When using devices that support max_segments > BIO_MAX_PAGES (256), direct IO tries to allocate a bio with more pages than allowed, which leads to an oops in dio_bio_alloc(). Clamp the request to the supported maximum, and return an error to catch future cases. Signed-off-by: David Dillow <dillowda@xxxxxxxx> -- On Fri, 2011-01-14 at 09:44 -0500, Jeff Moyer wrote: > David Dillow <dillowda@xxxxxxxx> writes: > > dio_bio_alloc() doesn't check the result of bio_alloc(), so it > > dereferences a NULL pointer. bio_alloc(GFP_KERNEL, ...) doesn't fail > > unless it gets called for an invalid number of pages, so it seems a bit > > like overkill to check for failure in dio_bio_alloc(), though it would > > have saved me some time tracking this down. > > We should always be checking return values. Mind respinning the patch > to include that? Here's the first alternate, that just checks the return value. I think I prefer the other one, though. diff --git a/fs/direct-io.c b/fs/direct-io.c index 85882f6..9fdbf99 100644 --- a/fs/direct-io.c +++ b/fs/direct-io.c @@ -332,6 +332,8 @@ dio_bio_alloc(struct dio *dio, struct block_device *bdev, struct bio *bio; bio = bio_alloc(GFP_KERNEL, nr_vecs); + if (!bio) + return -ENOMEM; bio->bi_bdev = bdev; bio->bi_sector = first_sector; @@ -583,6 +585,7 @@ static int dio_new_bio(struct dio *dio, sector_t start_sector) goto out; sector = start_sector << (dio->blkbits - 9); nr_pages = min(dio->pages_in_io, bio_get_nr_vecs(dio->map_bh.b_bdev)); + nr_pages = min(nr_pages, BIO_MAX_PAGES); BUG_ON(nr_pages <= 0); ret = dio_bio_alloc(dio, dio->map_bh.b_bdev, sector, nr_pages); dio->boundary = 0; -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html