3.2.97-rc1 review patch. If anyone has any objections, please let me know.
------------------
From: Al Viro viro@zeniv.linux.org.uk
commit 2b04e8f6bbb196cab4b232af0f8d48ff2c7a8058 upstream.
we need to take care of failure exit as well - pages already in bio should be dropped by analogue of bio_unmap_pages(), since their refcounts had been bumped only once per reference in bio.
Signed-off-by: Al Viro viro@zeniv.linux.org.uk [bwh: Backported to 3.2: adjust filename, context] Signed-off-by: Ben Hutchings ben@decadent.org.uk --- fs/bio.c | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-)
--- a/fs/bio.c +++ b/fs/bio.c @@ -976,6 +976,7 @@ static struct bio *__bio_map_user_iov(st struct bio *bio; int cur_page = 0; int ret, offset; + struct bio_vec *bvec;
for (i = 0; i < iov_count; i++) { unsigned long uaddr = (unsigned long)iov[i].iov_base; @@ -1019,7 +1020,12 @@ static struct bio *__bio_map_user_iov(st
ret = get_user_pages_fast(uaddr, local_nr_pages, write_to_vm, &pages[cur_page]); - if (ret < local_nr_pages) { + if (unlikely(ret < local_nr_pages)) { + for (j = cur_page; j < page_limit; j++) { + if (!pages[j]) + break; + put_page(pages[j]); + } ret = -EFAULT; goto out_unmap; } @@ -1074,10 +1080,8 @@ static struct bio *__bio_map_user_iov(st return bio;
out_unmap: - for (i = 0; i < nr_pages; i++) { - if(!pages[i]) - break; - page_cache_release(pages[i]); + bio_for_each_segment_all(bvec, bio, j) { + put_page(bvec->bv_page); } out: kfree(pages);