Home Phoronix Phoronix Forums X.Org Videos From FOSDEM 2008

Radeon IRC Logs For 2009-11-08

Search This Log:


Luzipher: still good to know there is at least something like ... printf ("Hello world\n"); ... :D
MostAwesomeDude: Why is SPI_INTERP_CONTROL notated as SPI_INTERP_CONTROL_0 if there's only one of them?
EruditeHermit: MostAwesomeDude, http://pastebin.com/m69b2b0b0
MostAwesomeDude: EruditeHermit: Don't use g3dvl. The only trackers I know to build and work right now are dri and egl.
EruditeHermit: ok
EruditeHermit: what about xorg?
MostAwesomeDude: It should build and work, but I don't test it regularly.
EruditeHermit: is egl required?
MostAwesomeDude: No.
soreau: compiz runs fine here except wall and cube is green/red gradient and lots of glitches
soreau: (glitches with wall and cube)
EruditeHermit: any ideas how far along the glsl rework is?
EruditeHermit: yay it finally compiled
EruditeHermit: g3dvl was the culprit
MostAwesomeDude: The green-red gradient is texcoords; it means that that texture was broken.
MostAwesomeDude: osiris has a patch that fixes some of them.
soreau: yea cool looking broken textures for show mouse too
soreau: I can't believe I was using r300g for this time.. and didn't know it
soreau: for compiz anyway
soreau: I thought the cube/wall red/green gradient was a glitch
MostAwesomeDude: No, what happens is that texcoords are indexed using STRQ.
MostAwesomeDude: And so 2D textures only use ST. (It looks like ST01, internally.)
MostAwesomeDude: And those happen to also be RGBA, for colors.
MostAwesomeDude: So if a texture doesn't get mapped, then the RG01 colors get used, which looks like...
MostAwesomeDude: ...gradients of red (R) and green (G) with no blue.
MostAwesomeDude: And now you know. :3
soreau: Well I am still impressed with compiz running using r300g
soreau: compiz water effects makes it look crazy cool :)
soreau: still with the green/red gradient type bug though
MostAwesomeDude: Yeah, it's probably a combination of texrect factors and NPOT brokenness.
soreau: Awesome though MAD :)
MostAwesomeDude: Yeah, I try. :3
MostAwesomeDude: git commit -a
MostAwesomeDude: Dammit.
soreau: \o/
soreau: :D
soreau: cd MostAwesomeDude; git push
EruditeHermit: I am looking forward to running it
EruditeHermit: brb reboot to test
soreau: I'm getting all kinds of spam in the terminal when running compiz: X Error: BadMatch (invalid parameter attributes) 8 Extension: 149 (Uknown extension) Minor opcode: 6 (Unknown request) Resource id: 0x340000e
soreau: just receently
soreau: don't know where it's coming from
MostAwesomeDude: That's probably bad.
soreau: With 'normal' KMS
soreau: but compiz seems to be running ok
soreau: only started happening recently
EruditeHermit: MostAwesomeDude, it works
EruditeHermit: what do you want me to test?
MostAwesomeDude: EruditeHermit: Make sure glxgears works right, then test whatever you like.
EruditeHermit: yeah it works fine
MostAwesomeDude: If you find incorrect rendering, let me know or file a bug report.
soreau: EruditeHermit: Here on my rv350 glxgears are not working right
EruditeHermit: oh really?
EruditeHermit: works fine here
soreau: s/working right/rendering correctly
EruditeHermit: renders fine here
EruditeHermit: what is wrong for you?
soreau: EruditeHermit: How are you running with r300g?
MostAwesomeDude: soreau: How so? Can you get me a screenshot?
soreau: MostAwesomeDude: yea
EruditeHermit: OpenGL renderer string: Gallium 0.3 on RV350
soreau: EruditeHermit: Here I am running kms but using LIBGL_DRIVERS_DIR
EruditeHermit: yeah same
EruditeHermit: LIBGL_DRIVERS_PATH=
soreau: MostAwesomeDude: http://omploader.org/vMnEwMw
EruditeHermit: is it in compiz?
soreau: yes
EruditeHermit: try not in compiz
soreau: but this happens regardless
MostAwesomeDude: soreau: You're not running head.
soreau: MostAwesomeDude: ok
MostAwesomeDude: I fixed this already; make sure you're using a commit at or after my merge of r300g-vbo.
soreau: Yea, I seem to not be on any branch for some reason
MostAwesomeDude: :T
MostAwesomeDude: That's no good.
soreau: MostAwesomeDude: Thanks for pointing that out, sorry for misreporting
EruditeHermit: MostAwesomeDude, its just slow
soreau: I have no idea how that happened
MostAwesomeDude: soreau: No worries, I expected it'd happen to some people.
soreau: EruditeHermit: What is slow?
EruditeHermit: gallium
MostAwesomeDude: EruditeHermit: "Slow" is such a relative term. :3
EruditeHermit: lol
EruditeHermit: its a LOT faster than it used to be
soreau: Works same speed here
MostAwesomeDude: I bet it's faster than forcing swrast/softpipe.
EruditeHermit: but compared to dri1
EruditeHermit: its slower
soreau: until segfaults or whatever
EruditeHermit: yeah its faster than softpipe
EruditeHermit: it used to be slower than softpipe
EruditeHermit: =p
soreau: EruditeHermit: dri2 is slower than dri1 without gallium
MostAwesomeDude: Speed improvements *are* planned.
EruditeHermit: soreau, true
soreau: But here it doesn't seem to matter much if at all
EruditeHermit: dri2 is about 1/2 as fast on openarena
MostAwesomeDude: And yes, DRI2 is currently slow for a variety of reasons, some of which will be fixed and some which will not.
soreau: I suspect you're using some kind of benchmarking utility :P
EruditeHermit: openarena
EruditeHermit: the anholt timedemo on openarena
soreau: does not benchmark anything since it's unreliable and subjective
EruditeHermit: fire is working
EruditeHermit: wow
EruditeHermit: thats pretty cool
EruditeHermit: ah well
EruditeHermit: actually it is corrupted a little
soreau: It should be corrupted a lot
EruditeHermit: the fire is fine
soreau: Try cube, expo or show mouse
EruditeHermit: the background textures are not so good
EruditeHermit: soreau, oh not compiz
soreau: oh my bad
EruditeHermit: the stuff in the demos dir
soreau: thought you were talking about compiz fire :)
EruditeHermit: yeah
EruditeHermit: =p
EruditeHermit: sorry my bad
soreau: On that note, you should test compiz
EruditeHermit: i'm afraid
EruditeHermit: ok
EruditeHermit: it works
EruditeHermit: just really slow/laggy
soreau: I wonder why
EruditeHermit: cube rotate is broken as you mentioned
EruditeHermit: =p
soreau: Is your rv370 better or worse in terms of speed to my rv350?
EruditeHermit: mine is an rv350
EruditeHermit: its a laptop card
EruditeHermit: M10/11
soreau: Ahh
soreau: teh suck?
EruditeHermit: its a little slow
EruditeHermit: but still probably 10fps
soreau: I always though your rv370 was an AGP card in a PC
MostAwesomeDude: Yeah, the new fastpath isn't as fast as it could be.
EruditeHermit: its impressive still
MostAwesomeDude: For starters, PSC is calculated twice.
EruditeHermit: last I tested gears was broken
MostAwesomeDude: I actually found out how gears was broken. It's amusing.
soreau: Here it is nice and speedy, kms that it. With or without r300g it's the same
soreau: s/it./is,
MostAwesomeDude: I still need to unbreak the gears for SW TCL people.
soreau: MostAwesomeDude :)
soreau: waits for mesa to finish building
EruditeHermit: WARNING: Application calling GLX 1.3 function "glXDestroyPixmap" when GLX 1.3 is not supported! This is an application bug!
soreau: EruditeHermit: It's Just a MESA Warning!
soreau: That's been in mesa for over a month now
MostAwesomeDude: EruditeHermit: That's just a compiz bug when you're doing DRI1.
EruditeHermit: what happened to firecube?
soreau: MostAwesomeDude: Yea that's fixed but resizing the gears window is still broken
soreau: MostAwesomeDude: It happens with DRI2 too (the compiz/mesa warning)
MostAwesomeDude: soreau: Ah, maybe with slightly older Xserver. New Xserver should advertise GLX 1.4 properly.
soreau: EruditeHermit: progs/demos/fbo_firecube ?
soreau: MostAwesomeDude: X 1.7.1?
MostAwesomeDude: Heh, firecube? I'll be impressed if it works. Or gearbox.
soreau: I mean..
EruditeHermit: ah
MostAwesomeDude: soreau: I think the patch is in the 1.7 series, yes.
soreau: MostAwesomeDude: I am using X 1.7.1
EruditeHermit: firecube is really broken
MostAwesomeDude: Either way, I really wouldn't worry about it; compiz knows what it's doing.
soreau: EruditeHermit: Fire works 'better' without r300g
soreau: and it has been working with 'normal' dri2 for some time now
soreau: MostAwesomeDude: Indeed
EruditeHermit: you guys have a lot to fix before bug reports are that useful for r300g
EruditeHermit: =p
soreau: MostAwesomeDude: The biggest issue I had with running X components from git (required by the server) was header files renamed with a couple of packages. So other packages that were looking for those headers would fail to build when looking for the 'old' header in whatever location
soreau: EruditeHermit: They're covering a lot of ground fairly quickly afaict
MostAwesomeDude: EruditeHermit: Yeah, I'm aware. :3
eosie: MostAwesomeDude: you asked about texture format differences between r500 and r600 many hours ago... well, there are also real 32bit integer textures
eosie: I just got up ;)
zhasha: yay, splitz
eosie: and GLSL with r300g works most of the time (tested before r300g-vbo was merged)
zhasha: eosie: I don't think r300 can support GLSL though
eosie: 3D application developers are very aware of that, nobody with a brain would use integer arithmetics in GLSL1.2
eosie: zhasha: not fully, but mesa should
zhasha: I know, but software fallback is such a terrible shame :(
eosie: it's ARB's failure, not ours, we just have to ingest their mess
eosie: I am dreaming about realtime switching between r300g and llvmpipe, which ironically wouldn't be realtime at all
airlied: llvmpipe vertex shader failover should be fine in theory
MostAwesomeDude: And Gallium can get a more fine-grained API to properly expose the HW caps.
EruditeHermit: eosie, r300g has GLSL?
eosie: yes, it has
EruditeHermit: eosie, none of the GLSL tests I ran worked
MostAwesomeDude: ...And which tests did you run?
EruditeHermit: demos in the glsl dir
EruditeHermit: glxinfo reports 1.5
eosie: that's too old
EruditeHermit: I just compiled it today
EruditeHermit: it says Already up to date
EruditeHermit: when I do git pull
eosie: some piglit glsl tests have been working for a week I guess
EruditeHermit: which ones?
eosie: wait
MostAwesomeDude: piglit tests will work; most of the Mesa GLSL demos are rather mean and test esoteric features.
MostAwesomeDude: e.g. *no* driver, proprietary or open, does noise right.
EruditeHermit: so where is the GLSL support?
EruditeHermit: also when will the renderer string say 2.1 support?
MostAwesomeDude: When I fix mesa-st.
MostAwesomeDude: And GLSL support is incremental, but the GLSL hooks magically appear when 2.0 support is enabled.
MostAwesomeDude: This is a series of mesa-st bugs IMO.
airlied: I thought someone already made GL2.0 report
eosie: EruditeHermit: http://storm.unas.cz/eo_2009-11-06_18.34.23/index.html
EruditeHermit: wow
MostAwesomeDude: airlied: Yeah, it was done by faking PIPE_CAP_NPOT.
EruditeHermit: much better than I expected
MostAwesomeDude: As you might imagine, I am *not* enthused about doing that permanently.
EruditeHermit: apart from speed
MostAwesomeDude: eosie: Wow. That's actually really encouraging.
EruditeHermit: features are getting very complete
happycube: indeed
eosie: GLSL is even in GL1.5 through ARB_shader_object, ARB_fragment_shader, ARB_vertex_shader and ARB_shading_language
eosie: or something like that
MostAwesomeDude: eosie: Yes, but those entrypoints aren't automatic to use and most GLSL apps don't even bother checking for them.
eosie: but many glsl tests require GL2.0 apparently for no reason
MostAwesomeDude: Hell, the Mesa GLSL demos don't check for them.
MostAwesomeDude: And piglit certainly doesn't try to use them.
eosie: MostAwesomeDude: the test report is before r300g-fbo was merged
eosie: *vbo
MostAwesomeDude: eosie: Oh, really? That's surprising.
MostAwesomeDude: I'd expect a fair handful of the tests to still be broken.
eosie: yeah
eosie: there are a few other tests which pass but aren't on the list because they take too long to complete
zhasha: eosie: got a report from after the merge?
eosie: zhasha: I am testing right now
zhasha: awesome
eosie: something you should know about:
eosie: r300_vbo.c:98: warning: implicit declaration of function ‘format_is_supported’
zhasha: what about it?
eosie: it makes my rendering indirect
eosie: I commented that function out
zhasha: it should at runtime, throw a linking error in dlopen and fall back on software rendering
eosie: not format_is_supported, I meant setup_vertex_buffers which uses it and isn't called at all
zhasha: I know what you mean, and a git grep reveals the function isn't declared anywhere and thus will generate a linker error at runtime
zhasha: it's not desired, I'm just saying that's what'll happen
TooTea: Hello, I have a simple question, but after more than half an hour googling I still haven't found any anwer/docs: How do I set the KMS mode (resolution)?
zhasha: TooTea: it should do it by itself
dileX: from the backlog: "23:59 #radeon: < EruditeHermit> the anholt timedemo on openarena"
dileX: where is this timedemo?
TooTea: zhasha: well, it does, but it picks some completely silly resolution (huge resolution with extremely low refresh rate)
EruditeHermit: dileX, http://dri.freedesktop.org/wiki/Benchmarking
zhasha: TooTea: is this under an X session or just in the terminal in general?
dileX: EruditeHermit: thx, found already
TooTea: zhasha: in the terminal...
zhasha: then I have no idea
TooTea: zhasha: afterwards, when X tries to set the correct mode, I end up with a blank screen
dileX: argh "The requested URL /~anholt/benchmarking/anholt.cfg was not found on this server."
dileX: EruditeHermit: can you upload that two files somewhere?
EruditeHermit: dileX, know anywhere I can upload it to?
dileX: EruditeHermit: how big are the files, btw?
EruditeHermit: not very big
dileX: rapidshare?
EruditeHermit: dileX, http://pastebin.com/m39a61ec4 this is anholt.cfg that goes in ~/.openarena/baseoa/
pixelhaze: Hi, I cloned the latest drm and mesa from git repos. After building radeon driver (Gallium version), glxgears crashes if softpipe is not used.
pixelhaze: GDB reveals a problem at line 131 in "r300_context.c". An assert(0) is executed if variable "r300screen->caps->has_tcl" is false. Is this expected ?
EruditeHermit: dileX, http://rapidshare.com/files/304005471/anholt.dm_68.html
eosie: pixelhaze: could you paste GL_RENDERER here?
pixelhaze: glxinfo crashes as well
pixelhaze: r300_context.c:131:r300_create_context: Assertion `0' failed.
pixelhaze: debug_get_bool_option: GALLIUM_ABORT_ON_ASSERT = TRUE
pixelhaze: the above lines are printed when running glxgears
pixelhaze: or glxinfo
eosie: set 0 there
pixelhaze: eosie: Can you kindly elaborate?
eosie: pixelhaze: did you said in your emails that you have Xpress X1100?
eosie: *say
pixelhaze: yes
EruditeHermit: brb
eosie: according to notebookcheck.net, you have hwtcl but gallium thinks you don't, and I have no idea what's true
pixelhaze: oh!
rozzo: hello
rozzo: oi have a problem with my radeon driver
rozzo: at random the system get 100%
rozzo: and the system become instable
eosie: pixelhaze: let's assume you have hwtcl, r300_chipset.c: at the end of the file, add "caps->has_tcl = TRUE", try glxgears and pray
rozzo: and i reboot manually
pixelhaze: :-) ...ok, will try it right now
rozzo: anyone can help me?
pixelhaze: eosie: glxgears did not crash, but it did not render the output
pixelhaze: there was a corrupted patch across the screen though
pixelhaze: and the fps is reported as "4 frames in 5.2 seconds = 0.772 FPS"
eosie: turns out you don't have hwtcl
pixelhaze: ok
eosie: try master before r300g-vbo was merged
dileX: EruditeHermit: thx for providing the files. how long is that demo going?
EruditeHermit: dileX, what do you mean how long?
eosie: pixelhaze: git reset --hard 57d77c6a
pixelhaze: ok, will do it right away
EruditeHermit: I get 125fps in DRI1 and 66fps in DRI2/kms
dileX: EruditeHermit: time - secs mins hrs?
EruditeHermit: 6seconds approx
EruditeHermit: for dri1
EruditeHermit: is that what you mean?
pixelhaze: eosie: Do I need to build libdrm afresh for the reset version of mesa ?
eosie: no, you don't
pixelhaze: yes, autogen went through, it is buiilding now
EruditeHermit: dileX, 840 frames 6.8 seconds 124.1 fps
pixelhaze: still building ...
EruditeHermit: gnight all
dileX: EruditeHermit:
dileX: $ openarena +exec anholt 2>&1 | egrep -e '[0-9]+ frames'
dileX: 840 frames 9.0 seconds 93.7 fps 4.0/10.7/50.0/4.7 ms
EruditeHermit: dileX, DRI1 or DRI2?
dileX: GLX Renderer Mesa DRI R300 (RV515 714A) 20090101 x86/MMX/SSE2 TCL DRI2 GLX Version 1.5 Mesa 7.7-devel
dileX: EruditeHermit: dri2 ^^
EruditeHermit: yeah, thats faster than me
EruditeHermit: DRI 2 for me is about 66fps
pixelhaze: eosie: glxgears crashes with no output, only the window is shown with a blank area
EruditeHermit: dileX, got to sleep
EruditeHermit: gnight
EruditeHermit: back tomorrow
dileX: EruditeHermit: gn8
pixelhaze: GDB reports a different location this time
pixelhaze: at r300_state_derived.c:88
eosie: pixelhaze: the reason why it doesn't work might be that no one ever tested r300g without hwtcl
pixelhaze: ok
dileX: lol
dileX: the anholt oa-benchmark runs on r300g dri/st with freaky colours, but runs (normal oa start shows blank screen)
dileX: $ openarena +exec anholt 2>&1 | egrep -e '[0-9]+ frames'
dileX: 840 frames 36.9 seconds 22.8 fps 15.0/43.9/94.0/13.1 ms
dileX: $ inxi -G
dileX: Graphics: Card ATI M52 [Mobility Radeon X1300] X.Org 1.7.1.901 Res: 1280x800@59.9hz
dileX: GLX Renderer Gallium 0.3 on RV515 GLX Version 2.1 Mesa 7.7-devel
pixelhaze: eosie: what is the suggested way forward ?
pixelhaze: eosie: I just recollected that I had run glxgears with Mesa radeon-rewrite . I had got about 1400 fps .
pixelhaze: With fglrx I had got 1900 fps .
rozzo: pixelhaze, you have compile glxgear?
pixelhaze: yes, it came along with mesa
rozzo: what distro you use?
pixelhaze: sources from git. It is part of the demos.
rozzo: gh
pixelhaze: I use Zenwalk (Slackware)
rozzo: :Q__________
pixelhaze: ?
rozzo: i'm older user of slackware
rozzo: i love it
rozzo: but i don't have time
rozzo: dont*
rozzo: to follow or recompile dependencies
rozzo: now i'm using fedora but i have a tragig problem
rozzo: tragic*
rozzo: seems the drm is the problem
eosie: pixelhaze: well, not sure, r300g is quite unstable right now, you have 2 options: 1) make it work somehow 2) wait a few more months
pixelhaze: Zenwalk is good for general use as well as experimentation. I use Xubuntu otherwise.
pixelhaze: eosie: I would like to help identify the issues and report it regularly in that case.
rozzo: pixelhaze, eh yes but i remember, if i recompile kernel in slackware i don't have problem
rozzo: but in fedora the sistem become crazy
glisse: rozzo: never had issue on fedora
glisse: make, make install and everythings works
pixelhaze: It depends on the setup one makes I guess.
rozzo: glisse, i try to dowgrade kernel
rozzo: and the entire system
rozzo: not boot, syslog hang, hal hang
rozzo: and more
pixelhaze: eosie: Are there any docs on hwtcl which I can refer/ cross refer ?
eosie: zhasha, MostAwesomeDude: the latest version: http://storm.unas.cz/eo_2009-11-08_11.46.51/index.html
eosie: i changed the test set a little bit
eosie: it's not better nor worse
zhasha: aww
glisse: rozzo: downgrade to which point ?
glisse: you need to enable someoptions on old kernel
rozzo: kernel 2.6.30 to 2.6.27
glisse: stuff like sysfs worksaround
rozzo: i not recompile the kernel i get the rpm
glisse: well what you are trying to do is asking for pain
eosie: pixelhaze: the only useful doc is the r5xx acceleration guide, but in your case we should make it work _without_ hwtcl... one day
pixelhaze: ok
pixelhaze: Should I make a bugzilla entry in that case ?
eosie: that is not necessary
pixelhaze: ok
eosie: very experimental stuff
pixelhaze: eosie: Thank you for looking into the issue.
eosie: you're welcome
pixelhaze: signing out ..Catch u all later.
hoo-hah: hi all. any of you debian users? I'm planning on switching from gentoo to something more stable, but I do want to continue using libdrm/mesa/xf86-driver-ati from git, along with a new kernel that utilises it. Have any of you had experience with setting it up on debian testing?
_moep_: hoo-hah: it works ;)
_moep_: but i've some problems with the radeon driver and my card
osiris: I've pushed updated blit branch. it works, except it doesn't render anything :P
osiris: airlied, nha: care to take a look?
Ghworg: hoo-hah: Yeah, it can be done fairly easily. I make custom packages so I can easily swap out different versions, that takes a little extra work but still is fairly simple
zhasha: MostAwesomeDude: http://img43.imageshack.us/img43/7738/rsim.png <- I was bored
osiris: zhasha: what's the status of your CS dumper?
zhasha: osiris: what exactly do you mean?
zhasha: because rsim displays a cs stored in a file perfectly well
osiris: zhasha: can you show me sample output?
zhasha: osiris: what a CS file is supposed to look like or how rsim displays it?
dileX: zhasha: where to get rsim (radeon simulator)?
zhasha: I put it in cgit just an hour or so ago
osiris: zhasha: I want to see rsim output. I suppose CS file is just a list of uint32_t.
Kaapa: hello everyone. For my card (M76) I noticed that hibernate works fine. Suspend has some problems, though; it wakes up nicely but hangs when I switch to console
Kaapa: any tips?
zhasha: yep
zhasha: osiris: http://imagebin.org/70930
zhasha: that's just a CS with 1 packet0
zhasha: and also, the dest is not an actual register or the name would show. It's still missing a lot of the planned features
osiris: zhasha: aah, I was hoping it would pretty print the register value
zhasha: register value?
zhasha: you mean show all the bitfields etc? it's underway
osiris: yes
zhasha: actually nearly done
osiris: that's great
zhasha: I was just struggling with auto* to git it all to compile and work :P
zhasha: you shouldn't need vala to compile it by the way
osiris: agd5f: ping
eosie: I wish I could compile mesa under Eclipse
zhasha: http://imagebin.org/70933 <- looks just a little bit cool
osiris: MostAwesomeDude: do you know what the US_OUT_FMT_n regs are for?
osiris: MostAwesomeDude: there are already COLORFORMAT fields in RB3D_COLORPITCHn regs
glisse: osiris: i think the output of us stage doesn't directly go into the cb
glisse: thus you can choose which format you output to stage btw us & cb
osiris: glisse: hmm, but what stage is it?
glisse: dithering, blur, rop
glisse: also likely multisampling stuff
glisse: s/blur/blend
glisse: best to test this theory is to set us_out_fmt to somethings different that rb3d fmt
osiris: glisse: I still can't see a possible usecase for this
glisse: osiris: engineer at ATI surely did see a usecase :)
osiris: glisse: anyway, I'm trying to get accelerated blit to work but somehow nothing gets rendered
glisse: osiris: using 2d engine ?
osiris: glisse: nope, 3d
osiris: http://cgit.freedesktop.org/~osiris/mesa/log/
glisse: millions things can go wrong
osiris: glisse: I know :(
osiris: glisse: any ideas?
glisse: osiris: what i would do is a gl program which render from one texture to buffer
glisse: ie does the blit
glisse: than capture its cs
glisse: and then start comparing what's different
Kaapa: I've tried every option in pm-suspend but still can't get suspend to work... always crashes when I switch to console after a successful resume :S
Araneidae: What is the status of 3D support for the radeon drivers these days? (Just upgraded to Ubuntu 9.10, might try my old ATI card again...)
glisse: Kaapa: KMS ?
twnqx: Kaapa: module or static?
Kaapa: glisse: sorry?
Kaapa: module, I think
glisse: kernel modesetting but as you don't know it likely means no
spreeuw: Araneidae: r200 platinum r300 gold r5/6/7/00 silver/wip
glisse: we won't spend much resources in fixing suspend/resume for non kms case
dileX: osiris: your mesa-tree: "warning: remote HEAD refers to nonexistent ref, unable to checkout."
Araneidae: Hmm. Sounds promising. My card is a Radeon X1650, guess that's a r200? Or is that a 300?
Kaapa: glisse: and where/how can I know so that it's worth the try?
glisse: Araneidae: r500
Araneidae: Where do I go for bug reports and detailed status reports?
Araneidae: ah
glisse: Kaapa: test fedora live cd, nightly build
glisse: google will give you direct likn
Araneidae: I guess "silver" == a bit dodgy
glisse: fedora 12
spreeuw: yes
spreeuw: based on earlier progress it will tkae a year or two to get ready for endusers
spreeuw: but then again endusers are never satisfied
Araneidae: glisse, actually google for radeon is singularly unhelpful -- unless you mean http://www.x.org/wiki/radeon ?
spreeuw: and by that time they will start to reinvent xorg all over again
Kaapa: glisse: http://alt.fedoraproject.org/pub/alt/nightly-composes/ ?
spreeuw: and there nothing to work with again
stikonas: aren't r500 supported exactly as r300
glisse: Kaapa: http://fedoraproject.org/wiki/Releases/Rawhide
glisse: Kaapa: if you want to test if suspend/resume works better with KMS
glisse: try fedora 12 livecd
twnqx: hm
glisse: especialy the nightly build one
twnqx: failed horribly for me in-kernel
twnqx: freeze on resume
Kaapa: glisse: so all I need to do is test it?
Kaapa: or should I do some kind of debugging?
adamk_: stikonas, Yes, they are.
osiris: dileX: you need to pass branch name during clone
glisse: Kaapa: it should works with kms, iirc only non working one are thinkpad t60 X1400 & some X850/X800 GPU
glisse: Kaapa: what i mean is that we won't spend much time fixing non kms case
glisse: so for non kms you are on your own
glisse: but if it doesn't work with kms we will definitly look into it
osiris: dileX: or do: git checkout -b origin/r300-blit r300-blit
Kaapa: glisse: fair enough!
osiris: dileX: after the git-clone warning
Kaapa: if it does work I'll come back and ask how I can set that into my existing system
glisse: and today easiest thing to test kms is to use fedora 12 livecd which has everythings
dileX: osiris: 'git pull git://anongit.freedesktop.org/~osiris/mesa r300-blit' was OK, too
Araneidae: Is there a process for reporting bugs against the radeon driver?
Araneidae: So far I've found: http://www.x.org/wiki/radeon , http://xorg.freedesktop.org/wiki/RadeonFeature and http://xorg.freedesktop.org/wiki/RadeonProgram
osiris: glisse: are there any requirements on order in which we setup different units?
glisse: osiris: only requirement is to setup them before the draw packet
glisse: order doesn't matter
osiris: glisse: ok
osiris: glisse: what would be your best guess here? I will check it before I start comparing it with generated cs
glisse: osiris: nothings happen ? no lockup ? corruption on the dst ?
glisse: osiris: my first guilty things would be scissor
osiris: glisse: yeah, completely nothing. no lockup, I've fixed all cs checker complaints, the destination buffer doesn't change
glisse: than transformation of vertex+viewport
glisse: and last fragment shader
osiris: glisse: fragment shader is generated so it should be ok
glisse: osiris: you never know
osiris: and so is vertex shader
oliver3_: Hey, I'm trying to get the experimental 3D working for a HD 4670 (r7xx), glxgears bails with this: http://pastebin.ca/1662040
Araneidae: So no bug report database (at least as far as anybody here knows), no mechanism for user feedback, if I use the radeon driver I'm on the bleeding edge?
dileX: Araneidae: bugs.fd.o
dileX: See also dri-devel and mesa3d-dev ML
Araneidae: bugs.fd.o ? Sorry, lost me there.
glisse: bugs.freedestkop.org
glisse: then select mesa if it's opengl bug
glisse: or xf86-video-ati if it's Xorg bug
dileX: Araneidae: -> "Documentation and Support" lists the links for bugs.fd.o and the MLs :-)
oliver3_: My machine locked up, had to do a hard reset.. hence my absence
Araneidae: dileX, cheers. I'd missed that :(
oliver3_: adamk, hey
oliver3_: 2.6.31-14-generic
oliver3_: Answer to your question in #radeonhd
oliver3_: adamk, preparing the pastes now
dileX: Araneidae: no problem. and sorry for using "common" abbreviations (used here in the channel)
osiris: glisse: SC_SCISSOR0/1 value should be based on the values from glScissor, right?
oliver3_: adamk, http://pastebin.ca/1662047 <- xorg log
glisse: osiris: no it should be the size of the destination buffer
oliver3_: adamk, http://pastebin.ca/1662049 <- LIBGL_DEBUG=verbose glxgears
osiris: glisse: yeah, but I'm asking how is it in general
glisse: yeah in general its opengl scissor
osiris: glisse: what about SC_CLIP_x_y?
oliver3_: pokes adamk_
osiris: glisse: and how do I disable opengl scissors?
oliver3_: adamk_, ? :(
adamk_: oliver3_, I was under the impression that r6xx/r7xx was only available on the 2.6.32 RC kernels (or drm-next).
glisse: osiris: in mesa driver ?
adamk_: I could be wrong about that, but that's my understanding.
osiris: glisse: what reg need to be written with what value
oliver3_: adamk_, I have no idea, that wasn't mentioned in the wiki. It compiled file so maybe not? I really don't have the time nor the motivation to build a custom kernel right now. *sigh*
oliver3_: Looks like I might as well start reinstalling Windows :'(
adamk_: Well, that's why I recommended the wiki in the topic on this channel :-)
oliver3_: Oh /that/ wiki
dileX: mika: radeon-kms? with the right kernel, libdrm-from-git (build with --enable-radeon-experimental-api), mesa-from-git and xf86-video-ati-from-git. for r100-r500: 2.6.31 (activated drm-radeon-kms/staging, rebuild libdrm-from-sid, mesa-7.6, ddx-from-git. in any case, xorg-server >=1.6.2
oliver3_: visits
dileX: oops, wrong channel (its radeon-related, anyway)
glisse: osiris: scissor0/scissor1
glisse: got to go
glisse: bbl
adamk_: oliver3_, I understand your desire not to compile a custom kernel, but bear in mind that this *is* experimental 3D acceleration, after all. Even if you get it working, it might still be buggy or unable to run that application you talked about earlier.
oliver3_: Yeah I understand.. I'm just so desperate to /not/ run Windows, I guess I need to stop being unrealistic
oliver3_: :(
oliver3_: I guess I'll just ask in #ati about solving the problems with fglrx instead...
oliver3_: adamk, thanks a lot for your help anyway, it's much appreciated :)
adamk_: Good luck. There tend to be one or two people who answer questions on #ati, and I'm one of them :-)
dileX: osiris: 840 frames 9.2 seconds 91.3 fps 4.0/10.9/66.0/5.4 ms (with your tree and anholt's oa-benchmark)
dileX: [12:02:58] $ openarena +exec anholt 2>&1 | egrep -e '[0-9]+ frames'
dileX: [12:02:58] 840 frames 9.0 seconds 93.7 fps 4.0/10.7/50.0/4.7 ms
dileX: osiris: that was with master GIT 3hrs ago (also DRI2)
osiris: dileX: the blit branch is still WIP (and doesn't work - oa probably doesn't use it so you're lucky)
dileX: osiris: which software or game is using it? any proposals?
osiris: dileX: half life
osiris: portal
osiris: tomb raider legend
osiris: and many more
varanus: hi guys, i have installed the other day the radeon driver for the hd4850 i have and there is much of improvment than the fglrx i used. I would like to ask if there is a way to check the temperature and the fan speed of the card cause it seems to work at max making a lot of noise. ty in advance
dileX: osiris: any free game?
osiris: hmm
dileX: (sorry I am not a gambler)
dileX: enemy territory?
osiris: dileX: sauerbraten when some effects are enabled
osiris: dileX: don't know about et
dileX: lists also a benchmark
Zajec: varanus: i don't think we have temperature/fan info gathering anywhere
Zajec: varanus: the problem is probably your driver don't do downclocking
Zajec: varanus: if you use DDX (xf86-video-ati) you can try ForceLowPowerMode (man radeon)
Zajec: varanus: for KMS you will have to wait, still
varanus: Zajec: i konw that KMS is not supported yet but the only thing i would like to do is to have it work at low range cause it's making a lot of noise and i don't need it to atm. I've not thought of man radxeon, ty very much
varanus: Zajec: sorry if i'm disturbing but i'd like to ask some regarding the man radeon. there are 2 options "clickgating" and "ForceLowPowerMode" that are both off by default. should i enable both of them or it is sufficient just one, and in case only one which. Finaly, if i change the default, will it be enabled always even after reboot?
varanus: Clockgating* sorry
chithead: power management does not work with kms yet. the clockgating and forcelowpowermode options will have no effect in kms
biotube: varanus: with normal modesetting, only ForceLowPowerMode is enabled with r600 and r700 chips
dileX: can someone tell me where to get "radar" demo for enemy-territory? web-link is dead.
varanus: ty guys for you advices and information cause i'm a bit new to the radeon driver
dileX: (see )
zhasha: I think I'm going insane: http://imagebin.org/70942
eosie: me too
zhasha: I call it Linux McWindows
Zajec: varanus: set, i was afk
Zajec: varanus: as you were told, ClockGating doesn't change anything on r6xx+ AFAIR, so just use ForceLowPowerMode
Zajec: varanus: xorg.conf is read every time you start X... so if you just change xorg.conf and restart X (like CTRL+ALT+BACKSPACE double click, or "init 3 && init 5") you should see effect
varanus: Zajec: sorry i didn't notice you were afk and thx again
varanus: does ForceLowPowerMode lower fan speed?
DeX77: hi
DeX77: still: Bogus firmware length..
DeX77: thats my current kernel config http://pastebin.ca/1662189 can anyone using Linus' Tree Head with RV770 perhaps upload hers?
ShapeShifter499: I have a powerbook g4 with ubuntu 9.10 on it, I want to have s-video output but can't seem to get it working....what do I need to do to get s-video working?
Zajec: varanus: fan speed is controlled by GPU
Zajec: varanus: ForceLowPowerMode should slow down engine (and memory?) and in result GPU should slow down fan due to lower temperature
Zajec: varanus: did that option change anything to you?
Zajec: varanus: put your /var/log/Xorg.0.log on pastebin
varanus: Zajec: since i installed the radeon driver i had no xorg.conf, so i entered safe mode, at console to create one with sudo X -configure, but the only xorg.conf i have is named xorg.conf.original-0, could this corrispond to the classic xorg.conf?
varanus: just a sec
varanus: Zajec: here is my log http://paste.ubuntu.com/313386/
varanus: Zajec: the fact is that i don't have an xorg.conf under my /etc/X11. the only files i have are an xorg.conf.failsafe and the xorg.conf.original-0
osiris_: otaylor: what app are you seeing the "Assertion `bo->space_accounted' failed." assertion in?
mjt: hm. I'm suffering from https://bugs.freedesktop.org/show_bug.cgi?id=24430 here, but in 2.6.32-rc6 the mentioned patch is reverted. I had some patch applied in 2.6.32-rc4 to get it to work correctly, now it's wrong again... In particular, it is here again: https://bugs.freedesktop.org/attachment.cgi?id=30523 . Any pointers?
mjt: agd5f: I think it was you who pointed me to the right fix when at 2.6.32-rc4...
varanus: Zajec: the xorg.conf.original-0 is here http://paste.ubuntu.com/313392/ and the failsafe here http://paste.ubuntu.com/313394/
mjt: oh... it's not in linus try yet....
ShapeShifter499: anyone know what I could do? I need s-video out to work for a skype call me and my mom may be placing to relatives, please help
chithead: it might help if you say what you tried and how it failed, etc.
mjt: btw, skype isn't the best debugging tool out there
ShapeShifter499: I have tried to reconfigure the xorg config files and tried to google search for info, which both have come up dead.. so I do not know what else I could do
ShapeShifter499: the irc channels are my last bet
chithead: "reconfigure the xorg config files" is still not a very precise description
chithead: how about pastebin of logs, config files and such, along with a description what happens when toggling the tv related options described in the radeon manpage
ShapeShifter499: well the problem is I'm still a BIG noob on ubuntu, xorg, and how to use them so I'm at a loss on what to do...
ShapeShifter499: I've googled, but most info is for x86 systems not powerpc systems
ShapeShifter499: ... now ppl are going to ignore me now??
cxo: Do you know how to pastebin log files?
chithead: it is difficult to help you if you don't give any details
ShapeShifter499: yes...but where are the log files?
cxo: /var/log
ShapeShifter499: ohh...
ShapeShifter499: one sec then
ShapeShifter499: here is what my xorg.conf looks like http://paste.ubuntu.com/313434/
chithead: ShapeShifter499: did you read the radeon manpage?
ShapeShifter499: I have reseted the config so it look like it did before I started messing with it, and you mean the ubuntu radeon page
ShapeShifter499: ?
chithead: the document that comes when you type "man radeon" (without the quotes) in a terminal window
ShapeShifter499: ohh
ShapeShifter499: no
ShapeShifter499: looking now
ShapeShifter499: did you look at my config? do I need to change it so it uses the radeon driver or is the way it is now good?
cxo: if you dont have fglrx installed, it should be using radeon by default
ShapeShifter499: ok
chithead: Xorg.0.log will also tell you which driver is in use
ShapeShifter499: I was looking around the man page and found this http://paste.ubuntu.com/313440/ do I need this also?
agd5f: MostAwesomeDude: there aren't any data types greater than 12 in the psc regs
MostAwesomeDude: agd5f: Alright, that's what I suspected.
agd5f: mjt: try drm-next if you are using linus tree
MostAwesomeDude: Also how did you know that I just got on? That's just creepy.
agd5f: MostAwesomeDude: I didn't I was just reading the backlog
ShapeShifter499: I was also looking at the logs and in xorg log 0 it has these two lines in it http://paste.ubuntu.com/313443/ suggesting its loading XFree86 code, I thought that was only for x86 systems not powerpc
jcristau: you thought wrong
agd5f: osiris_: the US_OUT_FMT* regs define the format and sizzles of the data leaving the pixel shader
MostAwesomeDude: osiris_: The US_OUT_FMT regs are for swizzling the output from the fragment shader for e.g. BGRA buffers.
DeX77: agd5f: I know you have a rv770 working with lates linus tree.. can you perhaps give me the config?
DeX77: +t
agd5f: Zajec: the fan is actually controlled via an external thermal chip controlled via i2c in most cases. r6xx+ got a built in controller, but I'm not sure how many oems used it intially
agd5f: DeX77: have you tried booting into single user mode with nomodeset and loading the radeon module manually?
otaylor: osiris_: was a patch to mutter I was working on
DeX77: agd5f: nomodeset is an option for kernel? drm module?
ShapeShifter499: I'll be back
ShapeShifter499: testing new xorg settings
DeX77: Sarah Connor? ;)
agd5f: DeX77: might be fedora specific. radeon.modeset=0 is equivalent
DeX77: agd5f: thx, will give that a try
osiris_: otaylor: try this patch http://pastebin.ca/1662272
DeX77: <-- rebooting with modeset=0 ;)
otaylor: osiris_: not in that mode right now, but I'll try to test it tomorrow. It looks plausible that it will work\, assuming that it's OK to have r300ValidateBuffers called multiple times for a single command buffer with different color buffers.
otaylor: (Not really sure what the point of having state emitted at the end of a command buffer to begin with is anyways...)
agd5f: may need some cache flushing added as well
Zajec: agd5f: so that external thermal chip has to be controlled? by driver or by something inside r6xx+ gpu?
Zajec: agd5f: what kind of control this thermal chip needs? shouldn't that be 100% automatic like "enable fan when temp > X; disable when temp < X"?
agd5f: Zajec: by the driver. the actual thermal chips can do a lot of it by themselves
agd5f: eg, lots of the thermal chips have profiles to handle that
cxo: I flashed a new profile onto the card
Zajec: agd5f: anything except setting TEMP_VALUE to enable(disable) fan?
agd5f: cxo: not a good idea
agd5f: bbl. gotta run
cxo: Well my card was originally a Visiontek, the Asus cards had a better clock speed and profile, so i just flashed it with an Asus bios
cxo: seems to work most of the time
agd5f: cxo: if your card has a different thermal chip, the bios won't be able to control it
cxo: My card is one of those really early reference boards, its identical to the Asus physically, well at least it looks identical
agd5f: also the gpio setup and connector tables are defined by the bios
agd5f: any way really gotta go now
cxo: bye
DeX77: re
DeX77: agd5f: ok, now it realy gets weird: if I boot into runlevel 3 radeon gets loaded without error.. same kernel
mjt: agd5f: re that r600 screen corruption - i just applied the same patch as i used for -rc4 :) I thought I've lost it but found it in irc logs.. Thanks!
mjt: (it should definitely go to linus for 2.6.32, i think)
cxo: agd5f, said he stepped out just a minute ago
cxo: Isnt there a specific checkin window if you want to make 2.6.32?
stikonas: radeon KMS is still in stageing
stikonas: and checkin window does not apply for experimental stuff
stikonas: and it may also count as bugfix
cxo: EXPERIMENTAL
cxo: needs to go and look pretty for some guests now... later
rlj: i have a problem with suspend/resume on recent arch linux system with dri2 and kms. if i suspend for only a split second, the display works fine, but if i wait a few seconds before i resuming, the contents of the screen are garbled (depending on how long i suspend, it appears as snowy arteficts, gradually ending up as stripes)
rlj: to me, this looks a lot like video memory simply fading from being completely unpowered when the computer is suspended, and X failing to restore the video memory from some kind of backing store on resume
DeX77: if I get "drmDynamicMajor: failed to open /proc/ati/major" then..
rlj: is there a way to force the driver to assume video memory is outdated or plain wrong and make it refresh from either some ram backing store or from the x applications themselves?
rlj: is the video ram even supposed to be powered off on suspend? can the hardware be programmed to keep it powered?
MostAwesomeDude: The driver evicts all BOs from VRAM on suspend.
MostAwesomeDude: Which chipset is this?
rlj: (btw, simply logging out from the garbled screen Xorg, thus forcing X to restart works fine and results in a correctly working X. also, the VT:s seem unaffected by the problem, i haven't noticed any artifacts there)
MostAwesomeDude: Also, does the display slowly un-corrupt as you keep using it?
rlj: rs480 i believe. lspci says: 01:05.0 VGA compatible controller: ATI Technologies Inc Radeon XPRESS 200M 5955 (PCIE)
DeX77: rlj: do you suspend or hibernate=
rlj: DeX77: suspend to ram
DeX77: rlj: then I guess you should check your system memorry
DeX77: with memtest86+ or thelike
rlj: MostAwesomeDude: well, hovering some icons makes them appear correctly (the hovered version), making me believe they are simply lazily loaded
MostAwesomeDude: rlj: Well, which window manager?
MostAwesomeDude: bets kwin4
rlj: MostAwesomeDude: sure thing
rlj: MostAwesomeDude: sucky wm?
rlj: i could install metacity or something to see if it changes things
MostAwesomeDude: rlj: It's a bit buggy. Are you using opengl effects, or xrender?
rlj: think it appears even with the desktop effects disabled
MostAwesomeDude: Now *that's* surprising.
rlj: or at least "suspended". haven't tried with an xorg.conf with RENDER or COMPOSITE disabled manually
MostAwesomeDude: But yeah, check with another WM first, as kwin4 is kind of nvidia-oriented.
rlj: at least to me, it clearly looks like videoram being powered off. which i'm not even sure if it should on suspend...
rlj: sounds like a good ideas
rlj: i'll install metacity
rlj: DeX77: i'm using only sideport btw
mokoloko: Are there still lot's of gfx glitches with radeon driver under kde4? just tried latest fedora 12 snapshot but their gfx related packages are like month old...
rlj: DeX77: my system ram is fine (tested recently with memtest86+) and is clearly powered during suspend as i can resume correctly without crashes :)
rlj: DeX77: but all this makes me think the gfx chip's dedicated sideport ram gets powered off
rlj: since a really quick suspend (less than half a second) gives no visible artifacts at all
MostAwesomeDude: rlj: If KMS is working properly, all the BOs should be getting evicted to system memory, where they won't get lost.
rlj: MostAwesomeDude: kms does work i think. the VT's all have the same native LCD resolution as in X (1280x800) and switching is instantaneout
rlj: *instantaneous. plus xorg.0.log says kms works :)
rlj: ok, i've got metacity running now. gonna do some tests. be back in a few minutes with results
XenoPL: Hi all, I've got a problem similar to this https://bugzilla.redhat.com/show_bug.cgi?id=522271
XenoPL: Sometimes KMS forgets to enable display
MostAwesomeDude: XenoPL: You're on Rawhide?
XenoPL: yup, all latest bits from koji
XenoPL: The difference is it happends only on bot
XenoPL: *boot
MostAwesomeDude: Including kernel 122?
XenoPL: Yup, exactly
MostAwesomeDude: Oooh. That's not good.
XenoPL: and -127
MostAwesomeDude: Hm. Well, I'm not too well-versed in it. I was going to say that 122 fixed all my r600 KMS problems...
XenoPL: SOmetimes mks wont enable display, i have to power off and power on the LCD
XenoPL: and it's happend on RV350
XenoPL: I've been looking at dmesg comparing to F11, it differs a bit but nothing interesting in thwere
XenoPL: Is any good way to find out what's up?
XenoPL: I have no such problem with F11 (KMS enabled in both installs)
MostAwesomeDude: Ah, *that's* important.
MostAwesomeDude: Are you CC'd or otherwise attached to the bug you linked?
XenoPL: I guess I should add a comment here, what else would be usefull?
MostAwesomeDude: I'm honestly not sure. Kernel version, probably.
rlj: with an xorg.conf with RANDER, COMPOSITE and DRI disabled, i still get the same behavior
rlj: even at kdm, if i suspend with pm-suspend from the VT and wait a few secs before resuming, KDM has garbled graphics
XenoPL: Ok, sure. I can't remember if it worked right from F12 beta (-56), but probably not, any dmesg?
MostAwesomeDude: rlj: Hm. Interesting.
XenoPL: This bug is not so critical for me, but I guess lappy users won't be hapy when they hit that little bastard
MostAwesomeDude: Well, I'm not the most useful guy to talk to; I write 3D code, not KMS code.
XenoPL: I know, i'll add myself to the bug report and wait till it'll get fixed
dmb: should post a comment
varanus: when in the var/log/Xorg.0.log the option it with the ** in the beggining it means that it's enabled? like this (**) RADEON(0): Option "ForceLowPowerMode" "true"
varanus: it=is with
XenoPL: ok thanks for the tips MAD, bbl
zxczxvz: does ati have something like nhancer
zxczxvz: coming its way
zxczxvz: i cant buy a ati product because they have shit drivers
zxczxvz: and lack of support
Kaapa: glisse: Just tested with latest fedora nightly. Suspension works, and even glxgears work!
zxczxvz: kaa kaa
zxczxvz: hello ka ka
rlj: MostAwesomeDude: would you know where to start looking for BO eviction bugs. since a quick suspend/resume cycle only results in "snowy" artifacts, to me it seems like it's at least never restored, but possibly stored away. cause failing to store but succeding to restore should result in pure garbage. the other way around though would give my symptoms
zxczxvz: yea
zxczxvz: i got lots of bugs in my apartment
zxczxvz: i live in district 9
portn0k: ive got a radeon 7000 card on a windows machine, im using the DVI output for the first monitor, and VGA for the second. however: my first monitor randomly adjusts resolution, and only uses 75% of the monitor, rendering the other 25% blank/black. thats as best as i can explain my problem, i was wondering if anyone has had an experience like this?
zxczxvz: yea i did
zxczxvz: i threw the ati in the garbage
zxczxvz: slam dunked it
portn0k: hah, i would but im a poor man and its my best card ;/ that machines onboard is foobar anyways
portn0k: i can disable the second monitor and then re-enable it, and the full area of the monitor is used again, but it only lasts for 5-10 minutes and then reverts back to gaymode
osiris: are there any known bugs preventing running 32bit apps on 64bit system?
osiris: I'm getting 'ioctl32(glxgears:17922): Unknown cmd fd(4) cmd(80046402){t:'d';sz:4} arg(ff942dcc) on /dev/dri/card0' when trying to run 32bit glxgears
zxczxvz: good
zxczxvz: that means your ati drivers are working properly
zxczxvz: just as they were programmed
adamk_: zxczxvz, If you have nothing positive to contribute, then STFU.
zxczxvz: take oyur own advice
zxczxvz: cause you just contradicted yourself so STFU FIRST
zxczxvz: and twice
biotube: zxczxvz: just because "support" is listed first in the topic doesn't mean it's the primary purpose of the channel
zxczxvz: wtf
zxczxvz: are yapping about
adamk_: zxczxvz, Actually getting you to shut up would be a positive contribution, so I did not contradict myself.
zxczxvz: =|
zxczxvz: yea you did
zxczxvz: so your admitting to being a lying fuckin whore bitch
zxczxvz: sounds like more contradiction
zxczxvz: so you would lie to get something
Kaapa: can someone kick this idiot?
zxczxvz: like a whore bag does
zxczxvz: =]
zxczxvz: suck a dick before
zxczxvz: you ask for ops something
zxczxvz: like you do in life with your keebler buddys
Kaapa: Irssi: zxczxvz added to shitlist
Kaapa: problem solved
zxczxvz: add yourself
zxczxvz: and there wouldnt been a problem
zxczxvz: to begin with
zxczxvz: fuckin cunt
zxczxvz: i like that emo mentality you got there adam
zxczxvz: =]
zxczxvz: mtv has done fine job of educatin the youth of america
zxczxvz: see i'm a primary example
zxczxvz: ~_^
zxczxvz: as well
zxczxvz: as you are
zxczxvz: anyways this got lame
MostAwesomeDude: Goddamn trolls. Don't feed them, please.
zhasha: MostAwesomeDude: so anyway, his ISP is Optimum Online, his IP is 24.185.135.118 and we know when, and he lives in or near new york
zhasha: do we report it to his ISP's abuse dept. just for fun?
twnqx: make som dmca claims while you're at it
cxo: "lying fuckin whore bitch" haha
twnqx: and radeon >>>> fglrx
twnqx: it gets a picture on my screen. that already beats fglrx :P
cxo: yeah all i got with fglrx was a blank screen
adamk_: That's alright, he was on #ati earlier.
airlied: osiris: isn't there a mesa meta blit oepration we can steal?
osiris: airlied: there's it but it's using glReadPixels
osiris: airlied: I've already got it working (at least for progs/tests/subtexrate)
osiris: airlied: demos/gearbox doesn't work unfortunately
osiris: airlied: also I'm lacking of test apps
airlied: osiris: we could optimise glReadPixels instead ;-)
osiris: airlied: how? blit is the only solution I see
airlied: suggests using the 2D engine and runs ;-)
MostAwesomeDude: I agree.
airlied: well you don't need blit really for ReadPixels
osiris: airlied: how's so?
airlied: since overlappimg type stuff isn't likely
osiris: airlied: yeah, but you have to get the data out of VRAM somehow and currently it's very slow. with blit it's possible that the data won't be copied to RAM and back to VRAM at all
airlied: using the 3D engine to render to GTT from VRAM is fine except for the fact that like DFS it might break
osiris: osiris: the speed improvement is enourmous. current solution get's about 6MB/s, with accelerated blit it is ~380MB/s - that's over 60x faster
EruditeHermit: lol you're talking to yourself
airlied: osiris: yup definitely
osiris: EruditeHermit: it's a sign I should take a brake
EruditeHermit: haha
spreeuw: and after that a break
osiris: haha :P
spreeuw: did you have alcoholic beverages the past days?
airlied: osiris: btw RS needs one texture + one color setup
spreeuw: those always wreck my language memory buffers
spreeuw: so much that its starting to bugg me
osiris: spreeuw: nope
osiris: airlied: it seems to work fine with current setup
airlied: cool, when it breaks one some shit r300 start looking there ;)
airlied: just feels we should be able to do a lot of this at a higher level
airlied: using a meta op
airlied: to avoid hand-setting up all th state
airlied: like its just a texture operation
cxo: hey airlied , whats your mailing address? I want to send you some cheese cake
airlied: it might not make it one piece ;-)
osiris: airlied: yeah, we should but I'm not sure it will be possible without some major changes.
airlied: osiris: drivers/common/meta.c
Pallokala: now that I have KMS (2.6.32-rcX) + DRI2 (git mesa + ati, libdrm-2.4.15) which part/parts are responsible for a random corruptions which I'm getting f.ex when watching images in f-spot?
airlied: seems to do blit framebuffer with texture ops
osiris: airlied: only if there's a texture attached to current renderbuffer
osiris: airlied: otherwise it falls back to glReadPixels
airlied: ah yes suck
airlied: wonders why we don't fake a texture there if we can
airlied: might be worth asking Brian
airlied: I just hate any specific code because it makes changing the drvier a real pain
osiris: airlied: you can't fake it, mesa doesn't know what's under the driver's renderbuffer
airlied: osiris: the driver does ;-)
airlied: osiris: but if it works I'm happy to see it merged, I can then give out until I figure it out myself ;-)
doors: hi radeon, this is quite a basic support question, so please tell me which manual to go read :)
doors: I'm not getting any EDID data from my laptop screen
doors: so the radeon module is using a bizarre resolution
doors: how do I force the radeon module for drm to use a resolution or sync speed?
doors: so for the vga driver it was a vga= line as a kernel param
cxo: someone said the answer a few days ago
cxo: me forgot
doors: confused the hell out of me for the first hour or so
doors: "why is my screen fuzzy"
airlied: laptop panels are generally fixed internally
airlied: we normally just set the mode the BIOS has stored in it
airlied: setting the command line mode will just affet the scaled mode
airlied: we still use the panel native timings defined in the bios
doors: I see
doors: this could be interesting - bios settings only use the middle 1/4 of the screen
doors: I'll see what happens when I set x to the same settings
osiris: airlied: I've pushed the working version to private repo
airlied: osiris: btw that fix for bo accounting seems a bit suspect
airlied: osiris: I suspect its just another case of getting the state emission size wrong
osiris: airlied: no, it isn't
airlied: well it at least breaks that white-space ;-)
airlied: for the query finish emit
airlied: I'd really appreciate a better commit message
osiris: airlied: oopsie
osiris: airlied: sure
airlied: if you can sstate in the commit, what exactly it fixes and/or a piglit test ;-)
airlied: its just bo accounting is messy and I'd like to know whats wrong with it so I can rewrite it prorperly next time ;-)
airlied: office &
osiris: airlied: here is the situation when the bug appears:
osiris: you set new colorbuffer
osiris: then glTexImage which invokes following callchain radeonTexImage2D->radeon_teximage->radeon_firevertices->radeonFlush->radeonEmitState and there emit_cb_offset is called
osiris: airlied: and there we are trying to write bo reloc for colorbuffer which hasn't been validated yet
osiris: airlied: otaylor has described it even better in #24963
doors_: sorry radeon, I'm still stuck
spreeuw: doors_: with modelines
doors_: will google :)
spreeuw: and then you can swap them with xrandr
spreeuw: you can also see what xrandr made of it now
spreeuw: just enter xrandr in a console and see whats listed
spreeuw: actually you can set the modelines with xrandr too
spreeuw: define them
spreeuw: iirc
spreeuw: I used it once when there was something broken
doors_: cool, I'll give it a go!
Kaapa: glisse: I was so impressed with what I saw in the fedora nightly live cd that I'm trying to get kms on my laptop
EruditeHermit: airlied, are you still the blocker for F12?
airlied: EruditeHermit: not sure, I have to read the list again.
soreau: Am I the only one seeing this from the output of compiz? http://pastebin.com/m7f7572b2
EruditeHermit: airlied, got time to look at my screen not turning on after resume?
soreau: Everything seems to still run ok, but I just started seeing this within the last few days of updates
airlied: EruditeHermit: hit me with the bug link again and I'll review ;-)
EruditeHermit: airlied, https://bugs.freedesktop.org/show_bug.cgi?id=23103
EruditeHermit: thanks
airlied: EruditeHermit: what distro/kernel did you last test with?
EruditeHermit: airlied, Ubuntu karmic kernels 2.6.31 and 2.6.32-rc6 tested
EruditeHermit: also all 2.6.32 rc kernels in the series were tested at some point
airlied: EruditeHermit: okay I don't suppose you have enough bandwidth to downlaod and test an F12 live CD?
EruditeHermit: sure
airlied: or did we try that already -)
airlied: http://alt.fedoraproject.org/pub/alt/nightly-composes/desktop/
EruditeHermit: we didn't try that
airlied: has LiveCD images
EruditeHermit: but we can
airlied: it might be slow, not the best connected machine in the world
EruditeHermit: its fast enough
EruditeHermit: 700KB/s
EruditeHermit: can I install it to a USB stick?
airlied: there is a way http://fedoraproject.org/wiki/FedoraLiveCD/USBHowTo
airlied: not sure if Ubuntu has all the packages
chithead: if it uses isolinux-3.72 or later, you can use isohybrid and simply dd on the usb drive
dmb: btw, i just a warning, i would stay away from 2.6.32 if you use ext4
EruditeHermit: nah, using ext3
dmb: there are some known data loss issues
dmb: one of which, i experienced
twnqx: should be fixed in rc6 right?
chithead: supposedly the ext4 dataloss issues were fixed in 2.6.32-rc6
chithead: at least two of them
dmb: they arn't sure thats the issue though
dmb: even Linus was experiencing some issues
EruditeHermit: chithead, how do I check it it uses isolinux-3.72?
EruditeHermit: is it a package?
EruditeHermit: I don't see it
Ghworg: EruditeHermit: syslinux package I think
chithead: EruditeHermit: dd if=foo.iso count=1000 | strings | grep ISOLINUX
EruditeHermit: syslinux 2:3.63+dfsg-2ubuntu
chithead: important is the boot loader on the cd itself, not any package
EruditeHermit: so is that too old?
EruditeHermit: 3.63
chithead: 3.63 is too old, so you need to use unetbootin or other method
EruditeHermit: blah
EruditeHermit: unetbootin it is
chithead: EruditeHermit: it is a bit strange, I still have a fedora testday livecd image, which already comes with isolinux-3.74
EruditeHermit: chithead, the debian import merge is 6 months before the release of an Ubuntu
EruditeHermit: or 5 months
chithead: ah debian. that explains it
EruditeHermit: airlied, you just want me to test suspend/resume?
EruditeHermit: should I collect any data?
airlied: EruditeHermit: yup
airlied: just see if it does anything different
airlied: we can get logs etc if it breaks.
EruditeHermit: brb testing
krabador: i've trouble in ubuntu 9.10 and an ati x1600
krabador: can you help me?
chithead: nobody can help you until you say what the problem is
krabador: chithead, it's how i have the driver installed, but i can't use totally
chithead: ubuntu already comes with correctly installed driver
krabador: if i type glxinfo | grep direct , i've YES
krabador: chithead, i can't enable the composite manager
krabador: and i've an error message
chithead: "glxinfo | grep -i render"
krabador: it seems all right,
krabador: i've a pc with a R300 card, and a pc with this R500 card
krabador: i've installed ubuntu 9.10 on both
Zajec: krabador: if you don't plan to cooperate... maybe just don't ask?
Zajec: glxinfo | grep -i render
krabador: and the pc with r500 have the problem
krabador: Mesa DRI R500 (RV530) 20090101 AGP 4x x86/MMX+/3DNow!+/SSE TCL
krabador: direct rendering: Yes
logari81: krabador: if you have the composite of metacity enabled compiz won't start, check the output of the command "compiz"
krabador: logari81, i haven't the metacity composite enabled
soreau: krabador: Pastebin the complete output of 'compiz &' to pastebin.com
krabador: soreau, ok
krabador: http://pastebin.com/mdb9e4ee
soreau: That output looks fine
soreau: Which means there's nothing wrong with your drivers
soreau: If you need further help, ask in #compiz
krabador: ok, maybe it's a compiz problem....
chithead: no window decorator loaded maybe
soreau: he never said what the problem was
EruditeHermit: airlied, same problem with fedora livecd
airlied: EruditeHermit: good thanks for testing that
airlied: saves me blaming ubuntu distro stuff ;-)
EruditeHermit: yep
XenoPL: Hi again, is any KMS/2D guru around?
soreau: Xeno: Please just ask your real question
EruditeHermit: airlied, any ideas on what to do about it?
XenoPL: Ok, I've experienced a problem that was looking similar to this one https://bugzilla.redhat.com/show_bug.cgi?id=522271
XenoPL: Quite often kms forgets to turn on dusplay
XenoPL: *display
XenoPL: It's raden 9550, tested on kernels 2.6.31-23, 2.6.31.1-56, .5-122 and .5-127
airlied: EruditeHermit: just thinking, the fact that VGA doesn't come back either is wierd
XenoPL: It happends only at boot time, display turns off (no signal message) and never goes up
EruditeHermit: airlied, the VGA is hitting a different bug I think
EruditeHermit: airlied, that was a while back, let me try again with the VGA
airlied: EruditeHermit: also a dmesg after resume swould be nice
airlied: even with debugging disabled
EruditeHermit: ok
EruditeHermit: give me 10mins to get that all together
XenoPL: I have to cycle the power to get the picture
XenoPL: I found out it doesn't happend when grub timout is > 0, then display is initialised properly
XenoPL: Should I attach to this bug or file new one, it seems to be different case after all
XenoPL: Oh I forgot to add it's Fedora 12 but i have no such problem with F11
EruditeHermit: beb
EruditeHermit: brb
XenoPL: any sugestions on that matter? ^^
airlied: XenoPL: wierd
airlied: so it works if grub displays anything
XenoPL: yup, it's enough it shows the default option and the counter
XenoPL: if it tries to boot imidiately display wont come up
XenoPL: I've tried too find something in dmesg, but I couldnt see anything interesting, the hw initialisation messages seems to be right, the console res is set ok
XenoPL: On the other hand I boot F11 with 5 sec delay so maybe I should set it to 0 and see if it's working there, anyway I've never had a situation like this before F12
XenoPL: brb, I'll check if F11 is really unaffected
Kaapa: guys, I need some help setting up my config file for kms
Kaapa: I think it's looking good; drm and radeon load
Kaapa: and the console seems to be on the native resolution
Kaapa: but X won't start - I changed nothing on the config files, so that may explain it :p
airlied: you need an upgraded userspace usually
Xeno: Ok I'm back F11 unaffected
Kaapa: not sure what that means, where can I get it / check the one I have?
airlied: Kaapa: what are you running?
airlied: distro etc?
Kaapa: my distro is slackware
airlied: ah so you need to read the wiki page in the topic most likely
airlied: kms is a lot more than just kernel
Kaapa: but I just upgraded to .32-rc6, libdrm, mesa and xf86-video-ati from git
airlied: Xeno: 9550 with what monitor attached? VGa or DVI?
airlied: or laptop?
Kaapa: reading topic now
Xeno: VGA LCD Benq FP73G
airlied: Kaapa: okay if you build libdrm with the flags and -ati after that i
airlied: it shuold work
Kaapa: the erros I get when I startx is: Failed to load module dri2, dri and radeon
Kaapa: note: I may just have an invalid xorg.conf
airlied: oh wierd, it might just be too old an X server
Kaapa: maybe I could try to delete it and see if hal picks up the correct stuff
Kaapa: oh, indeed!
Kaapa: it starts
Kaapa: wrong definition.... but starts
Kaapa: dang, somethings not right
Kaapa: it's not detecting the correct resolutions
airlied: Xeno: bbl will try and reproduce it here on some machines
Xeno: ok no prob
Xeno: should I fill new bug on edhat bugzilla?
airlied: Xeno: yeah if you could, add dmesg and Xorg log
Kaapa: is there any xorg.conf sample I can look for somewhere?
Xeno: ok
Kaapa: wow, just found out, looking at the logs of the module looking, that I have hdmi in the laptop
EruditeHermit: airlied, results are same as in bug report with regards to external monitor and resuming from suspend
EruditeHermit: airlied, http://pastebin.com/m6e7e7800 for demsg
Kaapa: xorg.log is complaining about missing libdri and libdri2, who provides that?
Xeno: @airlied, if you'll read the log here's the link to the bug report: https://bugzilla.redhat.com/show_bug.cgi?id=533760
airlied: EruditeHermit: need to get it back further
airlied: that seems like AGP resume ios broken what kernel was on that livecd?
EruditeHermit: airlied, 2.6.32-rc6 on ubuntu
airlied: EruditeHermit: you might want to pull drm-next into that
EruditeHermit: airlied, has drm-next changed significantly since rc6?
EruditeHermit: I thought it was only released a few days ago
airlied: EruditeHermit: it wasn't pulled into rc6
airlied: since i'm still fixing bugs in it
EruditeHermit: ah
EruditeHermit: is it waiting for the 2.6.33?
EruditeHermit: or just skipped rc6
airlied: it depends on when I fix the bugs in it
airlied: its not automatic
EruditeHermit: ok
EruditeHermit: i'll try that
EruditeHermit: are there changes that will affect the display turning on?
Xeno: Hey airlied got my bug report?
airlied: EruditeHermit: no this is more to see if we can at least get VGA working
airlied: Xeno: can you ssh in okay btw?
airlied: I might get you to run a test
Xeno: I guess i could if I had another machine here, tell me what i have to do and i'll try when i get something to shh in
airlied: Xeno: git clone git://people.freedesktop.org/~airlied/radeontool
airlied: Xeno: cd radeontool; make
airlied: get that done may need some devel packages installed
airlied: then ./radeontool regs from a normal startup and one where its broken
airlied: and attach them to the log
Xeno: ok, i guess i could make it tommorrow afternoon, it's hard to get new hardware in sunday 2am :)
airlied: Xeno: heh no worries
airlied: I'll try and reproduce it here but I usually fail at thos attempts ;-)
EruditeHermit: airlied, is the drm-next in the nightly fedora builds?
airlied: EruditeHermit: it should have been, but I'd need to know what kernel was on that liveCD
airlied: it at least should have the AGP resume fixes
Dusque: hello, I have installed the latest radeon drivers on ubuntu 9.10, and I am wondering what are the best settings for this. I have the xorg-edgers ppa added.
Dusque: or how should the xorg.conf look?
EruditeHermit: airlied, it might be faster to just boot using that
EruditeHermit: airlied, i'll find out what kernel is on there too
airlied: EruditeHermit: boot it and s/r with it and get dmesg
airlied: if you can suspend resume from runlevel 3
airlied: without X running it might make it easier to get the bits of dmesg I want
airlied: Dusque: we generally don't require xorg.conf anymore
Dusque: airlied, Oh I see. So is there anything I can do the optimize it? or is it done?
airlied: Dusque: generally the only option that people might want to play with is on AGP cards to enable AccelDFS
Xeno: Murphy's law, things works in presence qualified staff :)
EruditeHermit: airlied, ok
EruditeHermit: will do
EruditeHermit: brb
Dusque: Well mine is a PCI-E, but.... So there is nothing I should ensure is set up with xorg.conf, I did use the proprietary drivers for a bit.
airlied: we don't generally hide go-faster-stripes in xorg.conf
Dusque: ok lol. Not quite what I meant, but thanks.
Dusque: What I mean, is since I did use the fglrx drivers, could they have changed the xorg.conf so that it might not be set up properly. But since you guys don't really use it anymore, I guess thats a no.
Dusque: Basically, I just want to ensure that I set up my system properly using the Radeon drivers for my HD4850.
airlied: you can just try removing the xorg.conf and letting it work automatically
airlied: it should pick up radeon then
Dusque: really? just delete it?
Dusque: or rename it
biotube: I'd rename it just in case
Dusque: http://dri.freedesktop.org/wiki/ATIRadeon So the Configuration part is basically moot on that page now? Funky.
biotube: xorg.conf isn't ignored, but the xserver will come up with good defaults most of the time
Dusque: OK, and by renaming it, it will create a basic default one then?
biotube: no, but one will be used
Dusque: Ok. I'm just trying to use this so I can somehow help out.
Dusque: The progress made so far is fanstastic.
EruditeHermit: airlied, hi
EruditeHermit: airlied, how in fedora do you start sshd?
EruditeHermit: I tried /etc/init.d/sshd start
EruditeHermit: and it seems to have worked
EruditeHermit: but I am unable to login from remote
airlied: EruditeHermit: is the network up?
airlied: ssh normally just works for me
EruditeHermit: yeah the network is up
ssieb: EruditeHermit: even simpler, just do "service sshd start"
ssieb: are you trying to login as root?
airlied: maybe services iptables stop
ssieb: actually, by default, root can login, so that doesn't matter
EruditeHermit: ah
EruditeHermit: that did
ssieb: referring to my comment, not airlied's
EruditeHermit: the iptables was screwing up
EruditeHermit: filtering out my ssh requests
ssieb: you turned on the firewall?
airlied: livecd
EruditeHermit: what is the livecd password?
ssieb: there isn't one
EruditeHermit: wow
EruditeHermit: Fedora passwd is really strict
ssieb: :-)
EruditeHermit: it keeps telling me my password is too short
EruditeHermit: too many characters the same
EruditeHermit: etc
ssieb: do it as root
EruditeHermit: ah
EruditeHermit: =p
ssieb: that reminds me, I need to test some things with the F12 beta on this other computer
EruditeHermit: airlied, if I go to runlevel 3, will it take my network down?
airlied: EruditeHermit: just vt switch then ;-)
EruditeHermit: so vt switch
EruditeHermit: type pm-suspend
EruditeHermit: and then resume
airlied: yup and then get dmesg
EruditeHermit: all should be fine?
airlied: hopefully network comes back ;-)
EruditeHermit: airlied, http://pastebin.com/m1784657a
EruditeHermit: kernel is 2.6.31.5-122.fc12.i686
airlied: so the VGA monitor is plugged in and covered in garbage?
EruditeHermit: yes
EruditeHermit: same as the screenshot I posted earlier in the bug
ssieb: runlevel 3 still has network
airlied: EruditeHermit: can you send me your video bios?
EruditeHermit: airlied, how would I do that?
airlied: cd /sys/bus/pci/devices/
airlied: echo 1 > rom ; cat rom > /tmp/romfile ; echo 0 > rom
airlied: and mail me romfile at airlied@gmail.com
EruditeHermit: can I do it while it is not behaving weirdly?
EruditeHermit: as in can I boot normally to my Ubuntu
airlied: doesn't matter when
EruditeHermit: ok
ssieb: yay! I see a bug got fixed on RS690!
EruditeHermit: I get permission denied when I try to do sudo echo 1 > rom
agd5f: EruditeHermit: need to be root
EruditeHermit: sudo won't cut it?
airlied: EruditeHermit: sudo doesn't work like that
ssieb: that's a tricky one :-) the file opened by > is before you are root
agd5f: EruditeHermit: sudo doesn't support redirects
ssieb: your local shell opens the redirect before running sudo
EruditeHermit: ah
ssieb: and another one fixed!!
EruditeHermit: airlied, emailed
EruditeHermit: airlied, can you tell me if it is indeed the correct rom
airlied: EruditeHermit: looks like it
ssieb: still faster with nomodeset though :-/
EruditeHermit: airlied, cool
airlied: EruditeHermit: can you boot to non-X with nomodeset?
airlied: EruditeHermit: not sure how you do that on Ubuntu
airlied: singleuser might be easiest
airlied: then run the radeontool regs all again
EruditeHermit: airlied, ok
EruditeHermit: airlied, is it radeontool regs
EruditeHermit: regs all complains
airlied: ah radeontool regs '*'
airlied: or whatever we got you to run before
EruditeHermit: radeontool regmatch '*' > working.regs
EruditeHermit: airlied, http://pastebin.com/m65874ff
EruditeHermit: is that what you wanted?
airlied: EruditeHermit: yup\
airlied: as long as X never ran before radeontool
airlied: EruditeHermit: have another couple of things, can you pastebin lspci -vv
EruditeHermit: sure
EruditeHermit: give me 5mins though
EruditeHermit: I had to take the screen off =p
EruditeHermit: airlied, http://pastebin.com/m77ae8efe
EruditeHermit: airlied, do you want me to attach all these to the bug report too?
airlied: EruditeHermit: not yet I've got a program for you to run next
airlied: EruditeHermit:
airlied: EruditeHermit: dl http://people.freedesktop.org/~airlied/scratch/vbetool-0.7-m10-c000.tar.bz2
airlied: untar it, run make (may need pciutils-dev or whatever its called)
airlied: then boot to single user (no X running ever)
airlied: then run ./vbetool post 2>m10.regs
airlied: and attach that m10.regs file
EruditeHermit: ok
EruditeHermit: airlied, http://pastebin.com/m5f00d0e
EruditeHermit: didn't compile properly
EruditeHermit: i have libpci-dev installed
airlied: edit the Makefile and -lz after the libpci.a
airlied: or just gcc -g -Wall -pedantic -std=gnu99 -g -O2 -o vbetool vbetool.o thunk.o x86-common.o x86emu/libx86emu.a -lpci -lz
EruditeHermit: ok I used that to finish the compile
EruditeHermit: but for future reference
EruditeHermit: which line of the makefile do I need to edit
EruditeHermit: with the -lz
EruditeHermit: there are many libpci.a references
airlied: ah I need to fix that makefile its just broken on !fedora
airlied: I don't think we'll need to run this command again anyways ;-)
EruditeHermit: ok
EruditeHermit: hrm
EruditeHermit: that vbetool post made my screen go blank
EruditeHermit: and its not coming back
airlied: ah thats not so good,
airlied: does typeing reboot reboot?
EruditeHermit: yes
airlied: so does your machine resume in non-kms mode?
EruditeHermit: yep
EruditeHermit: ah
EruditeHermit: perhaps the prompt was below the screen
EruditeHermit: let me try again
airlied: ah the file shuold be fine
airlied: if it reboots cleanly
EruditeHermit: when I typed reboot it scrolled the prompt up
EruditeHermit: yeah
airlied: cool attach the file then
EruditeHermit: to the bug report?
EruditeHermit: what if any of the other things should I attach as well?
airlied: just that
EruditeHermit: airlied, it won't let me attach the file as its 6.9MB
airlied: EruditeHermit: oops, just email it or put it up on a website if you have one
airlied: does it gzip well?
EruditeHermit: airlied, it becomes 6.4kb
EruditeHermit: =p
EruditeHermit: when I bz2 it
airlied: attach that then
EruditeHermit: done
EruditeHermit: please check it now to see if it is indeed what you wanted
EruditeHermit: so that I can rectify it if it is not
airlied: EruditeHermit: looks like what I wanted I think
airlied: will need to analyse it a bit more
EruditeHermit: ok
EruditeHermit: has any of this helped to pinpoint where the problem is?
EruditeHermit: or does it require a lot more analysis
airlied: now should have enough info to do some analysis ;)
airlied: how much I need to do is unknown
EruditeHermit: ok
EruditeHermit: I have to drive somewhere an hour away
EruditeHermit: so I will be away for a while
EruditeHermit: thanks for taking the time to go through all this with me and for looking at all the dumps
airlied: EruditeHermit: cool I'll try and figure it out today but it might take a couple more ;-)
EruditeHermit: bonus pts for figuring out https://bugs.freedesktop.org/show_bug.cgi?id=19129
EruditeHermit: though this is unrelated to the kms bug
EruditeHermit: it has been annoying me for a year
EruditeHermit: =p
EruditeHermit: but I realize that its probably my crappy cards fault
EruditeHermit: ok
EruditeHermit: bbl