Skip to contentSkip to conversation

O_NOLINK for open()

DISCUSSION Bodo Eggert 3 participants 6 messages Sep 12, 2007 – Sep 14, 2007 3 messages not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

Brent Casavant <[email protected]> wrote:

[...]

Quoted 6 lines

I could mmap a temporary tmpfs file (tmpfs so that if there is a machine crash no sensitive data persists) which is created with permissions of 0, immediately unlink it, and pass the file descriptor through an AF_UNIX socket. This does open up a very small window of vulnerability if another process is able to chmod the file and open it before the unlink.

If the process can chmod the file, it can ptrace the daemon, too. Or, using CAP_DAC_OVERRIDE, it can patch the daemon.

Both will void any security.

Quoted 4 lines

However, it occurs to me that this problem goes away if there were a method create a file in an unlinked state to begin with. However there does not appear to be any such mechanism in Linux's open() interface.

Having no window for creating stale temp files is nice to have. We only need a clever fool to implement it.-) But since it's hard to get killed just in the right moment for having a stale temp file, there is very low interest for this feature.

Signature · 4 lines
You know you're in trouble when packet floods are competing to flood you.
        -- grc.com

Friß, Spammer: [email protected] [email protected]
#2 lore raw
Quoted 9 lines
Quoted 4 lines

However, it occurs to me that this problem goes away if there were a method create a file in an unlinked state to begin with. However there does not appear to be any such mechanism in Linux's open() interface.

Having no window for creating stale temp files is nice to have. We only need a clever fool to implement it.-) But since it's hard to get killed just in the right moment for having a stale temp file, there is very low interest for this feature.

I don't think this is a problem. The file is simply created with link
count 0. As soon as the process closes the file, it gets deleted. So there would be no stale files... Or did you mean anything else?

Honza

Signature · 2 lines
Jan Kara <[email protected]>
SuSE CR Labs
#3 lore raw
Quoted 5 lines · Jan Kara
Quoted 5 lines
Quoted 4 lines

However, it occurs to me that this problem goes away if there were a method create a file in an unlinked state to begin with. However there does not appear to be any such mechanism in Linux's open() interface.

Quoted 8 lines
Quoted 4 lines

Having no window for creating stale temp files is nice to have. We only need a clever fool to implement it.-) But since it's hard to get killed just in the right moment for having a stale temp file, there is very low interest for this feature.

I don't think this is a problem. The file is simply created with link
count 0. As soon as the process closes the file, it gets deleted. So there would be no stale files... Or did you mean anything else?

This feature does, AFAIK, not yet exist. Therefore we'd need a code monkey.

Signature · 3 lines
Top 100 things you don't want the sysadmin to say:
42. Hey Fred, did you save that posting about restoring filesystems
    with vi and a toothpick?  More importantly, did you print it out?

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#6 lore raw

Brent Casavant <[email protected]> wrote:

[...]

Quoted 4 lines

Hmm. This will work as long as the peer process is running setuid to it's own unique user. Excellent idea! Since I need to make the program setuid to avoid non-priveleged ptrace attacks, this is a terrific solution.

Tried that:

~ > cd tmp
~/tmp > cp /bin/sleep .
~/tmp > chmod u+s sleep
~/tmp > ./sleep 2147483647 &
[1] 2823
~/tmp > strace -p 2823
Process 2823 attached - interrupt to quit
setup(

Signature · 4 lines
Top 100 things you don't want the sysadmin to say:
27. You can do this patch with the system up...

Friß, Spammer: [email protected]
#7 lore raw
Quoted 7 lines · Bodo Eggert

~/tmp > cp /bin/sleep .
~/tmp > chmod u+s sleep
~/tmp > ./sleep 2147483647 &
[1] 2823
~/tmp > strace -p 2823
Process 2823 attached - interrupt to quit
setup(

You didn't change the owner, so this is not a setuid execution.

Andreas.

Signature · 4 lines
Andreas Schwab, SuSE Labs, [email protected]
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."
#8 lore raw
Quoted 11 lines · Andreas Schwab
Quoted 7 lines · Bodo Eggert

~/tmp > cp /bin/sleep .
~/tmp > chmod u+s sleep
~/tmp > ./sleep 2147483647 &
[1] 2823
~/tmp > strace -p 2823
Process 2823 attached - interrupt to quit
setup(

You didn't change the owner, so this is not a setuid execution.

I expected that, but I wanted to be sure before telling bull****. Besides that, if the suid program was owned by the suid-to user, that user could modify the binary in order to prepare a future attack.

Signature · 2 lines
Top 100 things you don't want the sysadmin to say:
16. find /usr2 -name nethack -exec rm -f {};