Skip to content

execve memory exhaust of argument-copying fixes

KOSAKI Motohiro linux-fsdevel lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest2 patches 6 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/2oom: don't ignore rss in nascent mmfs/compat.c fs/exec.c include/linux/binfmts.h +2 more+47 -105 replies
2/2execve: check the VM has enough memory at firstfs/compat.c fs/exec.c+62 -144 replies

Reviews trailers per patch

1 of 2 patches has a Reviewed-by or Acked-by.

PatchOleg Nesterov
1/2R-b
2/2

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/2] execve memory exhaust of argument-copying fixes
Quoted 6 lines

This is my take on parts of the execve large arguments copying issues that Kees posted about, and Brad and others have been discussing. I've only looked at the narrow area of the argument copying code itself. I think these are good and necessary fixes. But I'm not addressing the whole OOM killer/mm accounting issue, which also needs to be fixed (and I have the impression others are already looking into that).

Now, we have two OOM-Killer/mm acounting problem.
1) OOM-killer doesn't track nascent mm and It may kill innocent task
2) When execve argument-copying, our __vm_enough_memory() doesn't protect any wrong plenty argument. then, execve() invoke OOM instead return failure value when larger argument than system memory.

The patch series addressed this two issue.