[Fakeroot-commits] New commit in fakeroot--main--0.0--patch-26

[Clint Adams]

Revision: fakeroot--main--0.0--patch-26
Archive: fakeroot@packages.debian.org--fakeroot
Creator: Timo Savola <tsavola@movial.fi>
Date: Thu Sep 30 14:35:35 EEST 2004
Standard-date: 2004-09-30 11:35:35 GMT
Modified-files: communicate.h libfakeroot.c
New-patches: fakeroot@packages.debian.org--fakeroot/fakeroot--main--0.0--patch-26
    tsavola@movial.fi--scratchbox/fakeroot--scratchbox--0.0--patch-8
    tsavola@movial.fi--scratchbox/fakeroot--scratchbox--0.0--patch-9
    tsavola@movial.fi--scratchbox/fakeroot--scratchbox--0.0--patch-10
    tsavola@movial.fi--scratchbox/fakeroot--scratchbox--0.0--patch-11
    tsavola@movial.fi--scratchbox/fakeroot--scratchbox--0.0--patch-12
    tsavola@movial.fi--scratchbox/fakeroot--scratchbox--0.0--patch-13
    tsavola@movial.fi--scratchbox/fakeroot--scratchbox--0.0--patch-14
Summary: Support for setuid/setgid/etc.
Keywords: 

Fake implementations for the setuid family of functions (fakeroot used to
only provide dummy implementations).  The fake IDs are inherited by child
processes via environment variables.  Attempts to behave according to the
POSIX standard (= Linux man pages).

Issues:
* Privileges are not checked, which results in incorrect behaviour.
  Example: process changes its (real, effective and saved) uid to 1000
  and then tries to regain root privileges.  This should normally result
  in an EPERM, but our implementation doesn't care...
* If one of the setenv calls fails, the state may get corrupted.
* Not thread-safe.

Implemented functions: setuid, setgid, seteuid, setegid, setreuid, setregid,
setresuid and setresgid.