**** BEGIN LOGGING AT Wed May 03 03:00:01 2017 May 03 09:17:38 VartiWork, hi May 03 09:17:58 finally I decided to ask for user input May 03 09:18:12 ah, tell me more May 03 09:18:16 so you'll just need to put an empty file called repartition May 03 09:18:47 if thi sis found then you see the actual partitioning, it asks the new value (in MB), it flashes May 03 09:18:57 ok May 03 09:19:11 it checks for the max size as before May 03 09:20:14 what if, instead of using the empty file, it just checks for any of the expected files to be flashed, if it founds none, it asks if you want to repartition the nand? May 03 09:20:15 I am now thinking about erasing the partitions before May 03 09:21:20 so, if you want to do the repartition, boot with just updater.sh on the SD and nothing else May 03 09:21:48 hm, as it is now, it exits May 03 09:22:00 http://cgit.openembedded.org/meta-handheld/tree/recipes-bsp/zaurus-utils/zaurus-updater/updater.sh#n395 May 03 09:22:13 yes, it could be easy May 03 09:22:43 instead of adding a case, do it after that May 03 09:23:12 maybe a little menu: 1 repartition, 2, erase May 03 09:23:47 yes, that sounds like a good idea May 03 09:24:01 I don't remember if the binary /sbin/eraseall does add the jffs2 ceanmarklers May 03 09:24:32 i.e. ./flash_eraseall -j /dev/mtd2 May 03 09:24:45 no idea :/ May 03 09:24:46 I'll have to check some blocks May 03 09:25:56 I hope it doesn't May 03 09:26:30 this could be the source of the old jffs cruft... May 03 09:32:01 https://github.com/silid/tuxbox-cvs-apps/blob/master/misc/tools/mtd/eraseall.c May 03 09:33:04 http://www.infradead.org/pipermail/linux-mtd-cvs/2003-February/002767.html May 03 09:38:18 so, I see that eraseall is a tool coming from tuxbox, an OS for set top boxes May 03 09:39:18 or it might be older than tuxbox itself... May 03 09:39:39 yes, unless crazy dev customized it, it does NOT pass -j automatically May 03 09:42:39 then it seems ok to eraseall on mtd2 / mtd3 : I am dubious about doing it after repartitioning...the kernel has the old sizes but finally it should just erase May 03 09:43:26 remember that doing it means 'one life less' for the mtd counters May 03 09:43:51 best is ubiformat -f which preserves the erasecounters May 03 09:43:59 anywway.. May 03 09:59:30 what happens when that counter reaches zero? the partition becomes read-only? May 03 10:00:12 I have done many reflashes so far, I wonder if I might reach that condition soon... May 03 10:00:16 no, it is that you lose the stats of use for each block, how many erase.. May 03 10:00:32 see, never happened to me after 1000 flashes... May 03 10:00:40 still, ubi is way better May 03 10:01:27 the starting number must be pretty high then :) May 03 10:01:27 note that in OE we do have jffs2.sum images, with precalculated erasecounters May 03 10:01:47 err. s/erasecounters/cleanmarkers May 03 10:02:11 I guess we can't switch to ubi since it's not supported by all the older, NAND based distros like pdaxrom, isn't it? May 03 10:02:52 well, there is an hackish way May 03 10:02:52 http://mediawiki.compulab.com/w/index.php5/JFFS2_images_over_UBI_layer May 03 10:03:12 normalyy you use UBIFS on UBI containers May 03 10:03:29 brb May 03 11:46:07 pdaxrom's author InSearchOf was some month ago evaluating to resume pdaxrom, UBI's support might be one incentive more to do it :) May 03 11:46:43 yes May 03 11:47:17 there were some people interested, too bad he hasn't posted again since then May 03 11:47:27 btw we might just give the option for both fs May 03 11:47:49 in OE we build both May 03 11:47:59 yes May 03 11:48:05 once you have the tarball you can creat ethe image May 03 11:48:17 argh..this May 03 11:48:32 crappy kb May 03 11:48:41 have to go, bbl May 03 14:02:41 VartiWork, surprise surprise May 03 14:03:02 I descrambled cacko's updater.sh and I see May 03 14:03:18 ERASEALL="/tmp1/tools/bin/eraseall" May 03 14:03:18 ERASEALL_OPT="-j" May 03 14:03:30 this ^ May 03 14:07:05 so in the cacko's installer the updater created the jffs2 cleanmarkers, and this was removed in the later versions? May 03 14:16:58 this is 1.23 May 03 14:29:33 imho he copied the same bug May 03 14:31:31 of yesterday May 03 14:31:39 but not the other :) May 03 14:47:03 heh May 03 20:47:11 Varti, so there are also bad news... May 03 20:48:29 it appears, the mainte kernel does screw if you repartition, thus you cannot use updater.sh for flashing May 03 20:49:21 that means the last bug is pointless... May 03 20:56:42 I'm trying a workaround... **** ENDING LOGGING AT Thu May 04 03:00:03 2017