--- Log opened Птн Авг 18 00:00:05 2006 --- Day changed Птн Авг 18 2006 00:00 < pmdata> can you post a dump url? 00:01 < KoalaBR> I need to redump, because I split the whole test in smaller parts 00:01 < pmdata> "yesterday: 23:57 < pmdata> nv40[0x1ee8] seems to be the point sprite parameter" 00:01 < KoalaBR> Yes 00:02 < KoalaBR> regl.TexEnvi(GL_POINT_SPRITE_NV, GL_COORD_REPLACE_NV, GL_FALSE); leads to 00:02 < KoalaBR> NV30_TCL_PRIMITIVE_3D [0x1ee8/4] = 0x00000801 | UNKNOWN = 00000801 00:03 < KoalaBR> NV30_TCL_PRIMITIVE_3D [0x1ee8/4] = 0x00000001 | UNKNOWN = 00000801 00:03 < KoalaBR> Last one is the correct one 00:03 < KoalaBR> 8 is seen if you use GL_TRUE 00:03 < KoalaBR> Bit 0 is only 0 if setup runs / reset perhaps 00:05 < KoalaBR> Bit 1 is set for GL_R and Bit 2 = GL_S 00:05 < KoalaBR> That's what I found out, but wait I'll post a complete test 00:07 < KoalaBR> pmdata: http://sh.nu/p/2759 00:08 -!- sturmflut [n=sturmflu@2001:6f8:992:1337:213:d4ff:feaa:8a9f] has joined #nouveau 00:09 < KoalaBR> pmdata: was that all you were looking for in that test? 00:15 < KoalaBR> pmdata: Please look here: http://sh.nu/p/2760 00:15 < KoalaBR> pmdata: the first dump, was with GL_FALSE and did not have 1ee8 commands ... 00:15 < KoalaBR> This one is with GL_TRUE and has 1ee8 in it 00:19 < KoalaBR> Well, that's it for today. Good night all 00:20 -!- KoalaBR [n=KoalaBR@port-83-236-15-108.dynamic.qsc.de] has quit ["ChatZilla 0.9.61 [Mozilla rv:1.7.13/20060417]"] 00:22 < pmdata> don't forget that 1ee8 maybe set to disable in test_startup 00:24 < pmdata> try to copy the nv20_tcl_primitive_3d_point_sprite in nv_objects.h/objects.c and put in in nv30_tcl as 1ee8 to see if it matches, I think so 00:24 < pmdata> arg, he's gone 00:25 < Duke`> I'll read the log 00:26 < pmdata> duke> do you have nv30 or nv40? 00:29 < Duke`> nv30 but not mounted at this time (nv10 installed) 00:37 < pmdata> can you softquadro it? 00:41 < Duke`> the nv10 I failed, the nv30 I haven't tried yet 00:55 < atcl> hi there, does somebody here have a dump, in which the program is rendering a texture with a fragment shader, i like to find out how that works, and may be how the gpu is moved to use the fp. 00:55 < pmdata> there is a test_nv_texture_shader in renouveau 00:55 < atcl> ok 00:55 < atcl> ill check 00:55 < pmdata> just look in the main.c test_* calls 00:56 < pmdata> there is also a glsl one 00:59 < atcl> ok iam probably to stupid finding it.., 01:02 < pmdata> there are also some test_fragprog in it 01:02 < atcl> yes, i found them in the main,... 01:03 < darktama> if you're on a >=nv40 card, just run test_tex() 01:03 < atcl> oki 01:03 < atcl> i do, actually, (nv43) 01:04 < darktama> ok, then it'll be using a fragment program regardless of whether you tell it to or not :) 01:06 < atcl> got it now, 01:06 < atcl> thx 01:07 < darktama> np 01:09 -!- sturmflut [n=sturmflu@2001:6f8:992:1337:213:d4ff:feaa:8a9f] has quit [Remote closed the connection] 01:12 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has quit [Remote closed the connection] 01:17 -!- atcl [n=Universe@Lb465.l.pppool.de] has quit [Read error: 104 (Connection reset by peer)] 01:20 -!- pmdata [i=patrice@ANantes-154-1-68-160.w86-195.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"] 01:23 -!- predatorfreak [n=predator@adsl-69-213-80-253.dsl.sfldmi.ameritech.net] has joined #nouveau 01:31 -!- maxtoo [n=maxtoo@berryx.homedns.org] has quit [Remote closed the connection] 01:54 -!- Duke` [n=gnu@ANantes-251-1-142-44.w86-210.abo.wanadoo.fr] has quit [Remote closed the connection] 02:22 -!- K [i=hazel@tor/session/external/x-bb78ab9175d1edbb] has quit [Remote closed the connection] 02:55 < ddl> hey! 02:55 < ddl> home again :) 03:00 < marcheu> hey ddl, nice to have you back 03:02 < ddl> thanks :) 04:36 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has joined #nouveau 05:14 * Myrizio will now try to softquadro its card... 05:15 -!- Myrizio [n=Myrizio@host222-102.pool80104.interbusiness.it] has quit ["Goodbye Ruby Tuesday (Perl Friday)"] 05:26 -!- Myrizio [n=Myrizio@host124-101.pool80104.interbusiness.it] has joined #nouveau 05:36 * Myrizio was unable to softquadro his card (10de:0110 (rev a1)), MISSION FAILED. 07:07 -!- Myrizio [n=Myrizio@host124-101.pool80104.interbusiness.it] has quit ["Goodbye Ruby Tuesday (Perl Friday)"] 07:08 -!- predatorfreak [n=predator@adsl-69-213-80-253.dsl.sfldmi.ameritech.net] has quit [""] 08:43 -!- stillunknown_ [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau 08:46 -!- Netsplit calvino.freenode.net <-> irc.freenode.net quits: stillunknown, Aexoden, ddl, cptn, hiyuh 08:47 -!- Netsplit over, joins: ddl 08:55 -!- Aexoden [n=Aexoden@207-118-82-54.dyn.centurytel.net] has joined #nouveau 08:55 -!- cptn [n=jw@217-162-119-116.dclient.hispeed.ch] has joined #nouveau 09:03 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has joined #nouveau 09:57 -!- predatorfreak [n=predator@adsl-69-213-80-253.dsl.sfldmi.ameritech.net] has joined #nouveau 10:46 -!- EdB [n=EdB@ARennes-251-1-16-116.w83-195.abo.wanadoo.fr] has joined #nouveau 10:47 -!- predatorfreak [n=predator@adsl-69-213-80-253.dsl.sfldmi.ameritech.net] has quit [""] 11:59 -!- Unavowed [n=silent@host81-151-27-126.range81-151.btcentralplus.com] has joined #nouveau 12:16 -!- Duke` [n=gnu@ANantes-251-1-102-30.w86-203.abo.wanadoo.fr] has joined #nouveau 12:40 -!- maxtoo [n=maxtoo@berryx.homedns.org] has joined #nouveau 13:29 -!- Duke` [n=gnu@ANantes-251-1-102-30.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"] 13:33 -!- Duke` [n=gnu@ANantes-251-1-102-30.w86-203.abo.wanadoo.fr] has joined #nouveau 14:06 -!- Myrizio [n=Myrizio@host10-101.pool80104.interbusiness.it] has joined #nouveau 17:26 -!- Myrizio_ [n=Myrizio@host115-101.pool80104.interbusiness.it] has joined #nouveau 17:32 -!- Myrizio [n=Myrizio@host10-101.pool80104.interbusiness.it] has quit [Read error: 113 (No route to host)] 18:13 -!- K [i=hazel@tor/session/external/x-fd67a977a03b6c16] has joined #nouveau 19:37 -!- hiyuh [n=hiyuh@ZL050248.ppp.dion.ne.jp] has quit ["Leaving"] 19:37 -!- Unavowed [n=silent@host81-151-27-126.range81-151.btcentralplus.com] has quit ["leaving"] 19:40 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau 20:00 -!- littlesniper [n=guillaum@68.174.97-84.rev.gaoland.net] has joined #nouveau 20:04 -!- pmdata [i=patrice@ANantes-154-1-58-15.w81-53.abo.wanadoo.fr] has joined #nouveau 20:43 -!- Myrizio_ [n=Myrizio@host115-101.pool80104.interbusiness.it] has quit ["Goodbye Ruby Tuesday (Perl Friday)"] 20:43 -!- pmdata [i=patrice@ANantes-154-1-58-15.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"] 20:45 < darktama> mm, quiet in here tonight :) 20:45 < marcheu> hey darktama 20:46 < marcheu> darktama: are you currently doing anything with drm/ddx ? otherwise I'll start to break as much stuff as I can as quickly as possible 20:46 < marcheu> in as many ways as I can find 20:47 < darktama> nope, I've started work on >=NV30 EXA Composite stuff; but until we get 3D going it's useless anyway 20:47 < darktama> so, break away :) 20:48 < marcheu> do you plan to do it with the 3D engine ? 20:48 < darktama> yup 20:48 < marcheu> because the alpha blit available on nv10+ is premultiplied 20:48 < marcheu> I was hoping we could find one that isn't, and extend support down to nv10 20:49 < marcheu> btw does my latest drm work for you ? 20:49 < darktama> I recall reading on the Xorg ML at some stage that the original writer (Lars?) had ideas on how to do it, maybe we should contact him 20:49 < darktama> I haven't tried it yet actually 20:49 < marcheu> yeah, I met lars at fosdem 20:50 < marcheu> maybe we could fire up a couple of emails, he doesn't hang on irc much those days.. 20:51 < darktama> does the current composite hook work for you? it hangs e17 very quickly on my card 20:51 < marcheu> it's deadly slow 20:51 < marcheu> and it timeouts very often 20:51 < marcheu> and, yeah, crashes after some seconds 20:52 < marcheu> we still have to fix that stoopid timeout 20:52 < darktama> hm, perhaps we should figure out the FIFO error reporting ASAP.. might give us a better idea of where these issues are 20:53 < marcheu> hmm, it's fixed in cvs ? 20:53 < marcheu> darktama: you did that ? 20:53 < darktama> error reporting? 20:54 < marcheu> the NVDoSync 20:54 < darktama> oh, I just made it kill X if the timeout is too long.. I got sick of rebooting 20:54 < marcheu> at least the timeout is not CPU-dependent anymore 20:55 < marcheu> I really think my drm update is the culprit for my libdri crashes 20:58 < darktama> mm, I'll update from cvs and see what happens here 21:07 < darktama> mm, something is borked 21:07 < darktama> XAA shows a garbled screen, then hangs 21:07 < darktama> EXA shows the X grid thingo, then hangs 21:07 < darktama> oops, wrong channel :S 21:07 < darktama> just typed that in #winehackers by mistake... 21:07 < marcheu> hehe, yeah I can't even get that far 21:08 < marcheu> I'll se what I did wrong in there 21:14 -!- pmdata [i=patrice@ANantes-154-1-58-15.w81-53.abo.wanadoo.fr] has joined #nouveau 21:17 < pmdata> hello 21:18 < Duke`> hi 21:24 < darktama> marcheu: [87155.258559] [drm] DMA command buffer is 1024KiB at 0xfff00000(VRAM) 21:24 < darktama> that could be a problem, my vram is at 0xE0000000. 21:26 < marcheu> yeah, same happens here 21:27 < marcheu> but 21:29 < darktama> hm, it looks like the code that initialises fb_usable_size is #if 0'd out.. the cmdbuf base is calculated using it 21:32 < marcheu> yeah, that's what I was thinking :) 21:33 < marcheu> it was dumb on my part to update everything at once - drm, ddx, Xorg... 21:34 < darktama> mm, makes it hard to find bugs that way :) 21:37 < darktama> so, I guess for now we should leave it at BIOS setup mem_amount()-(560*1024) etc. until we can use the mman to alloc it 21:38 < marcheu> hmm ? 21:39 < marcheu> ah ok 21:44 < darktama> ok, that fixed it for me 21:45 < darktama> EXA at least, didn't bother trying XAA 21:45 < marcheu> cool, I'll try to move X to use the new memory manager in the next couple of hours 21:45 < marcheu> see what happens... 21:45 < darktama> can we do that yet? I didn't think EXA had the hooks to let the driver alloc memory itself? 21:46 < marcheu> well... I'll put in a hack to allocate a big exa chunk 21:46 < marcheu> and also have the front buffer, cursor and friends allocated through drm 21:46 < darktama> ah, good idea. 21:46 < marcheu> btw any idea why our DDX has that 256Mb limit ? 21:47 < darktama> which limit? 21:47 < marcheu> on vidram 21:47 < darktama> no idea, where is that? 21:47 < marcheu> that's from memory let me grep.. 21:47 < marcheu> nv_setup.c 21:48 < marcheu> if(pNv->RamAmountKBytes > 256*1024) 21:48 < marcheu> pNv->RamAmountKBytes = 256*1024; 21:49 < marcheu> oh also, I'm making all offsets 64 bit 21:49 < marcheu> that'll help a lot in the future, for a number of reasons 21:49 < darktama> 2005-08-04 Alan Coopersmith 21:49 < darktama> * programs/Xserver/hw/xfree86/drivers/nv/nv_setup.c: 21:49 < darktama> Bug #3987 21:49 < darktama> Patch #3264 21:49 < darktama> Workaround for 512 Meg video cards (can only map 256 Meg of it). 21:49 < darktama> (Mark Vojkovich) 21:50 < marcheu> "can only map 256 Meg of it" ?? 21:50 < darktama> yes, it's a bit vague.. possible like radeon's max-128MB CPU accessible limit? 21:50 < darktama> possibly* 21:50 < marcheu> wel... radeon has a hw limitation 21:50 < marcheu> maybe geforces have one as well.. 21:51 < marcheu> if that's the case, we have to take it into account in the memory allocator, because X likes to have pixmaps mmaped, while we don't care for 3D 21:52 < marcheu> hmm we can get around by adding a 3rd heap with non-mapable vram 21:53 < darktama> With EXA direct access is wrapped by PrepareAccess()/FinishAccess()? Can we use that to move pixmaps around if it's in a non-accessible area? 21:54 < marcheu> yeah, if you like moving things around, sure :) 21:54 < marcheu> but the other way is allocating mmapable ram if possible 21:55 < darktama> that's probably the better way actually :) 21:55 < marcheu> I'm not starting to add bounce buffers for VRAM anyway :) 21:55 < darktama> hehe 21:56 < darktama> so, on the topic of the ddx.. can we kill XAA? *hopeful glare* 21:56 < marcheu> hmm, you don't have any EXA crashes ? :) 21:57 < marcheu> did you try to VT switch ? ;) 21:57 < marcheu> ah wait, you've got a 6600 right ? 21:57 < darktama> 6800GT.. I *think* VT switching worked 21:57 < marcheu> wow 21:57 * darktama tries 21:57 < marcheu> I have to fix stuff here before this works 21:58 * marcheu says bye bye to darktama 21:58 < darktama> yup, I'm still here.. survived one VT switch.. 21:58 * darktama tries again 21:58 < marcheu> ah well 21:58 < darktama> just survived a couple.. now back to nvidia's driver so I can have both screens back :( 21:59 -!- predatorfreak [n=predator@adsl-69-213-80-253.dsl.sfldmi.ameritech.net] has joined #nouveau 22:08 -!- lumag_offline [n=mitya@chimpanzee.school.ioffe.ru] has joined #nouveau 22:08 -!- Topic for #nouveau: http://nouveau.freedesktop.org | open source 3D acceleration for nvidia cards | card dumps at http://nouveau.sourceforge.net/tests/ | http://perso.orange.fr/patrice.mandin/images/nv-kitten.jpg 22:08 -!- Topic set by sturmflut [] [Sun Aug 13 17:02:24 2006] 22:08 [Users #nouveau] 22:08 [@ChanServ] [ ajmitch ] [ Duke` ] [ marcheu] [ pmdata ] [ tibbs] 22:08 [ Aexoden ] [ cptn ] [ EdB ] [ maxtoo ] [ predatorfreak] 22:08 [ ag ] [ dagb ] [ K ] [ nano- ] [ stringfellow ] 22:08 [ airlied ] [ darktama] [ lumag_offline] [ philv ] [ swany ] 22:08 -!- Irssi: #nouveau: Total of 21 nicks [1 ops, 0 halfops, 0 voices, 20 normal] 22:08 -!- Channel #nouveau created Sun Jun 25 07:52:56 2006 22:08 -!- Irssi: Join to #nouveau was synced in 1 secs 22:09 < marcheu> hmm, libdri still broken :) 22:09 < darktama> not good :( 22:13 -!- gajownik_away [i=gajownik@zspswidwin.pl] has joined #nouveau 22:18 -!- ddl [i=erikw@montezuma.acc.umu.se] has joined #nouveau 22:21 * pmdata can not connect to nouveau.cvs.sf.net: "ssh: connect to host nouveau.cvs.sourceforge.net port 22: Connection refused" 22:22 -!- [freenode-info] help freenode weed out clonebots, please register your IRC nick and auto-identify: http://freenode.net/faq.shtml#nicksetup 22:22 < marcheu> webcvs doesn't work for me 22:22 < marcheu> maybe that's related 22:23 < darktama> mm, I checked out a fresh copy of the drm and ddx about 1/2 hour ago 22:23 < marcheu> although I don't have X up & running now so I can't check the sf.net page about errors 22:23 < marcheu> but on the upper right corner, there is a green/yellow ball that has the site status 22:25 < darktama> hmm, the status says that CVS is working fine.. it isn't though :) 22:26 < marcheu> yeah, that's what I could gather from running links on their page.. 22:29 -!- gajownik_away [i=gajownik@zspswidwin.pl] has quit [Read error: 110 (Connection timed out)] 22:34 < darktama> sometimes it's easier to read the HTML than to use links :) 22:41 -!- KoalaBR [n=KoalaBR@port-83-236-13-53.dynamic.qsc.de] has joined #nouveau 22:42 < KoalaBR> Hi all 22:42 < darktama> hey KoalaBR 22:43 < KoalaBR> Does anyone here know, whether the Wiki can use precreated html pages? 22:43 < KoalaBR> So that I can simply upload them? 22:44 < KoalaBR> pmdata: Read your suggestions and questions you offered after I logged off. If you haven't done it yourself, I will have a look.. 22:45 < KoalaBR> And finally: Yes, I can't currently connect to cvs on SF too 22:46 < pmdata> koala> I can leave some work for other people, I prefer to finish nv10 tcl 22:46 < pmdata> I wait for nv1x quadro dumps 22:48 < KoalaBR> pmdata: Ok, I will just do the tests you asked me to do and will report back 22:48 < marcheu> pmdata: you mean you're waiting for me ? 22:48 < marcheu> pmdata: or someone else ? 22:49 < marcheu> pmdata: I could plug the nv1x into my machine if you like, I just have to remove the nv28 22:50 -!- K [i=hazel@tor/session/external/x-6d6e619018345bca] has quit [Remote closed the connection] 22:51 < KoalaBR> CVS works again, at least for me 22:55 -!- Netsplit over, joins: marcheu, airlied 22:56 < marcheu> pmdata: which one you prefer ? nv11, nv15 or nv18 ? 22:57 -!- stillunknown [n=madman20@82-168-177-167.dsl.ip.tiscali.nl] has joined #nouveau 23:02 < KoalaBR> pmdata: You were right, except for the different address, NV30_TCL_PRIMITIVE_3D_POINT_SPRITE works the same as NV20_TCL_PRIMITIVE_3D_POINT_SPRITE 23:04 < marcheu> darktama: yeah, EXA without composite is stable here 23:04 < marcheu> maybe we can slash XAA already 23:05 < marcheu> darktama: also, maybe we could slash fbdev support, I don't know, that's up for discussion 23:05 < darktama> sounds good to me :) 23:05 < stillunknown> there is actually a half working driver with a new interface? 23:06 < marcheu> darktama: I think the new sync loop is helping a lot here 23:06 < darktama> I'm a little worried right now.. I've either gotten the 3D engine to work, or somehow my code is using the normal EXA codepath and reporting otherwise in Xorg.0.log 23:06 < marcheu> I wouldn't say "worried" 23:06 < marcheu> but maybe you ran the proprietary driver before, which setup some magic state 23:07 < marcheu> nvidia cards have some state that survive even reboots (like the objects) 23:07 < darktama> well, I was doing that before.. and it still didn't work.. I think this must be the normal EXA path.. it just doesn't make sense that it'd work now "magically" 23:08 < KoalaBR> darktama, marcheu: If you want me to write some docs (how to get "test ready") tell me how to do it and I will 1. try it out myself 2. write a small intro doc 23:08 < KoalaBR> Whenever you feel it would be helpful. Just ask me at that time 23:09 < pmdata> marcheu> nv18 , in softquadro mode please (and a cup of coffee) 23:10 < marcheu> KoalaBR: ok 23:11 < marcheu> pmdata: ok, I'll eat and switch the cards when I'm back. but... I don't have coffee since I never ever drink coffee 23:11 < darktama> you don't drink coffee!!! 23:12 < KoalaBR> He is not alone :) 23:12 < KoalaBR> I prefer tea, and I'm not british :) 23:12 < darktama> well, it's 5am here.. I *need* coffee 23:12 < darktama> :) 23:12 < stillunknown> there are even somewhat normal people who don't drink coffee :-) 23:13 < KoalaBR> Darktama: You are excused :) 23:13 < stillunknown> darktama: why aren't you in bed? 23:13 < KoalaBR> He drank too much coffee 23:13 < darktama> haha :) 23:13 -!- Myrizio [n=Myrizio@host180-101.pool80104.interbusiness.it] has joined #nouveau 23:13 < KoalaBR> SCNR 23:14 < darktama> no, I was deciding on how I'm going to implement the shader compiler.. then I ended up messing with the ddx somehow :S 23:15 < stillunknown> sleep is good, because we need someone to maintain the driver later on:-) 23:15 < pmdata> marcheu> oh I was waiting for t0ma7 dumps in fact, but I need to check some stuff on nv1xgl (softquadro or not) 23:29 < darktama> KoalaBR: what docs did you want to write? 23:32 < KoalaBR> darktama: If you think the code is advanced enough, what you need to check out, build, how to configure the system to use your new code (Driver, DRM?) 23:33 < KoalaBR> In the same vein, as the intro to renouveau I wrote 23:34 < KoalaBR> So that we have a standard answer for that code too "Read docs blabla and report back" or something like that 23:35 < darktama> ahh, ok 23:37 < KoalaBR> Just ask me, if you think others (devs at first) can try it too. I will play the guinea pig and if I survive, I will write some HOWTO doc :) 23:42 < pmdata> yep, would be nice to know if the nouveau stuff is barely tryable 23:44 < darktama> ok, so it wasn't the 3D engine.. I didn't fill exa_major/exa_minor and it fell back to software :( but, I do get some rendering and then a hang with them filled in.. I wish I knew if they were sw-fallbacks or not 23:51 < marcheu> pmdata: ok, I've made it into a nVidia Corporation NV18GL [Quadro4 380 XGL] (rev c1) 23:51 < pmdata> thanks 23:52 < marcheu> however, the nv18 doesn't use single-vertex stuff 23:52 < marcheu> nv18hl 23:53 < marcheu> nv18gl 23:54 < marcheu> hmm, maybe that variant does actually 23:54 < marcheu> it looks like quadro NVS cards behave like geforces, though --- Log closed Сбт Авг 19 00:00:17 2006