Am 10.02.2011 12:53, schrieb Gennady Kupava: > The ubifs has different options to decrease stability and decrease > performance, like disabling bulk read and enabling jffs-style > synchronous writes. I guess it's better to try em first.
OK > Also i talk to ubifs guys on irc, they look like quite open to > questions, but first we need something reproducible. > > Guy who'll have ubifs trubles please try to: > > 1. disable bulk read in fstab and check if problem will still exist. > 2. do nanddump and post it or try to reproduce bug with nandsim. (i > guess ubifs guys can fix problem in case if such dump exist quite fast) Is nanddump required? I tried to make a copy of my old destroyed nand partition and because loop-back mounting failed I saved my data with dd. Following http://www.linux-mtd.infradead.org/faq/ubifs.html#L_ubifs_nandsim: # modprobe nandsim first_id_byte=0x20 second_id_byte=0xaa \ third_id_byte=0x00 fourth_id_byte=0x15 # dd if=OpenMoko.Flash.Backup.April1 of=/dev/mtd0 bs=2048 # modprobe ubi mtd=0 FATAL: Error inserting ubi (/lib/modules/2.6.37+/kernel/drivers/mtd/ubi/ubi.ko): Invalid argument Tried also modprobe ubi mtd=1 ... Kernel log: [33221.063532] [nandsim] warning: write_byte: command (0x90) wasn't expected, expected state is STATE_READY, ignore previous states [33221.063543] [nandsim] warning: read_byte: unexpected data output cycle, state is STATE_READY return 0x0 [33221.063550] [nandsim] warning: read_byte: unexpected data output cycle, state is STATE_READY return 0x0 [33221.063556] [nandsim] warning: read_byte: unexpected data output cycle, state is STATE_READY return 0x0 [33221.063563] [nandsim] warning: read_byte: unexpected data output cycle, state is STATE_READY return 0x0 [33221.063571] NAND device: Manufacturer ID: 0x20, Chip ID: 0xaa (ST Micro ) [33221.063590] flash size: 256 MiB [33221.063594] page size: 2048 bytes [33221.063598] OOB area size: 64 bytes [33221.063602] sector size: 128 KiB [33221.063605] pages number: 131072 [33221.063609] pages per sector: 64 [33221.063612] bus width: 8 [33221.063616] bits in sector size: 17 [33221.063619] bits in page size: 11 [33221.063623] bits in OOB size: 6 [33221.063626] flash size with OOB: 270336 KiB [33221.063631] page address bytes: 5 [33221.063635] sector address bytes: 3 [33221.063639] options: 0x8 [33221.066121] Scanning device for bad blocks [33221.070456] Creating 1 MTD partitions on "NAND 256MiB 1,8V 8-bit": [33221.070475] 0x000000000000-0x000010000000 : "NAND simulator partition 0" [33239.745490] UBI: attaching mtd0 to ubi0 [33239.745499] UBI: physical eraseblock size: 131072 bytes (128 KiB) [33239.745505] UBI: logical eraseblock size: 129024 bytes [33239.745510] UBI: smallest flash I/O unit: 2048 [33239.745514] UBI: sub-page size: 512 [33239.745519] UBI: VID header offset: 512 (aligned 512) [33239.745524] UBI: data offset: 2048 [33239.745634] UBI error: validate_ec_hdr: bad VID header offset 2048, expected 512 [33239.745641] UBI error: validate_ec_hdr: bad EC header [33239.745646] UBI error: ubi_io_read_ec_hdr: validation failed for PEB 0 [33239.745675] UBI error: ubi_init: cannot attach mtd0 [33239.745853] UBI error: ubi_init: UBI error: cannot initialize UBI, error -22 [33245.933633] UBI error: ubi_init: UBI error: cannot initialize UBI, error -19 # modprobe ubi # mount -t ubifs /dev/ubi0_0 /mnt/usb/ mount: wrong fs type, bad option, bad superblock on /dev/ubi0_0, Is this useful? Jens _______________________________________________ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community