Much appreciate your suggestion. I've modified the patch based on your
advice and sent out a new patch with new subject "Btrfs: send, apply
asynchronous
page cache readahead to enhance page read".
Filipe Manana 於 2017-09-13 18:45 寫到:
On Wed, Sep 13, 2017 at 7:38 AM, peterh <pet...@synology.com> wrote:
From: Kuanling Huang <pet...@synology.com>
By analyzing the perf on btrfs send, we found it take large
amount of cpu time on page_cache_sync_readahead. This effort
can be reduced after switching to asynchronous one. Overall
performance gain on HDD and SSD were 9 and 15 respectively if
simply send a large file.
Besides what was pointed before, about saying what those 9 and 15 are,
the subject mentions incremental send, but there's nothing here that
is specific to incremental sends, as it applies to full send
operations as well, so please also change the subject.
Signed-off-by: Kuanling Huang <pet...@synology.com>
---
fs/btrfs/send.c | 21 ++++++++++++++++-----
1 file changed, 16 insertions(+), 5 deletions(-)
diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c
index 63a6152..ac67ff6 100644
--- a/fs/btrfs/send.c
+++ b/fs/btrfs/send.c
@@ -4475,16 +4475,27 @@ static ssize_t fill_read_buf(struct send_ctx
*sctx, u64 offset, u32 len)
/* initial readahead */
memset(&sctx->ra, 0, sizeof(struct file_ra_state));
file_ra_state_init(&sctx->ra, inode->i_mapping);
- btrfs_force_ra(inode->i_mapping, &sctx->ra, NULL, index,
- last_index - index + 1);
while (index <= last_index) {
unsigned cur_len = min_t(unsigned, len,
PAGE_CACHE_SIZE - pg_offset);
- page = find_or_create_page(inode->i_mapping, index,
GFP_NOFS);
You based this patch on an old code base. Currently it is GFP_KERNEL
and not GFP_NOFS anymore.
Please update the patch.
+ page = find_lock_page(inode->i_mapping, index);
if (!page) {
- ret = -ENOMEM;
- break;
+ page_cache_sync_readahead(inode->i_mapping,
+ &sctx->ra, NULL, index,
+ last_index + 1 - index);
+
+ page = find_or_create_page(inode->i_mapping,
index, GFP_NOFS);
+ if (unlikely(!page)) {
Please avoid polluting the code with unlikely/likely macros (unless
there's really a significant performance win, which isn't the case
here I bet).
+ ret = -ENOMEM;
+ break;
+ }
+ }
+
+ if (PageReadahead(page)) {
+ page_cache_async_readahead(inode->i_mapping,
+ &sctx->ra, NULL, page, index,
+ last_index + 1 - index);
}
if (!PageUptodate(page)) {
--
1.9.1
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs"
in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
