rx__: it's friday.. is it end of the week for AMD yet?
MrCooper: TZ=Toronto date
MrCooper: there won't be many people in the office before 9AM...
rx__: wishful thinking :)
MrCooper: err, make that TZ=America/Toronto date
rx__: 3:40A? ;)
airlied: slackers..
rx__: 3d code.. exciting stuff
rx__: hope it isn't some teaser :)
airlied: I thought they said only more 2D docs this week..
airlied: rs690 etc..
MrCooper: I read something about DRM sample code to go with it
rx__: 2 docs regarding mode setting and such
rx__: 2 other docs which are code samples
airlied: I think it might have r600 2D stuff..
airlied: microcode and blit shaders..
airlied: but no 3d info
rx__: r600 and m76
airlied: rx__: 2d info for those..
rx__: yes
airlied: ah tcore might come out as well..
MrCooper: tcore?
airlied: ah no, simple chipsetup stuff.
rx__: i read something about tcore too
airlied: some internal driver they use for bringup..
rx__: how they program the cards
rx__: in testing or something
rx__: yeah
airlied: so no shader info.. just flushing etc..
airlied: the 2D on r600 is the only really new thing..
MrCooper: never heard of that, so it must be relatively new or ex-top secret :)
airlied: some of it might be useful to fixup r300..
rx__: The main thing that tcore brings is examples on how to set up the command processors, ring buffers for the engines, including IDCT as well as the 3d engine << all this is known already?
airlied: rx__: yes..
rx__: :(
airlied: rx__: we already set the CP up for r500..
rx__: oh well
MrCooper: everything except IDCT
rx__: it'll make good for radeonhd to catch up
airlied: rx__: radeonhd could catch up anytime they wanted..
airlied: rx__: that stuff is all well known..
rx__: well i'm curious what's taking them then ;)
airlied: the r600 CP + 2d setup will be useful...
airlied: rx__: don't want to copy code from radeon, need to reinvent wheel :)
airlied: whereas I'll take code from anywhere I can find it..
rx__: haha
airlied: hmm I wonder can I get r600 2d in time for RHEL..
rx__: so you could add experimental hdmi
airlied: rx__: I think we have some of that..
rx__: oh right
rx__: agd5f committed some of that
airlied: I'll get an r500 laptop at LCA.. which will help..
airlied: but my development deadline for r500 for RH is next week ..
airlied: maybe I can make some bugfixes later.
rx__: is that some feature freeze deadline?
airlied: ah well fun fun fun.. I'll see what turns up in the morning..
airlied: rx__: not really, just have other things I really need to get done :)
rx__: oh
airlied: -ETOOMANYTASKS :)
MaximLevitsky: Anybody know when AMD will release 3D docs? (I don't expect that you know it, but what do you think)
airlied: MaximLevitsky: this year....
airlied: :P
MaximLevitsky: No bad :-)
MaximLevitsky: s/No/Not
rx__: this week if you want to be technical :)
MaximLevitsky: This week will be only 2D docs
MaximLevitsky: At least this is what I have seen on the web
airlied: rx__: all 2D docs this week...
airlied: http://www.phoronix.com/forums/showthread.php?t=7077
airlied: is the best summary
rx__: http://www.phoronix.com/forums/showpost.php?p=21283&postcount=4
rx__: :)
rx__: i'd consider the tcore to be 3d docs :P
MaximLevitsky: Well, this is what I have read
MaximLevitsky: :-)
airlied: rx__: it isn't .. no info on the 3D engine
airlied: just info on the chip setup ...
rx__: alright alright
airlied: as I said we have that for r500 already..
airlied: I don't see anyone getting 3D working :)
MaximLevitsky: Also, a code example isn't a good document
MaximLevitsky: It is good as an addition helper
MaximLevitsky: But not as a primary source
airlied: MaximLevitsky: you've obviously never read gpu docs :)
MaximLevitsky: Yes, but I did read lot of datasheets
airlied: a working piece of code can often be more valuable....
airlied: I expect the 3D docs to be just the shading languages..
airlied: so like vertex/frag shader info for r500, unifed shader for r600
MaximLevitsky: I say that both are critical
airlied: granted the Intel docs I have for 965 are pretty good...
MaximLevitsky: Both code and docs
airlied: but Intel have to write good docs for external driver writers..
airlied: AMD don't erally have that requirement..
MaximLevitsky: Generally code tends to hide lot of info
MaximLevitsky: like for example it writes some registers, but doesn't explain why or what bits do what
airlied: MaximLevitsky: if the sample code is any good it'll use descriptive register names and bit fields.
airlied: like they aren't going to try and release something like the "nv" driver
airlied: open source but seriously lacking in info..
MaximLevitsky: Lets hope....
MaximLevitsky: The day AMD release all 3D docs, I am going to take my nvidia card, throw it to the trash can, and buy an AMD card, even without the driver
rx__: don't throw it
rx__: send it over
rx__: :)
airlied: donate it to nouveau :)
MaximLevitsky: good idea.
umtc: someone told me that all radeons ever made had a builtin mpeg2 decoder, is that true?
airlied: umtc: pretty much ...
airlied: umtc: I think original radeon had mpeg1 at least..
airlied: rv200 had mpeg2 from reading docs.
airlied: I started writing RE'd code a few years ago..
airlied: but it was the disassembled type of RE as opposed to the really legal one :)
umtc: does running a buggy/damaged video through the decoder lock the card (in theory)? i've never heard much good about any card that was being used to decode the mpeg2
airlied: umtc: well you would mostly be only doing the idct and motion comp
airlied: umtc: so you probably need to validate the inputs a bit more..
rx__: well.. that makes the doc release utterly unexciting for me then :(
cbrill: I updated to latest git and now my system locks up when starting X11
cbrill: the bad commit (according to bisect) is 2e4473b63d65801ae8ac5a8391de232b2201d958
rx__: huh?
rx__: of what package?
cbrill: xf86-video-ati
rx__: oh that's the latest commit
cbrill: really?
rx__: checks for updates
rx__: okay i lied
cbrill: 94d5a432f72801f821d1c4ce952baba17552659d is
cbrill: rx__ :-)
cbrill: at least this commit causes my system to lockup
rx__: i must be lucky then :)
cbrill: anyone can tell me how to view (using git commandline) what a commit changed?
rx__: git diff commit
rx__: git diff 94d5a
rx__: well that compares head to the commit you picked
cbrill: rx__, thanks! works perfect!
cbrill: rx__: your git-foo seems to be quite good so here's another question: how to I update to a revision?
rx__: i use git reset --hard 94d5a
rx__: that'll nuke all your local changes
rx__: and revert to that commit
cbrill: rx__: I currently don't have any changes so that's fine
kokachev: airlied: i swithed tv timings , as you recommended yesterday. After that i even don't see cursor. Picture is not sync.
Terman: git question: how do I switch from initial-randr branch to master?
MrCooper: git-checkout master
Terman: MrCooper: thanks!
MrCooper: np Terman
rbmorse: FWIW, 94d5a432f72801f821d1c4ce952baba17552659d locks up my X, too. I'm here on vesa, and it sucks.
airlied: rbmorse: 94d5 or 2e44?
rbmorse: 94d5...I think. I pulled abvout 30 min ago. Let me check the log.
rbmorse: airlied: it was 94d5
agd5f: rbmorse: got a log?
airlied: wierd it shouldn't have broken things..
airlied: famouse last words.
rbmorse: Ubuntu at work...ate the log whilst trying to get vesa going before I understood what was going on. I'll try the ati driver again.
rbmorse: back in a minute...we hope.
rbmorse: Hi guys. I'm back. agd5f: You asked about log when the ati driver fails to load. Funny thing about Ubuntu is this "failsafe" feature that forces the vesa driver to load in reduced res, but also overwrites the X.org log file where the failure occurs.
rbmorse: If it's really important, I can indice the failure then kill the machine and use a live CD to try and retrieve the log before it gets overwritten.
agd5f: rbmorse: restart in single user mode then try startx as root
airlied: I think we can temporarily turn off failsafe..
airlied: but I'm not sure how..
Magnade: doesnt it copy the old log out of the way to a .old file or something?
rbmorse: agd5f: This would be after resetting xorg.conf to load the "ati" driver instead of "vesa" Can do.
agd5f: yeah
rbmorse: Magnade: It does make a backup, but the "failsafe" does what I would describe as a double-bump so that wipes the
rbmorse: log where the ati driver fails to load.
rbmorse: back in a minute...
damentz: agd5f: anything new on the mode changing issue?
Magnade: airlied: gdm.conf has several failsafe items removing or commenting them out might give desired results
agd5f: damentz: not yet
Magnade: airlied: i dont see anything related tho in the gui config of gdm however
airlied: Magnade: yeah I'm not sure how it works.. I'm not running ubuntu..
Magnade: airlied: well gdm side of things i would think how the config of failsafe works should be universal
airlied: I think moving the/etc/gdm/failsafeXServer aside might work..
egore: agd5f, you read that 2e4473b63d65801ae8ac5a8391de232b2201d958 causes my system to lockup when starting X?
agd5f: egore: no. missed that
agd5f: egore: got a log?
egore: agd5f, I bisect'd it this morning and if I didn't get something wrong it's this commit
egore: agd5f, unfortunately not yet. I need to figure out how to get a serial console working :-)
agd5f: why? can't you use a vga console?
egore: agd5f, I tried via ssh. the lockup occured without any helpful output. I heard that serial console might give more
agd5f: egore: does commenting out line 3644 of radeon_driver.c help?
egore: agd5f, texturesize = -1?
egore: doh!
rbmorse: Hi guys. Back again. The operative error message from the log seems to be RADEON (0) No valid modes. THat is followed by "Screens found but none have a usable configuration."
agd5f: egore: pScrn->pScreen = pScreen;
airlied: rbmorse: wierd...
airlied: can you post the log?
airlied: rbmorse: pastebin.ca
egore: agd5f, I guess it's another line for me ... 3644 is info->BlockHandler = pScreen->BlockHandler;
rbmorse: How does that work, just www.pastebin.ca ?
airlied: rbmorse: yup you can upload a file..
egore: double-doh! I was on the revision befor that commit ;-)
egore: I also found out I could lock up my system when removing the CailWriteATIRegister/CailReadATIRegister output
egore: brb, testing
rbmorse: airlied: pastboard.bin http://www.pastebin.ca/841910
agd5f: rbmorse: are you using DVI?
agd5f: it's no longer getting a ddc line
airlied: hmm a linemux of 0 must be valud..
egore: agd5f, still locks up when I comment out line 3644
agd5f: egore: ok, about reversing lines 3646 and 3648?
airlied: rbmorse: can you try the latest git?
airlied: rbmorse: I've just pushed the fix..
egore: agd5f, commentinh out 3644, 3646 and 3648?
agd5f: no switch their places
agd5f: move 3646 to 3648 and vice versa
egore: and still commenting out 3644?
agd5f: 3644 doesn't matter
egore: ok
egore: airlied, latest git doesn't compile for me ...
egore: airlied, radeon_atombios.c:1772: error: expected identifier or '(' before 'for'
rbmorse: airlied: on the way. Thanks. You guys are great.
egore: brb, testing
rbmorse: How does that work, just www.pastebin.ca ? /leave
egore: agd5f, still locks up :-/
agd5f: egore: what hw?
egore: rv535, iirc
egore: agd5f, yeap RV535
agd5f: egore: what xserver?
egore: 1.4
egore: 1.4.0.90
egore: to be exact
egore: no latest git
rbmorse: Hi guys. I'm all fixed, now. Thanks.
agd5f: egore: rever all your local changes and try this patch: http://www.botchco.com/alex/xorg/egore.diff
egore: brb, testing
egore: agd5f, it does no longer lock up
egore: agd5f, but now I get a 640x480 resolution during login
egore: but xrandr'ing the correct resolution works fine
agd5f: egore: git a log?
egore: http://www.egore911.de/vype/Xorg.0.log
damentz: hey guys
damentz: another person had the same resolution problem
damentz: a bit different though
damentz: here's the thread: http://sidux.com/index.php?name=PNphpBB2&file=viewtopic&t=8047&start=0&postdays=0&postorder=asc&highlight=