[PATCH] mm/filemap: fix page end in filemap_get_read_batch

[PATCH] mm/filemap: fix page end in filemap_get_read_batch

[coolqyj]

From: Qian Yingjin <qian@ddn.com>

I was running traces of the read code against an RAID storage
system to understand why read requests were being misaligned
against the underlying RAID strips. I found that the page end
offset calculation in filemap_get_read_batch() was off by one.

When a read is submitted with end offset 1048575, then it
calculates the end page for read of 256 when it should be 255.
"last_index" is the index of the page beyond the end of the read
and it should be skipped when get a batch of pages for read in
@filemap_get_read_batch().

The below simple patch fixes the problem. This code was introduced
in kernel 5.12.

Fixes: cbd59c48ae2b ("mm/filemap: use head pages in generic_file_buffered_read")

Signed-off-by: Qian Yingjin <qian@ddn.com>
---
 mm/filemap.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/mm/filemap.c b/mm/filemap.c
index c4d4ace9cc70..b7754760c09a 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2371,7 +2371,7 @@ static void shrink_readahead_size_eio(struct file_ra_state *ra)
  * clear so that the caller can take the appropriate action.
  */
 static void filemap_get_read_batch(struct address_space *mapping,
-		pgoff_t index, pgoff_t max, struct folio_batch *fbatch)
+		pgoff_t index, pgoff_t last_index, struct folio_batch *fbatch)
 {
 	XA_STATE(xas, &mapping->i_pages, index);
 	struct folio *folio;
@@ -2380,7 +2380,11 @@ static void filemap_get_read_batch(struct address_space *mapping,
 	for (folio = xas_load(&xas); folio; folio = xas_next(&xas)) {
 		if (xas_retry(&xas, folio))
 			continue;
-		if (xas.xa_index > max || xa_is_value(folio))
+		/*
+		 * "last_index" is the index of the page beyond the end of
+		 * the read.
+		 */
+		if (xas.xa_index >= last_index || xa_is_value(folio))
 			break;
 		if (xa_is_sibling(folio))
 			break;
@@ -2588,6 +2592,7 @@ static int filemap_get_pages(struct kiocb *iocb, struct iov_iter *iter,
 	struct folio *folio;
 	int err = 0;
 
+	/* "last_index" is the index of the page beyond the end of the read */
 	last_index = DIV_ROUND_UP(iocb->ki_pos + iter->count, PAGE_SIZE);
 retry:
 	if (fatal_signal_pending(current))
-- 
2.34.1

Re: [PATCH] mm/filemap: fix page end in filemap_get_read_batch

[Matthew Wilcox]

On Wed, Jan 04, 2023 at 11:21:24AM +0800, coolqyj@163.com wrote:
> From: Qian Yingjin <qian@ddn.com>
> 
> I was running traces of the read code against an RAID storage
> system to understand why read requests were being misaligned
> against the underlying RAID strips. I found that the page end
> offset calculation in filemap_get_read_batch() was off by one.
> 
> When a read is submitted with end offset 1048575, then it
> calculates the end page for read of 256 when it should be 255.
> "last_index" is the index of the page beyond the end of the read
> and it should be skipped when get a batch of pages for read in
> @filemap_get_read_batch().
> 
> The below simple patch fixes the problem. This code was introduced
> in kernel 5.12.

Thanks for diagnosing & sending a patch.  However, I'd really prefer
to work in terms of 'max' instead of 'last_index' in that function.
Would this work for you?

+++ b/mm/filemap.c
@@ -2595,13 +2595,13 @@ static int filemap_get_pages(struct kiocb *iocb, struct iov_iter *iter,
        if (fatal_signal_pending(current))
                return -EINTR;

-       filemap_get_read_batch(mapping, index, last_index, fbatch);
+       filemap_get_read_batch(mapping, index, last_index - 1, fbatch);
        if (!folio_batch_count(fbatch)) {
                if (iocb->ki_flags & IOCB_NOIO)
                        return -EAGAIN;
                page_cache_sync_readahead(mapping, ra, filp, index,
                                last_index - index);
-               filemap_get_read_batch(mapping, index, last_index, fbatch);
+               filemap_get_read_batch(mapping, index, last_index - 1, fbatch);
        }
        if (!folio_batch_count(fbatch)) {
                if (iocb->ki_flags & (IOCB_NOWAIT | IOCB_WAITQ))

[PATCH] mm/filemap: fix page end in filemap_get_read_batch

[coolqyj]

From: Qian Yingjin <qian@ddn.com>

I was running traces of the read code against an RAID storage
system to understand why read requests were being misaligned
against the underlying RAID strips. I found that the page end
offset calculation in filemap_get_read_batch() was off by one.

When a read is submitted with end offset 1048575, then it
calculates the end page for read of 256 when it should be 255.
"last_index" is the index of the page beyond the end of the read
and it should be skipped when get a batch of pages for read in
@filemap_get_read_batch().

The below simple patch fixes the problem. This code was introduced
in kernel 5.12.

Fixes: cbd59c48ae2b ("mm/filemap: use head pages in generic_file_buffered_read")

Signed-off-by: Qian Yingjin <qian@ddn.com>
---
 mm/filemap.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/mm/filemap.c b/mm/filemap.c
index c4d4ace9cc70..0e20a8d6dd93 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2588,18 +2588,19 @@ static int filemap_get_pages(struct kiocb *iocb, struct iov_iter *iter,
 	struct folio *folio;
 	int err = 0;
 
+	/* "last_index" is the index of the page beyond the end of the read */
 	last_index = DIV_ROUND_UP(iocb->ki_pos + iter->count, PAGE_SIZE);
 retry:
 	if (fatal_signal_pending(current))
 		return -EINTR;
 
-	filemap_get_read_batch(mapping, index, last_index, fbatch);
+	filemap_get_read_batch(mapping, index, last_index - 1, fbatch);
 	if (!folio_batch_count(fbatch)) {
 		if (iocb->ki_flags & IOCB_NOIO)
 			return -EAGAIN;
 		page_cache_sync_readahead(mapping, ra, filp, index,
 				last_index - index);
-		filemap_get_read_batch(mapping, index, last_index, fbatch);
+		filemap_get_read_batch(mapping, index, last_index - 1, fbatch);
 	}
 	if (!folio_batch_count(fbatch)) {
 		if (iocb->ki_flags & (IOCB_NOWAIT | IOCB_WAITQ))
-- 
2.34.1