Skip to contentSkip to conversation

setrlimit incorrectly allows hard limits to exceed soft limits

TRIVIAL Peter Chubb 2 participants 4 messages Dec 9, 2002 1 message 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]

Peter Chubb

author
#1 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Rik" == Rik van Riel <[email protected]> writes:

Quoted 5 lines · Rusty Trivial Russell
Quoted 5 lines
Quoted 1 line

Just try "ulimit -H -m 10000" for memory limits that were not >

previously set. You end up with (hard limit = 10000) < (soft limit = > unlimited).

+ if (new_rlim.rlim_cur > new_rlim.rlim_max) + return -EINVAL;

Rik> Wouldn't it be better to simply take the soft limit down to Rik> min(new_rlim.rlim_cur, new_rlim.rlim_max) ?

Single unix spec says to return EINVAL in this case.

[EINVAL]
An invalid resource was specified; or in a setrlimit() call, the new rlim_cur exceeds the new rlim_max.

#2 lore raw
Quoted 8 lines · Peter Chubb

Rik> Wouldn't it be better to simply take the soft limit down to Rik> min(new_rlim.rlim_cur, new_rlim.rlim_max) ?

Single unix spec says to return EINVAL in this case.

[EINVAL]
An invalid resource was specified; or in a setrlimit() call, the new rlim_cur exceeds the new rlim_max.

So how about "the old rlim_cur exceeds the new rlim_max" ? ;)

Rik

Signature · 3 lines
Bravely reimplemented by the knights who say "NIH".
http://www.surriel.com/		http://guru.conectiva.com/
Current spamtrap:  <a href=mailto:"[email protected]">[email protected]</a>
author
#3 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Rik" == Rik van Riel <[email protected]> writes:

Rik> On Mon, 9 Dec 2002, Peter Chubb wrote: Wouldn't it be better to Rik> simply take the soft limit down to min(new_rlim.rlim_cur, Rik> new_rlim.rlim_max) ?

Quoted 4 lines
Quoted 4 lines

Single unix spec says to return EINVAL in this case.

[EINVAL] An invalid resource was specified; or in a setrlimit() call, the new rlim_cur exceeds the new rlim_max.

Rik> So how about "the old rlim_cur exceeds the new rlim_max" ? ;)

You always have to set both, so the old value is irrelevant, except in so far as rlim_max may not be increased except by a privileged process.

setrlimit may return EINVAL if the actual usage is above the new rlim_cur.

--
Dr Peter Chubb [email protected]
You are lost in a maze of BitKeeper repositories, all almost the same.

#4 lore raw
Quoted 1 line · Peter Chubb

You always have to set both,

Duh, forgot about that. Nevermind.

Rik

Signature · 3 lines
Bravely reimplemented by the knights who say "NIH".
http://www.surriel.com/		http://guru.conectiva.com/
Current spamtrap:  <a href=mailto:"[email protected]">[email protected]</a>