diff options
author | pjd <pjd@FreeBSD.org> | 2012-09-27 16:43:23 +0000 |
---|---|---|
committer | pjd <pjd@FreeBSD.org> | 2012-09-27 16:43:23 +0000 |
commit | 3646977f282c6d3373dd9fb627f5cfc7c84e95a2 (patch) | |
tree | b9c7e417c97be25e2780675785b8bb8cb61cd4d1 /lib/libc/rpc/svc_auth_unix.c | |
parent | 52a6b7ce30ec585392dda7f2e15e64087a4335aa (diff) | |
download | FreeBSD-src-3646977f282c6d3373dd9fb627f5cfc7c84e95a2.zip FreeBSD-src-3646977f282c6d3373dd9fb627f5cfc7c84e95a2.tar.gz |
Revert r240931, as the previous comment was actually in sync with POSIX.
I have to note that POSIX is simply stupid in how it describes O_EXEC/fexecve
and friends. Yes, not only inconsistent, but stupid.
In the open(2) description, O_RDONLY flag is described as:
O_RDONLY Open for reading only.
Taken from:
http://pubs.opengroup.org/onlinepubs/9699919799/functions/open.html
Note "for reading only". Not "for reading or executing"!
In the fexecve(2) description you can find:
The fexecve() function shall fail if:
[EBADF]
The fd argument is not a valid file descriptor open for executing.
Taken from:
http://pubs.opengroup.org/onlinepubs/9699919799/functions/exec.html
As you can see the function shall fail if the file was not open with O_EXEC!
And yet, if you look closer you can find this mess in the exec.html:
Since execute permission is checked by fexecve(), the file description
fd need not have been opened with the O_EXEC flag.
Yes, O_EXEC flag doesn't have to be specified after all. You can open a file
with O_RDONLY and you still be able to fexecve(2) it.
Diffstat (limited to 'lib/libc/rpc/svc_auth_unix.c')
0 files changed, 0 insertions, 0 deletions