bricky: eh how do I max out my fan
bricky: rovclock doesnt have that does it?
bricky: is anybody around by chance?
rah: can someone make a 6.13.1 release with the latest fixes?
KotH: makes a 6.13.1 release of his vhdl parser in perl with all the latest fixes
rah: KotH: that would be an appropriate response within the context of #kothsvhdlparser
Zaba: somehow, since several (two, I think) days, 3d stuff on r600 is messed up for me.
Zaba: or rather, opengl
Zaba: any program that uses it is especially noticeable messed up when exits, because it stays visible until something draws over it, and then the same effect applies to everything
Zaba: (even non-gl stuff)
Zaba: and the same failure to erase previous things happens in, say, reflections, in 3d programs, too
Zaba: it used to work fine..
Zaba: oh, and restarting X has no effect
KotH: rah: SCNR :-)
sannes: Using xorg-server and radeon from git I get an assertion trying to allocate a 0 sized pixmap in AllocatePixmap (trace says it is KMS init in the radeon driver) .. anyone experienced that?
rah: hrmph
rah: sannes: yes
Zaba: what should I do?
rah: Zaba: meditate
Zaba: rah, I've already done that, twice, just to prevent myself from going insane from the lack of quake 3. :P
rah: Zaba: file a bug
Zaba: I need to bleeding register.
jcristau: poor you.
ossman: glisse, I don't suppose you've had any time to look at opengl vsync issues?
sannes: rah: Know of a quickfix for it? :P
rah: sannes: nope
rah: sannes: let me know if you find one though
sannes: rah : hm, what kind of setup do you have? I got a rv770 myself, with two screens active, using KMS (kernel 2.6.34)
rah: sannes: rv770, two screens active, KMS in kernel 2.6.34
sannes: rah: sounds familiar :P I upgraded a lot of things at the same time, know what you upgraded that caused it to stop?
rah: sannes: I upgraded everything from X git
rah: yesterday
sannes: rah: same as me, hehe, but I did it 4 days ago, not running gentoo also? :P
rah: no, debian
Zaba: Okay, the problem I'm having starts happening after c29157bbf5b0dd26857675282ab094082fbaed0d of the radeon driver, that is 'r6xx/r7xx: remove unnecessary state emit'
Zaba: could it be that I need a newer kernel, or even xorg, with some newer stuff in it? Because I update mesa and libdrm from git, too.
Zaba: I can't really come up with any other theory, as I have no idea whatsoever how the stuff works.
xmikos: Hello, 10 days ago I have asked here if somebody can fix this bug (at least temporary solution - patch to revert bad change - is there): https://bugs.freedesktop.org/show_bug.cgi?id=28284
xmikos: agd5f told me that it is not a real fix and that Will is working on proper fix. But according to bug report, it is still not fixed.
xmikos: Please shouldn't that change be reverted if there is not proper fix for that long time? Celestia is still broken :-(
glisse: Zaba: did you updated mesa too ?
adamk: Has anyone tried doom3 with r300g? There's serious texture problems http://thorn.visualtech.com/doom3-shot1.jpg It looks much like doom3 on r300c when S3TC is enabled.
mcgreg: Zaba: hey, you were right, I get corruptions with windowed 3d applications too
xmikos: Anyone, please?
adamk: xmikos: According to the bug reporter, the attached patch does fix it. If it's not a real fix, fine, but at least it's a temporary one that you can use till the real fix is produced.
xmikos: adamk: Yes, I know, but I am saying that this fix should be IMHO merged into Git repository till the real fix is produced.
adamk: I guess that all depends on how long the developers in question think it will take for the real fix to be developed.
MichaelLong: I have a problem with my setup consiting of a nvidia 8800 gts using nouveau and a HD 4350 (rv710) using radeon/KMS. the radeon card is not working at all, the module is getting unloaded Xorg.log: http://pastebin.com/LR5r1iYU
xmikos: adamk: Like I said I asked about it here about 10 days ago and the bug themself is 22 days old. This is long time in my opinion... this is why I am advocating temporary fix to go into Git till the real fix is produced.
adamk: Heh.. 22 days is definitely not long :-)
adamk: But you can always make that suggestion in the bug report itself.
xmikos: adamk: If it is regression and there is simple fix (revert that bad change), it is long time IMHO. Linus also reverts commits is there are regression till the fix is released in Linux Git tree.
adamk: Perhaps there are downsides to the temporary patch that we are not aware of.
adamk: But, again, given that there's already a bug report open, I'd recommend following up in the bug report.
xmikos: Ok, I will post my opinion there...
MichaelLong: how do I define a certain connector for a device section?
MostAwesomeDude: MichaelLong: Easiest way is to just go use Xorg -configure. It will generate a multicard xorg.conf automatically for you.
MichaelLong: my problem is that xorg unloads the radeon module. I don't know why.
MostAwesomeDude: By default, X will only do one card.
MostAwesomeDude: And the first card it finds is the nV card.
MichaelLong: what is missing here in terms of configuration?
MostAwesomeDude: Oh, wait. You're already using a multicard xorg.conf apparently?
MichaelLong: two separat screens
MichaelLong: wait
MichaelLong: http://pastebin.com/WGJtAumB
MostAwesomeDude: Take the Screen options out of the Device sections.
MostAwesomeDude: Additionally, you probably want something like this in ServerLayout:
MostAwesomeDude: Screen 1 "Screen1" RightOf "Screen0"
MostAwesomeDude: If you overlap the screens like you're doing, things won't work well.
MichaelLong: ok I'll give it a try
MostAwesomeDude: goes back to bed
Jonimus: enjoy your sleep MostAwesomeDude
MichaelLong: no effect radeon is getting unloaded. I guess the problem is on a another level.
Zaba: glisse, my mesa is also from git master, yes
Zaba: glisse, my kernel is 2.6.34, and not from git anything, though
glisse: well ddx simplification is likely the cullprit
MichaelLong: hm I'm really wondering why /dev/dri/card1 is not accessed. after loading the radeon module, xorg doesn't mention any activities like detecting devices and such until it is unloaded again
Zaba: glisse, you mean, the 'remove unnecessary state emit' commit that causes the issue for me? What is ddx?
glisse: yes this commit
glisse: ddx is xf86-video-ati
Zaba: ah
Terrax121: Hi guys
Terrax121: I have made an overview over the vdpau implementation
Terrax121: Though I have to ask if anyknow some documentation for creating a cdpay hw-backend
Terrax121: vdpau*
Terrax121: I can find documentation howto implement a vdpau in general (libvdpau.so.1), but howto to create something like libvdpau_radeon.so.1.. If I could implement the later, I didn't have to create the first one.
glisse: i don't think there any such documentation
Terrax121: glisse, darn.
glisse: most coder hate writting doc
Terrax121: Maybe I should just create libvdpau.so.1 then, and implement the hardware calls within that
glisse: i think what we would have like is a state tracker that could produce a libvdpau
glisse: thus other driver like nouveau could also have accel for free
glisse: also would be easier to debug using the sofware rendering of gallium
glisse: Terrax121: for an example http://www.bitblit.org/gsoc/g3dvl/
Terrax121: glisse, thx
Terrax121: Well I think I will implement it in OpenCL at first
Terrax121: Its easier for me to fix bugs in something I am used to
Terrax121: glisse, afterwards I can port it to galium
Terrax121: Then I'll drop the idea of having a libvdpau_'insert favorite vendor here' library
glisse: Terrax121: i think there is already opencl implementation
glisse: i might saw that in student project of some uni
Terrax121: glisse, do you have a link?
glisse: no, it was likely some uni project
Terrax121: glisse, okay
rindolf: Hi all.
xmikos: Hello again today, does anybody know if there is some known problem with RS690M (on laptop) and second monitor (Full HD television) connected over HDMI?
xmikos: Well, it simply doesn't work for me. It is detected OK, but no matter what resolution I try, TV screen stays always black and says "unknown format". Btw. it is Philips LCD HDTV.
xmikos: But my main desktop computer/HTPC is also connected over HDMI to this TV and it works without problems (it has RS780G and also opensource radeon drivers).
xmikos: So this must be something with RS690M and/or laptop and/or dual screen setup...
xmikos: Anyone?
soreau: xmikos: IIRC, the RS* cards are more problematic than others. Have you tried manually adding a mode or changing the refresh rate etc with xrandr?
xmikos: soreau: xrandr is reporting same modes/refresh rates on RS690M laptop as on RS780G desktop (where it works without problems)
soreau: and using same inputs on the tv?
glisse: i wonder if there is encoder/crtc limitation not allowing lvds+hdmi but that would soon broken design
glisse: soo*
xmikos: No, second HDMI input, but I have set it same way as first one (to "PC"). I have tried it even on first one with same results.
xmikos: Is there something else I can try?
xmikos: glisse: I can see this in dmesg: radeon 0000:01:05.0: Could not find HDMI block for 0x19 encoder
xmikos: (btw. I am using kernel 2.6.35-rc2-git5)
glisse: sounds like the bug is there
xmikos: glisse: So what can I do? Can I help you somehow debug it?
glisse: sounds like rs6xx hdmi got broken along the way
xmikos: glisse: I have tried it with stable drivers / kernel / UMS in the past (about year ago) but it didn't work back then...
xmikos: glisse: So it doesn't seems like some new bug or regression...
glisse: xmikos: can you build your own kernel ?
xmikos: glisse: Yes, I can, but it will take some time (at least 30 minutes)
xmikos: glisse: Should I try some patch or git branch?
glisse: xmikos: try with this patch http://people.freedesktop.org/~glisse/0001-drm-radeon-kms-assign-hdmi-block1-to-R5XX-avivo.patch
xmikos: glisse: On which kernel should I apply it?
glisse: 2.6.35 should be fine
glisse: maybe even 2.6.34
xmikos: glisse: Ok, I am going to compile it, please be patient and wait for results :-)
markit: hi, I'm looking at a laptop. Is Ati HD 560v supported by radeon drivers? I mean, 2D and composite effects in KDM? Hope for 3D?
markit: or better choose the HD 5650?
markit: or none of them? :)
markit: (Dell Studio 17)
glisse: not yet supported
glisse: hopefully will be soon
glisse: during summer
markit: glisse: so will it work like vesa in the meantime?
markit: I'm noticing, with big difficoulties, that dell seems also to use broadcom wifi chip
markit: so bad, as far as I know :(
markit: (they call the choice "dell xxxx", and you have to google to discover which chipset it uses)
xmikos: glisse: Are you still there? I have tested your patch.
xmikos: glisse: ping
soreau: xmikos: Patience
soreau: In the meantime you can explain what happened with the patch
xmikos: glisse: Sorry :-) Your patch unfortunately doesn't work... it makes things worse, after connecting HDMI, HDTV has been detected without problems (same as before). But after setting it to clone mode through xrandr, LVDS on laptop went garbage and graphics went completely frozen.
xmikos: But I have managet to get in system with SSH and there is kernel bug in dmesg:
xmikos: http://pastebin.com/NXLYkcnj
xmikos: Eh, sorry for my typing errors (and bad english, I must be
xmikos: tired)
xmikos: I have also made copy of Xorg.0.log, but I can't see anything relevant there... but if you want, I can upload it to pastebin too...
xmikos: Only last 2 lines seems relevant (but I cant see anything bad in it):
xmikos: (II) RADEON(0): Allocate new frame buffer 1920x1088 stride 1920
xmikos: (II) RADEON(0): VRAM usage limit set to 254246K
airlied: wow hdmi on rs690 looks to have some issues
glisse: yeah
glisse: i don't have doc on it
glisse: i think r5xx are in same state when it comes to hdmi
glisse: thought i don't know if any of r5xx gpu had hdmi connector
xmikos: glisse: What do you think about that bug in dmesg, does it help you in any way?
muep: hello
airlied: glisse: no r5xx had hdmi
airlied: we need to avoid hdmi audio completely on rs690
glisse: xmikos: i think it's simply r6xx hdmi polling trying to callback into r6xx stuff
glisse: but r6xx stuff is not init on rs690
glisse: airlied: only rs690/rs600 had such ?
glisse: and i guess rs740
airlied: glisse: from what I can see yes
airlied: glisse: btw I think that patch is wrong you did
glisse: yes well it bet on hdmi reg being at same place :)
airlied: there are special cases for original avivo I think
glisse: and forgot about audio
airlied: they are just not being used
glisse: zzzZZZZzzz time
xmikos: airlied or glisse: If any of you make new patch now, I can test it right away. But it is late here, so if it is more work, I can test it tomorrow. Will any of you be tomorrow here?
xmikos: Btw. should I make bug report on freedesktop bugzilla?
airlied: xmikos: please open a bug
airlied: I'll try and write a patch later, have to go play in the sandpit right now
xmikos: airlied: Where I should look for that patch? I will have time to test it tomorrow...
xmikos: airlied: It should go to kernel bugzilla, not freedesktop bugzilla, right?
rhodan: http://nopaste.info/2f731f7379.html
rhodan: I know that PM is experimental, but should it make my sound click every time it switches states?
airlied: xmikos: either bz is fine, I'll attach it there if my brain wake up
rhodan: I'm not sure whether it is a HW problem. My interrupts seem to have been fucked from the first day.
rhodan: Thinkpad T60. Sometimes the sound goes crccsss under Windows, too. At least Windows XP did, three years ago.
MichaelLong: any clue about this one: https://bugzilla.kernel.org/show_bug.cgi?id=16140 ?
airlied: MichaelLong: what was last kernel you tested on?
rhodan: Power usage (ACPI estimate): 315.7W (0.2 hours)
MichaelLong: the last kernel was .35-rc2 no rc3 yet. rc2 did not even resume properly, gave me a black screen. on .34 and maybe .33 also I was able to switch to a console as describe. before that. the screen was fading to an ugly green :)
rhodan: Oh noes D:
airlied: MichaelLong: is that dmesg from rc2?
MichaelLong: no .34
MichaelLong: [ 253.948171] Pid: 2415, comm: bash Not tainted 2.6.34 #2 oh theres an call trace seems ieee80211 related, didnt saw that before :)
DanaG: 315 watts? What system is giving you that?
DanaG: And is this a laptop or a desktop?
DanaG: ah
DanaG: Thinkpad?
DanaG: Considering my system with an HD3650, even when playing Crysis (on Windows) only goes up to like 70 watts... that 300 sounds bogus.
rhodan: You don't say ;)
cxo: My system freezes every few seconds, I see this in dmesg, [518918.822396] radeon 0000:02:00.0: ffff880077911e00 reserve failed for wait
rhodan: It's 15-30W normally. I have no Idea why it's that high all of a sudden.
mjg59: 320W is "Melt my face" kind of production. It sounds unlikely.
xmikos: airlied: Bug filled here: https://bugzilla.kernel.org/show_bug.cgi?id=16249
chrisign: hello I have a radeon x1300 in my thinkpad t60 and use debian testing. I have the problem that my display
chrisign: is this a problem with the new radeon driver?
DanaG: er, looks like you lost the description of your problem.
rah: airlied: https://bugs.freedesktop.org/show_bug.cgi?id=28500
rah: airlied: is that being dealt with?
soreau: heh
soreau: chrisign: Your first message cut off at '...debian testing. I have the problem that my display' <-- stopped there
MostAwesomeDude: I think he accidentally his display.
soreau: hahahaha
MostAwesomeDude: soreau: It's not funny. I accidentally my display last year, and it really sucked.
soreau: roflh
chrisign: ohh yes, my display flickers from time to time
chrisign: and if this starts it don't end unless I make a reboot
chrisign: and what differs is that the whole screen flickers, its imposible to work on then
DanaG: http://www.phoronix.com/scan.php?page=news_item&px=ODM0NQ
DanaG: hmm.
DanaG: KMS on GLINT? I wish they would've done it on S3, instead.
DanaG: At least that'd be a bit more useful.
MostAwesomeDude: "They" is a bit misleading. Matt's in here; his nick is mattst88.
MostAwesomeDude: And I think this is plenty useful. He's documenting the entire process, and he picked chipsets that don't need RE.
mattst88: DanaG, when I'm done, I'll be glad to send you all the documentation I've created so you can learn the process.
DanaG: Ah, S3 ones would need RE?
mattst88: could write a KMS driver for S3 then.
DanaG: I didn't think of that, actually. Needing RE, that is.
MostAwesomeDude: mattst88: I've got a Savage over here; I'll do S3.
DanaG: reverse-engineering.
MostAwesomeDude: I've also got a Gamma, so I'll add that on top of your driver. :3
mattst88: DanaG, the difference is (since both have open source drivers already) is that I've got documentation for glint.
mattst88: MostAwesomeDude, sounds good :)
DanaG: I have an HP Visualize! FX6 (or maybe it's an FX4) card here.
mattst88: I'll be back later to talk after I get something to eat.
DanaG: Seems to need a AGP Pro slot immediately adjacent to a PCI slot.
DanaG: ah, yes: http://cgi.ebay.com/HP-Visualize-FX6-VIDEO-BOARD-5064-3654-AGP-PCI-E3-/390055217626
chrisign: can anybody tell me if this flicker on my display is a known problem of the new radeon driver
chrisign: or it is just too sketchy
airlied: someone needs to do uvesakms first
sftp: фыв
sftp: GArik -> hello!
GArik_: sftp, ÔÙ -- ÈÕÊ
sftp: GArik_ -> use utf-8
pazof: use Strict;
sftp: use power
sftp: Luke
pazof: :)
T_UNIX: hey there
T_UNIX: glisse: you asked me to try the new_pll switch the other day
T_UNIX: turns out that it doesn't matter whether I put it to 0 or 1 the problem is fixed :-)
T_UNIX: but if I don't put that keyword, problem persists
T_UNIX: okay
T_UNIX: wrong assumption
T_UNIX: iit just took some time to show up
T_UNIX: new_pll=1 leeds to the flickering
LeidenFrost: wow, the driver has become very amazing
LeidenFrost: I managed to play UrT 4.1 with an HD3200
LeidenFrost: a question there: which component manages all the graphics? Im confused if this is mesa credit or xf86-video-ati
adamk: LeidenFrost: mesa handles 3D, xf86-video-ati handles 2D, the kernel DRM does both these days.
LeidenFrost: adamk: thanks
LeidenFrost: adamk: is there a feedback option in mesa site? i want to help a bit with the drivers. Specially some basic features like gamma correction (many games uses this opcion to set brightness)
adamk: LeidenFrost: mesa had a development mailing list that sounds appropriate.
LeidenFrost: adamk: ok, i will see. Thanks
adamk: http://lists.freedesktop.org/mailman/listinfo/mesa-dev
LeidenFrost: yeah, i saw that page, but i think mesa-users is a better option (i am not a programmer)
LeidenFrost: thanks again anyway
exw: I'm hoping that I can enable HyperZ on my card ( Radeon IGP 345). wikipedia and the manucaturers site says that it's a HyperZ capable card, but when I set it with driconf I get: "DRM version 1.0 too old to support HyperZ, disabling."
exw: Any ideas?
exw: Interesting. I found on this page http://homepage.hispeed.ch/rscheidegger/atilinux/ati_linux_comp_page2.html from '03 "dri driver does not support any HyperZ features, since ATI doesn't want to have them exposed in an open source driver"
exw: Is this still true today?
airlied: no now its just more we haven't implmenetde them in kms
exw: airlied: ok thanks