[Bug 252787] fstyp heuristics can misidentify UFS file systems created by newfs, goofs up autofs's special_media map

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=252787

--- Comment #3 from Richard M Kreuter <kreuter at progn.net> ---
By inspection, it appears that the improvements in 13 do not address this
issue: I tested with a checkout that includes commit ddf61156, using my
attached reproduction to create a suitably goofy disk image, and it still
reports exfat after newfs. 

The UFS superblock starts at 8192 bytes, i.e., 16 sectors, from the start of a
volume, right? If that's correct, then the exfat code's examination of the
first 12 sectors won't avoid this false positive.

Anyhow, if someone wishes to consider this a defect in fstyp, one potential
minimally disruptive approach would be to reorder the checks in fstyp.c so that
the more common file system formats are checked for last (right now things are
checked for in alphabetical order). For the purpose of heuristically guessing
the format of a removable medium, I'd say that EXFAT or FAT is most likely
(since commodity media is often pre-formatted that way), and so "falling
through" to check for those after ruling out UFS, ZFS, etc., will arrive at the
correct answer while avoiding this particular false positive.

That said, I'd be satisfied considering the possibility of a false positive
from fstyp just a documentation issue. It's easy enough to overwrite the first
sector of a disk with dd, for instance, so long as one knows that that's the
cause of the false positive.

-- 
You are receiving this mail because:
You are the assignee for the bug.