It was decided to remove the page_endio() as per the previous RFC
discussion[1] of this series and move that functionality into the caller
itself. One of the side benefit of doing that is the callers have been
modified to directly work on folios as page_endio() already worked on
folios.
As Christoph is doing ZRAM cleanups[4] which will get rid of
page_endio() function usage, I removed the final patch that removes
page_endio()[5]. I will send it separately after rc-1 once the zram
cleanups are merged.
mpage changes were tested with a simple boot testing and running a fio
workload on ext2 filesystem. orangefs was tested by Mike Marshall
(No code changes since he tested).
Changes since v2:
- Dropped the zram patch
- Dropped the patch that removes page_endio() function from filemap
- Also split mpage_submit_bio into read and write counterparts (Christoph)
Changes since v1:
- Always chain the IO to the parent as it can never be NULL (Minchan)
- Added reviewed and tested by tags
Changes since RFC 2[2]:
- Call bio_put in zram bio end io handler (Still not Acked by hch[3])
- Call folio_set_error in mpage read endio error path (Willy)
- Directly call folio->mapping in mpage write endio error path (Willy)
[1] https://lore.kernel.org/linux-mm/[email protected]/
[2] https://lore.kernel.org/linux-mm/[email protected]/
[3] https://lore.kernel.org/linux-mm/[email protected]/
[4] https://lore.kernel.org/linux-block/[email protected]/T/#t
[5] https://lore.kernel.org/lkml/[email protected]/
Pankaj Raghav (3):
orangefs: use folios in orangefs_readahead
mpage: split submit_bio and bio end_io handler for reads and writes
mpage: use folios in bio end_io handler
fs/mpage.c | 66 +++++++++++++++++++++++++++++++--------------
fs/orangefs/inode.c | 9 ++++---
2 files changed, 51 insertions(+), 24 deletions(-)
--
2.34.1
Use folios in the bio end_io handler. This conversion does the appropriate
handling on the folios in the respective end_io callback and removes the
call to page_endio(), which is soon to be removed.
Signed-off-by: Pankaj Raghav <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
---
fs/mpage.c | 28 ++++++++++++++++++----------
1 file changed, 18 insertions(+), 10 deletions(-)
diff --git a/fs/mpage.c b/fs/mpage.c
index d9540c1b7427..242e213ee064 100644
--- a/fs/mpage.c
+++ b/fs/mpage.c
@@ -45,24 +45,32 @@
*/
static void mpage_read_end_io(struct bio *bio)
{
- struct bio_vec *bv;
- struct bvec_iter_all iter_all;
+ struct folio_iter fi;
+ int err = blk_status_to_errno(bio->bi_status);
- bio_for_each_segment_all(bv, bio, iter_all)
- page_endio(bv->bv_page, REQ_OP_READ,
- blk_status_to_errno(bio->bi_status));
+ bio_for_each_folio_all(fi, bio) {
+ if (err)
+ folio_set_error(fi.folio);
+ else
+ folio_mark_uptodate(fi.folio);
+ folio_unlock(fi.folio);
+ }
bio_put(bio);
}
static void mpage_write_end_io(struct bio *bio)
{
- struct bio_vec *bv;
- struct bvec_iter_all iter_all;
+ struct folio_iter fi;
+ int err = blk_status_to_errno(bio->bi_status);
- bio_for_each_segment_all(bv, bio, iter_all)
- page_endio(bv->bv_page, REQ_OP_WRITE,
- blk_status_to_errno(bio->bi_status));
+ bio_for_each_folio_all(fi, bio) {
+ if (err) {
+ folio_set_error(fi.folio);
+ mapping_set_error(fi.folio->mapping, err);
+ }
+ folio_end_writeback(fi.folio);
+ }
bio_put(bio);
}
--
2.34.1
Convert orangefs_readahead() from using struct page to struct folio.
This conversion removes the call to page_endio() which is soon to be
removed, and simplifies the final page handling.
The page error flags is not required to be set in the error case as
orangefs doesn't depend on them.
Reviewed-by: Matthew Wilcox (Oracle) <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Tested-by: Mike Marshall <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
fs/orangefs/inode.c | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/fs/orangefs/inode.c b/fs/orangefs/inode.c
index aefdf1d3be7c..9014bbcc8031 100644
--- a/fs/orangefs/inode.c
+++ b/fs/orangefs/inode.c
@@ -244,7 +244,7 @@ static void orangefs_readahead(struct readahead_control *rac)
struct iov_iter iter;
struct inode *inode = rac->mapping->host;
struct xarray *i_pages;
- struct page *page;
+ struct folio *folio;
loff_t new_start = readahead_pos(rac);
int ret;
size_t new_len = 0;
@@ -275,9 +275,10 @@ static void orangefs_readahead(struct readahead_control *rac)
ret = 0;
/* clean up. */
- while ((page = readahead_page(rac))) {
- page_endio(page, false, ret);
- put_page(page);
+ while ((folio = readahead_folio(rac))) {
+ if (!ret)
+ folio_mark_uptodate(folio);
+ folio_unlock(folio);
}
}
--
2.34.1
Split the submit_bio() and bio end_io handler for reads and writes similar
to other aops.
This is a prep patch before we convert end_io handlers to use folios.
Suggested-by: Christoph Hellwig <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
fs/mpage.c | 54 ++++++++++++++++++++++++++++++++++++------------------
1 file changed, 36 insertions(+), 18 deletions(-)
diff --git a/fs/mpage.c b/fs/mpage.c
index 22b9de5ddd68..d9540c1b7427 100644
--- a/fs/mpage.c
+++ b/fs/mpage.c
@@ -43,23 +43,41 @@
* status of that page is hard. See end_buffer_async_read() for the details.
* There is no point in duplicating all that complexity.
*/
-static void mpage_end_io(struct bio *bio)
+static void mpage_read_end_io(struct bio *bio)
{
struct bio_vec *bv;
struct bvec_iter_all iter_all;
- bio_for_each_segment_all(bv, bio, iter_all) {
- struct page *page = bv->bv_page;
- page_endio(page, bio_op(bio),
+ bio_for_each_segment_all(bv, bio, iter_all)
+ page_endio(bv->bv_page, REQ_OP_READ,
blk_status_to_errno(bio->bi_status));
- }
bio_put(bio);
}
-static struct bio *mpage_bio_submit(struct bio *bio)
+static void mpage_write_end_io(struct bio *bio)
{
- bio->bi_end_io = mpage_end_io;
+ struct bio_vec *bv;
+ struct bvec_iter_all iter_all;
+
+ bio_for_each_segment_all(bv, bio, iter_all)
+ page_endio(bv->bv_page, REQ_OP_WRITE,
+ blk_status_to_errno(bio->bi_status));
+
+ bio_put(bio);
+}
+
+static struct bio *mpage_bio_submit_read(struct bio *bio)
+{
+ bio->bi_end_io = mpage_read_end_io;
+ guard_bio_eod(bio);
+ submit_bio(bio);
+ return NULL;
+}
+
+static struct bio *mpage_bio_submit_write(struct bio *bio)
+{
+ bio->bi_end_io = mpage_write_end_io;
guard_bio_eod(bio);
submit_bio(bio);
return NULL;
@@ -265,7 +283,7 @@ static struct bio *do_mpage_readpage(struct mpage_readpage_args *args)
* This folio will go to BIO. Do we need to send this BIO off first?
*/
if (args->bio && (args->last_block_in_bio != blocks[0] - 1))
- args->bio = mpage_bio_submit(args->bio);
+ args->bio = mpage_bio_submit_read(args->bio);
alloc_new:
if (args->bio == NULL) {
@@ -278,7 +296,7 @@ static struct bio *do_mpage_readpage(struct mpage_readpage_args *args)
length = first_hole << blkbits;
if (!bio_add_folio(args->bio, folio, length, 0)) {
- args->bio = mpage_bio_submit(args->bio);
+ args->bio = mpage_bio_submit_read(args->bio);
goto alloc_new;
}
@@ -286,7 +304,7 @@ static struct bio *do_mpage_readpage(struct mpage_readpage_args *args)
nblocks = map_bh->b_size >> blkbits;
if ((buffer_boundary(map_bh) && relative_block == nblocks) ||
(first_hole != blocks_per_page))
- args->bio = mpage_bio_submit(args->bio);
+ args->bio = mpage_bio_submit_read(args->bio);
else
args->last_block_in_bio = blocks[blocks_per_page - 1];
out:
@@ -294,7 +312,7 @@ static struct bio *do_mpage_readpage(struct mpage_readpage_args *args)
confused:
if (args->bio)
- args->bio = mpage_bio_submit(args->bio);
+ args->bio = mpage_bio_submit_read(args->bio);
if (!folio_test_uptodate(folio))
block_read_full_folio(folio, args->get_block);
else
@@ -356,7 +374,7 @@ void mpage_readahead(struct readahead_control *rac, get_block_t get_block)
args.bio = do_mpage_readpage(&args);
}
if (args.bio)
- mpage_bio_submit(args.bio);
+ mpage_bio_submit_read(args.bio);
}
EXPORT_SYMBOL(mpage_readahead);
@@ -373,7 +391,7 @@ int mpage_read_folio(struct folio *folio, get_block_t get_block)
args.bio = do_mpage_readpage(&args);
if (args.bio)
- mpage_bio_submit(args.bio);
+ mpage_bio_submit_read(args.bio);
return 0;
}
EXPORT_SYMBOL(mpage_read_folio);
@@ -577,7 +595,7 @@ static int __mpage_writepage(struct folio *folio, struct writeback_control *wbc,
* This page will go to BIO. Do we need to send this BIO off first?
*/
if (bio && mpd->last_block_in_bio != blocks[0] - 1)
- bio = mpage_bio_submit(bio);
+ bio = mpage_bio_submit_write(bio);
alloc_new:
if (bio == NULL) {
@@ -596,7 +614,7 @@ static int __mpage_writepage(struct folio *folio, struct writeback_control *wbc,
wbc_account_cgroup_owner(wbc, &folio->page, folio_size(folio));
length = first_unmapped << blkbits;
if (!bio_add_folio(bio, folio, length, 0)) {
- bio = mpage_bio_submit(bio);
+ bio = mpage_bio_submit_write(bio);
goto alloc_new;
}
@@ -606,7 +624,7 @@ static int __mpage_writepage(struct folio *folio, struct writeback_control *wbc,
folio_start_writeback(folio);
folio_unlock(folio);
if (boundary || (first_unmapped != blocks_per_page)) {
- bio = mpage_bio_submit(bio);
+ bio = mpage_bio_submit_write(bio);
if (boundary_block) {
write_boundary_block(boundary_bdev,
boundary_block, 1 << blkbits);
@@ -618,7 +636,7 @@ static int __mpage_writepage(struct folio *folio, struct writeback_control *wbc,
confused:
if (bio)
- bio = mpage_bio_submit(bio);
+ bio = mpage_bio_submit_write(bio);
/*
* The caller has a ref on the inode, so *mapping is stable
@@ -652,7 +670,7 @@ mpage_writepages(struct address_space *mapping,
blk_start_plug(&plug);
ret = write_cache_pages(mapping, wbc, __mpage_writepage, &mpd);
if (mpd.bio)
- mpage_bio_submit(mpd.bio);
+ mpage_bio_submit_write(mpd.bio);
blk_finish_plug(&plug);
return ret;
}
--
2.34.1
Looks good:
Reviewed-by: Christoph Hellwig <[email protected]>