Ben NanoNote

tested with 20100113 image, ubi-utils 1.2.0 from openwrt-packages.

reproduction steps:

1. boot from microSD

2. run 'ubiformat /dev/mtd2 -f openwrt-xburst-rootfs.ubi'

3. On devices with bad blocks in the first 50 MB of NAND, ubiformat 
will first run into a bad block saying:
"
cannot write 524288 bytes to mtd2 (eraseblock XX, offset 0)
error 5 (Input/output error)
libmtd: run torture test for PEB XX
libmtd: PEB XX passed torture test, do not mark it bad
ubiformat: mark it as bad? (Continue (yes/no)
"

4. say "y"

5. you see "ubiformat: marking block 17 bad"

6. ubiformat continues flashing, but at the end it will say
"
ubiformat: flashing eraseblock 78 - 101% complete
ubiformat: error!: eof reached; 524288 bytes remaining
ubiformat: error!: failed to read eraseblock 77 from 
"openwrt-xburst-rootfs.ubi"
error 5 (Input/output error)
"

(note the 101%)
Looks like a problem with ubiformat getting confused about offsets 
after the bad block. Also it's not clear whether the bad block 
should have been marked or not, seems libmtd and ubiformat disagree? 
Probably unrelated to the offset problem in ubiformat, and 101%.

Ben NanoNote

Issue 31: ubiformat runs into i/o error at EOF after encountering bad block

interactive