tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: openat/fstatat functions implementation

In message <200909171923.PAA02794%Sparkle.Rodents-Montreal.ORG@localhost>, der 
Mouse writ
>Unless I've totally missed something, that is exactly what fstatat()
>and openat() do - clone externally-invented and -specified semantics.
>(Or at least what they're _intended_ to do; bugs are a fact of life.)


>That's also why the horrible names, I suspect: someone said, upthread,
>that these are POSIX-specified.

That would not surprise me.

FWIW:  I loathe the names, the semantics are a bit dodgy at best,
and in particular, "fstatat" is just plain misnamed.  However, they
do seem to work and be sorta useful.

(I ran into them while writing a library which, inserted in front of glibc
using LD_PRELOAD, intercepts and changes the semantics of all filesystem
operations.  I actually want to port it to BSD some day just out of
curiosity.  It makes non-root programs able to believe that they are
root, including being able to issue chmod() and see the same results
from stat(); think like "fakeroot", only relatively more bulletproofed.)

Home | Main Index | Thread Index | Old Index