--- Log opened Вск Июн 11 00:00:25 2006
00:18 < marcheu> hmm it looks like bios parsing happens in the kernel nvidia module
00:20 < EdB> and free drivers don't do that ?
00:21 < marcheu> the haiku driver can parse bioses up to nv30
00:21 < marcheu> and ddl has investigated it on nv40
00:21 < marcheu> and I'm trying to get all that to work inside the DDX right now
00:22 < EdB> and nv from x.org
00:22 < EdB> ?
00:22 < marcheu> nope, it doesn't do that
00:22 < marcheu> it doesn't use the bios at all
00:22 < EdB> really a basic drver so :o)
00:31 -!- pmdata [n=patrice@ANantes-154-1-54-97.w81-53.abo.wanadoo.fr] has quit ["using sirc version 2.211+KSIRC/1.3.11"]
01:32 -!- EdB [n=EdB@ARennes-251-1-8-250.w83-195.abo.wanadoo.fr] has quit ["Konversation terminated!"]
01:46 -!- Lumag [n=Lumag@chimpanzee.school.ioffe.ru] has left #nouveau []
02:59 -!- Duke` [n=gnu@ANantes-251-1-155-73.w86-203.abo.wanadoo.fr] has quit ["Fatal signal: Segmentation Fault"]
03:28 -!- |pedro [n=kvirc@c906e04c.virtua.com.br] has joined #nouveau
03:35 -!- CoronaLVR [n=alex@80.178.203.74.adsl.012.net.il] has joined #nouveau
09:59 -!- |pedro [n=kvirc@c906e04c.virtua.com.br] has quit ["KVIrc 3.2.0 'Realia'"]
10:01 -!- pedro [n=pedro@c906e04c.virtua.com.br] has joined #nouveau
10:05 -!- pedro [n=pedro@c906e04c.virtua.com.br] has quit [Read error: 104 (Connection reset by peer)]
10:13 -!- CoronaLVR [n=alex@80.178.203.74.adsl.012.net.il] has quit [Remote closed the connection]
10:55 -!- Lumag [n=Lumag@chimpanzee.school.ioffe.ru] has joined #nouveau
11:10 -!- pedro [n=kvirc@c906e04c.virtua.com.br] has joined #nouveau
11:19 -!- Duke` [n=gnu@ANantes-251-1-104-166.w86-203.abo.wanadoo.fr] has joined #nouveau
11:25 -!- pedro [n=kvirc@c906e04c.virtua.com.br] has quit [Remote closed the connection]
11:50 -!- EdB [n=EdB@ARennes-251-1-16-96.w83-195.abo.wanadoo.fr] has joined #nouveau
11:50 -!- pedro [n=kvirc@c906e04c.virtua.com.br] has joined #nouveau
13:07 -!- Duke` [n=gnu@ANantes-251-1-104-166.w86-203.abo.wanadoo.fr] has quit [Remote closed the connection]
13:54 -!- pmdata [i=patrice@ANantes-154-1-96-128.w86-210.abo.wanadoo.fr] has joined #nouveau
14:10 -!- pmdata is now known as pmdata_f1
14:48 -!- Duke` [n=gnu@ANantes-251-1-104-166.w86-203.abo.wanadoo.fr] has joined #nouveau
16:45 -!- pmdata_f1 is now known as pmdata
17:14 < Lumag> I have made a preliminary version of patch that switches renoveau from 'commands' to 'objects' which is more correct representation of nVIdia state. The patch is available at http://lumag.spb.ru/nouveau/objects.diff .
17:48 < pmdata> marcheu> apart from various fosdem 2006 reports, and the article I wrote for linuxfr.org, do you know other sites which talked about nouveau project?
18:15 < EdB> pmdata, some people comme from nvidia forum :o)
18:22 < marcheu> Lumag: looks good, we still have to port cmds.h though
18:22 < marcheu> pmdata: no, I don't know more than that
18:23 < marcheu> oh yes, people from nvnews.net
18:23 < darktama> what does the change do exactly? give us the ability to translate commands based on the object bound to the FIFO channel?
18:26 < marcheu> darktama: AIUI at least, yes
18:26 < marcheu> I didn't try it yet
18:26 < darktama> me neither (hacking on my e17 taskbar module atm)
18:33 < Lumag> darktama: 1) it corrects mask used for size/channel/etc. 2) allows you to bind 'commands' to objects, not subchannels. A subchannel can be bound to various objects and a single object can be used in various subchannels.
18:34 < Lumag> and 3) it allows pretty printing bitmap fields.
18:35 < darktama> 3) could be useful to get rid of PRINT_FUNC for the texture commands :)
18:36 < darktama> it can't be used for shaders though, as the layout changes for some opcodes
18:59 < pmdata> another thing with commands: not all parameters have the same size/structure (i.e. you can have a 4*8bit parameter then a 32bit one)
19:06 -!- stringfellow [n=stringfe@dsl220-96-100.fastxdsl.nl] has quit [Remote closed the connection]
19:08 < darktama> Lumag: what does the NVType00[] lines mean? is that what we were prevously calling commands?
19:09 < darktama> (and the parameters to them)
19:14 < Lumag> More or less so. an NVTypeXX represents unknwon object name and the object fields were called commands.
19:14 < Lumag> There is one object defined (Dx5TextureTriangle). I'll upload sample dump in a moment.
19:15 < darktama> I'm looking at a dump at the moment, I guess NV40 doesn't use Dx5TexturedTriangle
19:16 < marcheu> heh, even nv10 doesnt :)
19:16 < darktama> ah :)
19:16 < marcheu> although nv10 can use it, but that one doens't make use of TCL
19:17 < Lumag> :)
19:17 < Lumag> Only old hw at hand currently :)
19:17 < Lumag> lumag.spb.ru/nouveau/dumps/sample_renoveau
19:19 < Lumag> pmdata: well... yes, but those wries can be split to uint32 writes.
19:24 < darktama> where does the 0x54 value come from for tho NV04_DX5_TEXTURED_TRIANGLE define?
19:24 < darktama> the*
19:25 < Lumag> I got it from ioctl param and the program decodes it from the instance memory.
19:25 < darktama> here, all the values pulled from RAMIN are 0x00000000 for some reason
19:26 < darktama> I can post a dump if it's useful
19:26 < Lumag> strange... What card do you have?
19:26 < darktama> NV40 (6800 GT)
19:27 < Lumag> pci-e?
19:27 < darktama> nope
19:27 < Lumag> strange... Yes, please post a dump.
19:28 < darktama> http://rafb.net/paste/results/ocjNnA58.html
19:28  * pmdata will try lumag patch on his nv15
19:28  * Lumag thanks pmdata
19:28 < marcheu> darktama: does it work the 5th time if you read the value 5 times in a row ?
19:29 < marcheu> darktama: some 6600/6800 have that problem with PROM, maybe it also happens with other areas
19:29 < darktama> so, modify the code to read RAMIN 5 times?
19:29 < marcheu> yup
19:30 < darktama> wow, that'd be quite a hardware bug if that was the case :)
19:30  * darktama tries
19:30 < marcheu> also, objects on nv40 are 8 dwords and not only 4
19:31 < darktama> ah, that's true
19:31 < marcheu> I think there's a number of other differences in that area as well
19:31 < marcheu> that might explain those 0s
19:32 < pmdata> lumag> it works, do you want a dump?
19:32 < Lumag> Yes, please
19:35 < darktama> ok, reading 5 times doesn't change anything..
19:36 < marcheu> you did to hammer 5 times the same value in a row, right ?
19:36 < marcheu> not read the whole area 5 times
19:36 < darktama> yup, one reg 5 times in a row
19:36 < marcheu> ok
19:37 < Lumag> darktama: could you please provide a full registers dump of you card?
19:37 < pmdata> http://pmandin.atari.org/download/lumag-nv15-draw_pixels_test.txt
19:37 < marcheu> darktama: you have to read nv_dma.c then, and adapt the code accordingly
19:37 < darktama> yeah, I figured as much :)
19:37 < Lumag> pmdata: thanks
19:38 < darktama> Lumag: do you just want the RAMIN area? or everything else
19:38 < Lumag> RAMIN probably is enough, but full dump will be wellcomed.
19:38 < pmdata> lumag> could you indent the output of your stuff? it would make it easier to read
19:39 < Lumag> ok.
20:01 < Duke`> huhu http://www.datasheetcatalog.net/cgi-bin/helo.pl?text=RIVA128&action=Search !!
20:03 < Lumag> Those datasheets don't seem to be usefull... There was once nv programming reference, but I couldn't find it...
20:04 < Duke`> yep they seems poor
20:05 -!- pmdata [i=patrice@ANantes-154-1-96-128.w86-210.abo.wanadoo.fr] has quit [Remote closed the connection]
20:22 -!- ChanServ [ChanServ@services.] has quit [Shutting Down]
20:22 -!- ChanServ [ChanServ@services.] has joined #nouveau
20:22 -!- ServerMode/#nouveau [+o ChanServ] by irc.freenode.net
21:27 -!- _Demo_ [n=Demo@modemcable031.147-200-24.mc.videotron.ca] has joined #nouveau
22:39 < marcheu> well, it's not poor if you have nv03 chips and want to design a board... it all depends on what you do :)
22:50 -!- pmdata [i=patrice@81.53.201.177] has joined #nouveau
23:10 -!- stringfellow [n=stringfe@ip56503c9f.direct-adsl.nl] has joined #nouveau
23:40 < marcheu> Lumag: I think you should commit your patch soon to handle objects, we'll convert commands as time goes
23:40 < marcheu> Lumag: if needed, we can fix the nv40+ stuff too
23:41 < marcheu> it's really documented in nv_dma.c
23:41 < Lumag> marcheu: There are some small problems, that I want to fix. I'll commit it today or tomorrow morning.
23:43 < marcheu> that would give us nv40 owners the occasion of fixing it for our cards
23:43 < Lumag> you can already provide a fix :)
--- Log closed Пнд Июн 12 00:00:25 2006
