VGA do all three heads at once?
airlied: actually no
airlied: I think you have to have DP on at least one
DanaG: Does the VGA hack work, at least?
airlied: vga hack?
airlied: yeah I think that should work in theory
airlied: DanaG: that article mentioned all that if you read it ;-0
gcarter: I have a simpler question. Where is the git/cvs/subversion links for mesa development?
gcarter: Can't seem to find it.
EruditeHermit: airlied, hey, any progress with the scrolling bug?
gcarter: Oh there it is! FYI: http://www.mesa3d.org/
soreau: gcarter: http://cgit.freedesktop.org/mesa/mesa/
MostAwesomeDude: gcarter: GSoC?
dileX: MostAwesomeDude: you linked your gallium-doc somewhere in a xorg-wiki?
dileX: speaking of http://people.freedesktop.org/~csimpson/gallium-docs/
MostAwesomeDude: dileX_: It's built from src/gallium/docs. Nothing special.
MostAwesomeDude: And I think it's out of date.
dileX_: MostAwesomeDude: cool, which software do you use to create the "website" (html, css?) stuff?
MostAwesomeDude: dileX_: rsync.
MostAwesomeDude: $ make -C src/gallium/docs html; rsync ...
dileX_: MostAwesomeDude: thx
dileX_: hmm, sphinx-build
dileX_: (python-sphinx pkg for debian)
dileX_: MostAwesomeDude: whats "pickled environment"? http://paste.debian.net/64812/
suokko: grr. unchecked kmalloc call in cs_parser_reloc or something like that :/
MostAwesomeDude: Oh, it'll cache previous runs.
MostAwesomeDude: Updated my version; now it's up-to-date with master.
Nightwulf|work: hi all
dileX_: hi Nightwulf|work
dileX_: MostAwesomeDude: yeah. saw it in a 2nd run. installed dvipng pkg. now it looks fine.
Nightwulf|work: hi dileX_
dileX_: MostAwesomeDude: remains two warnings :-) http://paste.debian.net/64813/
MostAwesomeDude: dileX_: Those last two have been there for a while and I haven't bothered to get them.
MostAwesomeDude: They're tricky. :3
dileX_: ReStructuredText (rst) - http://docutils.sourceforge.net/docs/user/rst/quickstart.html
dileX_: rst has also a spec :-) http://docutils.sourceforge.net/docs/ref/rst/restructuredtext.html
MostAwesomeDude: I know ReST. :3
TuomasT: Is it mesa, which provides the GLSL support for HD radeons?
suokko: I was able to cause relocations take 128k memory which caused kernel allocation failure :/
dileX_: agd5f: r300g dri/st (mesa master GIT) is playable - but has color corruptions. after anholt oa benchmark finished - I could return to desktop. this is with pm2 (0001..0030)
dileX_: MostAwesomeDude: could you try oa with latest mesa?
MostAwesomeDude: dileX_: Tomorrow; I need sleeps.
dileX_: hehe, yoh I fell asleep last night while watching tv - short nights before (testing pm2 stuff)
conathan: Greetings. I was having problems diagnosing why my system is using swrast_dri.so instead of the radeon one.
conathan: Using a LFS based system, Xorg 7.5, the xserver 7.5.902, and mesa 7.8-rc1
suokko: conathan: check xorg.log and dmesg for errors
conathan: according to xorg.0.log, Screen0 is not DRI or DRI2 capable
conathan: suokko: nothing I could note as a problem... [albet, more used to working with intel's dri]
conathan: actually, I'm going to revert to a stable xserver/mesa, noting a few other issues. (firefox segfaulting)
suokko: conathan: pastebin your logs
conathan: it'll be a few minutes...
suokko: and I don't think mesa can cause firefox to segfault
suokko: or xserver either
conathan: about a week ago, upgraded a coupple protocol packages and libraries (dri2, and libX11), and the rc of the xserver. it's bin a bit unstable since. Most likely one of the libraries, albet never had firefox crash about 10seconds after opening... [just today, was experimenting w/ a non intel card to see how dri worked]
conathan: (it's more for tinkering. I can fix the unstability part later)
conathan: there we go.
conathan: xorg.conf, dmesg, and Xorg.0.log
conathan: xorg.conf is just there to add Virtual 2960 1050, for xorg
suokko: conathan: You need to recompile all apps that depend on updated library
suokko: If there is header changes
conathan: did that with firefox, and it started working agin (at least, until I threw the radeon card in). blender still segfaulted, but that I can handle on my own.
conathan: it's the dri driver that I was at a loss for
dileX_: conathan: yoh man - LFS - doing from scratch. have look into http://cgit.freedesktop.org/~whot/X11R7.5/ (packages list) and http://cgit.freedesktop.org/xorg/util/modular/tree/build.sh for build-order.
TuomasT: glGenFramebuffers(1, &fbo); <-- is this segfaulting something that is normal with Mesa DRI R600 2.0 Mesa 7.9-devel ? OpenGL 1.5 works, but I have issues with more advanced calls..
suokko: conathan: RADEON(0): GPU accel disabled or not working, using shadowfb for KMS
conathan: dileX_: http://www.linuxfromscratch.org/blfs/view/svn/x/installing.html works farely well for me, I've been using it the last 6 years for Xorg builds, although this current system was rebuilt about 1-2 months ago
conathan: suokko: ah, something I can trace. Thanks
suokko: conathan: It is mostlike kernel side error
suokko: conathan: Do yo uhave R700_rlc.bin?
conathan: suokko: I dont, just whatever the kernel has builtin
conathan: r600_cp: Failed to load firmware "radeon/R700_rlc.bin"
suokko: conathan: see build how to from topic
conathan: suokko: thanks, and sorry for bothering you
dileX_: conathan: we have a radeon-build-howto
conathan: dileX_: thanks
dileX_: conathan: if you update - 1st libdrm - build mesa and xf86-video-ati against this version of libdrm. check if both (mesa and ddx) use libdrm_radeon. and of course you need RLC firmware files for r700 as suokko pointed already.
conathan: I think it's 2.4.19 that I have, and I recompiled all the Xorg libraries, mesa, xserver, and the drivers after installing it. I should be set (about to do a reboot).
conathan: but thanks
conathan: and working. Thank you
conathan: does seem a bit faster then intel.
dileX_: conathan: you are welcome
airlied: suokko: libdrm already resubmits ioctls on EINTR or EAGAIN
suokko: airlied: good
airlied: MostAwesomeDude: you soc student looks to have rs480
suokko: airlied: Should I CC stable for the NULL pointer dereference in cs parser?
loki_666: what should be the server glx version
loki_666: here i have 1.2
MrCooper: it can be 1.4 with current xserver and DRI2
loki_666: i'm using dri2 with xorg-server 1.7.6
loki_666: i have 1.4 for CLIENT version and 1.2 for SERVER vesion
jcristau: the bump to 1.4 was post-1.7
edwin: loki_666: are you sure you don't have stuff from binary drivers laying around?
edwin: can you paste your Xorg.0.log?
loki_666: edwin: http://pastebin.com/5CD2QBWQ
loki_666: edwin: no binary
MrCooper: edwin: jcristau explained it
loki_666: glxinfo: http://pastebin.com/mmpGPhaW
MrCooper: it's only available in 1.8 or Git snapshots
edwin: well I get 1.4 with this Xorg: X.Org X Server 126.96.36.1992 (1.7.6 RC 2)
edwin: 1.7.6 is newer than that, isn't it?
jcristau: edwin: your distro could have backported the patches
edwin: ah ok
jcristau: stole them from fedora
loki_666: jcristau: 1.4 is enabled for xorg-server 1.8?
jcristau: loki_666: yes
loki_666: ah ok
jcristau: with dri2
jcristau: or swrast
BeteNoire: starting to have this enough
BeteNoire: now when built firmware into kernel and disable kms i can't get to tty
BeteNoire: when enabled [ ] Enable modesetting on radeon by default - NEW DRIVER i couldn't get to gdm...
Pallokala: did you remember to compile/load fbcon?
suokko: BeteNoire: Do you have KMS log too? What about X log from UMS?
BeteNoire: http://pastebin.com/ZZ3jYWMM ← Xorg.0.log
BeteNoire: how do you get kms log?
suokko: BeteNoire: (II) RADEON(0): XAA Render acceleration unsupported on Radeon 9500/9700 and newer. Please use EXA instead.
BeteNoire: with exa i couldn't get to gdm
BeteNoire: is x 200m newer than 9500?
suokko: yes- But x200m is problematic card
BeteNoire: believe me, i know about this
suokko: I think you will need some fb device to render the console with UMS X driver (vesafb maybe)
mjr: mm, this has been talked about here earlier so, prototype of multiple X servers on one GPU using KMS: http://www.phoronix.com/scan.php?page=news_item&px=ODA3MQ
spreeuw: hey very nice developement with the multigpu
spreeuw: put the integrated video chips to work for 2D accell ;p
spreeuw: UMS would be dri1?
mjr: yeah IIRC the radeon UMS stack doesn't do dri2
BeteNoire: suokko: should i compile vesafb into kernel?
airlied: suokko: i can add it to the commit when I push it
suokko: BeteNoire: I use modules because of easier recompile
MarcoPau: hello, after my release upgrade to ubuntu lucid, althou I only have xserver-xorg-video-radeon installed, xorg runs vesa drivers. what am I supposed to do?
adamk: MarcoPau: Pastebin your /var/log/Xorg.0.log file please.
adamk: Install xserver-xorg-video-ati
adamk: Alternatively, create an xorg.conf file that tells the X server to use the radeon driver.
BeteNoire: lucid is out?
Tommeh: BeteNoire: beta1 was supposedly released yesterday, according to the schedule.
Tommeh: Meaning that it's getting sensible enough to start testing :)
Tommeh: I'm personally wondering how it's going to go, upgrading from Karmic to Lucid with all my extra packages from xorg-edgers installed.
adamk: It's still only at alpha2.
adamk: Oh, sorry, make that alpha3.
Tommeh: That's completely inaccurate - alpha3 was already released.
Tommeh: Ah, sorry.
adamk: And it's still March 19th here, which is the scheduled release date for Beta1.
BeteNoire: i have computer with kubuntu karmic which is not intended for testing so i'll wait for stable release
Tommeh: http://www.ubuntu.com/testing/lucid/beta1 doesn't give a 404 anymore, so I guess they're updating it :)
Tommeh: adamk: I'm certain it was the 18th yesterday.
Tommeh: I guess they had to push it back.
Tommeh: (I mean, I'm certain beta1 was due on the 18th)
Tommeh: But as you say, it's scheduled for the 19th now.
adamk: Well even yesterday the folks on #ubuntu+1 were talking about it being released on the 19th.
adamk: So I'm not sure when it changed or if you're just mistaken.
adamk: I vote for the latter :-)
Tommeh: adamk: I remember specifically speaking to a colleague about the release date of beta1, and that it was Thursday 18th (UK dates)
Tommeh: Therefore I'm not mistaken.
adamk: Yeah, yeah :-)
Tommeh: If I could view revisions I would! I can't see how.
adamk: Or you were both mistaken during this conversation :-)
adamk: Not that it matters, anyway.
Tommeh: Indeed, it's not massively important.
Tommeh: soreau: one of my favourites :)
Tommeh: adamk, from the mailing-list (I've just checked my personal inbox): Dear developers,
Tommeh: Owing to some late image rebuilds for a couple of critical issues that we
Tommeh: didn't want to leave our beta users subjected to, validation of 10.04 Beta 1
Tommeh: is still in progress. This means that Beta 1 will not be released on
Tommeh: Thursday, March 18 as planned; it will be released tomorrow, March 19.
Tommeh: is quite happy that he's not going mad ^_^
adamk: Lies... All lies!
GNU\colossus: hi guys
GNU\colossus: should a RADEON X300SE 128MB be enough to drive a 1680x1050x24bpp display, using a compositing wm?
GNU\colossus: fine :)
svenstaro: not much else though
GNU\colossus: the card is for my grandpa's PC, I guess he'll get along without fast-paced 3D shooters ;)
suokko: GNU\colossus: You never know ;)
Pallokala: If he took part in WWII then perhaps some Day of Defeat oslt. would be a good choice?
Pallokala: I don't really play so I don't know.
MostAwesomeDude: airlied: Yeah, that's...iffy.
MostAwesomeDude: I'd put a card in the mail for him, but the only spare I have is an rv410, which has that damn errata.
MrCooper: MostAwesomeDude: couldn't you work around that?
MostAwesomeDude: MrCooper: I got 2D to be relatively stable and sent a patch, but 3D still locked up after a bit.
MrCooper: that's probably not related
MostAwesomeDude: On the rv410?
MostAwesomeDude: Well, I was using r300c to test, so it sucks no matter which part's bad.
MostAwesomeDude: I suppose I could pop it back in and see if I can get it to behave...
MrCooper: yeah, if the workaround has any effect at all the erratum should be taken care of
MarcoPau: adamk: back to my radeon drivers issue. isn't xserver-xorg-video-ati something different?
adamk: The ati driver is a wrapper that loads the correct driver for your card.
adamk: When Xorg sees an ATI/AMD card, it loads the ati driver by default, and ati loads the radeon driver.
MarcoPau: I see
adamk: Since it's trying to load 'ati_drv.so' and that driver doesn't exist, it loads vesa iiiiiiiiiiiiiiiiiiiiiinstead.
MarcoPau: awesome "i" :D
MarcoPau: adamk: I'm still there, with vesa
adamk: Pastebin the new log file.
MarcoPau: adamk: I don't have ati module
adamk: Are you sure you installed xserver-xorg-video-ati ?
MarcoPau: ii xserver-xorg-video-ati 1:6.12.191-1ubuntu2 X.Org X server -- ATI display driver wrapper
adamk: Does /usr/lib/xorg/modules/drivers/ati_drv.so exist?
MarcoPau: adamk: nope
adamk: Then it's not installed :-)
MarcoPau: well you just saw dpkg -l | grep video-ati output. let me reinstall...
adamk: /usr/lib/xorg/modules/drivers/radeon_drv.so exists?
MarcoPau: wait, it's there
MarcoPau: adamk: it's there, but still not loading
adamk: /usr/lib/xorg/modules/drivers/ati_drv.so is there?
MarcoPau: (II) LoadModule: "ati"
MarcoPau: (WW) Warning, couldn't open module ati
MarcoPau: (II) UnloadModule: "ati"
MarcoPau: (EE) Failed to load module "ati" (module does not exist, 0)
MarcoPau: yes, it's there
MarcoPau: both ati_drv.so and radeon_drv.so
adamk: You're using the 32-bit version?
loki_666: what's the maximum texture resolution supported?
adamk: iirc, Mesa supports a maximum of 4096. Some video cards support less.
loki_666: i'm wirting a small game, but when i try to draw the the track texture (2048x2048) it hang
loki_666: i'm on RV610
adamk: All cards support at least 2048, iirc.
adamk: Oh, in that case it's 4096.
MarcoPau: adamk: yes
adamk: MarcoPau: Then I really don't know what the problem is. Sounds like an issue with the Ubuntu packages, perhaps.
loki_666: well i don't even have an error message, since the whole system seems to hang
adamk: MarcoPau: You could always create an xorg.conf file and load the 'radeon' driver directly.
adamk: MarcoPau: Then it won't try to load 'ati' at all.
MarcoPau: adamk: should I pick a xorg.conf file from anywhere and edit it?
MarcoPau: or just write a driver section?
adamk: I think just a Device section is fine.
adamk: Section "Device"
adamk: Driver "radeon"
adamk: I *think* that should work.
adamk: I have to run out for a bit.
MarcoPau: try to believe
MarcoPau: adamk: no identifier?
adamk: Oh, yeah, an Identifier is probably necessary.
edwin: loki_666: a game is not supposed to be able to DoS the system ;)
edwin: does it work with mesa in software mode?
loki_666: gonna try
loki_666: edwin: yes it does works un software
edwin: does it work in hw-accel mode, if that is the only texture you load?
edwin: maybe you run out of memory if you load lots of big ones
loki_666: i'm loading 3 texture the big track (which cover the whole screen 2048x2048) and 2 small textures for the cars (64x64)
loki_666: no compiz running
spreeuw: does exa still divide memory up into sectionS?
spreeuw: maybe theres the problem
spreeuw: have you tried using texture slightly smaller?
loki_666: w8 gonna try
loki_666: spreeuw: yes with 1024x1024 it seems to works
loki_666: no hang yet,
loki_666: spreeuw: what did you mean by exa divide memory up into sections?
spreeuw: it chops up the memory is sections
spreeuw: earlier it used to do this very ineffective
loki_666: so a texture could be split
mjr: you mean a tiled framebuffer?
spreeuw: 50% texture mem and 50 video
spreeuw: or so
loki_666: ah ok
loki_666: but if that would be the case, i should have an error from gl, when loading the texture
loki_666: and not a hang when trying to draw it
MarcoPau: adamk: xorg.conf won't help
adamk: MarcoPau: Oh?
adamk: Same result?
MarcoPau: adamk: it's still trying to load ati. yes, same result
MarcoPau: also tried with dection monitor and other, but that won't have any effect
adamk: MarcoPau: Please pastebin the output while you are using the xorg.conf specifying the radeon driver.
MarcoPau: adamk: http://pastebin.com/QKJ7ya7V
adamk: That's not using an xorg.conf file.
adamk: (==) Using default built-in configuration
MarcoPau: adamk: how would I tell him to use xorg.conf? it's there in /etc/X11/
adamk: MarcoPau: As long as /etc/X11/xorg.conf exists, Xorg will try to use it.
MarcoPau: adamk: well it's there...
adamk: Then your Xorg is majorly screwed up.
adamk: But I already suspected that.
adamk: I suggest reinstalling it all.
adamk: Or the entire distribution.
adamk: Your choice :-)
MarcoPau: oh well... :P let me try with an apt-get install --reinstall... which packages?
adamk: I would say anything and everything related to Xorg.
MarcoPau: I have both xorg and xserver-xorg. correct?
edwin: MarcoPau: dpkg -l|grep ii.*xorg
Obscene_CNN: edwin, try my updated patch yet?
MarcoPau: well, I'm stuck
edwin: Obscene_CNN: I tried your latest patch on the 2d driver only now, no GPU lockups, but there is still something wrong. When I login to LXDE and open a menu no items can be seen. When I mouseover them, the item's text appears.
adamk: MarcoPau: ?
MarcoPau: adamk: I don't know what more to try
adamk: Reinstalling Xorg didn't help?
MarcoPau: there don't seem to be bugs around
MarcoPau: related to my situation
adamk: I'd guess that something went wrong during the upgrade.
adamk: I did a fresh install of alpha2 a few days ago (using the amd64 version) and haven't had any of these issues.
Obscene_CNN: hmmmm, I guess thats progress :/ Thanks edwin
Obscene_CNN: On the bright side on my machine with my code I tried raising the amount of textures used in torcs. I went from a setting of 128 to 2048 with not much degradation in performance
Jonimus: MostAwesomeDude: are you there?
MostAwesomeDude: Jonimus: I am now. What's cookin'?
Jonimus: Did you ever get a chance to look at setting up a driver for the Wii gpu, I thought you had mentioned it in the past I was just wondering if you had actually looked into it?
Jonimus: I only bring it up as the distro I use on my Wii has upgraded to xorg 1.7 and the existing driver we have doesn't compile against it.
MostAwesomeDude: I haven't really had time or incentive, to be honest.
Jonimus: ok, well I was just wondering since you had mentioned it int he past, thanks anyway
MostAwesomeDude: Well, I'd like to. I've got a GC, and there's a Wii I can use too.
MostAwesomeDude: You should bug Isobel about it though; you should be able to fbdev it with the kernel fb.
Jonimus: MostAwesomeDude: I think that is basically what they did, a modified fbdev to convert between colorspaces before writting.
Jonimus: Isobel is not interesting in video sadly, she is mainly a console person from what I understand, I'll maybe email the guy who made the original driver about it again.
MostAwesomeDude: Well, it *should* be pretty easy to get a simple KMS driver up and running, but again, I haven't had the time.
MrCooper: still it should be muuuch simpler to get an existing driver to build against xserver 1.7 :)
Jonimus: If I knew C I'd gladly do it but alas I only know a bit of Java(ew), Python, and z80 asm.
Dr_Jakob: Jonimus: abit of Java is all you need.
Dr_Jakob: it shouldn't be that hard to see what the errors are.
Jonimus: Dr_Jakob: enough has changed that I'm pretty sure it needs added code to work right
dileX: (build.log for 7.8 GIT is 300k smaller than for mesa GIT master)
Obscene_CNN: edwin, I just sent you a new patch with a new bug fix. It also should improve performance of r600 textured videofunctions a tad
MrCooper: Jonimus: not sure what between 1.6 and 1.7 would require significant driver changes
Jonimus: all I know is last time i tried I was able to get it to comple but X still complained but it still refused to work.
edwin: Obscene_CNN: still not working, http://img709.imageshack.us/img709/3941/snapshot1l.png
edwin: Obscene_CNN: see that screenshot for the corruption I get
edwin: the grey/black areas are where other windows used to be
edwin: and the elements in that menu show up as I mouseover
edwin: not fully ok, but readable
Jonimus: edwin: nice theme there btw
edwin: heh, its the default
edwin: and it starts faster than kde for testing ;)
Obscene_CNN: does kde do something different?
Jonimus: lol, I use xfce but that panel looks rather nice
edwin: no, I started ksnapshot to take this snapshot
edwin: it was black
edwin: when I mouseovered I've seen the buttons
edwin: I don't know if xfce uses compositing or not
edwin: anyway I only built 2D driver this time, since your mesa patch locks up the GPU ;)
Obscene_CNN: alright thanks, I'll continue to look for bugs
edwin: there are some commented out stuff in your patch
Jonimus: it can composite yes
edwin: are those meant that way?
edwin: things like R600_CLEAR(cs)
edwin: - CLEAR (cb_conf);
edwin: - CLEAR (vs_conf);
edwin: - CLEAR (ps_conf);
edwin: these ones I mean
Obscene_CNN: yes, I inlined those functions
edwin: doesn't look like you inlined at that particular location
edwin: @@ -223,16 +226,16 @@ R600PrepareSolid(PixmapPtr pPix, int alu, Pixel pm, Pixel fg)
edwin: I'm talking about this hunk
Jonimus: you'd think gcc would do those optimisations on its own though...
edwin: well they are macros, it does...
edwin: I thought you commented those out because they were redundant
Obscene_CNN: creating a structure , doing a memset to zero, then changing just a few variables in it , and finally handing it off to a function that parses all of the structure relevant or not isn't very efficient
Obscene_CNN: one moment......
edwin: ah you removed the entire cb_conf stuff
Obscene_CNN: yes, the CLEAR functions are just macros that call memset
cheako: Hello, can any one play doom3 on an RV670?
Obscene_CNN: the vs_setup, ps_setup, and set_render_target functions are in the file r6xx_accel.c . By looking at the values that are to be set in their corresponding data structures and tracing through those functions you can see how much code can be cut
Obscene_CNN: I'm betting that the bug is in the composit functions some place
UbuntuUser: hello, I'm using ArchLinux and have a question to the xorg.conf file:
UbuntuUser: should I use 'ati' or 'radeon' as driver?
UbuntuUser: or is it the same?
stikonas: UbuntuUser: ati is just the wrapper, so these are basically the same
UbuntuUser: ah ok
stikonas: but most people use radeon
UbuntuUser: I ask because I read in a Ubuntu wiki ati and in a Arch one radeon ;)
UbuntuUser: ok, thx for the answer. cu
stikonas: I myself don't use xorg.conf at all
UbuntuUser: keep up the good work :)
UbuntuUser: the opensource driver gets better and better :)
MostAwesomeDude: agd5f, airlied : Around? I'm sitting here with one of the guys with the Dell RV350 with S/R issues.
airlied: MostAwesomeDude: oh nice
MostAwesomeDude: Backlight doesn't come back on, external monitor is corrupted, fdo bug 24097.
airlied: MostAwesomeDude: you can ssh in after resume?
MostAwesomeDude: airlied: Yep. Not hardlocked, VT switch works.
MostAwesomeDude: Cursor works, although all VRAM's hosed.
airlied: MostAwesomeDude: cn you unload radeon, vbetool post and reload radeon?
MostAwesomeDude: Sure, one sec.
airlied: you'll need to unbind console
soreau: rv350 \o/
soreau: MostAwesomeDude: My dell+agp rv350 works fine for s/r last I checked
airlied: MostAwesomeDude: is it a Lenovo?
Jonimus: "Dell RV350 with S/R issues."
airlied: oh its a Dell also
MostAwesomeDude: airlied: Dell. Unbind, rmmod, vbetool post, modprobe, doesn't work.
airlied: MostAwesomeDude: radeon.agpmode=-1?
MostAwesomeDude: Haha, what.
MostAwesomeDude: Modprobing with agpmode=-1, removing it again, and re-posting causes the backlight and cursor to come on.
MostAwesomeDude: VGA mode! Whoohoo!
MostAwesomeDude: ...but loading the module again causes everything to go away.
airlied: MostAwesomeDude: nice
glisse: reloading with agp desactived ?
DanaG: hmm, the only place I had agpmode issues was on an nforce2 board.
DanaG: It gave a null-pointer dereference.
DanaG: had to disable agp on nforce2.
glisse: i think here it's better to disable agp all time until you fix pci case
DanaG: AGP actually worked fine for me on a VIA board, though. =þ
DanaG: GPU was R350.
airlied: yeah the M10 problem is in laptops, I'm not sure if its not powering off the whole way or not setting up the VRAM or something on resume
MostAwesomeDude: Whoo, this is fun.
MostAwesomeDude: agpmode=-1 followed by rmmod and post gets VGA back, right?
MostAwesomeDude: Doing UMS after that works fine.
MostAwesomeDude: Just not KMS.
MostAwesomeDude: So there's a UMS quirk missing from KMS, I guess?
airlied: MostAwesomeDude: no juust sounds like luck ;-)
MostAwesomeDude: airlied: If it's reproducible, is it still luck? :3
airlied: MostAwesomeDude: yeah ;-)
airlied: MostAwesomeDude: we should try and get radeon.agpmode=-1 + vga monitor to resume first I think
MostAwesomeDude: Well, that's what I meant.
MostAwesomeDude: Rebooted with agpmode=-1, s/r, rmmod, vbetool post, UMS works.
airlied: MostAwesomeDude: yeah without vbetool post and UMS steps
MostAwesomeDude: Hm. So what should my next step be?
Curan: just as a little feedback (I'm not sure if it's a bug in KDE/Konsole or in radeon, so first here): when I enable KMS color tiling (by setting the option in xorg.conf), Konsole doesn't have a transparent background anymore, as soon as I disable color tiling (ie. remove the option from xorg.conf), the background is transparent again
Curan: nothing is log in Xorg.0.log or visible via dmesg
Curan: so I'm not sure who is to blame, but if somebody has pointers where to look, I'm happy to provide that information)
airlied: MostAwesomeDude: get mmapr/mmapw and on resume poke random bits of VRAM
airlied: to see if writes are sticking
MostAwesomeDude: mmapr and mmapw were nuked a while back. Would regsetting a scratch reg work?
Obscene_CNN: edwin, could you back out my patch just for the file r6xx_accel.c and see if that fixes things?
airlied: MostAwesomeDude: no its not regs its VRAM
airlied: MostAwesomeDude: mmapr.c is in my fd.o
airlied: mmapw.c is in google
MostAwesomeDude: Hey, nifty. Do I just grab the offsets from lspci and go to town? :3
airlied: MostAwesomeDude: use /sys/bus/pci/
airlied: MostAwesomeDude: also owrk out what is in theory being scanned out of VGA
airlied: debugfs + dri/0/radeon_vram_mm
airlied: should be possible to find the frontbuffer object
airlied: and check 0x23c, 0x33c, + 0x224, 0x324
MostAwesomeDude: Writes stick. Found something that looks a lot like a frontbuffer.
airlied: MostAwesomeDude: does hte VGa monitor come on?
airlied: does writing random misc to VRMA change whats on the monitor?
nike: Hi there!
nike: I am trying to figure out if I have KMS enabled, or at least why 2-D acceleration seems to be broken on my HD 4670
nike: I have 188.8.131.52 and radeon driver 6.12.192
MostAwesomeDude: Hm. Writing trash doesn't appear to do anything.
nike: My CONFIG_DRM_KMS_HELPER=m
nike: So perhaps I need to run a command to enable KMS?
nike: Would enabling KMS make my 2-D accel (and maybe 3-D as well) start working? Right now dragging windows around or scrolling in web pages is extremely laggy.
edwin: what is CONFIG_DRM_RADEON_KMS?
airlied: MostAwesomeDude: so its still scanning out garbage?
nike: edwin: Uh-oh... not set? It has a # mark in front of it!
nike: Is this why my acceleration is broken, or is it only have to do with VT and resolution switching?
edwin: UMS should have accel too (2D)
nike: I went from an r500 card which was working to this r700 one...
edwin: ah for r700 you need a new mesa
nike: I thought by using 2.6.33 and the radeon RC of the upcoming 6.13 version would make it work with acceleration
edwin: for 3D
edwin: and for 2D you probably need the latest -rc of the 2D driver too
MostAwesomeDude: airlied: Well, we restarted, so fbcon is -- ooh.
nike: edwin: I have 7.7 mesa
edwin: nike: for 3D you need mesa 7.8 (or git)
nike: Is 6.12.192 not the latest RC?
edwin: yeah thats fine, that is what I got too
edwin: I use KMS though
edwin: on boot
edwin: make sure you got the firmware too
edwin: without it there is no 3D
edwin: see the wiki in the topic for some instructions
edwin: on building mesa, etc.
edwin: I have to leave now
Obscene_CNN: goodbye edwin
nike: I have the firmware installed
nike: That and flash are the only non-free things i have
nike: I don't care as much about 3-D as I do 2-D
nike: edwin said UMS should enable 2-D accel?
nike: How can I check if I have UMS?
Obscene_CNN: if you don't have KMS running thhen UMS is running
nike: Okay, well, why do I not have 2-D accel?
nike: I am running my WM without compositing
nike: I am powering 2 displays at 1080p though
nike: Is that... too much for the driver to handle or something?
dileX_: nike: might have a look into wiki (see topic)
Obscene_CNN: okay type glxinfo
Obscene_CNN: in a terminal
Obscene_CNN: terminal window
nike: direct rendering is YES
nike: GLX version is 1.4
nike: what else should I look for?
Obscene_CNN: what does OpenGL renderer string:
nike: Obscene_CNN: Software Rasterizer
nike: That's not good.
Obscene_CNN: what version of the kernel do you have?
dileX_: nike: check dmesg and Xorg.log (or paste somewhere)
nike: Obscene_CNN: 184.108.40.206
Obscene_CNN: okay good.
nike: Oh shoot
nike: failed to load firmware radeon/RV730_pfp.bin!
nike: in dmesg
nike: Is that why I have no 2-D accel?
Thunderbird: direct rendering alone is meaningless; you always have to check the opengl renderer string as well
nike: I have /lib/firmware/radeon/RV730_cp.bin
nike: What is looking for RV730_pfp.bin? The driver?
MostAwesomeDude: airlied: Found it!
dileX_: you are missing RLC firmware file
MostAwesomeDude: It wasn't at the right location.
nike: Ah... okay so my distribution does not have this. I do not know what RLC is, but can I download it from somewhere?
MostAwesomeDude: Gonna go ahead and S/R and see if it still works.
MostAwesomeDude: Hm, looks like the FB moved. Tryin' to find it.
nike: Okay I got the pfp firmware, do I need to reboot or just restart X?
Jonimus: nike: rebooting would be the best bet
MostAwesomeDude: airlied: Alrighty, I scribbled over where the CRTCs are pointing, but the VGA still shows garbage.
MostAwesomeDude: Hm. Okay, so scribbling on the fb works fine before S/R, not after S/R. :T
MostAwesomeDude: Meh, calling it quits for today.
Obscene_CNN: buys MostAwesomeDude a virtual beer
kees: MostAwesomeDude, airlied: thanks again for poking at this poor laptop. :)
Obscene_CNN: gets ready to go buy himself a real beer and flirt with the pretty waitresses :D
spreeuw: dont hit on personell man ;p
Obscene_CNN: Actually I sit next to the terminals at the bar and they come over and use it so they can talk to me. Its not like I'm forcing my self on them.
agd5f: MostAwesomeDude, airlied: maybe we don't set DISP(2)_BASE correctly
MostAwesomeDude: agd5f: it was 0x0 on both regs, ith
airlied: agd5f: I think I check that on EH's laptop, but I'm sorta lost to what would stop VGA working
airlied: unless it some display wratermark type stuff
airlied: those M10s had a few wierd errata,
Fisiu: I tried kms with radeon hd4770, but something is wrong: http://img233.imageshack.us/img233/3778/xchatf.png
Fisiu: any idea?
Fisiu: it look like this only when I make a screenshot
Fisiu: mesa, ati driver and all stuff taken from git
spreeuw: so during normal operation everything is correct?
Fisiu: almost, some artefacts are visible randomly
spreeuw: screenshot with import looks ok here
spreeuw: try it: import shot.png
spreeuw: maybe something is broken with compositing
spreeuw: I dont use that here
Fisiu: import do the same thing: http://img221.imageshack.us/img221/3514/scrn.png
Fisiu: glxinfo looks ok, almost ok, without first 2 lines: http://wklej.org/id/300203/
Fisiu: disable xfce4 compositor didn't help at all
Fisiu: with mesa and radeon from git (1 week old) it worked fine
Fisiu: no strange artefacts
Fisiu: but after update those 2 packages yesterday, it just happened
Fisiu: I can provide any log if any is necessary
Jonimus: Fisiu: did you update libdrm as well?
Fisiu: yes, libdrm was updated too
Fisiu: after update libdrm I rebuilded mesa and ati package
Fisiu: my Xorg.0.log: http://wklej.org/id/300207/
scarabeus: Fisiu: wrong firmware?
Fisiu: now I have only: http://wklej.org/id/300208/
Fisiu: scarabeus, do I need firmware from http://people.freedesktop.org/~agd5f/radeon_ucode/ ?
scarabeus: Fisiu: you do
Fisiu: ok, done
Fisiu: I need relogin, brb
Fisiu: firmware R600 didn't help
Fisiu: ok, need go now
Fisiu: we'll be in touch
Fisiu: good night
[Enrico]: http://www.phoronix.com/scan.php?page=article&item=intel_atom_ionamd&num=2 <---- what to say? not bad at all! radeon kms on a r700 can't compete with an nvidia propetary driver (for now) but result are not so bad. good work guys keep improving radeon. Thank you very much