Radeon IRC Logs For 2008-5-20

airlied: agd5f: I get gears crash unless I run xv app after a poweroff
timofonic: airlied, ohhh thanks for the explanation 🙂
airlied: MostAwesomeDude: so still missing some regs somewhere.
airlied: if I clean boot, gears hangs.
airlied: I have to do some xv before it comes good.
MostAwesomeDude: airlied: Hmm.
airlied: I’ve found some other tirival bits but nothing with a big sign pointing to a bug yet.
airlied: agd5f: any ideas on what a user color is?
airlied: agd5f: and any more info on what 2180 reg does..
airlied: MostAwesomeDude: so the tex/output stuff not sure what you did there.
airlied: I see you adding one to counter, but how does that make the out work?
MostAwesomeDude: airlied: Look at emit_tex.
MostAwesomeDude: I had to split it since counter is passed by value and I didn’t feel like messing around with references.
MostAwesomeDude: But emit_tex now takes care of putting to the output fifo by adding an OUT if necessary.
MostAwesomeDude: So we don’t need that hackish thing at the end of the parser, and we can have multiple TEX-OUT pairs.
airlied: MostAwesomeDude: ah cool.. didn’t look too deeply 🙂
bridgman: airlied: ping
MostAwesomeDude: POOONG!
MostAwesomeDude: Oh wait, I’m not airlied.
airlied: bridgman: pong sorta.
airlied: needs to do dishes 🙂
bridgman: Morning/afternoon; looks like there may be an error in the 5xx docs re: user color
bridgman: front of the doc says only 1 user color and enable reg’s gone but register section
bridgman: still talks about user color 0/1 and enable bits, will ask at work
bridgman: looking up 2180 now
bridgman: ahh, I see – it is in the 5xx doc, you’re asking “why is it inconsistent ?” 😉
bridgman: can’t do that from home ;(
MostAwesomeDude: Sleep time. Night.
bridgman: I was about to say thank you for the shorter nick but I still have to type in sleeeeeep ?
MrCooper: bridgman: no, you only have to type ‘mad’ and hit the Tab key 😉
airlied: bridgman: cool. I actually just think I’m avoiding the real problem, by finding lots of other small ones 🙂
airlied: I did some secondary color traces with revenge on fglrx and they didn’t make much sense.
airlied: not sure whether to blame revenge or not 🙂
airlied: bridgman: actually what a user color is vs a color has confused me 🙂
airlied: granted I might just need to read vp/fp specs or something..
bridgman: I didn’t see it properly covered in the specs; looking through HW internals now
bridgman: I bet poor Alex needs to do a v1.4 😉
bridgman: OK, looks like :
bridgman: color0-color7 are for the colors specified in GLMaterial()
bridgman: user color 0 is for the openGL user color state
bridgman: user color 1 is for the openGL “current color state”
bridgman: make any sense ?
bridgman: 2180 controls the 10:8 muxing of those colors into the 8 available vertex colors
airlied: not sure where this gels with glSecondaryColorEXT
airlied: bridgman: glMaterial is the VAP level, the secondary color is at the FP stage.
airlied: anyway time for be to log off.. can’t use computer too close to bed time 🙂
bridgman: that’s good, right ?
bridgman: ok, g’night
airlied: yeah I think it means I can ignore the VAP color stuff 🙂
airlied: g’night.
glisse: bridgman: btw can we have insigh on r700 memory functionalities ?
bridgman: glisse: trying to find a suitable way to give you insight re: the difference between
bridgman: unreleased r6xx memory functionality and unreleased r7xx memory functionality, when I’m
bridgman: not sure we are going to be releasing either of them 😉
glisse: 🙂
bridgman: that said, I do want to give some guidance re: current memory manager discussions
glisse: bridgman: in 1 year linux will hold 90% of the market so you will have to document this 😉
bridgman: just haven’t figured out how (or fully understood either 6xx or 7xx memory mgmt yet ;))
bridgman: not really, we don’t use it for windows or linux today
bridgman: it’s just stuff in the chip we aren’t using yet
bridgman: and probably don’t test
glisse: you put too much things in your chip 🙂
bridgman: hey, you’re shooting a moving target and it takes 3 years for the bullet to arrive
glisse: yes that’s true, but i would be too easy if the targest wasn’t moving 🙂
glisse: is wondering what is more painfull gpu programming or heavy bandwidth udp program debugging…
bridgman: life is pain, get used to it
glisse: i am likely one the few people on earth which daily experience the Murphy law 😉
bridgman: true, you do tend to work on widely different things at the same time 😉
bridgman: which is like saying “yo, Mr. Murphy, over here !!!”
glisse: 🙂
bridgman: Anyways, time to go to work…
bridgman: over to agd5f… wake up !
icewaterman: hi, 3d effects on gnome do not work well with 3d applications, right?
glisse: icewaterman: yes
glisse: bottom line no dri2 -> no luck
icewaterman: ok
icewaterman: it is more of an xorg question, but how do i disable ctrl+alt+ +|- to change the resolutioni?
icewaterman: -i
icewaterman: because it screws with my display and i cannot get back without logging out (it will not go back to 1024×768)
icewaterman: uhm back to 1280×1024
agd5f: airlied: what bridgman said 🙂
agd5f: I guess I need to wake up earlier 🙂
z3ro_: agd5f: you might be able to answer this… what would make the type3 bitblt packet lockup? (actually I’m not quite sure this is a full hard lockup since I’m running on a single box atm)
z3ro_: I suppose the only thing is invalid addresses?
icewaterman: glisse: i take it, dri2 is not gonna be stable anytime soon
glisse: icewaterman: dri2 need a memory manager
glisse: icewaterman: so far for radeon only hack i have done exist as memory manager
glisse: so dri2 won’t happen for radeon until there is a memory manager
agd5f: z3ro_: yeah invalid addresses I would guess
z3ro_: it looks like r300_demo is detecting the FB as 0xd0000000, but perhaps this should be 0xd7ffd000?
z3ro_: (II) RADEON(0): MC_FB_LOCATION : 0xd7ffd000 0xd7ffd000
glisse: z3ro_: no that fine
glisse: d0000000 is start of fb
glisse: d7ffffff is end
z3ro_: I see
glisse: z3ro_: did you adjusted display_width ?
z3ro_: it’s set as: display_width=(get_int(RADEON_CRTC_PITCH) & 0x3ff)*8;
z3ro_: display_width=1728 (actually that looks a bit weird…)
z3ro_: this monitor is 1680×1050
glisse: z3ro_: could be padding
z3ro_: hmm
z3ro_: src_pitch = 6912
z3ro_: dst_pitch = 6912
z3ro_: src_offset = 0
z3ro_: dst_offset = 0
z3ro_: look okay?
glisse: yes
glisse: z3ro_: you disabled aiglx ?
glisse: you should disable all 3d while playing with r300 demo
z3ro_: aiglx is still enabled, but it’s using the software path afaik.
glisse: z3ro_: you better disable it
glisse: you never know
z3ro_: Option “AIGLX” False?
glisse: iirc yes
z3ro_: ok, got it disabled (was a ServerFlags option)
z3ro_: here goes…
glisse: it works ?
z3ro: well, that didn’t work. I also tried logging the output of r300_demo (maybe catch an error) but there is nothing interesting.
tulcod: z3ro: and Option “Composite” “false” too?
tulcod: in serverflags
tulcod: or something like that
timofonic: Anyone here with experience on uvesafb under radeon hardware? http://rafb.net/p/DAtzDw20.html
funda3: what parameters are you passing to the kernel on startup?
otaylor: agd5f: Shouldn’t it be (!((op == Src || op == Clear) && (PICT_FORMAT_A(dst) == 0)) (or some de morgan’s rule permutation of that) ?
MrCooper: I was just about to point that out as well 🙂
agd5f: otaylor: oh yeah, I think so
agd5f: fixed
Borgvall: hi
MostAwesomeDude: Borgvall: Howdy.
Borgvall: i am currently trying the r500-support mesa branch
MostAwesomeDude: Oh, cool.
MostAwesomeDude: How are things?
Borgvall: but while compiling i am get this error message
Borgvall: radeon_screen.c: In function ‘radeonCreateScreen’:
Borgvall: radeon_screen.c:843: error: ‘RADEON_PARAM_FB_LOCATION’ undeclared (first use in this function)
Borgvall: radeon_screen.c:843: error: (Each undeclared identifier is reported only once
Borgvall: radeon_screen.c:843: error: for each function it appears in.)
Borgvall: any suggests?
MostAwesomeDude: Borgvall: You have DRM from git, right?
Borgvall: kernel module yes, libdrm 2.3.0
MostAwesomeDude: Did you install the libdrm and libdrm development libs from git?
MostAwesomeDude: Or just the kernel modules?
Borgvall: just the kernel modules
MostAwesomeDude: Alright.
MostAwesomeDude: You should “make && make install” in your DRM directory; you need the libdrm from git as well.
MostAwesomeDude: That’s where RADEON_PARAM_FB_LOCATION and various other things are defined.
Borgvall: ok i try
Borgvall: thanks now it compiles..
Borgvall: …however glxgears said Error: couldn’t get an RGB, Double-buffered visual
MostAwesomeDude: Borgvall: Hmm.
MostAwesomeDude: Could you run “LIBGL_DEBUG=verbose glxinfo”?
MrCooper: AIGLX must be enabled with xserver Git
glisse: Borgvall: likely X & mesa GL missmatch
MostAwesomeDude: glisse: I didn’t know that you had to use the same Mesa for X. I certainly don’t.
glisse: MostAwesomeDude: not what i am saying
glisse: well X, mesa, dri dependency is a nightmare
MostAwesomeDude: Oh, yeah.
glisse: there have been a lot of change to it lately and when visual are missing often its because one part of the chain doesn’t reconize another
Borgvall: sounds bad
glisse: “LIBGL_DEBUG=verbose glxinfo will give a good hint
glisse: at top of the output
glisse: also check AIGLX is enable like MrCooper said
MostAwesomeDude: I mean, could be a missing chip ID, or modules didn’t start right, or something non-X.
Borgvall: ahh it seems i am using wrong env-var
Borgvall: LD_PRELOAD=./lib64/libGL.so.1 DRI_DRIVERS_PATHS=./lib64/ LIBGL_DEBUG=verbose glxinfo
Borgvall: name of display: :0.0
Borgvall: libGL: XF86DRIGetClientDriverName: 5.3.0 r300 (screen 0)
Borgvall: libGL: OpenDriver: trying /usr/lib32/dri/r300_dri.so
Borgvall: libGL error: dlopen /usr/lib32/dri/r300_dri.so failed (/usr/lib32/dri/r300_dri.so: wrong ELF class: ELFCLASS32)
Borgvall: libGL: OpenDriver: trying /usr/lib64/dri/r300_dri.so
Borgvall: libGL error: driver exports no extensions (/usr/lib64/dri/r300_dri.so: undefined symbol: __driDriverExtensions)
Borgvall: Error: couldn’t find RGB GLX visual
glisse: pastbin
Borgvall: DRI_DRIVERS_PATH how is it right?
MrCooper: LIBGL_DRIVERS_PATH
glisse: is your X 32 or 64 ? it sounds like you are asking for trouble
Borgvall: im using x86-64
MostAwesomeDude: Huh.
glisse: guess it’s just some old hardcoded path to 32 that lie around
MostAwesomeDude: Sounds like you’re mixing multilib stuff.
MostAwesomeDude: Did you set the prefix when you configured?
MrCooper: to me it looks like the 64 bit libGL picking up stale drivers because $LIBGL_DRIVERS_PATH isn’t set
MostAwesomeDude: glisse or anyone who knows: In glxinfo, what is “Slow” in the caveat column?
glisse: iirc its a hint we set in visual config in dri driver
glisse: so you can tell your client which mode it should use
MostAwesomeDude: Hmm. Looks like it’s marked because of the accum buffer.
MrCooper: yeah, basically it means some of the visual features can’t be accelerated
Borgvall: after setting the correct drivers path, i got still “could not find RGB GLX visual” error
Borgvall: and 3D HAHAHA Warning *g
Borgvall: should I try to restart x with aiglx enabled?
leio: aiglx is enabled by default if requirements for it are met
MostAwesomeDude: Borgvall: You did restart after installing new DRM and libdrm?
Borgvall: @madude no
MostAwesomeDude: Okay.
MostAwesomeDude: That’s it, right there.
MostAwesomeDude: You need to either restart, or stop X, go down to a console, and reload the DRM modules.
MostAwesomeDude: After that, Mesa should be able to directly render, and glxinfo should stop whining.
Borgvall: ok cu in a minute
MostAwesomeDude: Whoo, temp temps work for r5xx.
MostAwesomeDude: tests/texrect looks better now.
agd5f: MostAwesomeDude: cool!
Borgvall: ok a can run glxgears ONCE after starting X
Borgvall: … then i get the “couldn’t find RGB GLX visual” problem again
MrCooper: Borgvall: oh yeah, there’s a bug report about that on 64 bit
MrCooper: krh’s GLX visual reworks introduced some nice regressions :}
MostAwesomeDude: agd5f: It is cool. I only wish I knew why SIN/COS/SCS still doesn’t work.
MostAwesomeDude: Although it looks pretty psychedelic now, with the swiz fixes.
mcgreg: hi
MostAwesomeDude: Hey.
mcgreg: just switched to EXA .. and tsted with gtkperf, as some ppl stated.. and it also improved speed here. bu the most important part is, it fixed gfx problem I had with XAA with current X-git
mcgreg: hmm compiling latets mesa git failes here
mcgreg: multitex.c:32:21: error: readtex.h: No such file or directory
egbert: agd5f: i’ve just pushed the rmx patch we talked about.
agd5f: egbert: cool. i’ll test it out
egbert: agd5f: thx 🙂
arekm: what’s “rmx”?
egbert: arekm: i don’t know what it stands for but it is the screen scaler that scales the screen to a resolution of an fp for example.
rx__: ATI also applies an RMX (Ratiometric Expansion) Filter that makes text in non-native screen resolutions appear sharper.
rx__: aka.. as egbert said.. scaler
MostAwesomeDude: rx__: Ooh, delicious.
rx__: 🙂
rx__: indeed
MostAwesomeDude: What does it do for us in practice, and how do I get it?
rx__: last time i heard about it.. seemed like it was just automatic
arekm: btw. does this “3D driver claims to not support visual XYZ” mean lack of some support in mesa?
egbert: MostAwesomeDude: if you own a laptop and you want to display a screen of the size other than the native.
MostAwesomeDude: oOH.
egbert: rx__: on older radeons you needed to program the scaling ratio. on later ones (ie >= r5xx) it’s pretty automatic. you just turn it on.
rx__: ah
arekm: or maybe this is already fixed and I just need to find the fix in git?
rx__: i don’t remember seeing a driver option.. so i assumed automatic 🙂
arekm: hm e7ad10587ada74570a13c503aae274bc187910d3 but that’s from 2006 and I already have it
timofonic: Hi
timofonic: I’m having some problems with my radeon X700 under Gentoo Linux with 2.6.25.4 and xorg 7.3. lspci info about my radeon X700 http://rafb.net/p/5DoOsk45.html xorg.conf http://rafb.net/p/soZVqc86.html Xorg.0.log http://rafb.net/p/m74o0L97.html
timofonic: AMD64 dual Opteron 242, 1gb ram and nforce chipset
PSYCHO___: what are those problems?
timofonic: I don’t understand clearly what happens, it tried to show the X but before showing somethng the CRT monitor goes to disconnected mode. It occoured me something similar with uvesafb too
timofonic: Now I’m using vesafb 😛
PSYCHO___: hmm
PSYCHO___: uvesafb should be independent on ga…
PSYCHO___: what version of driver
PSYCHO___: for x
timofonic: PSYCHO___, 4.3.0
timofonic: Xorg 7.3
timofonic: “compiled for 1.4.0.90, module version = 4.3.0
PSYCHO___: nono xf86-video-ati
PSYCHO___: but if you have testing as i can see you have 6.8.0
timofonic: 6.8.0-r1*
timofonic: PSYCHO___, do you need more info to know?
PSYCHO___: nop
PSYCHO___: thinking
timofonic: Thanks for thinking, it’s very difficult for me hehe
PSYCHO___: it hang also uvesafb?
timofonic: It did the same, “disconnected” monitor
timofonic: You know, disconnected mode of monitor
PSYCHO___: it shows nothing
PSYCHO___: and says no input on blabla
timofonic: yep, shit like that 😛
timofonic: It’s quite frustrating, no idea what happens
PSYCHO___: well in log is nothing
timofonic: yep
PSYCHO___: does the monitor allows that resolution?
PSYCHO___: and freq.
PSYCHO___: try smaller one
PSYCHO___: just for testing
eboettcher: finished output detect: 1
eboettcher: (EE) RADEON(0): No connected devices found!
eboettcher: below…
eboettcher: (II) RADEON(0): Detected non-DDC Monitor Type: 0
eboettcher: (II) RADEON(0): Output DVI-0 using initial mode 1024×768
eboettcher: when he does the VT switch, it enables it again
eboettcher: then disables
eboettcher: (WW) RADEON(0): No crtc mode list for crtc 1,continuing with desired mode
eboettcher: after that it gets disabled until the VT switch
eboettcher: try adding a modeline to your monitor section and using it?
timofonic: re, sorry
timofonic: eboettcher, how to generate a modeline?
timofonic: Sorry, I forgot to say a friend was talking to me by phone…
eboettcher: there’s utilities out there, but a google search may quickly give you the modeline you seek
eboettcher: man xorg.conf may tell you how to calculate certain timmings, I don’t recall though.
eboettcher: I do remember seing it in a couple of man pages in the past
timofonic: ok 🙂
rschmidt: you can easily create modelines in GTF format (xorg.conf compatible) using: ‘gtf width height VerticalRefresh’ from the command-line
rschmidt: ie: ‘gtf 1280 720 60’ should give you an xorg.conf compatible modeline for 1280×720 @ 60Hz
eboettcher: rschmidt: I didn’t know you could do that 🙂
eboettcher: last I saw you had to calculate those yourself :/
rschmidt: yes, it’s possible, and it’s very useful 🙂
eboettcher: that’s cool 🙂
rschmidt: this may not be possible in all versions of Linux, I use Fedora mostly, but in most cases, it should be part of some xorg package or other
rschmidt: eboettcher: if you have to calculate a modeline more than once every 6 months, it’s great
rschmidt: if not, you can poke around on google, and things like that, but I do systems integration for a living, and I got really tired of looking for compatible modes when I had a monitor’s specs but not the specific modelines needed
timofonic: rschmidt, what section to add the modeline?
rschmidt: should be under ‘Monitor’, timo
rschmidt: make sure you reference it properly in the screen section… gtf will give you a modeline called “XxY_Refresh”, and you have to use the exact same name under the “Screen” section
rschmidt: so either change the name to something easy to use, or cut & paste it into “Screen”
timofonic: disconnected screen again 🙁
timofonic: My monitor is DVI, not VGA
timofonic: Here is the modified xorg.conf http://rafb.net/p/nDQSp554.html and here is the latest Xorg.0.log http://rafb.net/p/Julq5e92.html
agd5f: timofonic: you need to assoicate your monitor with the output. so add the following to your device section: Option “Monitor-DVI-0” “Monitor1”
eboettcher: (WW) RADEON(0): No crtc mode list for crtc 1,continuing with desired mode
eboettcher: oh, I see, thanks agd5f
timofonic: agd5f, but my monitor is VGA and no DVI 😛
eboettcher: timofonic: that’s the name of the output
otaylor: Hmm, compiz with r500-support is almost approaching usability 🙂
eboettcher: (II) RADEON(0): Output DVI-0 connected
eboettcher: (II) RADEON(0): Output S-video disconnected
eboettcher: timofonic: from your log
MostAwesomeDude: otaylor: I’m scared to try it.
timofonic: I see
timofonic: eboettcher and agd5f, same problem 🙁
timofonic: Latest Xorg.0.log http://rafb.net/p/QSYqsh83.html
otaylor: MostAwesomeDude: it’s actually no that bad .. it shows up OK for a while, some noise when windows show up, things eventually go entirely op-art after you move things around enough
MostAwesomeDude: otaylor: XD. Okay, I’m gonna try it.
agd5f: timofonic: I thought you said it was DVI earlier
otaylor: MostAwesomeDude: it looks like there’s probably only 2 or 3 separate bugs, maybe just 1 that are causing problems
MostAwesomeDude: otaylor: I’m missing GLX_EXT_tex_from_pixmap.
MostAwesomeDude: No idea how to get it.
otaylor: MostAwesomeDude: that may well be a function of how you started compiz
timofonic: agd5f, sorry
otaylor: MostAwesomeDude: I think you have to run it with LIBGL_ALWAYS_INDIRECT=1 ?
agd5f: MostAwesomeDude: r300 already support that so it should just work on r5xx
otaylor: MostAwesomeDude: I switched to it using desktop-effects from fedora, I never have much luck trying to run compiz directly
TobiasTheCommie: compiz on r500, nice…
TobiasTheCommie: guess i should give it a go on my r520 at some point this week
eboettcher: agd5f: his bios connector table reports 3 connectors, types 0, 0x64, and 0x60 — but the section below only reports the 0x64(DVI-I) and 0x0(STV) ports
timofonic: Should I update the AtomBios of my Radeon X700?
timofonic: If yes, how to do it?
timofonic: Something quite similar happens me with uvesafb too 🙁
timofonic: vesafb runs correctly
MostAwesomeDude: otaylor: No good. I just get a white screen.
MostAwesomeDude: Killing compiz makes it work again.
timofonic: compiz is evil! 😛
agd5f: timofonic: the problem is the driver is not able to detect what monitor is attached
MostAwesomeDude: How do we get GLX_EXT_tex_from_pixmap working?
timofonic: agd5f, old AtomBios?
agd5f: timofonic: nothing to do with atombios
timofonic: agd5f, any possible solution? Please, I have no ideas
timofonic: agd5f, oh ok
otaylor: MostAwesomeDude: texture from pixmap seems to be working for me
otaylor: MostAwesomeDude: Is AIGLX on in your server?
agd5f: timofonic: what connectors does your card actually have?
timofonic: agd5f, DVI and VGA
timofonic: err wait
timofonic: DVI, s-video and VGA
otaylor: MostAwesomeDude: If the server isn’t running with and finding the new dri modules, AIGLX and tfp aren’t going to work
eboettcher: timofonic: is the display connected to the VGA port then (which is reported by the bios but not radeon driver)?
timofonic: I use VGA, it’s an old CRT monitor
MostAwesomeDude: (==) AIGLX enabled
MostAwesomeDude: (EE) AIGLX error: dlsym for __driCreateNewScreen_20050727 failed (/usr/lib/dri/r300_dri.so: undefined symbol: __driCreateNewScreen_20050727)
MostAwesomeDude: (EE) AIGLX: reverting to software rendering
otaylor: MostAwesomeDude: OK, that’s the issue, I think
MostAwesomeDude: simpson@celes ~ $ nm /usr/lib/dri/r300_dri.so | grep driCreateNewScreen
MostAwesomeDude: 00021690 t driCreateNewScreen
MostAwesomeDude: I think I have to rebuild my X.
agd5f: timofonic: looks like a bad connector table
MostAwesomeDude: With my current Mesa tree.
otaylor: MostAwesomeDude: Maybe it’s just linking against the wrong libdrm?
MostAwesomeDude: otaylor: There’s only one libdrm on my comp.
timofonic: agd5f, a bug? 😛
agd5f: busted bios in your case
agd5f: I can work around it in the driver though
agd5f: timofonic: can you send me your bios?
otaylor: MostAwesomeDude: I’m not sure where driCreateNewScreen is supposed to come from. /usr/lib/dri/r300_dri.so is, I assume the r500 support r300 driver?
timofonic: agd5f, that’s rocket science to me. How to do it!?!?!?
otaylor: MostAwesomeDude: The weird things is that driCreateNewScreen as a ‘t’ is a local symbol to the module
eboettcher: timofonic: Rocket science in a nutshell: go back to newton’s formula but don’t assume dm/dt=0, and assume the speed of light is constant in all intertal reference frames.
agd5f: timofonic: cd /sys/bus/pci/devices/; echo 1 > rom; cat rom > /tmp/video.rom; echo 0 > rom
agd5f: timofonic: find the pci slot from lspci
MostAwesomeDude: otaylor: Yeah, r300 powers r3xx/r4xx/r5xx.
eboettcher: (or xorg.log 🙂
agd5f: lspci | grep VGA
MostAwesomeDude: otaylor: Yeah, that’s it, really. It should be exported, right?
timofonic: agd5f, it’s this? 03:00.1
otaylor: No, it’s not exported on my module either … and note the driCreateNew screen vs. __driCreateNewScreen_20050727
eboettcher: 15:39 < agd5f> busted bios in your case
eboettcher: oops, sorry
eboettcher: timofonic: 03:00.0
agd5f: timofonic: cd /sys/bus/pci/devices/0000:03:00.0 probably
timofonic: cd /sys/bus/pci/devices/0000\:03\:00.
timofonic: 0000:03:00.0/ 0000:03:00.1/
timofonic: There’s two!
otaylor: MostAwesomeDude: I think you might be right though that it’s a server vs. drm problem
eboettcher: timofonic: .
eboettcher: 0
MostAwesomeDude: Hmm.
otaylor: MostAwesomeDude: driCreateNewScreen is a entry point in the drm driver used by only by the server
MostAwesomeDude: Well, I am using an older X. 1.3.0
timofonic: agd5f, where’s the rom file /tmp/video.rom ?
otaylor: MostAwesomeDude: that’s pretty old, yeah. I was testing with a build of xserver master
agd5f: timofonic: yes
bgoglin: agd5f: is there any big commit pending for tonight? I’d like to push a snapshot of xf86-video-ati with your exa rotation fix into debian
timofonic: agd5f, dcc send?
MostAwesomeDude: otaylor: I can’t really update; there’s a nasty hardlock bug further upstream.
timofonic: agd5f, can’t receive it?
agd5f: bgoglin: nothing major. some small stuff maybe. I was trying to do a release soon, but stuff kept coming up, and then the keys thing
bgoglin: ok
agd5f: timofonic: gues not
otaylor: (some gyrations to get my system session to launch with a X server / mesa / xf86-video-ati from /opt/X .. ended up making /usr/bin/Xorg a wrapper script that set PATH and LD_LIBRARY_PATH and then called /opt/X/bin/Xorg)
otaylor: MostAwesomeDude: in xf86-video-ati?
timofonic: agd5f, do you have some ftp for sending the file then?
agd5f: timofonic: just email me alexdeucher gmail.com
MostAwesomeDude: otaylor: No, in the main server.
MostAwesomeDude: I suppose I’ll do it anyway.
otaylor: MostAwesomeDude: a hardlock in the main server?
timofonic: agd5f, uhhhhh that’s difficult without X and barely configured system
otaylor: MostAwesomeDude: how does that happen?
timofonic: eboettcher, can you send it to him?
eboettcher: sure
eboettcher: let me forward a port here for dcc first
eboettcher: don’t send yet
MostAwesomeDude: otaylor: I don’t know, only that I’m running modular X and using xf86-video-ati from git.
timofonic: eboettcher, ok thanks a lot!
MostAwesomeDude: I’m going to switch to git master for xserver.
agd5f: eboettcher: thanks!
MostAwesomeDude: The build order is xserver w/Mesa tree, then DRM, libdrm, Mesa, right?
agd5f: MostAwesomeDude: I’ve got a perl script if you want it
eboettcher: timofonic: send it 🙂
timofonic: eboettcher, sent
timofonic: eboettcher, 😀
otaylor: MostAwesomeDude: ordering shouldn’t matter much. libdrm first, any order after that
otaylor: MostAwesomeDude: The mesa tree does need to be there and pointed to when you configure xserver
MostAwesomeDude: otaylor: ‘k. This will be…fun.
agd5f: well some of the libs have weird dependencies if your are bootstapping X
MostAwesomeDude: Oh, and has anybody figured out synaptics yet? I remember it didn’t work last time.
eboettcher: timofonic: 183b7f25be527b781a5a06432de60568 /home/foo/video.rom
eboettcher: can you check that md5sum please?
timofonic: eboettcher, 183b7f25be527b781a5a06432de60568 /tmp/video.rom
timofonic: eboettcher, same here it seems 🙂
otaylor: MostAwesomeDude: it went pretty smoothly to me, well, until I spent an hour why the X server was crashing … selinux … grrr
agd5f: timofonic: is your card dual DVI by any chance?
MostAwesomeDude: otaylor: Building now.
agd5f: timofonic: this should do it: http://www.botchco.com/alex/xorg/timofonic.diff
timofonic: agd5f, externally has DVI s-video VGA
timofonic: agd5f, What to do with that?
agd5f: timofonic: should be fixed now in git
agd5f: timofonic: you’ll have to download the source from git and build it
timofonic: agd5f, I see
timofonic: agd5f, so I found a bug? 🙂
timofonic: “found”
agd5f: timofonic: yeah 🙂
timofonic: Hehe
timofonic: agd5f, I use Gentoo. I’ll check later if there are an overlay with that xorg stuff or whatever
eboettcher: yes
eboettcher: timofonic: emerge layman; layman -a x11
timofonic: Why naming it xf86-video-ati and not xorg-video-ati ?
timofonic: eboettcher, nice 😀
MostAwesomeDude: timofonic: Historical reasons.
rx__: there was some debate about that on ML
timofonic: MostAwesomeDude, so it will be changed?
rx__: but came down to historical reasons 🙂
rx__: no
MostAwesomeDude: timofonic: No, for historical reasons. :3
agd5f: timofonic: it was originally an xf86 driver as is most of the xserver
timofonic: I don’t understand why, but then…
timofonic: “historical reasons” seems not reasons to me at all, but well
timofonic: agd5f, other drivers have been renamed to xorg- I think
agd5f: timofonic: I don’t thnk so 🙂
rx__: none has
rx__: just look at the git repo
timofonic: agd5f, uhm then I got confused
timofonic: ok
eboettcher: so what happened to the discussion of moving xorg drivers back into xserver?
eboettcher: I don’t think it’s such a great idea, but then it seems like there needs to be better versioning when a driver starts using a new feature.
agd5f: eboettcher: when you try and build a driver it will check and see if you have the mimimum requirements
MostAwesomeDude: I think that the current system of having a list of extensions and an ABI versioning system works pretty well.
agd5f: radeon already does quite a bit of that
agd5f: compile time checks that is
MostAwesomeDude: Yep.
MostAwesomeDude: And modular X permits us to develop drivers without the official blessing of Xorg.
agd5f: does anyone remember the command to delete a local branch?
agd5f: with git
timofonic: Going to eat, will install that new driver after
libv: agd5f: -d, or -D when it would mean losing information
timofonic: agd5f, do you have the bug in some bugtracker? I want to send the bug to the guy over uvesafb 🙂
agd5f: libv: thanks!
MostAwesomeDude: agd5f: git-branch -d
MostAwesomeDude: Hey, does anybody know how to deal with “No rule to make target `fbpseudocolor.c’, needed by `libfb_la-fbpseudocolor.lo’. Stop.” or should I go to the other channel?
timofonic: Away, I’ll back later
timofonic: I’m quite hungry…
otaylor: MostAwesomeDude: you got that when building the xserver?
MostAwesomeDude: Yes.
otaylor: MostAwesomeDude: Were you building in an old tree that you just updated? sometime hyou can get things like that from leftovers when you update a previously built tree
MostAwesomeDude: Already made distclean.
otaylor: MostAwesomeDude: Not sure that necessary helps
MostAwesomeDude: Hmm.
otaylor: MostAwesomeDude: try rm -rf fb/.deps
otaylor: You’ll probably need to re-run configure then
MostAwesomeDude: otaylor: Still not going.
MostAwesomeDude: I’m cloning a new tree.
MostAwesomeDude: Think maybe it didn’t fast-forward right, or I had some kind of tweaked configuration going on.
agd5f: if you had a previous tree checked out and configured maintainer mode is fail
agd5f: wow compiz does sort of work
rx__: compiz on what?
shosca: it works here with x1600
agd5f: rx__: r5xx hw
rx__: oh
rx__: that’s nice.. for compiz lovers
MostAwesomeDude: rx__: It’s an important landmark.
rx__: eyecandy isn’t for me 🙂
rx__: but congrats anyway
rx__: feature parity to fglrx is near!
MostAwesomeDude: Mm, but there’s more features in there than eyecandy.
rx__: jk
MostAwesomeDude: And technically, if you consider EXA, RENDER, and Composite, we’ve already passed fglrx in the important features.
MostAwesomeDude: Just a bit further to perfect 3D.
rx__: oh.. don’t get me wrong.. it’s definitely usable
MostAwesomeDude: otaylor: Appears to be building correctly now.
otaylor: MostAwesomeDude: Those autodep issues are a pain in the neck
MostAwesomeDude: otaylor: Back in a few.
simpson: Howdy.
simpson: Wobbly windows are wobbly.
simpson: All we need now are working textures.
MostAwesomeDude: otaylor: Got it.
otaylor: MostAwesomeDude: working x server again?
MostAwesomeDude: otaylor: synaptics doesn’t work. Aside from that, it all works.
MostAwesomeDude: AIGLX makes everything feel a tiny bit slower, but compiz runs nicely
MostAwesomeDude: At least, until I need a tex. Then it goes glitchy.
MostAwesomeDude: airlied: Any ideas/progress on tex fixing?
MostAwesomeDude: *fixing texs, even
mattmatteh: anyone familiar with radeonfb ? i know its off topic, i can go away if you like.. but not sure where else to ask. sometimes there is blue pixel noise with the framebuffer or black back ground in X. a few reboots did nothing, but i powered down and it seems to have gone away. perhaps something is not initialized correctly ?
mattmatteh: darn, starting X and leaving X the blue pixel noise has returned
agd5f: mattmatteh: radeonfh is bitrotten
agd5f: it only ever really support chips up to the early r3xx chips, everything past that was a hack
mattmatteh: agd5f, thanks for being patient with that
mattmatteh: agd5f, for the most part it works fine
mattmatteh: just blue dots
mattmatteh: and radeonfb is the reason i replaced my nvidia.
mattmatteh: i have an old computer that doesnt work well with widescreen
mattmatteh: radeon works better
mattmatteh: agd5f, you suggested to change the resolution, but not sure how to do that
MostAwesomeDude: XD, X just died on me.
agd5f: mattmatteh: fbset IIRC
agd5f: mattmatteh: also weren’t you having problems with Xv? I think http://bugs.freedesktop.org/show_bug.cgi?id=16001 is relevant in your case
mattmatteh: agd5f, i read the help and manual before. that was kinda confusing
mattmatteh: agd5f, that is exactly the problem i am having
mattmatteh: steps out for food, be back soon
airlied: MostAwesomeDude: hmm no ideas yet but it seems to be one problem at a time stuff 🙂
timofonic: From the x11 Gentoo overlay… fatal: Unable to look up anongit.freedesktop.org (port 9418) (Name or service not known)
timofonic: It’s the address wrong? Some server bug?
MostAwesomeDude: airlied: ‘k.
MostAwesomeDude: Just the fact that it runs, and stuff goes wobbly, is exciting.
timofonic: Does anongit works correctly?
timofonic: Ok, now connected 🙂
airlied: MostAwesomeDude: yeah, need to write more tests 🙂
timofonic: agd5f, it works! Thanks a lot 🙂
agd5f: airlied, MostAwesomeDude: the FP analyzer is pretty neat. requires windows though: http://ati.amd.com/developer/gpusa/index.html
agd5f: timofonic: cool!
airlied: MostAwesomeDude: I suspect we’ll keep fixing bugs like the temps one and texrect will keep getting better 🙂
agd5f: shows the FP instructions and RS instructions, pixsize, counts, etc
airlied: agd5f: it looks very nice..
airlied: wonders does it run under Wine 🙂
airlied: does it need an ATI driver installed?
agd5f: needs the dx runtimes for dx fps
agd5f: not sure
agd5f: if you want I can run FPs through it and send you the results
agd5f: or pastebin them
airlied: MostAwesomeDude: its really unusal that texrect doesn’t work with a single texture 🙂
airlied: like if I swap the textures I can see the girl one..
airlied: so which textures is texture one wins.
airlied: which sounds like some blending fail or something.
agd5f: airlied: are we setting the tex id in the tx regs? I’d assume we are
airlied: agd5f: hopefully 🙂
airlied: all r300 code..
airlied: and the fp stuff looks fine
agd5f: yeah true
airlied: though I really should test r500-support on an r300 🙂
airlied: we might have broken something..
airlied: damn no r300 in the office yet.
agd5f: heh 🙂
agd5f: I’ve got one here
airlied: even an rs690 would be cool
airlied: agd5f: btw one thing I don’t like is we seem to now emit a bunch of null state.
dli: wah, compiz almost working! r5xx
airlied: agd5f: so we send clear state, bunch of null state, real state.
airlied: maybe we always did it but it feels wrong.
agd5f: airlied: the whole driver needs work…
airlied: agd5f: true 🙂
airlied: re-work..
agd5f: yeah, re-org by block
agd5f: gallium 😉
dli: great! the fullscreen xv bug also fixed.
agd5f: dli: what bug was that?
airlied: agd5f: does seccolor work on r300 also 🙂
dli: agd5f, up to last week, whenever I choose “fullscreen” with “mplayer -vo xv”, whole machine hangs
airlied: really needs to get the rs690 into my remote control rack.
agd5f: airlied: swapping the r300 in now
airlied: I only have an i815 under remote control now.
airlied: which seems kinda a waste
agd5f: airlied: what fun is that 🙂
agd5f: airlied: what’s a good demo for seccolor?
airlied: tests/seccolor
agd5f: yeah works
airlied: so r500-support still works on r300. good to know.
airlied: must finished the instruction decoder for tex instrs 🙂
agd5f: yeah, just ran a bunch of tests in progs/tests and all is well
mattmatteh: agd5f, i was reading that bug. you didnt commit that patch yet ?
mattmatteh: agd5f, i was going to test
mattmatteh: if you like
mattmatteh: and i can add a picture if you want too ?
agd5f: mattmatteh: already committed part of it
mattmatteh: i got a calibration dvd that shows the grey scales
mattmatteh: and might be able to get a good comparison with x11 and xv
mattmatteh: agd5f, i only have a slight understanding of the gamma curve, but isnt that something that would be calculated ? seems like its being guessed
mattmatteh: agd5f, or could be an option like in xvattr ?
agd5f: mattmatteh: it’s a hardcoded table for like 6 or 8 gamma settings. ideally we calculate it as per the patch in that bug
mattmatteh: agd5f, hmm. perhaps ill have to read up on gamma and the source code. thanks for telling me about the bug, much appreciated
agd5f: mattmatteh: np
timofonic: How it’s going the radeon vs radeonhd fight? Who’s winning? 🙂
MostAwesomeDude: agd5f: Thanks for the shader assembler. Unfortunately, it doesn’t quite work in Wine since it needs to link to D3D9.
agd5f: MostAwesomeDude: bummer
MostAwesomeDude: agd5f: It’s just a runtime link; if I could patch Wine somehow, we could make it work, I’m sure.
timofonic: It’s funny Wine is going 1.0 when there’s lots of important stuff in not working status or very high compatibility 😛
airlied: MostAwesomeDude: wine should have bits of d3d9 🙂
MostAwesomeDude: timofonic: The Wine 1.0 release is specifically targeted at a handful of programs.
timofonic: versioning marketing is sad
airlied: it really does look like we are getting a bad color and its saturating the textures
MostAwesomeDude: airlied: wine: Call from 0x7b84346e to unimplemented function d3dx9_36.dll.D3DXAssembleShader, aborting
airlied: MostAwesomeDude: pwned.
MostAwesomeDude: That’s what we’re missing.
timofonic: MostAwesomeDude, what ones? backgammong and those games included in windows? 😀
MostAwesomeDude: timofonic: Photoshop CS2/CS3, Microsoft Office 2004/2007, and a few others.
timofonic: solitaire and that game with wines… 🙂
timofonic: Microsoft Office? Bahh
timofonic: wines=mines 😛
timofonic: Where’s interesting stuff like OCR and CAD? 😛
MostAwesomeDude: timofonic: Where, indeed.
MostAwesomeDude: agd5f: I’m going to go grab a Wine git tree and hack on it a bit.
MostAwesomeDude: Unless you feel like putting out an OpenGL-only version of the tool… :3
agd5f: MostAwesomeDude: heh 🙂
MostAwesomeDude: Is that a “no?”
agd5f: i’d say so
airlied: agd5f: hmm we might need to program 0x46c0/4 at some point..
agd5f: airlied: 0x4e14 as well
MostAwesomeDude: agd5f, airlied: Wait, we don’t set those?
MostAwesomeDude: How do we do the color buffer clear, then?
agd5f: MostAwesomeDude: we might, I didn’t check
MostAwesomeDude: agd5f: Checking right now.
MostAwesomeDude: 0x4E14 isn’t used.
MostAwesomeDude: 300_reg.h:#define R300_US_ALU_RGB_ADDR_0 0x46C0
MostAwesomeDude: That could be a problem.
agd5f: MostAwesomeDude: r3xx works
MostAwesomeDude: agd5f: Well, I mean, we are definitely not using any of those registers.
MostAwesomeDude: 0x46C0 is only being used in r3xx for the shaders.
airlied: we do color buffer clear with a texture from what I know..
MostAwesomeDude: airlied: Icky. Is that faster?
airlied: MostAwesomeDude: its what fglrx did I assume.
airlied: I think its a big point or something.
airlied: wierd it appears fglrx uses MAD for MOV 🙂
airlied: hmm my revenge seccolor test was made of fail.. time to give it another go.
MostAwesomeDude: airlied: Yeah, I know they use MAD, but it’s recommended to use CMP or MAX.
MostAwesomeDude: And it’s not any more or less expensive.
airlied: MostAwesomeDude: yeah but they recommend it but don’t use it 🙂
spstarr: looks at airlied
airlied: I think I have it.
airlied: my bug damn 🙂
airlied: I’ve just pushed the greatest fix in the world.
Saist: ? … funny… Nvidia’s site is still up… so it can’t be that…
airlied: I think r500 compiz should be working now..
Saist: !! ooo. nice 🙂
airlied: I just need to boot to F9 to test it..
airlied: hmm I can move the cube, but when it returns to normal it garbages the screen.
airlied: however moving it again make it look okay.
airlied: windows all wobble nicely..
airlied: and the background is made of fail, has a border repeating.
eboettcher: meh
eboettcher: I wish I could put both an r300 card in here and and r500
eboettcher: and just disable the other when not used rather than having to swap
eboettcher: (or better yet: run different servers on each)
eboettcher: so that I could leave my programs open in the one I’m not playing with atm
eboettcher: but one card is here, the other is far away :/
airlied: hmm we have a fail point around 2048 wide windows, we get some sort of repeat.
airlied: with compiz
rx__: must be nice having a 30″ ?
airlied: it would be nice if I had two like ajax 🙂
eboettcher: O.O
eboettcher: my 24″ are far more than enough!
eboettcher: s/are/is/
eboettcher: how does ajax send signals to that from one machine?
airlied: eboettcher: using r500 and -ati
eboettcher: you don’t need dual link DVI?
airlied: you need dual dual-link DVI
eboettcher: (for each display)
airlied: yup two DVIs each with dual-link…
airlied: I think he has a FireGL of some sort, my X1900XTX also does it
eboettcher: I need to win the lottery so I can get two digital cinema projectors up, each cover 100″ surfaces 😛
eboettcher: that would use 8192px wide scanlines 🙂
eboettcher: how much bandwidth can displayport offer?
airlied: eboettcher: no idea what it tops out at .
airlied: nobody has given me any DP hw yet 🙁
eboettcher: I have yet to see and DP hardware in the online and offline retailers I sometimes browse
eboettcher: but then again I have not been looking 😉
eboettcher: well I’ve seen video cards
eboettcher: but not displays
roh: sighs
roh: wasted about 8 hours of googling the last week for a sane board which supports 2x dvi, core2quad with recent fsb of 1.3/1.6ghz and having a reasonable nic (e1000)
roh: seems i have to go for a seperate gpu on pciE again.. had hoped for intel graphics (only need 2D, multihead, 2x dvi and working xvideo) since the ati drivers are still quite a hassle to get running atm
eboettcher: I thought Intel was finally moving to drop the front side bus… (HOORAY! I may buy an intel again in the future… maybee)
roh: i have a X1600 here which ‘works’ with git versions of mesa, xfree-driver-ati and drm..
airlied: roh: do you need dual-link DVI?
roh: but _stable_ and bugfree is much different 😉
airlied: roh: I have an intel board with an add-in dual DVI card.
roh: airlied you mean for higher resolutions? dunno. i have a SGI1600sw (vga/dvi 1600×1024) as second head and would want to go to 1920×1200 for the primary
airlied: roh: that shoudl work..
roh: currently the primary is the 160×1050 inside the T60p
roh: eh 1650×1050
airlied: roh: you can pretty much get an Intel 965 based machine with a dual dvi add2 card
roh: but io performance on a notebook always sucks somehow.. disks are too slow
roh: airlied yeah.. but it seems all boards with integrated graphics do either not write anythin about add2 cards anymore in the docs
eboettcher: roh: SSDs if you have the money ;P
airlied: roh: its called SDVO
eboettcher: battery power and speed 🙂
airlied: roh: and the driver supports it.
roh: airlied there are dual-dvi aswell?
roh: only saw single dvi to pci2-connector with a tdms transmitter on it
airlied: roh: don’t think intel chipset has the bandwidth to do dual-dvi 🙂
roh: but the boards were all fsb 800/1080 or so
roh: *yikes* ..*sigh* so its radeon again… gnah
airlied: if they do, nobody has produced a card that does it.
eboettcher: airlied: not even i965?
airlied: eboettcher: it might be able, but I’ve never seen the ADD2 card.
roh: eboettcher does i965 support fsb1.6 and recent quad cpus?
airlied: roh: you don’t need dual-dvi for those monitors.
airlied: sorry dual-link
airlied: I have a card at home with 2X DVI connectors ADD2
airlied: roh: you don’t need dual-dvi for those monitors.
airlied: dog.
airlied: doh.
airlied: http://www.wintecindustries.com/orderdesk/products/productInfo.aspx?category_id=2129&product_id=22508
airlied: ^^
eboettcher: arg
eboettcher: this box is dead
roh: airlied i understand what you mean.. i just need to find a board which exists, is supported by software and reasonable priced… this industry makes it quite hard
eboettcher: IRC is about all it can do…
eboettcher: 00:06:39 up 46 days, 4:56, 6 users, load average: 1.06, 1.02, 1.00
eboettcher: my laptop.. O.O
eboettcher: I should start playing with the rs480 in this thing.
roh: since about a week and the hardy update this R530 is quite nice
roh: only it eats cpu unreproducible from time to time, still has that flicker when using xrandr and gl is broken now (only a black box with some flickering rendering from time to time)
roh: but xvideo works
eboettcher: what kind of issues could I expect from running two different family radeons (IE r300 and r500) on this box? Would they ignore eachother and get along?
roh: real progress for me.
roh: also the blocking when switching workspaces is gone and scrolling in FF is really fast now
eboettcher: roh: I’m having scrolling issues in FF now that I need to look into
eboettcher: using R423
roh: eboettcher how recent is you X core?
roh: and are you using EXA or are you still at XAA
airlied: realises the x1900 card was given to him by a SuSE employee..
airlied: about 2 years ago..
roh: hrhrh
roh: thats what happened to dok also around 2000.. soon everybody was giving him a gpu which had no drivers yet 😉
eboettcher: roh: May 18 2008 02:10Z-0400
eboettcher: using EXA
eboettcher: it’s not *slow* per say
roh: eboettcher i had it on some websites more than on others…
roh: try http://kor.gazaxian.com/index.php/category/dj-sets/ and scroll down
roh: when it blocks and jumps its like it was for be before i changed my machine from gutsy to hardy (used both with git head of mesa, drm and xorg-ati)
eboettcher: that site is fine
eboettcher: but some places it’s “jittery”
eboettcher: very slight, but there
eboettcher: if I turn of smooth scrolling that’s gone
roh: i guess ff is doing loads of extra expose events on some pages due to sucky rendering
roh: http://yamato.hyte.de/xorg-ati/flicker-bug/ is my ‘flicker bug’
roh: thats still there with hardy xorg-core and recent drivers (< 1 week old)
agd5f: airlied: nice!
agd5f: airlied: I think we need to revisit clipping
airlied: agd5f: yeah its a wierd effect when you drag a window border across 2048
agd5f: r3xx had that weird offset thing with the clipping to support 2560, so r5xx shouldn’t need it
airlied: agd5f: I think I turned that off already.
agd5f: ah ok
airlied: I remember doing something in the drm..
airlied: openarenea plays well
airlied: I need to enable scaler on DVI for this monitor..
agd5f: output attribute
eboettcher: speaking of clipping, I should revisit that xv clipping issue I encountered the other night
eboettcher: sure enough a test case is comming up — roh’s flicker video is at 93% 😛
roh: eboettcher the interresting thing is that it goes away when i clone the second head and not make it independant
eboettcher: agd5f: I’ve found that when I drag a xv window off the edge of the screen, segfault!
eboettcher: testing right edge this time…
roh: note that my second head is currently on VGA-0 and the primary on LVDS
eboettcher: O.O, nothing
eboettcher: brb
agd5f: eboettcher: overlay or textured video?
airlied: wow 87fps openarena at 2048×1536
agd5f: eboettcher: compositing manager running
agd5f: airlied: sweet!
agd5f: ppracer has texture issues, but generally works
eboettcher: hm
eboettcher: right now I’m unable to reproduce it, yet I did several times the other night…
eboettcher: the only difference in the enviornment I can see from the logs is where the ring buffers…
eboettcher: oh wait I just did it again
airlied: damn that typo, I did a typo search the other day and I missed it.
eboettcher: okay so it only happens when mplayer is using ffodivx
eboettcher: and I drag off the LEFT side of the screen
eboettcher: http://rafb.net/p/vEVP5234.html is the log
eboettcher: it does not happen with any other video codec
eboettcher: and in all cases, VO: [xv] 512×384 => 512×384 Planar YV12

Radeon IRC Logs For 2008-5-20

Rig Lab Hardware

You Might Also Like

RadeonHD.org | Radeon IRC Logs For 2008-6-03

RadeonHD.org | Radeon IRC Logs For 2008-5-22

RadeonHD.org | Radeon IRC Logs For 2008-11-04