Martin Husemann <mar...@duskware.de> wrote: > Cause - I don't think so. > It probably has made the race conditions more likely to fail.
This is not obvious: unmount(2) performs a VFS_SYNC before VFS_UNMOUNT, hence the sync(2) call in umount(8) was supposed to be useless. > I haven't been able to reproduce it now in a debuggable envrionement, > unfortunately. Can you just reinstall umount(8) binary from 3 days ago and try it out? -- Emmanuel Dreyfus http://hcpnet.free.fr/pubz m...@netbsd.org