Messages in this thread | | | Date | Thu, 28 May 2009 12:20:35 +0200 | From | Jens Axboe <> | Subject | Re: [PATCH 0/12] Per-bdi writeback flusher threads v7 |
| |
On Wed, May 27 2009, Andrew Morton wrote: > On Wed, 27 May 2009 17:05:47 +0200 > Jens Axboe <jens.axboe@oracle.com> wrote: > > > I ran testing with artificial > > bdi_work allocation failures to verify the oom path > > I bet you didn't use lib/fault-inject.c :(
I didn't, it was a 2-second hack to add a
static int i;
if (!(i++ % 4)) return NULL;
so I didn't look further...
> For tricky pieces of core kernel like this, there's a good case for > including and permanently maintaining fault injection testing > capabilities.
I agree.
-- Jens Axboe
| |