writeback: track number of inodes under writeback
Patch series "writeback: Fix bandwidth estimates", v4.
Fix estimate of writeback throughput when device is not fully busy doing
writeback. Michael Stapelberg has reported that such workload (e.g.
generated by linking) tends to push estimated throughput down to 0 and as
a result writeback on the device is practically stalled.
The first three patches fix the reported issue, the remaining two patches
are unrelated cleanups of problems I've noticed when reading the code.
This patch (of 4):
Track number of inodes under writeback for each bdi_writeback structure.
We will use this to decide whether wb does any IO and so we can estimate
its writeback throughput. In principle we could use number of pages under
writeback (WB_WRITEBACK counter) for this however normal percpu counter
reads are too inaccurate for our purposes and summing the counter is too
expensive.
Link: https://lkml.kernel.org/r/20210713104519.16394-1-jack@suse.cz
Link: https://lkml.kernel.org/r/20210713104716.22868-1-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Michael Stapelberg <stapelberg+linux@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 4c33705..7439ecd 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -406,6 +406,11 @@ static bool inode_do_switch_wbs(struct inode *inode,
inc_wb_stat(new_wb, WB_WRITEBACK);
}
+ if (mapping_tagged(mapping, PAGECACHE_TAG_WRITEBACK)) {
+ atomic_dec(&old_wb->writeback_inodes);
+ atomic_inc(&new_wb->writeback_inodes);
+ }
+
wb_get(new_wb);
/*
diff --git a/include/linux/backing-dev-defs.h b/include/linux/backing-dev-defs.h
index 1d7edad..06fb8e1 100644
--- a/include/linux/backing-dev-defs.h
+++ b/include/linux/backing-dev-defs.h
@@ -116,6 +116,7 @@ struct bdi_writeback {
struct list_head b_dirty_time; /* time stamps are dirty */
spinlock_t list_lock; /* protects the b_* lists */
+ atomic_t writeback_inodes; /* number of inodes under writeback */
struct percpu_counter stat[NR_WB_STAT_ITEMS];
unsigned long congested; /* WB_[a]sync_congested flags */
diff --git a/mm/backing-dev.c b/mm/backing-dev.c
index f5561ea..b4c707d 100644
--- a/mm/backing-dev.c
+++ b/mm/backing-dev.c
@@ -293,6 +293,7 @@ static int wb_init(struct bdi_writeback *wb, struct backing_dev_info *bdi,
INIT_LIST_HEAD(&wb->b_dirty_time);
spin_lock_init(&wb->list_lock);
+ atomic_set(&wb->writeback_inodes, 0);
wb->bw_time_stamp = jiffies;
wb->balanced_dirty_ratelimit = INIT_BW;
wb->dirty_ratelimit = INIT_BW;
diff --git a/mm/page-writeback.c b/mm/page-writeback.c
index 9f63548..e1aa1c9 100644
--- a/mm/page-writeback.c
+++ b/mm/page-writeback.c
@@ -2731,6 +2731,16 @@ int clear_page_dirty_for_io(struct page *page)
}
EXPORT_SYMBOL(clear_page_dirty_for_io);
+static void wb_inode_writeback_start(struct bdi_writeback *wb)
+{
+ atomic_inc(&wb->writeback_inodes);
+}
+
+static void wb_inode_writeback_end(struct bdi_writeback *wb)
+{
+ atomic_dec(&wb->writeback_inodes);
+}
+
int test_clear_page_writeback(struct page *page)
{
struct address_space *mapping = page_mapping(page);
@@ -2752,6 +2762,9 @@ int test_clear_page_writeback(struct page *page)
dec_wb_stat(wb, WB_WRITEBACK);
__wb_writeout_inc(wb);
+ if (!mapping_tagged(mapping,
+ PAGECACHE_TAG_WRITEBACK))
+ wb_inode_writeback_end(wb);
}
}
@@ -2794,8 +2807,13 @@ int __test_set_page_writeback(struct page *page, bool keep_write)
PAGECACHE_TAG_WRITEBACK);
xas_set_mark(&xas, PAGECACHE_TAG_WRITEBACK);
- if (bdi->capabilities & BDI_CAP_WRITEBACK_ACCT)
- inc_wb_stat(inode_to_wb(inode), WB_WRITEBACK);
+ if (bdi->capabilities & BDI_CAP_WRITEBACK_ACCT) {
+ struct bdi_writeback *wb = inode_to_wb(inode);
+
+ inc_wb_stat(wb, WB_WRITEBACK);
+ if (!on_wblist)
+ wb_inode_writeback_start(wb);
+ }
/*
* We can come through here when swapping anonymous