--- Log opened Птн Авг 25 00:00:12 2006
--- Day changed Птн Авг 25 2006
00:00 < phh> pff
00:00 < phh> it's hard to add a new test
00:02 < phh> (ok because it uses X call which need to get the Display and the Drawable from the SDL)
00:03 < marcheu> stillunknown: oh, what are the symptoms ?
00:04 -!- shenki [n=shenki@121.44.69.112] has quit [Read error: 60 (Operation timed out)]
00:04 < stillunknown> marcheu: i forgot to install the dri module
00:04 < stillunknown> or library
00:04 < marcheu> stillunknown: you don't need nouveau_dri.so yet, it'll work without
00:04 < marcheu> because it doesn't even compile at the moment :)
00:05 < marcheu> so what are the errors you have ?
00:05 < stillunknown> the system totally freezes up
00:05 < stillunknown> no EE's
00:05 < marcheu> do you have the /var/log/Xorg.0.log from it ?
00:05 < marcheu> could you put it online ?
00:05 < stillunknown> yes
00:05 < marcheu> btw you got everything from cvs, right ?
00:06 < stillunknown> no
00:06 < marcheu> no ?
00:06 < stillunknown> i only have the ddx, drm and libdrm from cvs
00:06 < marcheu> ok, that's correct
00:08 < stillunknown> http://pastebin.co.uk/651
00:08 < stillunknown> this log shows glx being loaded, also tried without it (lost that log) and with ddx
00:08 < marcheu> ah, you ahve 2 screens ?
00:08 < stillunknown> r/ddx/xaa
00:08 < stillunknown> no
00:09 < stillunknown> only one screen
00:09 < marcheu> what is your setup like in /etc/X11/xorg.conf ?
00:11 < stillunknown> Section "Device"
00:11 < stillunknown>     Identifier  "Gigabyte 6600GT (nouveau)"
00:11 < stillunknown>     Driver      "nv"
00:11 < stillunknown>     VideoRam    131072
00:11 < stillunknown>     Option "HWCursor" "true"
00:11 < stillunknown>     Option "FlatPanel" "true"
00:11 < stillunknown>     Option "AccelMethod" "EXA"
00:11 < stillunknown> EndSection
00:11 < stillunknown> anything else you need (in that case i'll post my xorg.conf)
00:12 < marcheu> nope
00:13 < stillunknown> http://pastebin.co.uk/652
00:13 < stillunknown> xaa
00:13 < stillunknown> this time i remembered to disable the glx module
00:15 < stillunknown> that one is somehow incomplete, i will repost
00:16 < stillunknown> strange they all end too soon, will post the last few hundred lines seperately
00:17 < marcheu> well, there are is a with 6600s and the original "nv" driver. we probably inherited it, I guess that's the occasion to fix it
00:18 < stillunknown> http://pastebin.co.uk/654
00:18 < stillunknown> maybe the end part is interresting
00:19 < marcheu> do both crash the same way ?
00:19 < stillunknown> yes
00:20 < stillunknown> any idea what the problem is?
00:20 < marcheu> I'll look into it
00:21 < marcheu> it's good to have someone who can reproduce :)
00:21 < stillunknown> you mentioned 6600 have a problem, does this mean someone else had this problem too?
00:22 < marcheu> yup, there are bugs in the fd.o bugzilla
00:23 < marcheu> https://bugs.freedesktop.org/buglist.cgi?query_format=specific&order=relevance+desc&bug_status=__all__&product=xorg&content=nv
00:24 < marcheu> I have to dig through those
00:25 < marcheu> stillunknown: is it pcie or agp ?
00:25 < marcheu> I'd guess pcie
00:25 < stillunknown> pci-e
00:26 < stillunknown> need more info?
00:26 < marcheu> no, will you be there in the next days ?
00:26 < marcheu> I'll look at the code and try to understand what happens, and might need some testing :)
00:26 < stillunknown> vacation ends in 10 days, i expect to be around during the day then
00:26 < stillunknown> *until then
00:27 < marcheu> see https://bugs.freedesktop.org/show_bug.cgi?id=6038
00:27 < marcheu> a pcie 6600, which doesn't start...
00:28 < marcheu> actually, that one is just out of sync
00:28 < marcheu> is yours out of sync or crashed ?
00:28 < stillunknown> just a black screen and then freeze
00:29 < marcheu> no ctrl-alt-backspace ?
00:29 < stillunknown> no
00:30 < marcheu> ah, theres almost a backtrace at the end
00:31 < marcheu> well, I'll be looking at the code over the night and ping you when I have something... it's probably something we overlooked WRT non-agp cards
00:31 < stillunknown> (II) NV(0): AGP: aperture is 0MB
00:31 < marcheu> yeah :)
00:32 < marcheu> well, technically that's true since you don't have AGP
00:32 < phh> lol
00:34 < stillunknown> away for a while, will look later for questions
00:35 < marcheu> ok, found it, will fix it
00:36 < phh> boum.
00:36 -!- phh [n=phh@moi.phhusson.info] has quit ["Quitte"]
00:37 -!- phh [n=phh@moi.phhusson.info] has joined #nouveau
00:50 -!- skrull [i=skr@200.96.100.169] has joined #nouveau
00:52 < phh> is the object beef0028, type 307e known ?
00:52 < marcheu> http://nouveau.freedesktop.org/wiki/NvObjectTypes
00:53 < pmdata> nice, it's unknown
00:53 < marcheu> where does it come from ?
00:53 < phh> i suspect it to be something with rotation
00:54 < marcheu> ah
00:54 < marcheu> or something with 2D, sqince you're looking at the 2D fifo
00:54 < phh> marcheu, there are "many" things for rotation but yes it can be something else
00:54 < phh> (this object is created 4 time (0 90 180 and 270)
00:55 < phh> and when i try (successless so maybe not that) to change it recreate object
00:55 < marcheu> well, if you did the changes I told you, you're actually seeing all 2D commands used to draw your desktop
00:55 < marcheu> which is quite a lot
00:55 < phh> aie
00:55 < phh> ok
00:55 < phh> bah i try and i will see after
01:02 < marcheu> stillunknown: is it better with current CVS ?
01:03 < phh> bon
01:03 < phh> ca fait peur.
01:07 < phh> WAHOU.
01:08 < phh> When it changes the orientation it segfaults
01:08 < phh> but when i call the XLib but no change to be made, it works
01:08 < phh> (it's in dump_after :()
01:09 < marcheu> then you have to fire gdb and debug it I guess
01:09 < phh> 0x0804aaae in dump_after (dump_startup=0) at re.c:631
01:09 < phh> 631                             memcpy(mappings[i].after,mappings[i].start,mappings[i].size);
01:10 < phh> mm i maybe don't know how to call dump_after & so on
01:10 < phh> i will look others tests
01:10 < marcheu> did you call dump_before ?
01:10 < phh> yep
01:10 < marcheu> is any of the memcpysrc/dst invalid ?
01:11 < phh> (gdb) print mappings[i].after
01:11 < phh> $1 = (unsigned int *) 0xad904008
01:11 < phh> (gdb) print mappings[i].start
01:11 < phh> $2 = (unsigned int *) 0xadb09000
01:11 < phh> (gdb) print mappings[i].size
01:11 < phh> $3 = 1048576
01:11 < phh> (gdb) 
01:11 < phh> looks ok
01:12 < phh> but i don't know how to check if a region is rightly allowed in gdb
01:12 < marcheu> print *region
01:12 < phh> 0
01:12 < marcheu> and it'll say if it can't access
01:12 < marcheu> otherwise it's ok
01:12 < phh> ah
01:12 < phh> start isn't ok
01:13 < phh> Cannot access memory at address 0xadb09000
01:13 < marcheu> hehe, so one of the mappings was destroyed during the test
01:13 < phh> huhu
01:13 < marcheu> the application probably didn't need it any more and removed it
01:14 < marcheu> that doesn't happen with OpenGL stuff, but with 2D that's quite common
01:14 < phh> aie.
01:14 < marcheu> I guess you have to google a bit, and figure out that funtion (whose name I don't remember) which determines wether a memory adress is valid
01:14 < marcheu> and then use it
01:15 < phh> access_ok in kernel but in user land mmm
01:24 < stillunknown> marcheu: will update now
01:25 < marcheu> stillunknown: ok, I don't promise it'll go too further, but thanks :)
01:25 < stillunknown> only drm?
01:25 < stillunknown> update
01:26 < marcheu> actually, the drm is only a warning
01:26 < marcheu> the fix is in the DDX
01:26 < stillunknown> i started emerge xf86-video-nv and no files were updated from cvs
01:26 < stillunknown> will look again
01:27 < stillunknown> maybe it's too soon
01:28 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Remote closed the connection]
01:29 < phh> marcheu, is it a problem if the way i will fix that is glibc specific ?
01:30 < marcheu> I don't really care
01:31 < marcheu> other way is looking at the list of mappings each time.. but it involves writing a lot of code
01:31 -!- darktama [n=darktama@gentoo/contributor/darktama] has quit [Read error: 113 (No route to host)]
01:31 < phh> marcheu, mappings[i].start comes from a mmap ?
01:32 < phh> or a malloc ?
01:32 < marcheu> mmap, it's a mapping :)
01:32 < stillunknown> going to try now, brb (hopefully)
01:32 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
01:32 < phh> marcheu, et zut.
01:32 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has joined #nouveau
01:33 -!- pmdata [i=patrice@ANantes-154-1-15-159.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
01:34 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
01:34 < stillunknown> let's just say, my computer doesn't reboot this fast :-)
01:34 < marcheu> oh, so it works ?
01:34 < stillunknown> yes
01:35 < marcheu> are you trying exa ?
01:35 < stillunknown> yes
01:35 < marcheu> could you send me your /var/log/Xorg.0.log ? that's still interesting to me
01:35 < stillunknown> just need to switch to softgl and see if that works
01:35 < b33fc0d3> anyone know if these drivers support suspend2 ?
01:36 < phh> b33fc0d3, power management isn't a priority i think :)
01:36 < marcheu> b33fc0d3: hahaha no
01:36 < b33fc0d3> heh
01:36 < phh> a day... maybe
01:36 < marcheu> power up after suspend is very complex
01:36 < b33fc0d3> i was expecting that
01:36 < marcheu> and it requires kernel-side code. a lot of code
01:36 < b33fc0d3> even nvidia can't do it without disabling agp
01:36 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit [Client Quit]
01:37 < marcheu> but since we have a kernel module, things are not impossible
01:37 < marcheu> in the long run maybe...
01:37 < phh> b33fc0d3, WRONG
01:37 < phh> b33fc0d3, nvidia does it !
01:37 < phh> but with agpgart
01:37 < b33fc0d3> phh: no it doesnt
01:37 < phh> b33fc0d3, so it's random
01:37 < b33fc0d3> 6800 doesn't work
01:37 < b33fc0d3> hangs on resume
01:38 < b33fc0d3> unless you disable agp altogether =)
01:38 < b33fc0d3> well i don't game in linux, so it's all good
01:38 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
01:40 < stillunknown> http://www.rafb.net/paste/results/Sk99Yx71.html
01:40 < stillunknown> my log
01:40 < stillunknown> is it normal for mesa opengl not to work?
01:41 < phh> mesa doesn't mean unaccelerated
01:41 < stillunknown> i mean software accelerated opengk
01:41 < stillunknown> *opengl
01:41 < marcheu> hmm, it should work, what's the error ?
01:41 < stillunknown> Xlib:  extension "GLX" missing on display ":0.0".
01:43 < stillunknown> Xv is broken (tested just to verify)
01:44 < marcheu> hmm, you probably only need to have the glx module loaded
01:45 < stillunknown> but isn't the glx module part of the nvidia binary dirvers?
01:45 < stillunknown> *drivers
01:46 < marcheu> no, glx is a module to do opengl with X, whatever the drivers
01:47 < stillunknown> brb
01:47 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
01:48 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
01:49 < stillunknown> that was the problem, i assumed something, which was just dependent on which opengl i had chosen
01:49 < stillunknown> glxgears does seem a bit slower
01:50 < stillunknown> i could get close to 1000 fps from nv driver
01:50 < phh> Oo
01:50 < stillunknown> about 700 now
01:50 < marcheu> well, glxgears is not really a benchmark you know... 
01:50 < stillunknown> this is NOT fullscreen
01:50 < stillunknown> i know
01:51 < phh> stillunknown, i fullscreen with proprio i have 70fps you know...
01:51 < marcheu> but yeah, EXA will be slower than XAA for software opengl
01:51 < marcheu> we'll have hardware opengl anyway in the end:)
01:51 < phh> and around 2000 in normal window
01:51 < stillunknown> marcheu: anything else you want/need to know?
01:52 < marcheu> no, thanks a lot four your testing
01:52 < marcheu> you're our first tester btw :)
01:53 < marcheu> also, EXA will get faster over time
01:53 < stillunknown> cvs/svn is pretty easy to intergrate into portage, otherwise i wouldn't have done it
01:53 < marcheu> EXA is a bit complicated, because most of the "core" EXA sits outside the drivers
01:53 < marcheu> and the "core" EXA will be improved before it's usable
01:53 < stillunknown> EXA is not common for drivers?
01:54 < marcheu> yes it is, a lot of code is common
01:54 < stillunknown> i mean, is it used a lot and is the nv just old?
01:54 < marcheu> and ATM that code is not as fast as possible. but in the future, it will become faster, without any changes to our driver
01:55 < stillunknown> are you french?
01:55 < marcheu> yes
01:56 < stillunknown> glad you can speak english :-)
01:56 < stillunknown> (i had a few years of french in school, but it was a disaster)
01:56 < marcheu> hehe, a lot of people are french in here
01:57 < stillunknown> maybe someone will fork the driver in the end and call it new :-)
01:57  * phh puts hand up
01:57 < marcheu> "new" doens't have "N" and "V" in the name :)
01:57 < zoeloelip> new version?
01:57 < phh> bwarf
01:57 < zoeloelip> :p
01:57 < stillunknown> NVidia
01:58 < phh> and New is a known nouns in whole world
01:58 < phh> whereas nouveau is mystic for many people
01:58 < phh> (and prononciation should be really fun ;o)
01:58 < zoeloelip> I like it, but it's not mystic
01:58 < zoeloelip> I had french for eight years
01:58 < zoeloelip> not that much remains, it's all fading away :)
01:58 < stillunknown> nouveau one of the easier french words
01:59 < phh> what is its plural ? :)
01:59 < marcheu> actually, you know how I came up with that name (or how I didn't, rather) ?
01:59 < stillunknown> there is more than one word in french
01:59 < stillunknown> no
02:00 < marcheu> I've got an automatic replacement in my irc client which turns "nv" into "nouveau" (for speaking in french channels) and I was saying on a xorg channel that I was working on a "nv" driver. and the replacement changed that to "nouveau", and people tole me they liked the name :)
02:00 < marcheu> s/tole/told
02:00 < phh> huhu
02:02 < stillunknown> do you mind if the wiki (eventually) is split into a (test) user part and developer part?
02:02 < marcheu> stillunknown: not at all
02:03 < stillunknown> i have made an account on the wiki and i'll start adding some stuff, like my ebuilds for gentoo (with a warning that they are just hacked ebuilds)
02:03 < stillunknown> my name on the wiki is my more common nickname, namely madman2003
02:05 < stillunknown> i have one recommendation as far as developing path goes
02:06 < stillunknown> you would make a lot of people happy (not me) with dualhead support for an open source driver
02:06 < stillunknown> (apart from 3d, this seems to pop up as the major problem with the nv driver)
02:07 < marcheu> hehe, yes. I think the most happy would be darktama :)
02:07 < phh> the best thing would be dynamic dual head :p
02:07 < phh> (as in fgflrx)
02:08 < marcheu> phh: yes, there's a new X api being designed for that (dual head, and maybe dynamic even). we're awaiting to see how it turns out before implementing dual head
02:08 < stillunknown> i will keep this driver on until at least tomorrow, maybe Xv will work then, later i will make an ebuild for renouveau and try some dumps
02:09 < marcheu> oh linuxfr journal about us :)
02:11 < phh> "Xorg et les modules proprio"?
02:11 < marcheu> yup
02:12 < phh> i'm the first answerer :D
02:12 < phh> (but not about nouveua)
02:12 < phh> nouveau*
02:18 < phh> marcheu, i wondered, are all EXA functions accelerated for NVidia cards ?
02:18 < marcheu> phh: all ? you mean even composite ? no.
02:18 -!- tibbs|h is now known as tibbs
02:18 < stillunknown> do the nvidia binary drivers use EXA?
02:18 < marcheu> no
02:19 < phh> and even not XAA :p
02:19 < phh> marcheu, nop, just basics one
02:19 < phh> and don't say basic nv driver does !
02:19 < marcheu> they use something that resembles EXA, but which is not EXA. they have the equivalent of EXA + the core EXA stuff in their driver
02:19 < phh> (comparing 2D with nvidia and nv is easy, nvidia is quicker)
02:19 < marcheu> phh: yes, we have all the functions except composite ones
02:20 < phh> hum
02:20 < marcheu> which will probably have to go through the 3D engine, because we can't find a corret alpha blitting operation
02:20 < marcheu> but I really hope to find something that works on nv04 or nv10 and that doesn't need 3D. but well.. 
02:20 < phh> why that ?
02:21 < phh> 3D is slow on it?
02:21 < marcheu> because 3D is more complex to program, I'd rather use 2D for this
02:21 -!- K [i=hazel@tor/session/external/x-85fbb9e5c24a50d1] has quit [Remote closed the connection]
02:21 < stillunknown> newer cards do support it?
02:21 < marcheu> stillunknown: I didn't find it yet, so I dont know :)
02:22 < stillunknown> is there still 2d improvements that can happen for "normal" things?
02:23 < marcheu> yeah, as I said most of the improvements will come from the core EXA which will be improved
02:23 < marcheu> so it's outside our business
02:23 < stillunknown> and that's good :-)
02:24 < stillunknown> will nouveau be moved to git later on?
02:24 < Duke`> (hum half an hour to read the entire irc buffer from today :x)
02:25 < marcheu> stillunknown: yup, at some point in the near future
02:26 < marcheu> stillunknown: actually, I'm awaiting for airlied to prepare that repository stuff I don't understand :)
02:27 < phh> Duke`, only ? :p
02:30 < stillunknown> goodnight everyone
02:33 < phh> Duke`, and you learned interesting things ? p:
02:34 < phh> (manquerait plus que ca)
02:35 < Duke`> yes, new nvidia's blob is out o/
02:36 < phh> :)
02:36 < phh> Duke`, you haven't read linuxfr.org's journal so :p
02:37 < Duke`> it's just open in my current tab
02:38 < Duke`> then I have the long discussion on lwn, the mega journal on tabaco, my wikipedia follow list to check, all that before sleeping !
02:38 < phh> hehe
02:39 < phh> Duke`, what on lwn ?
02:39 < Duke`> article about xorg and the fuckin' proprietary modules
02:40 < phh> Duke`, need to be more precise
02:40 < phh> and propriatery modules ON FEDORA
02:40 < Duke`> http://lwn.net/Articles/195351/
02:40 < phh> Duke`, i know
02:41 < phh> ++
02:41 -!- phh [n=phh@moi.phhusson.info] has quit ["Quitte"]
03:02 -!- Duke` [n=gnu@ANantes-251-1-124-144.w86-210.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
05:44 < marcheu> ok, it seems I broke the drm
06:07 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has joined #nouveau
06:13 < marcheu> ok, I fixed it as far as I could go, more work would require the DDX to allocate its memory properly
06:26 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has quit [Read error: 104 (Connection reset by peer)]
06:35 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has joined #nouveau
06:58 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has quit [Read error: 104 (Connection reset by peer)]
06:58 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has joined #nouveau
07:28 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
07:52 -!- Myrizio [n=Myrizio@host139-102.pool80104.interbusiness.it] has quit [Read error: 113 (No route to host)]
07:56 -!- darktama [n=darktama@gentoo/contributor/darktama] has joined #nouveau
08:47 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit ["<blank>"]
10:35 -!- phh [n=phh@moi.phhusson.info] has joined #nouveau
10:35 < phh> hello
11:11  * skrull is going to sleep
11:11 -!- skrull is now known as skrull_far_away
11:37 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
11:38 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
11:40 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
11:40 < stillunknown> i have bug to report in the drivers :-(
11:41 < phh> 3D doesn't works? :)
11:41 < phh> stillunknown, it seems that no one is awake so go to the bugtracker
11:44 < stillunknown> i didn't know there was a bugtracker
11:44 < phh> right it's nv
11:44 < phh> not nouveau
11:46 < phh> stillunknown, paste the bug here, i think marcheu will read it when he will be back
11:46 < stillunknown> but the last commit message sais: Fix the DRM to work properly with the allocated cmdbuf.  The DDX is still
11:46 < stillunknown> broken hovever because it's FIFO is overlapping the front buffer.
11:46 < stillunknown> so he probably knows it's broken
11:46 < phh> right
11:47 < stillunknown> marcheu: http://www.rafb.net/paste/results/PVw8jA83.html <-- my Xorg.0.log if it's of any use
11:48 -!- Myrizio [n=Myrizio@host166-101.pool80104.interbusiness.it] has joined #nouveau
12:06 -!- Myrizio [n=Myrizio@host166-101.pool80104.interbusiness.it] has quit [Read error: 104 (Connection reset by peer)]
12:35 -!- Unavowed [n=silent@host86-132-160-124.range86-132.btcentralplus.com] has joined #nouveau
12:46 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
12:58 -!- Duke` [n=gnu@ANantes-251-1-99-188.w86-203.abo.wanadoo.fr] has joined #nouveau
13:26 < airlied> okay I've got a git repo up for the DDX..
13:27 < airlied> git://git.freedesktop.org/git/nouveau/xf86-video-nouveau
13:27 < airlied> or
13:27 < airlied> ssh://git.freedesktop.org/git/nouveau/xf86-video-nouveau
13:27 < airlied> it'll be up on gitweb in 10-15 mins..
13:28 < airlied> I actually started from the nv git repo, then imported all the changes from nouveau CVS so we have got full histsory..
13:28 < airlied> I think it sohuld all build.. but some of the changes  may not have  made it over..
13:34 < airlied> okay it now all builds and has shown up in gitweb... 
13:39 < stillunknown> airlied: will the other modules be moved as well?
13:40 < airlied> stillunknown: well the DRM probasbly will probably no point moving the RE work off sf.net
13:41 < stillunknown> i though the drm was a modified version of the normal drm module?
13:44 < stillunknown> and with most versioning systems branching and merging work best on the same server
13:45 < stillunknown> i'm not really familiar with git
13:45 < stillunknown> airlied: trigger
13:55 < stillunknown> it seems to build from git
13:56 < stillunknown> airlied: let me rephrase (i know you haven't seen any of this yet), why does the ddx benefit from git and the drm does not?
14:10 < airlied> drm master is already in git we will be using it ..
14:11 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Remote closed the connection]
14:15 < stillunknown> so will changes directly happen to git drm or only when it's finished?
14:22 < marcheu> stillunknown: yeah I know it's broken
14:22 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
14:23 < marcheu> darktama: you around ?
14:23 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Remote closed the connection]
14:26 < marcheu> hmm he's not around, but he commited a couple of stuff... let's build upon that :)
14:27 < marcheu> arghh no DDX commit insight
14:28 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
14:31 < stillunknown> anyone know how to force a newline in the wiki?
14:32 < marcheu> wohoo git ddx
14:36  * marcheu learns git...
14:36 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit ["<blank>"]
14:38 < marcheu> airlied: btw could we hook the git repo to our commit list ?
14:38 < marcheu> nouveau-cvs@lists.sourceforge.net it is
14:43 < stillunknown> http://nouveau.freedesktop.org/wiki/UserStatus <-- i have begun on some wiki pages
14:45 < marcheu> ah cool
14:45 < phh> stillunknown, two new lines iirc
14:45 < phh> hum not
14:45 < phh> it's latex
14:50 < marcheu> stillunknown: btw I think the daily tarball will continue, I just need to change my script for the git DDX
14:54 < stillunknown> marcheu: i usually just grab for cvs or svn stuff, since tarballs are harder to deal with and not as fast
14:54 < stillunknown> but i will include that
14:54 < marcheu> hmm you're probably right, at least git/cvs is always up to date :)
14:58 < stillunknown> did darktama break drm yesterday?
14:58 < marcheu> I did
14:59 < marcheu> we have to make the jump to a DDX that uses the drm memory allocator at some point...
15:00 < marcheu> so DRM is not really broken, but waiting for the DDX to catch up
15:02 < stillunknown> i will keep an eye of ddx changes :-)
15:10 -!- Duke` [n=gnu@ANantes-251-1-99-188.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
15:18 -!- Duke` [n=gnu@ANantes-251-1-99-188.w86-203.abo.wanadoo.fr] has joined #nouveau
15:46 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit [Remote closed the connection]
15:49 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
16:05 < darktama> hopefully I'll have it back in a workable state soon (ie. next couple of hours)
16:05 < darktama> EXA works again already
16:07 < stillunknown> renouveau has no make install?
16:07 < darktama> no, it's easiest to just run it from the build dir..
16:08 < stillunknown> i rarely do manual cvs builds
16:08 < stillunknown> guess i'll have to copy the content into opt
16:08 < stillunknown> */opt
16:15 < stillunknown> darktama: EXA is not completely working yet
16:15 < darktama> not in CVS, no :)
16:15 < stillunknown> remember that ddx is now in git
16:16 < darktama> it is? hmm, I guess I'll read the backscroll before I commit that then :)
16:17 < stillunknown> git://git.freedesktop.org/git/nouveau/xf86-video-nouveau/
16:17 < stillunknown> i think developer acces is git+ssh://
16:17 < stillunknown> (i don't have the log anymore, so doing of the top of my head)
16:18 < darktama> yeah, I have logs so I'll check it soon
16:24 < stillunknown> does renouveau run all tests by default?
16:28 < stillunknown> the logfile is a little under 1000 lines
16:29 < stillunknown> and a white triangle appeared in a black sqaure for a little while
17:04 < phh> stillunknown, nop
17:04 < phh> for default only one test
17:05 < stillunknown> how do i activate other tests?
17:10 < phh> edit main.c
17:10 < phh> and drop all //
17:10 < phh> (not the 2 last tests)
17:10 < phh> and only for test :)
17:12 < stillunknown> that's going to be annoying
17:13 < phh> use a true text editor :p
17:14 < Unavowed> sed -i -e 's/^\/\/\(\s*\)test/\1test/' main.c
17:15 < Unavowed> ;-)
17:16 < phh> huhu
17:16 < stillunknown> sed can be real magic
17:17 < phh> :)
17:22 < stillunknown> the amount of lines is crazy in the complete dump
17:22 < phh> stillunknown, in re.c
17:22 < phh> uncomment output multiple files
17:22 < phh> (or something like
17:29 -!- pmdata [i=patrice@ANantes-154-1-54-150.w81-53.abo.wanadoo.fr] has joined #nouveau
17:56 -!- Unavowed [n=silent@host86-132-160-124.range86-132.btcentralplus.com] has quit ["leaving"]
18:07 -!- Myrizio [n=Myrizio@host73-96.pool80104.interbusiness.it] has joined #nouveau
18:35 < darktama> ok, the ddx should work again now for most things
18:35 < darktama> stillunknown: yes, including EXA+Xv
18:37 < phh> there is aboslutely no 3D ?
18:37 < darktama> no
18:38 < darktama> well.. there's software mesa.. but I don't think that's what you meant :)
18:42 -!- shenki [n=shenki@121.44.69.112] has joined #nouveau
18:49 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has quit []
18:51 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau
18:52 < stillunknown> confirmed EXA+Xv
18:53 < darktama> cool :)
18:59 < zoeloelip> are there any benchmarks done with something like gtkperf to see the improvements against nv and nvidia?
19:00 < zoeloelip> they already have some nvidia benchmarks on their site: http://gtkperf.sourceforge.net/index.php?page=testing&id=2
19:05 < stillunknown> plain nv would require downgrading some things, i can probably do a comparison between nouveau and nvidia
19:06 < zoeloelip> most interesting would be nv against nouveau, to see the progress
19:07 < darktama> I doubt there'd be a much (any?) difference at all
19:07 -!- ag [i=ag@caladan.roxor.cx] has quit ["BRB"]
19:07 < phh> darktama, maybe AGP against PCI helps ?
19:07 < phh> (Ok i don't think so)
19:08 < marcheu> EXA vs XAA will definitely make a difference, although I'm not sure which one will be better
19:08 < darktama> yeah, I was just about to mention that
19:09 -!- ag [i=ag@caladan.roxor.cx] has joined #nouveau
19:09 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has quit ["Mike Patton"]
19:09 < marcheu> there's definitely some tuning we could do in EXA (in particular, I think the flushes are delayed too much sometimes)
19:10 < marcheu> but I think it's better to just wait for EXA core to be stable and tune from there
19:10 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has joined #nouveau
19:11 < phh> bah gtkperf will show higher scores ^^
19:13 -!- b33fc0d3 [n=bifter@gentoo/contributor/b33fc0d3] has quit [Client Quit]
19:14 < phh> stillunknown, what's needed to build nouveau driver so ?
19:17 < darktama> marcheu: btw, NV10_SCALED_IMAGE_FROM_MEMORY has BLEND_AND and BLEND_PREMUL
19:17 < marcheu> darktama: and BLEND_AND is non-premultiplied blending ?
19:18 < darktama> no idea, I'd assume so if BLEND_PREMUL is premultiplied
19:18 < darktama> might be worth testing
19:18 < darktama> we're using 0x77 (NV_SCALED_IMAGE_FROM_MEMORY) at the moment
19:18 < marcheu> yup, although IIRC lars said he couldn't find a non-premul blend
19:23 < darktama> hmm, is there a RENDER conformance test somewhere?
19:23 < marcheu> better off asking that in #xorg-devel I'd say
19:25 < darktama> yeah, I will do at some point.  not a priority at this stage.. setparam and cleaning up the ddx properly (as unappealing as that is) come first
19:25 < marcheu> ah, you mean 0x89
19:25 < darktama> yup
19:26 < marcheu> yeah, it'd be nice to make a single piece of code do composite on nv10+
19:29 < darktama> btw, what is pDRIInfo->framebuffer* referring to?  the entire video memory, or just the front buffer?
19:29 -!- Duke` [n=gnu@ANantes-251-1-99-188.w86-203.abo.wanadoo.fr] has left #nouveau []
19:29 -!- Duke` [n=gnu@ANantes-251-1-99-188.w86-203.abo.wanadoo.fr] has joined #nouveau
19:30 -!- tibbs is now known as tibbs|h
19:31 < marcheu> darktama: the whole range of memory that a DRI app can access
19:31 < marcheu> darktama: so that can be the whole VRAM but for example not the fifo if it's in vram
19:31  * marcheu tries to make sense
19:31  * marcheu fails
19:32 < darktama> yup, I think I understand.
19:32 < darktama> if it was the visible screen it'd have been fun.. we can't allocate it until we've done drmOpen().. and DRIScreenInit() does that
19:33 < darktama> another question, DGA.. do we care about it?
19:34 < stillunknown> phh: libdrm, x11-drm, xf86-video-nv (from git)
19:34 < marcheu> darktama: no
19:35 < darktama> good, that's one chunk of code I wont have to worry about fixing then
19:35 < phh> stillunknown, for libdrm it's from the CVS, the drm module and simply configure && make && make install ?
19:35 < stillunknown> all three from cvs or git
19:36 < phh> ah.
19:36 < stillunknown> probably, i just intergrated the whole thing into my package manager
19:36 -!- pmdata [i=patrice@ANantes-154-1-54-150.w81-53.abo.wanadoo.fr] has quit [Read error: 110 (Connection timed out)]
19:36 < phh> x11-drm set into MesaLib ?
19:36 < stillunknown> which has an eclass for modular x, which does a lot for me
19:37 < darktama> marcheu: so, figured out git yet?
19:37 < stillunknown> phh: what do you mean exactly?
19:38 < marcheu> darktama: I didn't, I think of a cvs command and read at the end of http://www.freedesktop.org/wiki/UsingGit :)
19:38 < phh> stillunknown, right, in fact i don't see what's x11-drm
19:38 < darktama> hehe, yeah.. I read that to learn how to push changes into the main repo
19:38 < darktama> hopefully I didn't mess anything up
19:39 < marcheu> well it seems just fine to me, but I'm not reference on the matter
19:39 < darktama> I've been using it locally as a sort of patch control system for a while.. and have messed up the repo a couple of times
19:40 < stillunknown> phh: gentoo seems to use snapshots for x11-drm package, the url is dri.sf.net
19:40 < marcheu> yeah, well I just have to migrate my ssh key to my new linux install before I can commit :)
19:40 < phh> stillunknown, and download url? 
19:40 < stillunknown> phh: no, project url
19:40 < darktama> ahh
19:40 < phh> i love unreadable ebuilds .........
19:41 < darktama> where are commit messages going to btw?
19:41 < darktama> they didn't seem to appear on nouveau-cvs
19:41 < marcheu> darktama: all fd.o gits are here : http://gitweb.freedesktop.org/
19:41 < marcheu> yeah, the git commit is not hooked to the mailing list. I'll pester airlied about that :)
19:41 < darktama> cool :)
19:41 < stillunknown> not unreadable, just not a 100% sure where the official packages/pages are
19:42 < phh> stillunknown, on sf.net/projects/dri there is only docs :(
19:43 < stillunknown> the point is that modular xorg have all sorts of packages, which do not have clear project homes
19:43 < stillunknown> because they are one
19:44 < stillunknown> what distro do you use?
19:44 < phh> LFS :p
19:46 < stillunknown> you must have built modular xorg and know how it all works
19:46 < phh> i know
19:46 < phh> and i have
19:46 < darktama> hmm, I switched to Gentoo from LFS.. so much better
19:46 < phh> (in fact it isn't a true LFS)
19:46 < darktama> LFS is far too hands-on for my liking
19:46 < stillunknown> portage causes an overhead on the system
19:47 < phh> stillunknown, Oo
19:47 < darktama> stillunknown: btw, is it just me or is portage steadily becoming sloooooow
19:48 < phh> let's rebuild xorg server with libdrm....
19:48 < stillunknown> phh: i like portage though
19:48 < darktama> it takes almost as much time to "Calculating dependencies..." as it does to compile :S
19:48 < phh> darktama, Oo
19:48 < stillunknown> what portage do you use darktama?
19:48 < stillunknown> what filesystem?
19:48 < darktama> whatever the latest ~amd64 is, and ext3
19:48 < phh> on my "scripted LFS", managing depends is too quick to bi humanly viewed..
19:48 < phh> s/bi/be
19:49 < phh> (and yes deps are managed)
19:49 < stillunknown> your own package manager, neat
19:49 -!- pmdata [i=patrice@ANantes-154-1-28-224.w81-53.abo.wanadoo.fr] has joined #nouveau
19:49 < phh> with 2 users as best \o/
19:49 < phh> euh
19:49 < phh> at most*
19:49 < phh> (english is a hard language :D)
19:50 < stillunknown> portage-2.1 has shown speed improvements, you can get improvements by hooking a non flat file db into portage
19:50 < phh> they use flat file DB ?!?
19:50 < darktama> yeah, I might use that cdb patch again actually.. stopped using at when portage 2.1 first appeared
19:50 < darktama> s/at/it/
19:51 < stillunknown> cdb works with a few changes
19:51 < stillunknown> http://gentoo-wiki.com/TIP_speed_up_portage_with_cdb
19:52 < stillunknown> phh: yes
19:52 < stillunknown> the idea is to reduce the deps of portage to python only
19:52 < phh> stillunknown, for installed package too ?
19:52 < stillunknown> so in the case of breakage you don't have multiple things to deal with
19:52 < phh> stillunknown, choosing between speed /5 and having to build sqlite, i prefer sqlite.
19:52 < stillunknown> everything is flat file
19:52 < phh> (my distrib even depends on Qt)
19:52 < stillunknown> as far as i know
19:53 < stillunknown> but there is place to use an external database
19:54 < stillunknown> and many people choose to use one for the speed advantages it has
19:54 < stillunknown> because portage db is huge
19:54 < darktama> hm, I might use the sqlite module instead actually
19:54 < phh> stillunknown, with "only" around 800packages DB is something like 40Mbyte...
19:55 < stillunknown> i might trt the sqlite module too
19:57 < darktama> considering it comes with portage, seems like a better choice
19:58 < stillunknown> i'm doing emerge --metadata now, so i'll see how it turns out
19:58 < darktama> doing the same now, it's taking a while.. hopefully just because wine is compiling too
19:59 < stillunknown> emerge --metadata on portage-2.1 always takes long the first time
20:00 < stillunknown> i'm at 75% now
20:00 < darktama> 31% here
20:00 < darktama> I have about 10 overlays to do aswell
20:00 < stillunknown> 10 overlays :-|
20:01 < darktama> hehe, some of them are redundant now.. mainline portage overtook them, just haven't gotten around to killing them off yet
20:01 < phh> xorg-server building......
20:01 < phh> zzz
20:02 < stillunknown> why xorg-server?
20:02 < phh> stillunknown, mine doesn't know using libdrm
20:05 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has quit ["Leaving"]
20:08 -!- phh [n=phh@moi.phhusson.info] has quit [Remote closed the connection]
20:15 -!- phh [n=jayce@moi.phhusson.info] has joined #nouveau
20:15 < phh> how do you build drm driver ?
20:16 < phh> I use the script in libdrm
20:16 < phh> but don't know if it's the easier way
20:16 < marcheu> cd linux-core ; make
20:16 < phh> .....
20:16 < phh> Why do i make so hard ?
20:17 < stillunknown> drm is a modified module, it was already that way (otherwise i couldn't have modified the ebuild so easily)
20:20 -!- skrull_far_away is now known as skrull
20:20 < darktama> wow, that Xi Graphics pdf (mentioned on dri-devel ML) is quite.. interesting
20:20 < phh> i explosed my kernel sourcre tree \o/
20:20 < phh> darktama: mm?
20:21 < marcheu> phh: he means "a pile of bullshit"
20:21 < darktama> marcheu: ahh, so you've read it then :P
20:21 < phh> what's Xi Graphics ?
20:21 < marcheu> heh, I expected something interesting in there
20:22 < darktama> phh: they make a proprietary X server
20:22 < phh> ah
20:22 < marcheu> not the details on how to design an outdated graphics architecture
20:22 < phh> :)
20:22 < marcheu> including the detail flaws
20:22 < darktama> I haven't quite finished reading it yet, just got past the Xorg/XFree86 bashing.. onto the OpenGL part
20:23 < phh> marcheu: they don't have a nvidia driver ?
20:23 < marcheu> that's the best part
20:23 < marcheu> phh: no, for some invalid reason like "we poor guys don't have tech specs"
20:23 < phh> marcheu: hehe :)
20:24 < phh> let's start trying launching nouveau
20:25 < phh> (now you have to say "good crash")
20:25 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Remote closed the connection]
20:26 < phh> it launches..
20:26 < phh> i switch console
20:26 < phh> rego
20:26 < phh> BOUM
20:26 -!- `Duke` [n=gnu@86.72.170.74] has joined #nouveau
20:26 < darktama> hmm, from what I can tell so far they don't make any attempt to stop say.. DMA'ing over the top of system memory randomly..
20:27 < phh> i should have something not uptodate, it crashes too quick
20:27 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau
20:27 < stillunknown> Xorg.0.log?
20:28 < phh> DMA queue hang: dmaPut=f, current=0, status=0
20:28 < stillunknown> http://nouveau.freedesktop.org/wiki/UserStatus <-- is this acceptable?
20:28 < darktama> hm, can you post the full log somewhere?
20:29 < stillunknown> phh: did you get the driver from git?
20:29 < phh> stillunknown: nop, CVS
20:29 < darktama> ahh, that's why then
20:29 < stillunknown> outdated
20:29 < phh> maybe everything isn't updated
20:29 < phh> aie
20:29 < phh> wait i rego to X
20:29 -!- phh [n=jayce@moi.phhusson.info] has quit ["leaving"]
20:30 -!- phh [n=phh@moi.phhusson.info] has joined #nouveau
20:30 < phh> so.
20:30 < stillunknown> http://nouveau.freedesktop.org/wiki/UserStatus <-- link to git
20:30 < phh> i first have to dl & install git :p
20:31 < darktama> I updated the main page link to point to fd.o's gitweb also
20:31 < stillunknown> darktama: what do you think of the new wiki page?
20:31 < darktama> I'll have a look in a sec, finishing reading this pdf first :)
20:32 < phh> darktama, you won't go under the table because of laught ?
20:32 < phh> ('tin c'est plus facile  dire en francais)
20:32 < darktama> bleh ok, that was a waste of my time really.. :S
20:32 < phh> lol
20:33 < stillunknown> darktama: the pdf was a waste of time?
20:33 < darktama> yeah
20:33 < phh> XAA en EXA are memory interface Oo
20:33 < stillunknown> am i mistaken?
20:34 < phh> i thaught it was an "acceleration architecture"
20:34 < darktama> stillunknown: nice, good to have something to point people to when they ask for the current status.
20:34 < darktama> yeah, might be an idea to remove that bit
20:34 < stillunknown> i'll fix that
20:34 < phh> stillunknown, wait i may be wrong !
20:34 < phh> now, how do we use git ? :)
20:34 < darktama> git clone <url>
20:35 < darktama> url is git://anongit.freedesktop.org/git/nouveau/xf86-video-nouveau
20:35 < stillunknown> phh: you're right
20:35 < phh> only xf86-video-nouveau at git ?
20:35 < darktama> yup
20:36 < phh> ok
20:36 < phh> darktama, and to update ?
20:36 < phh> git update ?
20:36 < darktama> git pull
20:36 < phh> ok
20:36 < stillunknown> darktama: i think the main wiki page is messy, i like structure, so that's why i don't use big screamy letters and such
20:37 < phh> i thaught that git was more... "down"
20:37 < phh> (like diff/patch but a bit easier)
20:38 < stillunknown> phh: i was confusing drm and exa a bit, that's what caused the very funny situation :-)
20:38 < phh> still one ask before
20:38 < phh> how activate EXA ?
20:39 < phh> i know the begin and the end
20:39 < phh> not the middle :)
20:39 < phh> (Option .......... Exa
20:39 < darktama> Option "AccelMethod" "EXA" (in the device section of xorg.conf)
20:39 < stillunknown>  Option "AccelMethod" "EXA"
20:39 < phh> thks
20:39 < stillunknown> i will add that to the wiki
20:39 < darktama> we'll probably make it default at some point
20:40 < phh> the thing i "hate" with modular Xorg, is that we need to set --prefix=/usr
20:40 < phh> so let's try
20:41 -!- phh [n=phh@moi.phhusson.info] has quit ["Quitte"]
20:42 -!- phh [n=phh@moi.phhusson.info] has joined #nouveau
20:42 < phh> it works.
20:42 < phh> XV works too
20:44 < phh> and i confirm it's slower than nvidia driver :/
20:46 < phh> at least in konsole
20:46 < stillunknown> it's only nv driver using exa and a drm module (functionally not much different)
20:47 < stillunknown> (the foundation has been laid)
20:47 < phh> stillunknown, last time i tried nv driver was some months (maybe one year), so it could have changed
20:47 < darktama> nv never changes :)
20:48 < stillunknown> nv adds pci id's sometimes, but that's it
20:48 < phh> ah.
20:48 < phh> iirc i add nforce2 at a time :D
20:48 < phh> ho no it was framebuffer
20:50 < stillunknown> i'll be away for a few hours, i've added a link from the main page and added info how to enable exa
20:54 -!- phh [n=phh@moi.phhusson.info] has quit [Remote closed the connection]
20:55 -!- phh [n=phh@moi.phhusson.info] has joined #nouveau
20:58 < marcheu> hmm, what are the chances I can switch NV_SCALED_IMAGE_FROM_MEMORY for NV10_SCALED_IMAGE_FROM_MEMORY and it works...
20:58 < marcheu> lets see
20:59  * darktama crosses fingers
20:59 < marcheu> hehe, we've got the reg file anyway :)
20:59 < phh> marcheu, what would that change ?
20:59 < darktama> true
21:00 < darktama> phh: we may be able to accelerate RENDER better if it works
21:00 < marcheu> phh: because there a "10" which changes everything
21:00 < marcheu> there is
21:00 < marcheu> phh: yeah, alpha blended windows if it works
21:00 < phh> marcheu, ahum.
21:00 < phh> aahh
21:00 < marcheu> and shadows
21:00 < phh> darktama, better ?
21:00 < phh> in which ways ?
21:00 < marcheu> and other stuff I don't use either
21:00 < phh> marcheu, shadow are special use of alpha ..
21:01 < darktama> marcheu just mentioned the big one, alpha blending
21:01 < marcheu> I know :)
21:01 < darktama> which, I do use :)
21:01 < phh> marcheu, GO try !
21:01 < phh> i want alpha bending :p
21:01  * darktama is going to tackle redirected Xv at some point too
21:06 < phh> marcheu, so boom ?
21:07 < darktama> it'd be nice to find a way of doing OutReverse also, so we can have fast subpixel AA
21:11 < marcheu> crashes in the kernel, I guess I didn't update everything
21:12 < darktama> :S don't tell me I broke something
21:18 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has joined #nouveau
21:21 < marcheu> well, I think I did with my mem changes
21:21 < marcheu> I get a oops so that's traceable
21:22 < darktama> hmm, I wonder why we don't see it
21:25 -!- Myrizio [n=Myrizio@host73-96.pool80104.interbusiness.it] has quit [Read error: 113 (No route to host)]
21:25 -!- pef [n=loic@ubuntu/member/pef] has joined #nouveau
21:25 < pef> hello
21:27 < marcheu> hello
21:27 < marcheu> darktama: hmm old DDX it seems. I'll rebuild from scratch except from my trees
21:27 < darktama> the CVS ddx isn't up to date btw
21:32 < marcheu> darktama: I guess we should just empty that module
21:32 < marcheu> darktama: and I'll submit a request to sf.net to remove it
21:33 < darktama> good idea, will prevent some confusion in the future
21:40 -!- littlesniper [n=littlesn@113.31.98-84.rev.gaoland.net] has joined #nouveau
21:47 -!- shenki_ [n=shenki@121.44.69.112] has joined #nouveau
21:49 -!- Myrizio [n=Myrizio@host66-98.pool80104.interbusiness.it] has joined #nouveau
21:50 < phh> marcheu, so ?
21:52 < marcheu> I'll rebuild everything from scratch, and write a compiling guide while I'm at it
21:53 < phh> you're still at building it Oo
21:55 < marcheu> ùmy kernel module crashes, not sure if its because of some changes I have...
21:56 < phh> good luck
22:06 < marcheu> so I guess no one runs 2.6.17
22:06 < phh> marcheu, i use 2.6.18-rc3
22:06 < darktama> I use .18-rc4-ck1
22:06 < phh> darktama, zuteuh
22:06 < phh> i haven't the biggest
22:06 < darktama> haha!
22:06 < marcheu> because I have that weird warning when building the drm
22:07 < phh> marcheu, which one ?
22:07 < darktama> which?
22:07 < phh> darktama, first NA
22:07 < marcheu> on linking
22:07 < phh> marcheu, i think i know which one, can you copy/paste ?
22:07 < marcheu> WARNING: /home/steph/build/drm/linux-core/drm.o - Section mismatch: reference to .exit.text: from .text between 'drm_cleanup_pci' (at offset 0x36d4) and 'drm_version'
22:07 < marcheu> WARNING: /home/steph/build/drm/linux-core/drm.o - Section mismatch: reference to .exit.text:drm_exit from __ksymtab between '__ksymtab_drm_exit' (at offset 0x58) and '__ksymtab_drm_init'
22:07 < darktama> yeah, I get them too
22:07 < phh> marcheu, i had strictly same
22:08 < phh> i fixed it but don't know how :D
22:08 < phh> ho yes i know
22:08 < phh> i dropped from the kernel config
22:08 < phh> make
22:08 < phh> ande readd
22:08 < phh> then it worked
22:08 < phh> marcheu, but iirc i had same for another driver and it worked
22:08 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
22:09 < marcheu> well, I get Aug 25 20:05:38 tubular-brain kernel: BUG: unable to handle kernel NULL pointer dereference at virtual address 00000007
22:09 < phh> marcheu, backtrace ?
22:09 < marcheu> no
22:09 < phh> Oo
22:10 < marcheu> I got one with another build...
22:11 < marcheu> http://rafb.net/paste/results/28SMku19.html
22:11 < marcheu> and well
22:12 < phh> i had a warning about a lock mmm
22:12 < marcheu> I can almost directly relate the linking warning I get to that 
22:13 < phh> marcheu, if it can helps i got a warning when i tried nouveau, saying that it tried to unlock something locked
22:13 < phh> so i think it's a bug in nouveau, with a workaround in 2.6.18
22:13 < phh> or something like :D
22:15 < marcheu> phh: yeah, that warning is harmless
22:15 < phh> ha.
22:16 < phh> it's in sys_exit_group Oo
22:16 < phh> that's pretty strange.
22:16 < marcheu> god
22:16 < marcheu> wasn't that fixed already
22:16 < marcheu> airlied: ping
22:17 < marcheu> airlied: it's probably the __exit in front of drm_exit. we have to choose if it's there (drm_drv.c) or not (drmP;h)
22:18 < pef> i have a nv20 based nvidia card (geforce ti 4200), are tests logs valuable for you for this card ? I already see in the tests directory logs for a nv28 card
22:19 < phh> ho it remembers me to make dumps for nforce2
22:20 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
22:22 < phh> darktama, Is FIFO channel 2 every time used ?
22:22 < phh> or there are exceptions ?
22:22 < phh> (right i have one under eyes)
22:22 < darktama> it depends on what apps are running
22:23 < pef> is the matrix features up to date ?
22:23 < darktama> if you start glxgears first you'll probably see FIFO 3 being used
22:23 < darktama> pef: for NV40 it is fairly accurate at least
22:23 < phh> darktama, ok
22:24 < phh> darktama, in fact there is no other OpenGL app
22:24 < phh> but it's the second X server
22:24 < pef> darktama: I just want to know if my nv20 card can be usefull, or if all needed data have already been collected :)
22:24 < darktama> phh: Xorg uses FIFO 0, no idea what uses 1 (reserved for aiglx).. GL usually starts from FIFO 2
22:25 < darktama> pef: I think pmdata will have a better idea about NV20, he's done a bit of work on it
22:25 < pef> darktama: ok, thanks
22:26 < darktama> anyway, I must sleep
22:26 < darktama> night
22:27 < pef> pmdata: hello, do you need help with nv20 ? I have a geforce ti 4200, maybe can I help
22:27 < pef> darktama: night !
22:27 -!- pef is now known as pef_aw
22:27 < phh> darktama, good night
22:29 < pmdata> arf, pef is out, 2 minutes late
22:29 < pmdata> hum a gf4ti is nv25/28, so same as marcheu one I guess
22:30 < phh> lol
22:31 < phh> nforce 2 tests are running
22:31 < phh> zzzzzz
22:33 -!- predatorfreak [n=predator@adsl-69-212-32-15.dsl.sfldmi.ameritech.net] has quit ["<blank>"]
22:34 -!- littlesniper [n=littlesn@113.31.98-84.rev.gaoland.net] has quit []
22:55 < phh> BAM
22:55 < phh> freeze.
22:56 < phh> (while trying renouveau)
23:02 -!- pmdata [i=patrice@ANantes-154-1-28-224.w81-53.abo.wanadoo.fr] has quit [Read error: 60 (Operation timed out)]
23:11 < swany> Ok, so do I need a 2.6.18 kernel for the module to not crash right now. Or is there a workaround?
23:11 < marcheu> no, that won't change anything
23:11 < marcheu> I have the same bug, so rest assured that this issue will be investigated :)
23:12 < marcheu> swany: do you get those crashes ?
23:12 < swany> yes
23:15 < phh> why i haven't mm
23:17 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has quit []
23:18 -!- swany [n=swany@81-234-181-143-o1108.tbon.telia.com] has joined #nouveau
23:24 -!- pmdata [i=patrice@ANantes-154-1-28-224.w81-53.abo.wanadoo.fr] has joined #nouveau
23:34 -!- Myrizio [n=Myrizio@host66-98.pool80104.interbusiness.it] has quit [Read error: 113 (No route to host)]
23:34 -!- Myrizio_ [n=Myrizio@host27-102.pool80104.interbusiness.it] has joined #nouveau
--- Log closed Сбт Авг 26 00:00:22 2006
