Nightwulf|work: hi all
dileX_: hi Nightwulf|work
Nightwulf|work: hi dileX
MrCooper: adamk_: Workstation (version 7, not sure about 6.5) works with the i965 driver, so the problem probably isn't (just) the GLSL compiler
MrCooper: anyway it's a rather demanding app, r300g may still have some way to go before it'll work
Dr_Jakob: I also think that you need a r5xx card.
MrCooper: Dr_Jakob: why?
Dr_Jakob: MrCooper: npot
Dr_Jakob: and a bunch of other glsl features.
MrCooper: npot is emulated by the driver (except for mipmapping)
MrCooper: for GLSL it probably depends on the actual shaders, should be possible except for demanding guest apps
Dr_Jakob: hmm yeah
Dr_Jakob: yeah I forgot that r300g lies about what it supports.
MrCooper: not so much lie as emulate as well as possible
adamk_: There's an update to workstation 6.5 that I'm grabbing now.
adamk_: I also found something in vmware..log that might be more helpful regarding the VMGL panic.
adamk_: I have bigger fish to fry. I haven't used my X850 in this machine for a while, but since I decided to test something with r300g, this washed out color problem has become a real issue.
airlied: adamk_: try d-r-t
airlied: I think Alex fixed a few issues in that area
airlied: or even Linus tree
MrCooper: adamk_: you could also try a newer version of Player
adamk_: MrCooper: Oh, good point.
amarsh04: has anyone experienced loss of vt1-vt7 (text consoles) with kernel 2.6.33 compared with kernel 2.6.32? radeon 9200se and kernel modesetting here
airlied: amarsh04: wierd it should be quite hard to lose vts with kms
MrCooper: amarsh04: fbcon is built in or the module loaded? And there's no other conflicting framebuffer device?
adamk_: http://thorn.visualtech.com/good.jpg http://thorn.visualtech.com/bad.jpg
adamk_: Aside from good.jpg being blurrier, bad.jpg is washed out.
amarsh04: ok, I'll check on that MrCooper
adamk_: You can see the difference in the icons, primary. In bad.jpg, you can't even make out the check.
adamk_: These are two monitors, both hooked up to the x850. One of the DVI ports is always washed out like that (it doesn't matter which monitor is connected to the port, either, it's always that port).
adamk_: Only happens with KMS enabled.
airlied: adamk_: any DVi->VGA?
airlied: or all pure DVI?
adamk_: Yes, both are using DVI->VGA unfortunately, but I've tried swapping the adapters.
adamk_: I think...
adamk_: Let me double check.
airlied: adamk_: okay try 2.6.34-rc5 if you can
airlied: adamk_: probably http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=commit;h=643acacf02679befd0f98ac3c5fecb805f1c9548
adamk_: Ahhh, cool.
Tommeh: adamk_: I've had a very similar problem with washed-out images.
Tommeh: But only on my secondary screen, when compositing is enabled.
airlied: should also be in stable 2.6.33 soon
Tommeh: I'm yet to try 2.6.34-rc5 yet though so hopefully it's fixed as per airlied's commit.
adamk_: Yeah, even swapping the DVI->VGA thingy didn't help.
adamk_: I'll grab -rc5 now.
Tommeh: Mine are both VGA, sadly. Damn Dell :(
amarsh04: ok, I'm not seeing fbcon loaded in Xorg.0.log
airlied: Tommeh: same problem then
airlied: at least if its with pre r500 card
Tommeh: Strange that it was only the 2nd monitor.
Tommeh: Oo err, this is R600 :)
MrCooper: amarsh04: it's in the kernel
amarsh04: ok MrCooper
Tommeh: airlied: weirdly, the combination has to be KMS+compositing. Without compositing, it's fine. Without KMS, it's fine. And as above, it's then only the 2nd screen.
Tommeh: Both connect via VGA, identical screens.
zhasha: airlied: you're here! :D I'm not sure if you've addressed https://bugs.freedesktop.org/show_bug.cgi?id=25386 yet, but it should probably be a blocker for F13
airlied: zhasha: yeah we should look at that again, now is bed time though ;-)
zhasha: alright, thanks
amarsh04: ok, I have CONFIG_FB_RADEON=m in my kernel config, and can modprobe radeonfb successfully, but vt1-vt6 are still not showing anything.
amarsh04: other configs are:
airlied: amarsh04: don't use CONFIG_FB_RADEON
amarsh04: ok thanks airlied
airlied: its not kms compatible
amarsh04: ok, I've modprobe -r radeonfb
amarsh04: I see the getty's for tty1..tty6
MrCooper: you need to prevent it from loading in the first place
mjt: or just remove it
mjt: from kernel
amarsh04: ok, I don't think it was loading in the first place, but I could blacklist it
amarsh04: radeonfb is already in /etc/modprobe.d/fbdev-blacklist.conf
amarsh04: let me try my amd64 machine and see if it has the same problem with missing vt1-vt6
amarsh04: ok, the amd64 is worse - I can switch away from vt7 to a dark vt1-vt6, but I can't switch back to vt7
amarsh04: both the i386 and amd64 with kernel 2.6.33 list in dmesg: fb0: radeondrmfb frame buffer device
amarsh04: is this normal?
amarsh04: ok, I'm just wondering if there is a set of information I could provide that might shed some clues as to why I can't use vt1-vt6 under kernel 2.6.33
adamk_: vmplayer 3.0.1 disables 3D acceleration, telling me my video card doesn't support it.
adamk_: Despite loading the r300g driver.
adamk_: Well that's no good.
adamk_: Well, it's better than vmware hanging, I guess, but not exactly what I was looking for :-)
adamk_: http://slexy.org/view/s2m8puXPXG for anyone interested.
adamk_: I guess the requirements changed from vmware player 2 to 3 (and workstation 6.5 to 7)
adamk_: airlied: -rc5 fixed it.
amarsh04: the only other clue I am getting about blank vt1-vt6 in kernel 2.6.33 is that in kernel 2.6.33 I'm getting dmesg notification about kernel modesetting very late in the boot process
amarsh04: [ 143.506925] [drm] radeon kernel modesetting enabled.
amarsh04: [ 143.537358] [drm] radeon: Initializing kernel modesetting.
adamk_: OK, so time to update my e-mail to mesa3d-dev.
amarsh04: ok, if I'm getting the error
amarsh04: [dri] If using kernel modesetting, make sure your module is
amarsh04: [dri] loaded prior to starting X, and that this driver was built
amarsh04: [dri] with support for KMS.
amarsh04: [dri] Disabling DRI.
amarsh04: what module must be loaded prior to starting x (radeon, drm_kms_helper...?)
AndrewR: amarsh04, for me modprobe radeon before startx works .....
amarsh04: ok, thanks AndrewR
amarsh04: I'm adding radeon to /etc/modules and running update-initramfs
adamk_: But, with 2.6.34-rc5, three of the vmware modules won't compile.
adamk_: I just can't seem to catch a break :-)
adamk_: Shoot, when I updated vmware I got rid of the patched module source.
oga: argh. turns out I am *not* the only one seeing the odd stretched zaphod thing on 6.13.0 (with my zaphod patches)
oga: there is still something very wrong going on there.
amarsh04: ok, adding module radeon to /etc/modules then running update-initramfs then rebooting solved the missing vt1-vt6 problem
adamk_: Well now even an updated version of vmware workstation 6.5 will disable 3D acceleration.
adamk_: Ahhh, it requires S3TC.
Dr_Jakob: adamk_: you need to install the libdxt library.
Dr_Jakob: for some distros you can get it from sort of none free package.
adamk_: Dr_Jakob: I have it installed, though :-) /usr/lib/libtxc_dxtn.so
adamk_: LIBGL_DRIVERS_DIR=/home/adamk/saved/source/mesa/lib64/gallium glxinfo | grep s3tc
adamk_: Doesn't turn up anything.
Dr_Jakob: adamk_: its called inside of gallium DXT*
Dr_Jakob: its called DXT* inside of gallium
Dr_Jakob: adamk_: also which branch are you using? I think r300g only grew s3tc/dxt support recently.
adamk_: Sorry, I'm just not understanding you.
adamk_: I'm using mesa from git as of this morning.
adamk_: Built with --enable-gallium-radeon
adamk_: I need a specific branch?
Dr_Jakob: What I mean is that the 7.8 branch probably doesn't have DXT while master does.
adamk_: OK, but I'm using master and still not seeing s3tc support, with the external libtxc_dxtn.so library installed.
Dr_Jakob: hmm, you have comfirmed with glxinfo?
adamk_: Yeah: http://pastebin.com/F4C4SBEk
adamk_: Something strange.
adamk_: OK, no, that's right. It's not showing up. Here's the full glxinfo output: http://pastebin.com/bSkFTdxH
adamk_: In driconf, I am not disabling S3TC compression either.
Dr_Jakob: export LIBGL_DEBUG=true
adamk_: With LIBGL_DEBUG=true all I get for debugging is:
adamk_: libGL: Can't open configuration file /etc/drirc: No such file or directory.
adamk_: For verbose, I get that *plus* the line showing 'libGL: OpenDriver' (pointing to the r300g driver)
adamk_: Maybe my source tree is hosed.
adamk_: Receiving objects: 1%
adamk_: God this is slow.
MaDMaLKaV: hi all
MaDMaLKaV: is there any literature about how all the graphic driver system in linux works? I know most of it was recently refactored to move most of the parts to the kernel, but this is quite a jungle for someone not used to this. I will like to try to learn some stuff but I'm not sure where to start.
Tommeh: MaDMaLKaV: it took me *ages* to understand it, and I still don't get everything :)
Tommeh: I recently realised that some things are best left to developers.
MaDMaLKaV: I'm trying to figure what well be needed to do to get the r500 driver and modify it to support the Xbox 360 gpu so we can get a proper graphic driver . I'm aware that it will be probably far away for my skills, but at least I want to know why :)
MaDMaLKaV: I also thought I will be unable to get the old X framebuffer driver we had to work on latest X.Org and I finally got it working. It took me like 100x the time an skilled coder will require, but it was so nice to get it working, and I learned some interesting things in the process :)
Tommeh: You're likely way above my skill level then. ;)
MaDMaLKaV: if we put some effort, sooner or later we will improve dude :)
Tommeh: Programming's not really my thing to be fair, but I do agree. :)
Tommeh: I also fully support what you're doing! :)
MrCooper: MaDMaLKaV: you should probably start by making the kernel radeon driver work with kernel modesetting (KMS)
MaDMaLKaV: I think I better start getting to know what it is exactly KMS , how does it works, and trying to understand a little how the kernel driver works . As I said, I'm lagging behind a lot of knowledge before I can even think to start doing *something*
MaDMaLKaV: -googling KMS on google right now, thanks for the tip by the way :)-
MrCooper: np, have fun :)
MaDMaLKaV: for what I have seen, there is some good books about device drivers in linux, but they normally doesn't talk about graphic drivers beyond basic framebuffer and also they do not cover all this stuff that got recently into the kernel
mokoloko: Isn't 360's gpu more like r600? I mean with the unified shaders and all that.
Thunderbird: thought it was a r300 actually
MaDMaLKaV: if I understood correctly, it is a r500 derivative with some functions that later appeared on the r600 series + a framebuffer/modesetting chip
mokoloko: something like that probably
JonimusP: is there a set of PM patches that will apply against 2.6.34rc yet?
Ronis_BR: hi all
Ronis_BR: I use xf86-video-ati, libdrm and mesa from git, but kernel stable. Is it a problem?
Ronis_BR: where I'll notice more improvements: drm updates or xf86-video-ati updates?
JonimusP: I know agd5f was going to work on rebasing his current patches against 2.6.34 but I wasn't sure if he had done so or not yet
Tommeh: Ronis_BR: there are a _lot_ of updates in 2.6.34-rc* :)
Tommeh: drm updates, I mean.
JonimusP: I need 2.6.34 for a ACPI fix so I can actually suspend but I need PM more :(
Tommeh: Though I think I saw airlied say this morning that they might be rolling them back into 2.6.33 stable.
JonimusP: but that doesn't help me :(
Tommeh: Was @Ronis, sorry :)
JonimusP: yeah I know
Tommeh: Ronis_BR: also, looking at my scroll back - I think it was just one fix. It would be wise to test 2.6.34-rc5 if you can :)
Ronis_BR: Tommeh: hum
Ronis_BR: Tommeh: radeon is cool here with 2.6.33, I can hibernate, I can use compositing, the only problem is that some OGL software are really slow
Ronis_BR: does anyone have permission to edit http://www.x.org/wiki/RadeonProgram?action=edit?
Tommeh: Ronis_BR: is that with KMS?
Ronis_BR: Tommeh: with and without KMS
Tommeh: Ah :)
Ronis_BR: Tommeh: some software has poooor performance with radeon
Tommeh: I know there are a lot of improvements going into 2.6.34-rc5 .. I've heard in #phoronix that the performance differences can be massive
Tommeh: But it's #phoronix, so take it with a pinch of salt.
JonimusP: Yeah, I can hibernate most of the time with 2.6.33 but I need 2.6.34 to suspend with this dell laptop and its current BIOS.
Tommeh: ^ I knew I had that in my #phoronix scroll-back
Ronis_BR: I think I'll test 2.6.34_rc5 this weekend
Tommeh: Ronis_BR: I'm waiting until I've updated to Ubuntu 10.04 before doing so, as I do most things on a PPA/package basis
Ronis_BR: Tommeh: It think I'll install it now :D
Ronis_BR: Tommeh: why don't you remove ubuntu? :D kidding
Tommeh: Work machine. Can't muck about too much ;)
Ronis_BR: Tommeh: :D
Ronis_BR: Tommeh: nevermind me and ubuntu have some problems :D
Tommeh: What do you use now? :)
Ronis_BR: Tommeh: gentoo
JonimusP: Tommeh: you have sudo on a work machine, nice
Tommeh: JonimusP: I would have had Vista if I didn't install Ubuntu.
Tommeh: The difference is that I need something which I don't have to spend hours fiddling with. I've got servers that need my time :)
Tommeh: I used to manage a whole bunch of Gentoo servers though - never again ;)
Tommeh: And to be fair, the only real reason I use Ubuntu instead of something like Fedora is familiarity.
Ronis_BR: Tommeh: and that's my problem. I have been using gentoo since 2003, so I'm used to a very clean operational system. Ubuntu is very fine for begginers and people who just want to use, it is amayzing because more people are trying linux, but you have to agree with me that Ubuntu is the definition of mess just behind windows :)
Tommeh: Ronis_BR: I'm blissfully ignorant about the 'mess'.
Tommeh: On a server installation, it is messy - I agree.
Tommeh: But for my desktop machine I like to plug in hardware and have it 'just work' etc.
Ronis_BR: Tommeh: like, it uses debian, but it uses packages from stable, testing, unstable, experimental all together. You just can't update system, it always generates problems, so you have to format and reinstall
Ronis_BR: Tommeh: but it is just what I'm used to. I'm not saying it isn't right
Tommeh: Ronis_BR: it's fair enough - I have enough respect for anyone using Gentoo :)
Ronis_BR: Tommeh: here, it just work, but it have taken about 4 years :D
Tommeh: But upgrades in Ubuntu are one of the things I like - they only fail because I break them doing something silly.
JonimusP: I'm an Arch Linux user and yeah even with how simplified they try and keep it you can see how much of a Mess Linux in general can be, I tried setting my mom up with Ubuntu and getting it to play nice with a custom kernel was a pain
Ronis_BR: Tommeh: but, at other point I just love Ubuntu, because people are discovering Open Source world
Tommeh: JonimusP: why did your mum need a custom kernel? :/
Ronis_BR: gooood question
JonimusP: Tommeh: she had a radeon card :P
yoshi314: Ronis_BR: not quite as i had imagined though
Ronis_BR: gooood answer :D
Tommeh: Ronis_BR: it's great - my girlfriend uses Ubuntu and loves it. :)
Tommeh: JonimusP: hah.. Fair enough.
Ronis_BR: Tommeh: yeah, it is really something we have always needed
JonimusP: Her desktop has a HD3870 in it so yeah, its a backup for my Desktop at LANs.
Tommeh: Though if you want to use custom kernels in Ubuntu, just use the kernel PPA
Tommeh: No need to compile anything.
Ronis_BR: Tommeh: IMO, it was the very first desktop-only linux
Tommeh: Ronis_BR: I agree. :)
Ronis_BR: Tommeh: and then fedora, and others started to look this way
Tommeh: Because the focus is on the user, not geek.
yoshi314: what sucks about ubuntu is the developer community
Ronis_BR: Tommeh: even gentoo became to be more "desktop" :D
JonimusP: Tommeh: I tried to install it form a PPA but I couldn't get it to add the Kernel PPA or play well with me installing the package with dpkg...
yoshi314: there is nearly no mention of any other linux distro on their planet
Ronis_BR: Tommeh: yup
yoshi314: so it's pretty obvious that people will think ubuntu=linux
Tommeh: JonimusP: no, I couldn't get it to 'add' the PPA - I just downloaded the .deb's and installed with the package manager.
yoshi314: esp. the new ones
Ronis_BR: yoshi314: what? I didn't get it
Tommeh: Thinking back, it won't automatically update your kernels anyway.. They're all 'different' packages.
JonimusP: Tommeh: I tired that but when I ran an update it wanted to remove it.
Tommeh: Update, or upgrade?
Tommeh: Weird. I've never had this problem - they're separate packages.
yoshi314: Ronis_BR: it's strange that people involved in ubuntu are so careful not to mention other distributions on their blogs
JonimusP: thinks its stupid that there is a difference between those.
Tommeh: I run 2.6.34-rc1 and 2.6.31-21 side by side. The former is from the PPA the latter is from Ubuntu's repositories.
JonimusP: Tommeh: could it maybe be that I used dpkg to install rather than some higher level wrapper.
yoshi314: it's hard to even see them mention debian ;)
Tommeh: JonimusP: it's not stupid. Update updates the current system, Upgrade upgrades the current system to a newer system.
Tommeh: Seems more sensible than RHEL using 'update' and 'upgrade' interchangeably!
JonimusP: Tommeh: I'm just used to Arch's Rolling release system.
Tommeh: Ah, fair enough. :)
Tommeh: Between 'releases' though the distinction between update and upgrade is a nice one.
JonimusP: There is no such thing as a System upgrade.
Tommeh: Arch is forked from Gentoo, yes?
Ronis_BR: yoshi314: this is bad
Ronis_BR: Tommeh: no
yoshi314: Tommeh: not really
JonimusP: Tommeh: no not at all
yoshi314: i'd say arch is based on LFS
Tommeh: Oh, my mistake. :)
yoshi314: because that's what they supposedly started with
KotH: on LFS?
Ronis_BR: Tommeh: Arch is rolling release too but uses binary software
KotH: so it must be masochists heaven
yoshi314: before pacman got usable
Ronis_BR: Tommeh: gentoo is source-based
Tommeh: Ronis_BR: thanks for the clarification :)
yoshi314: arch is a decent distro because it doesn't pack too much stuff, and it's easy to manage manually built software there
yoshi314: and also, there are no fancy configurators
Ronis_BR: KotH: no, it has a wonderful package manager, that make all the job for you
yoshi314: dpkg-reconfigure drives me insane
JonimusP: Yeah basically a guy used LFS to make his own distro, and made a package manager to go with it
JonimusP: makepkg+pacman is heaven
Ronis_BR: KotH: a package manager that allows you do whatever you want (like downgrading :D)
Ronis_BR: JonimusP: makepkg+pacmas == portage?!
yoshi314: what sucks about arch is that there is usually one package version available
yoshi314: on mirrors
adamk_: Dr_Jakob: FYI, even after pulling a fresh copy of the mesa repo from git, I do not get the s3tc extensions showing up in glxinfo.
Ronis_BR: yoshi314: yes, that's why I'm on gentoo now :)
yoshi314: you have to keep older ones around, or grab older PKGBUILD's off svn
Ronis_BR: yoshi314: we have many versions of packages to choose, we can combine stable/testing/unstable branches they way we want
JonimusP: yoshi314: yes that is the point of the Arch Rollback Machine, a repo that archives old packages
adamk_: Dr_Jakob: I can at least continue testing vmware workstation using a suggestion from Brian on the mesa3d-dev mailing list, but I wanted to let you know that something is definitely up.
yoshi314: Janhouse: oh there is one such repo? i have to check it out
Ronis_BR: brb, I'll boot 22.214.171.124
JonimusP: yoshi314: just /msg phirk !arm
JonimusP: phirk is #archlinux's bot btw
Dr_Jakob: adamk_: I'll try to look into this week.
Dr_Jakob: look into it this week*
Dr_Jakob: from what I can see in the code, if the driver reports that it supports DTX formats we should enable it.
MrCooper: adamk_: it's in the extension list here with r300g; note that vmware-vmx may not pick up the driver you think
Dr_Jakob: Tho glxinfo also said it wasn't enabled...
MrCooper: maybe it still needs libtxc_dxtn.so?
adamk_: MrCooper: I'm setting LIBGL_DRIVERS_DIR, and LIBGL_DEBUG=verbose clearly shows it loading the r300g driver.
adamk_: And I have libtxc_dxtn.so installed in /usr/lib
MrCooper: adamk_: where does it show that? vmware.log?
Dr_Jakob: I think he is trying to solwe the first problem, it doesn't get enabled in glxinfo.
adamk_: MrCooper: both in vmware.log and on the terminal. vmware.log shows the glxinfo output and it's clearly using r300g.
adamk_: Yeah, I assume if it shows in glxinfo, then vmware will pick it up.
adamk_: I'm sorry guys.
adamk_: User error.
adamk_: I had the 32-bit version installed correctly, oddly enough, but the 64-bit version was missing from /usr/lib64/
Dr_Jakob: mesa should still warn about it missing..
adamk_: Well it definitely didn't do that: http://paste2.org/p/785581
adamk_: That's with the 64-bit library missing.
adamk_: With it in place: http://paste2.org/p/785589
Dr_Jakob: it reports it now..
Dr_Jakob: ah it doesn't warn on none debug builds...
Dr_Jakob: adamk_: set "MESA_DEBUG=true"
Dr_Jakob: that will print the info.
adamk_: Ahhh... Too many variables.
adamk_: MESA_GLSL, LIBGL_DEBUG, MESA_DEBUG :-)
JonimusP: lol I can get ~8FPS in glxgears with SSH X Forwarding :P
Dr_Jakob: well thinking about it LIBGL_DEBUG is for libGL, while MESA_DEBUG is for the mesa DDK.
adamk_: Yeah, Mesa complaints.
adamk_: complains, even.
Dr_Jakob: so does VMware accept the driver now?
adamk_: Yay... vmware starts up with 3D acceleration enabled with r300g now.
JonimusP: nice adamk_
Dr_Jakob: I guess so ;-)
adamk_: And then crashes when running the display tests in dxdiag :-)
Dr_Jakob: adamk_: the shader assert?
adamk_: Something different from my e-mail to mesa3d-dev
Dr_Jakob: well report any errors please.
Dr_Jakob: to mesa-dev or bugzilla
adamk_: vmware-vmx: state_tracker/st_manager.c:483: st_context_notify_invalid_framebuffer: Assertion `stfb && stfb->iface == stfbi' failed.
adamk_: Yeah, gonna reply to Brian now.
adamk_: Dr_Jakob: Any thoughts before I reply?
Dr_Jakob: Brian probably wont know what to do with that error, its all new code.
Dr_Jakob: I'll try to look into it this week
adamk_: ALright, well I'll send the update anyway, with your comment about hopefully looking into this week.
Dr_Jakob: Naw, leave that bit out, if somebody feels they want to fix it they can do it :)
adamk_: That works for me, too :-)
JonimusP: http://tinyurl.com/y5dl8rk :P so ATI's do infact scale better with crossfire, but once you reach 4GPU's it makes little to no difference :P
adamk_: So this is doom3 with or without s3tc with r300g: http://thorn.visualtech.com/doom3.jpg :-)
Ronis_BR: Tommeh: there isn't fbcondecor for 2.6.34 yet :(
Tommeh: Ronis_BR: I don't know what this is? :(
adamk_: actually, with s3tc it's actually worse.
adamk_: Too much actualizing.
Ronis_BR: Tommeh: something to make consoles (tty1, tty2,...) decorated :D
soreau: adamk: With classic it should render the planet correctly, yes?
adamk_: soreau: Yes, it works quite good with classic, in fact.
adamk_: ANd here's compiz: http://thorn.visualtech.com/comiz.jpg I get lots of those little white dots all over the place :-)
adamk_: Sorry: http://thorn.visualtech.com/compiz.jpg
soreau: adamk_: http://thorn.visualtech.com/comiz.jpg <-- 404
Dr_Jakob: what the, it looks almost like you managed to caputre broken pixels on a LCD..
adamk_: Yeah, that's actually the first thing that jumped to mind :-)
adamk_: It only happens to textures that are being manipulated.
adamk_: So if I don't move, resize, rotate, etc, and just let windows sit there, it's fine :-)
JonimusP: well my Torcs FPS seems to have gotten better and I was actually able to not place last on one race :P
Ronis_BR: Tommeh: I haven't seem much improvement
Ronis_BR: Tommeh: but kernel 2.6.34 correct a LED problem here!!!
Tommeh: LED? Heh
Ronis_BR: Tommeh: sound led, it should be orange when muted
Ronis_BR: Tommeh: now, the only problem is with wireless button, that when i press it turn off bluetooth :D
Tommeh: Ahh, it's a laptop. :)
Ronis_BR: Tommeh: yup
Wizzup: Woo, QuantZ seems to work much better now with the radeon driver. :-) (Last time I tested was a few months back)
soreau: I get a ton of these with gallium using cairo-dock when hovering over the dock "Mesa: User error: GL_INVALID_ENUM in glEnable(0x1702)"
soreau: Does this indicate a problem with cairo-dock?
MostAwesomeDude: Yes. User errors are errors with the GL-enabled app.
soreau: that is what I thought, thanks
lordheavy: nice :-) some work in progress :-p http://cgit.freedesktop.org/~glisse/mesa/
Obscene_CNN: my r600/r700 mesa patch with inlined assembly goodness for X86-64 http://www.phoronix.com/forums/showpost.php?s=7c41d5eedc37b355a182d81bcf06f24f&p=123402&postcount=58
Obscene_CNN: now to work on getting a few small portions that are mergeable into patches for the dri_dev mailing list :)
Ronis_BR: Obscene_CNN: I'm just curious, what it will do?
JonimusP: Obscene_CNN: inline asm is unlikely to be merged due to mesa being on other platforms as well unless your patches include checks for build arch
Ronis_BR: wonder what is an inline asm...
MostAwesomeDude: Icky is what it is.
MostAwesomeDude: And there's other parts of r600 that could be optimized instead, I'm sure.
MostAwesomeDude: But whatever.
psyke83: hi folks. I'm using a R100 card with the latest drivers/Xorg/KMS enabled, and I noticed that in my /proc/dri/0/gem_objects I *always* have 0 pinned/pin bytes/gtt bytes/gtt total. Is this normal or does it warrant a bug report?
psyke83: (object count and object bytes is reporting values)
soreau: Is this causing a problem for you elsewhere?
airlied: I don't think we update those numbers
airlied: with radeon
psyke83: soreau: perhaps indirectly, as I'm trying to troubleshoot a GEM memory leak. The object count/bytes increase even when applications are closed. See: https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/565981
psyke83: airlied: ok, thanks. Can you take a quick look of my object count here? https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/565981/comments/33
psyke83: that report is testing against the older Xorg/radeon driver/2.6.32 kernel, but I'm also testing against the latest versions and kernel 2.6.32-rc5 and see the same continual object count growth
airlied: nothing to do with me, I'm sure canonical are working on it
airlied: its a server bug
psyke83: airlied: right, but the output does indicate a leak (even if caused by the X server and not from a driver)?
airlied: I'll aait the Ubuntu guys analysis, since we haven't seen much reports of this from any other distro
psyke83: sure, I understand... thanks
amarsh04: added the following to /etc/modprobe.d/local to load radeon in time for everything that needed it: alias pci:v00001002d00005964* radeon