Messages in this thread | | | Date | Wed, 3 Oct 2001 17:30:16 +0100 | From | "Stephen C. Tweedie" <> | Subject | Re: lock_kiovec question |
| |
Hi,
On Sat, Sep 29, 2001 at 11:09:03AM +0200, Manfred Spraul wrote: > lock_kiovec tries to lock each page in the kiovec, and fails if it can't > lock one of the pages. > > What if the zero page is mapped multiple times in the kiobuf?
Don't Do That then. lock_kiobuf can be called if some caller really wants the kiobuf pages locked, but kiobuf page mapping is much cleaner in 2.4 than it had to be in 2.2 and there's no need to keep pages locked during the mapping. raw IO doesn't ever use lock_kiobuf in 2.4.
> AFAICS map_user_pages doesn't break zero page mappings if it's called > with rw==WRITE (i.e write to disk, read from kiobuf)
Correct. The raw IO code for 2.2 had to exempt zero pages from locking for this reason.
Cheers, Stephen - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |