**** BEGIN LOGGING AT Wed May 20 02:59:59 2015 May 20 08:47:50 gm lumag__ May 20 08:47:59 hello ant___ May 20 08:48:21 * ant___ keeps the long tail today May 20 08:49:01 lumag: about sharpsl partition parser, I have found some examples May 20 08:49:06 both are modules May 20 08:49:49 my doubt is now about the wear leveling/nandlogical May 20 08:50:09 I suppose the first sectors are not reallocated, I guess May 20 08:52:53 mtd/bcm47xxpart.c or mtd/redboot.c May 20 08:53:46 I'd suggest to check drivers/mtd/redboot.c or drivers/mtd/afs.c May 20 08:53:56 suppose that I read also the machine model (ID), this module should be loaded before boot to be useful :) May 20 08:53:59 Yeah :) May 20 08:54:14 no use of machine id at that point? May 20 08:54:30 'yeah' was a phrase to the names of your c files. May 20 08:54:40 got it :) May 20 08:54:58 One will specify the name of the partitions parser when registering an mtd. May 20 08:56:00 (looking for the example) May 20 08:56:11 yes, and there are pending changes right now... May 20 08:56:25 see linux-mtd mtd: part: add generic parsing of linux,part-probe May 20 08:57:28 ok, first hit at arch/blackfin/mach-bf537/boards/stamp.c May 20 08:57:28 finally : May 20 08:58:54 I see, #389 May 20 08:59:21 I was tellign that finally Brian wrote: May 20 08:59:44 OK, so this much seems to suggest that we can't really determine a set of May 20 08:59:44 parsers that apply to all MTDs, and we have to stick with some kind of May 20 08:59:44 platform information for choosing a set of custom parsers. May 20 09:01:45 That is quite logical. We had that for the board files. May 20 09:03:18 Hmm. May 20 09:03:25 grep cmdlinepart arch/ -r May 20 09:03:25 arch/blackfin/mach-bf537/boards/stamp.c:const char *part_probes[] = { "cmdlinepart", "RedBoot", NULL }; May 20 09:03:25 arch/mips/cavium-octeon/flash_setup.c: "cmdlinepart", May 20 09:03:25 arch/mips/netlogic/xlr/platform-flash.c:const char *xlr_part_probes[] = { "cmdlinepart", NULL }; May 20 09:03:25 arch/sh/boards/mach-rsk/setup.c:static const char *part_probes[] = { "cmdlinepart", NULL }; May 20 10:01:24 lumag: you once disassembled the Angel bootloader iirc May 20 10:02:18 Yes May 20 10:02:31 I imagine it relies on standard addresses in mtd1 so nandlogical stuff is unneeded May 20 10:02:54 I did not look into that part. May 20 10:03:14 I'd assume that it reads raw sectors, w/o any wear leveling May 20 10:06:11 I'm sorry, gtg now. May 20 10:06:55 (and btw, just to be sure. I assume that partitioning for collie is done in a completely different way). May 20 10:13:16 yes, only for sharpsl nand May 20 10:14:11 no work for collie ;) May 20 10:14:16 bbl May 20 21:13:26 lumag: I think I better understand the afs.c parser May 20 21:13:47 :) May 20 21:14:28 redboot.c does strange things and refers to jffs2 May 20 21:14:42 redboot is huuuge May 20 21:14:46 and old May 20 21:14:55 and lots of #ifdef 's May 20 21:15:22 and lot of swab32s May 20 21:15:31 (what's that?) May 20 21:16:55 swap bytes in 32-bit word May 20 21:16:58 ABCD <-> DCBA May 20 21:21:38 ok May 20 21:21:55 let's try and see what is read May 20 21:22:23 byte swap only if necessary May 20 21:22:31 (iirc yes...) May 20 21:23:13 hm. a good part of kexecboot/machine/zaurus.c can be used it seems May 20 21:48:48 The whole story about programming is about reusing and refactoring old code :)) May 20 21:53:50 heh, lookin now at pxa2xx-flash.c May 20 21:54:01 47 static const char * const probes[] = { "RedBoot", "cmdlinepart", NULL }; May 20 21:54:26 pxa2xxflash is for nor. May 20 21:54:29 I think May 20 21:55:57 Ah, yes. drivers/mtd/nand/sharpsl.c doesn't specify probe order (I think, that means {"cmdlinepart", NULL} May 20 21:56:48 Ah. "cmdlinepart", "ofpart", NULL May 20 21:59:52 ok, core-base-image built for poodle May 20 22:00:01 I have kernel and toolchain now May 20 22:00:12 tomorrow I'll cook up smthg May 20 22:00:45 (as with the userspace code in kexecboot I'll need someone verifying/rewriting the code...) May 20 22:00:52 :)) May 20 22:00:53 :p May 20 22:00:57 good night May 20 22:01:13 I'd really suggest to do developmnet kernel build outside of Yocto - to save time May 20 22:01:14 gn **** ENDING LOGGING AT Thu May 21 02:59:58 2015