Messages in this thread | | | Subject | Re: [PATCH] slub: prevent validate_slab() error due to race condition | From | Eric Dumazet <> | Date | Thu, 26 Apr 2012 21:44:11 +0200 |
| |
On Thu, 2012-04-26 at 12:20 -0700, Don Morris wrote:
> Note that he sets n here, hence the if() block on 2458 can not > be taken (!n fails) and the if(likely(!n)) is not taken for the > same reason. As such, the code falls through to the returns for > either the slab being empty (or not) where the node lock is > released (2529 / 2543).
Ah yes, you're right, thanks for clarification.
| |