Commit 249ffe18 authored Aug 18, 2020 by Bob Peterson Committed by Andreas Gruenbacher Oct 15, 2020

gfs2: don't lock sd_ail_lock in gfs2_releasepage

Patch 380f7c65 changed gfs2_releasepage
so that it held the sd_ail_lock spin_lock for most of its processing.
It did this for some mysterious undocumented bug somewhere in the
evict code path. But in the nine years since, evict has been reworked
and fixed many times, and so have the transactions and ail list.
I can't see a reason to hold the sd_ail_lock unless it's protecting
the actual ail lists hung off the transactions. Therefore, this patch
removes the locking to increase speed and efficiency, and to further help
us rework the log flush code to be more concurrent with transactions.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>

parent 36c78309

fs/gfs2/aops.c

+0 −3

Original line number	Diff line number	Diff line
		@@ -741,7 +741,6 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask)
		*/

		gfs2_log_lock(sdp);
		spin_lock(&sdp->sd_ail_lock);
		head = bh = page_buffers(page);
		do {
		if (atomic_read(&bh->b_count))
		@@ -753,7 +752,6 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask)
		goto cannot_release;
		bh = bh->b_this_page;
		} while(bh != head);
		spin_unlock(&sdp->sd_ail_lock);

		head = bh = page_buffers(page);
		do {
		@@ -779,7 +777,6 @@ int gfs2_releasepage(struct page *page, gfp_t gfp_mask)
		return try_to_free_buffers(page);

		cannot_release:
		spin_unlock(&sdp->sd_ail_lock);
		gfs2_log_unlock(sdp);
		return 0;
		}

Admin message