[PATCH 42/47] nfs: heuristics to avoid commit

From: Wu Fengguang
Date: Mon Dec 13 2010 - 01:51:54 EST


The heuristics introduced by commit 420e3646 ("NFS: Reduce the number of
unnecessary COMMIT calls") do not work well for large inodes being
actively written to.

Refine the criterion to
- it has gone quiet (all data transfered to server)
- has accumulated >= 4MB data to commit (so it will be large IO)
- too few active commits (hence active IO) in the server

Signed-off-by: Wu Fengguang <fengguang.wu@xxxxxxxxx>
---
fs/nfs/write.c | 31 ++++++++++++++++++++++++++-----
1 file changed, 26 insertions(+), 5 deletions(-)

--- linux-next.orig/fs/nfs/write.c 2010-12-09 12:19:24.000000000 +0800
+++ linux-next/fs/nfs/write.c 2010-12-09 12:20:20.000000000 +0800
@@ -1518,17 +1518,38 @@ out_mark_dirty:
return res;
}

-static int nfs_commit_unstable_pages(struct inode *inode, struct writeback_control *wbc)
+static bool nfs_should_commit(struct inode *inode,
+ struct writeback_control *wbc)
{
+ struct nfs_server *nfss = NFS_SERVER(inode);
struct nfs_inode *nfsi = NFS_I(inode);
+ unsigned long npages = nfsi->npages;
+ unsigned long to_commit = nfsi->ncommit;
+ unsigned long in_commit = atomic_long_read(&nfss->in_commit);
+
+ /* no more active writes */
+ if (to_commit == npages)
+ return true;
+
+ /* big enough */
+ if (to_commit >= MIN_WRITEBACK_PAGES)
+ return true;
+
+ /* active commits drop low: kick more IO for the server disk */
+ if (to_commit > in_commit / 2)
+ return true;
+
+ return false;
+}
+
+static int nfs_commit_unstable_pages(struct inode *inode,
+ struct writeback_control *wbc)
+{
int flags = FLUSH_SYNC;
int ret = 0;

if (wbc->sync_mode == WB_SYNC_NONE) {
- /* Don't commit yet if this is a non-blocking flush and there
- * are a lot of outstanding writes for this mapping.
- */
- if (nfsi->ncommit <= (nfsi->npages >> 1))
+ if (!nfs_should_commit(inode, wbc))
goto out_mark_dirty;

/* don't wait for the COMMIT response */


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/