It seems that I found the cause of that issue: after dd-ing image that
changes partition info I need to:
1. call blockdev --rereadpt <device>, else mkfs will create filesystem
at old partition table.
2. Wait until <device>1...4 nodes appeared in /dev, because blockdev
is async utility and I should sleep for 2 seconds.
Everything looked like busybox's mount couldn't mount mkfs-created
partition or mounted it incorrectly, because host machine used old
partition info (i have external card-reader and I did not unplug it,
but took flash from it), and embedded system used actual partitions.
So when partition's offsets were the same, host machine mounted
partition as ext2, and embedded system mounts it as vfat.
+ Host machine used caches, so everything mounted well, but if unplug
the reader and then plug it again, then partitions were not readable
even on the host machine. I knew that I had to use sync, but forgot
about rereading partition info.

Sorry and thank you for your attention.
_______________________________________________
busybox mailing list
[email protected]
http://lists.busybox.net/mailman/listinfo/busybox

Reply via email to