01:27.46 | *** join/#openjtag rd_ (n=dr@toi.yeu.phu.nu) |
05:23.26 | *** join/#openjtag pleemans (n=peter@leemanpe.ugent.be) |
12:58.36 | *** join/#openjtag p2-mate (n=p2@cable-87-244-173-194.upc.chello.be) |
13:36.15 | *** join/#openjtag prpplague (n=dave@mail.americanmicrosystems.com) |
13:40.31 | prpplague | morning all |
13:40.38 | prpplague | drath: whats cookin today? |
14:31.32 | *** join/#openjtag rd_ (n=dr@toi.yeu.phu.nu) |
14:35.25 | prpplague | drath: i really think that nand.c should be totally ignorant of the buswidth of the nand flash |
14:36.34 | drath | hey prpplague |
14:36.55 | prpplague | drath: greetings |
14:37.06 | prpplague | drath: beer-thirty there yet? |
14:38.19 | drath | prpplague: imho every hour with a "pm" qualifies for that ;) |
14:38.20 | prpplague | drath: imho all the data i/o from within nand.c should be done as u32, the controller code, i.e. the lpc3180 or s3c24xx should handle the proper breakdown of the data |
14:38.24 | prpplague | drath: hehe |
14:38.28 | prpplague | drath: good deal |
14:45.30 | drath | prpplague: well, in case of a manufacturer id read for example, we want exactly one bus-size data word - what would you return in that case? |
14:46.57 | drath | a 32-bit read could hold 2x16bit or 4x8bit - we need to know that bus-width to know what that data is |
14:47.06 | prpplague | drath: i'd still return a u32, just have some code to break out the ids |
14:47.33 | prpplague | drath: hmm |
14:48.21 | prpplague | drath: the problem i have is that even though this is a 8-bit buswidth, all the access to it is designed around operations doing x32 |
14:50.05 | prpplague | drath: i'll tinker around some more, but i still think the nand.c data i/o should be x32 and let the controller code do the actually low level buswidth control |
14:50.50 | drath | prpplague: heh, feel free to propose a patch that works for you and i'll have a look at it |
14:53.27 | prpplague | drath: will do, still testing and tinkering |
14:53.39 | prpplague | drath: just thought i'd break the ice on the topic |
14:54.49 | prpplague | drath: a good example for me is the fact that the address function call doesn't know if it it needs one write or four |
14:55.00 | prpplague | drath: thats handled in the controller code |
14:55.18 | prpplague | drath: i.e. in my s3c24xx code i do one x32 write |
14:56.04 | drath | prpplague: if this is only about performance: that's what the ->read_page and ->write_page functions are for |
14:56.51 | drath | prpplague: the controller is free to do whatever it wants in that case, it should just return the data |
14:57.08 | prpplague | drath: it is only partly related to performance |
16:18.29 | *** join/#openjtag pleemans (n=peter@d51A5E76A.access.telenet.be) |
20:10.31 | *** join/#openjtag drath_ (i=vmaster@p5B07F4C1.dip.t-dialin.net) |
21:56.44 | prpplague | drath: ping |
21:57.01 | prpplague | drath: nm |
21:57.07 | prpplague | unping |
23:46.10 | *** join/#openjtag pipomolo42 (n=alex@ALille-152-1-97-102.w90-18.abo.wanadoo.fr) |
23:54.04 | *** part/#openjtag pipomolo42 (n=alex@ALille-152-1-97-102.w90-18.abo.wanadoo.fr) |