Re: [PATCH] fchmodat, lchmod: port to buggy Linux filesystems

[Paul Eggert]

On 3/10/20 12:30 PM, Florian Weimer wrote:
> The glibc implementation needs /proc to avoid the race.  There is no
> way around that, otherwise we introduce a security vulnerability.

It is unfortunate that we have dueling paranoia here. coreutils mknod is 
paranoid so it uses lchmod to avoid a race, and then glibc lchmod is paranoid so 
it refuses to work with lchmod unless /proc is mounted.

Since we apparently cannot avoid a race unless /proc is mounted, I suppose we 
could change gnulib lchmod to consider the current glibc behavior to be a bug, 
and to fall back on lstat+chmod when /proc is not mounted. This would fix 
coreutils and every other Gnulib-using program that uses lchmod or fchmodat. But 
on the whole it would be somewhat cleaner if glibc lchmod and fchmodat were 
merely documented to have races when /proc is not mounted; that'd be simpler 
than manually adjusting all programs that use glibc lchmod so that they all 
explicitly have races when /proc is not mounted.