Hi,
Axel Beckert wrote:
> 10/0/0 root@acromantula:pts/4 18:53:47 [~] # ps auxwwwf | fgrep git
> abe 11352 77.1 0.0 4552 596 pts/2 R+ 18:53 0:10 |
> | \_ strace -f -e fork git status
> abe 11353 6.3 0.1 24632 6848 pts/2 S+ 18:53 0:00 |
> | \_ git status
> abe 11354 7.3 0.1 24632 6856 pts/2 S+ 18:53 0:00 |
> | \_ git status --porcelain
> abe 11355 8.0 0.1 24632 6856 pts/2 S+ 18:53 0:00 |
> | \_ git status --porcelain
> abe 11358 9.8 0.1 24632 6852 pts/2 S+ 18:53 0:00 |
> | \_ git status --porcelain
Yeah, this is a submodule issue. What is the output of
git ls-files -s | grep ^160000 | cut -d$'\t' -f2 | xargs ls -ld
?
[...]
> So this issue has quite some potential to bring down a system within
> minutes and trigger an OOM condition.
That's a general feature of fork bombs. I think we should try to
figure out why git is recursing into the same repository again and
again --- the fork bomb mitigation aspect is less interesting to me
(though certainly a worthwhile thing to work on for kernel hackers
or people coming up with default rlimits).
Thanks,
Jonathan
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
