Re: [PATCH v5 08/12] block: Introduce new bio_split()

[Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>]

Hello,

On Mon, Aug 13, 2012 at 02:55:11PM -0700, Kent Overstreet wrote:
> > If the condition is detectable, WARN_ON_ONCE() please.
> 
> I know I said I liked this idea, but I changed my mind.
> 
> Sticking a WARN_ON_ONCE() there is saying that passing __GFP_WAIT from
> under generic_make_request() is always wrong - it might as well be a
> BUG_ON() except warn is better for the user.

WARN_ON_ONCE(), as opposed to BUG_ON(), doesn't mean that something
isn't always wrong.  It's just better to always use WARN variant if
the system doesn't *need* to be put down immediately to avoid, say,
massive data corruption - limping machine is simply better than dead
one.

> If that's true, then an assertion is completely wrong because we can
> just do the right thing instead - mask out __GFP_WAIT if
> current->bio_list != NULL and document that it can fail in that
> situation.

That's worse because it confuses the reader.  Taking something which
is as well-known as __GFP_WAIT and then silently ignoring it isn't a
good idea.  Please just add a WARN_ON.

Thanks.

-- 
tejun