> On 07 Nov 2016, at 14:04, Christos Zoulas <chris...@astron.com> wrote: > > In article > <camnsw54_di9cbtc6ckmlnan6j--uljkhvswvne5q2y+u3c6...@mail.gmail.com>, > JaromĆr DoleÄ ek <jaromir.dole...@gmail.com> wrote: >> >> Anyway, IMO eventual change to allow fsck to DTRT after crash in >> ffs_truncate() should go separately from the crash fix. > > While fsck could be fixed, IMO, it is best for the kernel to try to leave > things as consistent as possible avoid scary noise from fsck and to keep > recovery simpler.
It is impossible to fix fsck here: fsck gets a file system where some inodes have the same block allocated. It is impossible to determine which allocation is wrong as its inode was in truncation. -- J. Hannken-Illjes - hann...@eis.cs.tu-bs.de - TU Braunschweig (Germany)
