airlied: agd5f, mjg59 : I think I'd like to leave the dynpm=1 option in with all that code
airlied: and have the default code just be the static stuff for now
airlied: so I'd rather not rip it out if we could avoid it, since its quite close to working
airlied: though we could probably to with squashing a lot of the fixes together
airlied: btw I'm off work for two weeks now, I might get some time to do some drm merging in that time since I'm not actually going anywhere but it'll be minimal
mjg59: airlied: Yeah, that works
Jonimus: I just built a kernel with his new patches and am going to try them in the morn, I'll let you guys know how it works out
DanaG: asdf
algol: what is meant by "posting" the video card?
MrCooper: algol: from POST (Power On Self Test), low-level initialization of the card
algol: MrCooper: oh, *that* post! :)
algol: MrCooper: thanks
MrCooper: np
Nightwulf|work: hi all
twiztid: hey anyone know how i would be able to upgrade my GLX version from 1.2 to at least 1.3?
MrCooper: by using xserver 1.8 or newer with KMS
twiztid: MrCooper: my fresh install of Lucid had KMS enabled by default and compiz and gnome ran very poorly, i had to turn effects off and ultimatly turn KMS off to get compiz and effects all the way up, but now cairo dock is calling for GLX 1.3
MrCooper: don't expect GLX > 1.2 to ever become available without KMS
twiztid: MrCooper: i dont, im trying to get 1.3 to satisify cairo; id love to have KMS enabled but it makes compiz and gnome slow with effects even at medium
MrCooper: I rather doubt cairo-dock really absolutely needs 1.3 anyway
twiztid: so could there be another reason it flickers so badly? http://pastebin.com/uuzk8ZrT line 29 shows it calling 1.3 but goes on to say it might be a bug???
MrCooper: translucency with OpenGL can only work properly with DRI2, which is again only available with KMS
MrCooper: same for proper integration of OpenGL apps with compositing
MrCooper: so I suspect your best way forward would be to overcome the KMS issues
[Enrico]: twiztid: cairo-dock has 2 backends, the opengl one (which flicker without dri2 as MrCooper said) and the cairo one which should work good
twiztid: Enrico: so sould i go ahead with gallium, fix kms, or use cairo backend (how?)
[Enrico]: twiztid: as you wish.... as of the cairo back end i have no idea, i used cairo-dock only months ago just for a try but in the menu i was getting 2 entry for it, one with opengl and the other with cairo (i guess you can try to run cairo-dock --help and read)
twiztid: MrCooper: what would you suggest my first troublshooting step be in smoothing out my pc's performance with KMS enabled?
MrCooper: describe in more detail what the problem is and provide logs
MrCooper: twiztid: r300g requires DRI2
twiztid: MrCooper: is DRI2 dependant on my video card? I have an x1300... Upon a first boot of a fresh install of Lucid, compiz on Normal and Extra was slow and gnome as well, i set effects to none and performance returned, but to get effects to extra i had to disable kms
MrCooper: twiztid: DRI2 is available with KMS
twiztid: MrCooper: how can i help you help me? what logs should i provide and so on... i really appreciate it
MrCooper: twiztid: actually I'm afraid I don't have time for that level of help right now, hopefully someone else can pick it up
twiztid: no worries, im just as curious about kms being choppy as the next guy... thank you for pointing me int the right direction tho
hnsr: when I try to switch to vt from X, screen freezes and on dmesg I get: [drm:drm_mode_getfb] *ERROR* invalid framebuffer id
hnsr: when I switch back to X vt X screen unfreezes and I can use my desktop normally again
hnsr: anyone know what might cause this? running latest stable versions of everything
hnsr: or specifically linux kernel 2.6.33, mesa-7.8.1, xorg-server-1.8.0 and xf86-video-ati-6.13.0
airlied: hnsr: wierd got fbcon built in?
hnsr: hmm not sure, how can I check?
airlied: check kernel config for CONFIG_FRAMEBUFFER_CONSOLE
hnsr: CONFIG_FRAMEBUFFER_CONSOLE=m
airlied: is it loaded?
hnsr: hmm i have loaded: radeon, ttm, drm_kms_helper, drm, fb, cfbcopyarea,cfbimgblt, cfbfillrect
mjt: lsmod | grep fbcon
hnsr: ah no, not loaded then
hnsr: yep and now it works after I load that by hand
mjt: heh
hnsr: weird though, I never had to load it by hand before, stuff just seemed to work :(
mjt: not weird actually
mjt: and it really brings up a qieston: is there any reason to NOT load it automatically?
airlied: we build it into the kernel on Fedora
mjt: ditto on debian
hnsr: holy crap I can actually run doom3
hnsr: some glitches, and huge fps drop when I look at some glass window (some shader issue I guess)
hnsr: but still, impressive :)(
Tommeh: Woohoo
Tommeh: Which card?
hnsr: radeonhd 4850
Tommeh: Nice. :D
hnsr: im actually not running git versions of anything so maybe its even better with those
mjt: i can play doom3 on a built-in amd780g ;)
mjt: "medium" quality, 800x600
mjt: it's Radeon HD 3200
hnsr: ah i hate those in my other pc
hnsr: but never used it, put a cheap nvidia in it as well :/
hnsr: s/hate/have/
Tommeh: I still wonder why my second screen flickers on boot/when I load a Wine app.
Tommeh: Ubuntu Lucid, xorg-edgers with all bleeding edge updates, 2.6.34-rc6 on an R600 card with KMS
Tommeh: I should probably look to see if a bug has already been filed.
mjt: maybe hd5450 will be able to replace my onboard graphics.. i dunno. My main concern is the noize level, which is the higher the more powerful the components inside, and the built-in gpu is one of the least-consuming (energy-wise), and it is sufficient for all my tasks so far...
MaximLevitsky: not related to radeon, but I am running the piglit, and it says that some tests were skipped. why?
algol: why does cail_mc_read and similar functions are named like that? what does "cail" mean?
edwin: isn't it call?
algol: edwin: no, it's spelled "cail"
edwin: maybe it is an opcode?
algol: the comment above those function says "ATOM accessor methods"
evil_core: hi all
algol: and they use ASIC dependent calls to perform read/write
glisse: iirc cail is something like C...A... Instruction load
glisse: likely commonadvancedinstructionload
glisse: but it maybe nothing
glisse: iirc name came from AMD
algol: glisse: ah, ok, I'm just trying to put things in place
algol: maybe it's better if I skip lowlevel things, like atombios related functions...
bridgman: I think CAIL is Common ASIC Initialization Layer or something like that
bridgman: it's the "set up the chip" code anyways
bridgman: glisse; ping
bridgman: glisse; unping, will email ;)
glisse: question is which email ;)
cafaro_: Hi, I'm trying to get X working with Debian lenny (card: HD3650), unfortunately, I get a black screen, xorg.log shows a bunch of "(EE) RADEON(0): Idle timed out, resetting engine..." errors, any could help me fix this? thanks
cafaro_: Here's the entire log: http://paste.ubuntu.com/429484/
cafaro_: xorg.conf is nothing special, just default with Driver "radeon" in the Device section
cafaro_: I wasn't quite sure what packages to install though, only package I installed was: xserver-xorg-video-radeon.
adamk_: cafaro_: Try disabling DRI. It's not going to work anyway on such an old kernel.
adamk_: And probably with such an old Xorg.
adamk_: Or, even better, upgrade to something that isn't ancient :-)
cafaro_: debian stable is ancient? I tried fglrx on debian squeeze, which didn't work as well
adamk: Take a look at the release date on that X server.
adamk: Nearly two years ago.
adamk: That's pretty old, especially when it comes to video drivers.
cafaro_: adamk_: add Option "DRI" "false" in xorg.conf?
jcristau: yeah
adamk: I *think* that'll do it if you put it in the Device section.
adamk: You won't get any 2D acceleration without a much newer kernel, just so you know.
cafaro_: It won't even start X
adamk_: Can you pastebin the log file?
cafaro_: sure 1 sec
cafaro_: adamk_: http://paste.ubuntu.com/429496/
cafaro_: I didn't let it finish, I manually turned of my laptop
adamk: LOoks like the same error.
cafaro_: yes it is
adamk_: Unfortunately, I really have no other suggestion other than to upgrade.
jcristau: maybe radeonhd fares better..
adamk_: Oh, that's true.
cafaro_: Would you think it'll work with squeeze (it has a pretty recent x/kernel version), and would you reccommend radeon or radeonhd?
cafaro_: I don't really know which one to choose, only differences I found were on this page: http://www.x.org/wiki/radeon
adamk: Well I would almost always recommend radeon over radeonhd. It gets more testing and development.
adamk: radeonhd is simply being recommended this time because radeon just doesn't seem to want to work for you.
jcristau: if it doesn't work with squeeze we'll want to fix that.
cafaro_: alright
cafaro_: back to squeeze then ;)
evil_core: Obscene_CNN patch speeds up KMS also?
adamk_: Is there a good utility for testing 2D performance?
evil_core: allegro games :P
Curan: adamk_: maybe so (older) 2D game? Or something more synthetic like cairo-benchmark (though I'm not sure that isn't more a benchmark for improving cairo): http://cgit.freedesktop.org/cairo-benchmarks/
Curan: s/so/some/
adamk_: Hmmm.
adamk_: Thanks, I'll take a look at cairo-benchmark.
algol: are there any docs regarding what drm_crtc_helper_funcs are supposed to do?
Curan: (for the games part I thought of something like "Jagged Alliance 2" (there is a Linux port:
algol: I mean, what operations should be performed in mode_fixup() or in mode_set_base()
Thunderbird: MostAwesomeDude, could you try this test app? http://wine.pastebin.com/KWZTUmw6 (you would have to enable s3tc)
Thunderbird: a few weeks ago I reported this bug for fglrx but I think other drivers are affected as well based on a appdb report
Obscene_CNN: evil_core, yes kms should be sped up too by my mesa patch
evil_core: Obscene_CNN: but I want more :P
Obscene_CNN: evil_core, I'll work on getting more speed out of the r300/r500 at lunch. Unfortunately I have no way to test it
Obscene_CNN: goes back to writing report
ickle_: adamk_: ignore cairo-benchmark, use cairo-traces for 2D performance testing of real-world behaviour
evil_core: Obscene_CNN: I ca ntest anything
ickle_: and cairo-benchmark is not a patch on the perf suite inside cairo anyway
Obscene_CNN: evil_core, okay
evil_core: you can even make somewhere dir with patches_to_test_for_r500, and I will be looking over forr anything new to test everytime I will get to internet
evil_core: I believe that in place I will move in few days people bought open and WEP AP's, not WPA, so still will have access to intenet ;)
evil_core: internet*
agd5f: algol: fix up makes any adjustments to the mode needed for the specfic output, e.g., adjustments for panel scalers, etc.
agd5f: mjg59: power_supply_is_system_supplied() seems to always report DC on desktop systems. is there a way to tell if that function is reliable, or should I just limit it's use to mobility chips?
ajax: you have a battery-powered desktop?
algol: agd5f: so, during a mode set, mode_fixup gets() called to give the driver a chance to make adjustments?
agd5f: ajax: nope
agd5f: algol: yes
ajax: agd5f: then i don't understand the question
agd5f: ajax: my desktop should report AC
ajax: oh. brain sense inversion.
evil_core: agd5f: no!
agd5f: evil_core: ?
evil_core: its power supply problem
evil_core: but you can switch to one using AC, i.e. for a car or soemthing
evil_core: you have only DC on the M/B side
agd5f: evil_core: what? my desktop is plugged into the wall
evil_core: who cares, it notchecka le what power supply you got
evil_core: and your laptop is connected to WALL also trought AC/DC adaptop
agd5f: evil_core: because the power management checks if the system is on ac or dc
evil_core: wtf?
agd5f: battery or wall
agd5f: if it's on battery, it makes different decisions
evil_core: hmm...stupid namecheme
evil_core: laptop is never conencted to AC
mjg59: agd5f: Hm. I meant it to return 1 on desktops, so should probably fix that
evil_core: or simply I dotn now such laptop
agd5f: evil_core: semantics. battery or wall power
evil_core: its stupidity and should be changed IMHO
agd5f: evil_core: my phrasing?
evil_core: I was thinking its namescheme in driver
agd5f: ac/dc isnt used at all in the driver
algol: uhm... I found some comments about crtc helper functions in drm_crtc.h
algol: probably the comment remained there
algol: after the structure was divided in two
glisse: agd5f: my kernel patch is needed for the ddx patch to work, vblank event need crtc to be either primary or secondary
glisse: this is broken but fixing it need a new interface for the event stuff
agd5f: glisse: yeah, sorry, I didn't see your ddx patch until after I read the drm patch
glisse: no problem, i don't like that info patch neither
spstarr: lolbuntu indeed
MostAwesomeDude: Thunderbird: In a bit.
Thunderbird: ok
jlkiller: Hi!
MostAwesomeDude: should /nick MAD|sick to warn people
jlkiller: I've got issues with the graphical driver @ Kubuntu 10.04, when i boot, or shut down the computer, the kernel gives errors. I use a ATI Radeon HD4850 on a 64-Bit-system. The error-message: [ 20.412455] [drm:radeon_gart_bind] *ERROR* trying to bind memory to unitialized GART !
jlkiller: Anyone an idea what the problem could be? When i use the non-opensource-driver my system doesn't boot. The proprierty driver isn't a solution.
jlkiller: The computer does boot, but gives this errors before shutdown en before booting in graphicalmode
zhasha: MostAwesomeDude: you're mad AND sick? poor you
dileX_: agd5f: if I looked over correctly
zhasha: jlkiller: I get this: "[drm:drm_mode_getfb] *ERROR* invalid framebuffer id", which locks up my GPU for a short while :P
agd5f: dileX: yes
odin_: I have performed a full set of tests now and have X.0.log output with this RV100, trying to show performance regression in the default 1.7.6 settings
agd5f: airlied, mjg59: ok, new patches that leave the dynpm stuff in: http://people.freedesktop.org/~agd5f/pm-drt/
mjt: doh. Why I read it as "pm-dirt" ? :)
mjt: it's much welcome actually
felipec: there's something strange on my vga output, it looks like some kind of timing issue, or interlacing or something
felipec: when I switch mode it's ok for a couple of seconds... then blanks, and then goes all weird
felipec: sounds familiar?
cafaro: Tried to get X working with radeon (Radeon HD3650 using Debian squeeze), but I only get a black screen... here's X's log: http://pastebin.com/adKVhf4x, any ideas?
NTU: post expired
NTU: nvm
NTU: comma was left in :P
cafaro: ah ok
cafaro: my radeon packages: http://pastebin.com/EX9BHJ3X
cafaro: I also installed the firmware package from the non-free repo
cafaro: I got a segfault error when doing X -configure, so I manually made a xorg.conf
cafaro: here's my xorg.conf: http://pastebin.com/kniimLPa
NTU: ....2.6.32
cafaro: ?
NTU: well i guess if you dont use KMS it doesnt matter. disregard that
NTU: can you post the seg fault?
cafaro: sure
cafaro: NTU: http://pastebin.com/UMckESHE
NTU: oh thats a known bug
cafaro: oh it is?
NTU: you have to remove the xorg server xmwgfx thing
NTU: jcristau said so
NTU: sudo apt-get purge xorg-xserver-vmwgfx or something
mjt: xserver-xorg-video-vmwgfx?
NTU: right
NTU: sry im not on debian
mjt: i'm on debian but i've no such package ;)
mjt: but ok
mjt: probably comes from vmware
NTU: i have no idea
mjt: it's not part of debian
NTU: i know
NTU: its an option when compiling xserver IIRC
NTU: --enable-vmwgfx or something
NTU: my memory stinks when im on windows 7.
mjt: actually i'd remove all extra/irrelevant video drivers
NTU: why?
chithead: vmwgfx is in libdrm, and used by xf86-video-vmware
mjt: starting with xserver-xorg-video-all
mjt: NTU: because it loads them all, whith hell knows which conflicts between diff. drivers :)
cafaro: I don't have such package!?
NTU: xserver-xorg-video-vmwgfx?
mjt: xserver-xorg-video-vmware
NTU: try that
NTU: what he said
mjt: actually it's xf86-video-vmware
NTU: thats what the offcial name is
NTU: debian likes being cute and renames everything to make things more confusing
chithead: some distros call it xserver-xorg-video-foo or invent other names for xf86-video-*
NTU: right i hate that
NTU: its annoying as shit
chithead: fedora does it too
NTU: yeah i dont use that either
mjt: it was after xfree fiasco
mjt: so it's not xf86 anymore
cafaro: My screen
MostAwesomeDude: Actually, xf86 is the correct name.
cafaro: My screen is messed up
cafaro: brb
MostAwesomeDude: xf86 is the name of the DDX interface within the X server that loads these modules.
MostAwesomeDude: And it hasn't changed radically in a while, so nobody's seen a good reason to change it externally.
MostAwesomeDude: The More You Know
NTU: right. xf86-video-* is the official name as used by cgit.freedesktop.org
NTU: i love that site. i should make it my home page! :)
mjt: ok, that was my guess ;)
cafaro: ok back again; couldn't read the last art
cafaro: part
cafaro: NTU: ?
NTU: yeah?
NTU: what part do you want me to repeat?
cafaro: from when I said i didn't have that package
NTU: xserver-xorg-video-vmware
NTU: try that
NTU: like.. try purging it
cafaro: okay done
cafaro: now X -configure?
NTU: yeah as sudo / root
cafaro: segmentation fault again :(
NTU: same one?
NTU: or a different seg fault error?
cafaro: NTU: It's nearly the same
NTU: oh but it changed!
cafaro: nah
cafaro: 99.99% the same
cafaro: let me upload it
NTU: haha
cafaro: old log: http://pastebin.com/UMckESHE
cafaro: new log: http://pastebin.com/7tLzA3V1
cafaro: backtrace is pretty much the same
NTU: fglrx...
NTU: post xorg.conf
NTU: wait you already did
NTU: nvm
NTU: let me edit it
mattst88: cafaro, if you're trying to use the open source driver, remove fglrx first.
NTU: but he has "radeon" set in driver..
cafaro: I think I did
NTU: this is so weird
mattst88: cafaro, did you read your log? It loads fglrx and detects your card and then dies.
NTU: it shouldnt be loading fglrx in the first place
cafaro: which one?
NTU: section "device" driver "radeon" is set
mattst88: the new one.
mattst88: it's also loading a _bunch_ of other drivers.
cafaro: maybe it's because i installed http://packages.debian.org/squeeze/firmware-linux-nonfree ?
NTU: no thats radeon ucode
NTU: i think.
NTU: all these cute names drive me nuts.. i think im gunna take off.
cafaro: alright
cafaro: I think I'll just stop trying to get linux working
cafaro: i'm pretty sure i won't get it working any time soon anyway
felipec: anybody with an R520?
Obscene_CNN: evil_core: are you still here?
evil_core: Obscene_CNN: I am 24h online
Obscene_CNN: evil_core, msg me an email address
Obscene_CNN: I got some code for you to try
evil_core: krzysztof.krakowiak@gmail.com
Obscene_CNN: okay
evil_core: Obscene_CNN: still no mail in mailbox, you wouldnt be a cowboy ;)
Obscene_CNN: just sent it
Obscene_CNN: had to talk to a co-worker for a moment
Obscene_CNN: get it yet evil-core?
evil_core: Obscene_CNN: yes
Obscene_CNN: okay
evil_core: buy I did make clean, so it wil ltake a bit time
Obscene_CNN: okay, thats fine. no rush
evil_core: Obscene_CNN: you wrote you are wating for results
evil_core: how to test it?
Obscene_CNN: I just want to know if it compiles and works
evil_core: I was thinking that benchmarksl ;)
Obscene_CNN: I'm hoping you will see small jump in performance over what you have now
evil_core: btw, cleaning is needed after patching or gitpulling?
evil_core: if its 10% then I will not see
evil_core: or even 25% could be hard to cath up on the eyes
Obscene_CNN: cleaning is best done after a patch or a git pull
evil_core: hmmm.maybe I shiuld return to ccace then :/
Obscene_CNN: try what your doing first
Obscene_CNN: if you have a problem then do a clean
evil_core: the worst it I cam compiling both 32/64bit at time ;)
evil_core: Obscene_CNN: I doubt theres not much improvement ;)
Obscene_CNN: So not much change for you?
evil_core: nit benchmarked it
evil_core: I was playing q3a and warsow
Obscene_CNN: okay, but it works fine though?
evil_core: yes
Obscene_CNN: okay I will post it
evil_core: Obscene_CNN: I see difference between system mesa and master with your patches
Obscene_CNN: okay good
evil_core: but dunno if because your patches or because of master ;)
Obscene_CNN: :)
evil_core: in system we got always last stablke release
evil_core: Obscene_CNN: why rou are not starting bringing Gallium to UMS classic state, etc?
evil_core: and only play with strange optimizations :P
Obscene_CNN: I'm still learning my way around mesa
evil_core: Obscene_CNN: maybe try fixing r5xx_pageflip, it should work for newer chips too
evil_core: I could test it and take locks on mine machine ;)
evil_core: Obscene_CNN: and I am using hard optimizations when building drm libdrm, ddx and mesa
evil_core: so its not really comparable with system, and maybe some of your optimizations are lost with such gcc flags ;)
Obscene_CNN: maybe
evil_core: and why Bloom suck so much on radeons?
Obscene_CNN: evil_core, pageflipping may be fixed soon
evil_core: Obscene_CNN: why?
evil_core: pageflipping was oly iplemented up to r400
evil_core: oficially
Obscene_CNN: its was mentioned on #dri-devel
evil_core: becaue it was ready in 99% for newer too
evil_core: KMS mesa classic is broken for warsow
evil_core: datafiles loads ver long up to timeout
evil_core: gallium and UMS dont got suchs problrm
evil_core: bit its damny slow and artifacted
NTU: evil_core are you talking about Obscene_CNN's patch he sent you in email or KMS?
NTU: *UMS
evil_core: KMS
evil_core: in UMS warsow works nearly perfect
NTU: are you using 2.6.32?
NTU: i know a lot of people who used the .32 kernel and have had nothing but problems.. me on the other hand.. i never had KMS issues
evil_core: d-r-t
evil_core: it has many, especially performance one
NTU: http://people.freedesktop.org/~agd5f/pm-drt/ ?
evil_core: ␁nopm
evil_core: read topic
evil_core: pm is special brabch with dynpm2
NTU: thanks :)
Obscene_CNN: off to deposit my paycheck then its beer time :D
NTU: heh
AstralStorm: hello
AstralStorm: seems that my r600 is suffering from double vision when dynpm or dynclks is active
AstralStorm: which one of those should I toggle first for the test?
AstralStorm: w/o any of those options, it works fine
NTU: i personally don't know. someone here will probably answer you shortly.
AstralStorm: using fresh drm-radeon-testing
AstralStorm: also I get:
AstralStorm: "[ 438.501044] You have old & broken userspace please consider updating mesa"
AstralStorm: again
AstralStorm: this is extremely fresh mesa and stuff, like, some minutes old
NTU: did you specify PATH?
NTU: or install to /usr?
AstralStorm: to /usr
AstralStorm: and yes, glxinfo says the right things
AstralStorm: glxgears works just as well, barring the message in syslog
AstralStorm: sounds like someone is sending invalid packets
NTU: make sure --with-dri-driverdir is set correctly too _and_ lib path.
NTU: --prefix=/usr isnt always enough
NTU: is really depends on the distro though
NTU: *it
AstralStorm: ...
AstralStorm: everything is set correctly
AstralStorm: it has worked before and works now
AstralStorm: and glxinfo works fine too
AstralStorm: plus some older 3D apps (damned lack of GLSL 1.2)
NTU: i have a patch that enables GLSL 1.2 if you need it for r600.
NTU: plus some gl extensions that are off that i turned on
AstralStorm: gimme, I'll be happy to test that
AstralStorm: I bet it's a large patch to the parser
NTU: r600_context.c
AstralStorm: hm
NTU: and no its not very big
airlied: NTU: did you do any work to actually support the extensions?
airlied: its kinda stupid just enabling things without writing the code to implement them
NTU: static const struct dri_extension card_extensions[] = { i edited those lines
AstralStorm: it'll explode then
NTU: AstralStorm, this isn
NTU: 't much but it gets GLSL 1.2 at least according to glxinfo http://pastebin.com/EYvk93JU
AstralStorm: so it "cheats" ;p
AstralStorm: that won't get me actual GLSL 1.2 working
NTU: {"GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, ?? that doesnt do it?
AstralStorm: not sure the driver supports those functions
AstralStorm: might get a software fallback and we know how fast that is
airlied: more likely to get a misc crash
AstralStorm: lack of PBO is bad too, but the app in question can use FBO
AstralStorm: even better ;p
NTU: hmm
NTU: so glxinfo isnt exactly honest
AstralStorm: it prints what the driver tells it to print
AstralStorm: :)
NTU: more recently ive been messing with r600g trying to port it to the new resources API. been awhile since i last looked at classic r600 dri driver
AstralStorm: porting won't help r600g all that much, unless someone actually added all the insns
AstralStorm: maybe that message was caused by libdrm?
NTU: its a good learning experience for me i think
AstralStorm: I do have it built with --enable-radeon though
AstralStorm: (and it's latest git as well)
agd5f: AstralStorm: try drm-radeon-testing plus the patches I just mailed to dri-devel or slightly older ones from here: http://people.freedesktop.org/~agd5f/pm-drt/
AstralStorm: ok
AstralStorm: although the effect was quite funny ;>
airlied: agd5f: care to update those ones ;-)
agd5f: airlied: np
airlied: makes it easier to pull into d-r-t from there, I'll do that, then you guys can decided on squashfest
agd5f: done
agd5f: airlied: I sent the disable encoder on separately since it's probably useful by itself
airlied: I'll put that in drm-next and rebase d-r-t on it at some point today
AstralStorm: agd5f: so do I still have to extract those patches from the ML?
AstralStorm: only Gmane carries those messages and it's not trivial to extract them from there
airlied: okay pushed out new d-r-t
airlied: should take the usual 5-10 mins
agd5f: AstralStorm: http://people.freedesktop.org/~agd5f/pm-drt/
AstralStorm: I'll wait those 10 mins :)
agd5f: also http://lists.freedesktop.org/archives/dri-devel/
AstralStorm: which are missing the patches ;p
AstralStorm: or rather, the messages
AstralStorm: oh, just went in, great
AstralStorm: oh btw, won't gui idle interrupt work on multiple screens correctly?
airlied: agd5f: you might want to make the sysfs files act like /sys/power/disk
airlied: where it prints out all hte options with the current sleceted one with [] aruond it
michaellarabel: airlied: Do you happen to have agd5f's pm patch 8 of 9?
airlied: michaellarabel: all should be http://people.freedesktop.org/~agd5f/pm-drt/
airlied: or in drm-radeon-testing now
michaellarabel: Ah okay, thanks, was going off the mailing list messages.
NTU: airlied: i dont see any changes in your kernel git tree
NTU: in drm-radeon-testing
airlied: NTU: takes 10 mins
airlied: maybe longer
NTU: oh ok
agd5f: airlied: accidently sent an older version of the disable encoders patch. newer one is here: http://people.freedesktop.org/~agd5f/0001-drm-radeon-kms-atom-disable-the-encoders-in-encoder.patch
agd5f: I can send an incremental patch against drm-next if you'd prefer
NTU: my dad has a laptop with an X1650 or X1600 in it and when in linux plugged into an LG 1080i HDTV, its all green on the HDTV but normal on the other.
NTU: any idea?
NTU: but i only tried it with s-video
NTU: an a hp compaq.. forget model number
NTU: *its an
agd5f: NTU: using kms?
NTU: yes it an nx series
NTU: like its not solid green. you can still make out objects and stuff
NTU: but its highly corrupted. i can upload pic if you want
agd5f: NTU: file a bug: https://bugs.freedesktop.org
NTU: alright. nvm about the pic. its not in linux right now and i rather not tinker with it ;)
sneakers_: Hi I am using a ATI Technologies Inc RV770 [Radeon HD 4850] with Fedora 12 and in one of the games I play I cannot adjust the brightness. Is that a known bug?
NTU: the backlight controls arent controlled in radeon DDX, mesa or libdrm
NTU: i dont think...
sneakers_: NTU, Was that to me?
NTU: yes
sneakers_: I guess "backlight controls" is the same thing as brightness?
Jonimus: no
Jonimus: they are different
NTU: hmm
sneakers: ug
sneakers: So what is the word on this issue?
NTU: i personally dont know
NTU: id recommend waiting a bit though
sneakers: ok I waited
sneakers: now what
sneakers: ;)
NTU: ...
NTU: sneakers, if you're in a rush, you can come back later when this channel is more active. its usually _not_ like this _at all._
NTU: at least when im here ;)
sneakers: not in a rush
sneakers: I'm using the mesa-dri-experimental package in Fedora 12
sneakers: mesa-dri-drivers-experimental.x86_64 : Mesa-based DRI drivers (experimental)
sneakers: ^ that one I mean
sneakers: version 7.7-4.fc12 (x86_64)
AstralStorm: hmm, should I merge drm-linus on top of drm-radeon-testing?
AstralStorm: oh, wait, it's already mainlined.
AstralStorm: ignore me then
AstralStorm: sneakers: that's no 7.8
AstralStorm: try git
AstralStorm: esp. if you're using r700
sneakers: AstralStorm, How do I use git?
NTU: git clone
NTU: :)
AstralStorm: find that old howto
sneakers: I need a walk-through I don't even know how to do it
AstralStorm: you can install your new mesa in another prefix
sneakers: sounds like greek to me
NTU: ok yum install git-core or whatever, then git clone
NTU: like tree would be git://anongit.freedesktop.org/mesa/mesa
AstralStorm: afterwards run the ./autogen.sh, then ./configure --prefix=
NTU: why somewhere not usr?
NTU: i always install to /usr
AstralStorm: to not overwrite the original mesa and make the package manager explode
AstralStorm: yum isn't too smart ;p
NTU: oh right.. fedora
sneakers: I think any package management system would have that same problem....
NTU: right which is why i use arch and lock everything up
NTU: nothing gets in my way! :)
AstralStorm: almost any. this excludes Arch, Gentoo and Slackware
NTU: hehehe
AstralStorm: the former two have scripts to install this
sneakers: I tried Arch for a few months. Not my cup of tea. I like stuff delivered to me on a platter.
NTU: meg i even bypass the arch pacman system
AstralStorm: in slackware you can easily package that yourself
NTU: i use holdpkg
sneakers: I'm still doing this step: git clone git://anongit.freedesktop.org/mesa/mesa
AstralStorm: NTU: ugh
AstralStorm: ok
AstralStorm: sneakers: http://www.x.org/wiki/radeonBuildHowTo
NTU: hey! you're not giving me hits!
NTU: http://neo-technical.wikispaces.com/radeonhd-3d :P
sneakers: AstralStorm, Those directions are way too advanced for me. I have no idea what they are talking about.
NTU: my guide is more more suited for beginners
AstralStorm: uh, yeah
AstralStorm: sneakers: not really, they're just more detailed
NTU: detailed = dumbed down
AstralStorm: just skip the branch step
AstralStorm: to use the master branch
sneakers: AstralStorm, What is the next step?
AstralStorm: it's mentioned in the guide
AstralStorm: it's that configure thing :)
AstralStorm: or rather, ./autogen.sh :)
AstralStorm: btw, which kernel version do you use?
NTU: sneakers, uname -r
sneakers: NTU, 2.6.32.11-99.fc12.x86_64
AstralStorm: not too fresh
NTU: heh
AstralStorm: should be enough though
AstralStorm: 2.6.33 would be better, and 2.6.34-rc better still
NTU: 2.6.32 had quirks in ubuntu 10.04 lol. i laughed my ass off when i had LVDS quirks on my dad's laptop out of the box. hehe!! damn ubuntu devs cant get anything right