lkml.org 
[lkml]   [1999]   [Dec]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Can't hardlink in different dirs. (BUG#826)


On Fri, 3 Dec 1999, Andrea Arcangeli wrote:

> On Fri, 3 Dec 1999, Klaus Espenlaub wrote:
>
> >But don't do this if the system administrator is anyone else than your best
> >buddy.
>
> Unless the sysadm has process accounting enabled he has no way to guess
> who did the link and kiked me out of quota. If he has process accounting
> enabled you must hope to have not run `ln` too at the directory mtime.
>
> If instead I am out of quota because a malicious task is taking a file
> opened, then the admin will find the malicious user immediatly.

Oh, really? RTFM on use of AF_UNIX for passing file descriptors. Notice
that there's nobody to associate them with if sender had closed the fd
after sendmsg(). They are in flight - pointers to struct file stored in
SCM_RIGHTS packet that hangs in receiving queue of a socket. Like hell
you'll see them in procfs or in fuser output. About the only thing you can
do is to start killing everything with receiving AF_UNIX socket. And
looking which kill will result in space getting freed. Great. Especially
since you'll have to figure out which space increase came from _that_ and
which was the result of the fscker finally leaving the old version of
libfoobar.so to die. And which was the result of postponed activity - if
we'll get s-u expect it big way.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:55    [W:0.061 / U:0.244 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site