Revert "nfs: replace nfs_add_stats with nfs_inc_stats when add one"
authorNicolas Iooss <nicolas.iooss_linux@m4x.org>
Thu, 16 Apr 2015 10:48:39 +0000 (18:48 +0800)
committerTrond Myklebust <trond.myklebust@primarydata.com>
Thu, 23 Apr 2015 19:16:15 +0000 (15:16 -0400)
This reverts commit 5a254d08b086d80cbead2ebcee6d2a4b3a15587a.

Since commit 5a254d08b086 ("nfs: replace nfs_add_stats with
nfs_inc_stats when add one"), nfs_readpage and nfs_do_writepage use
nfs_inc_stats to increment NFSIOS_READPAGES and NFSIOS_WRITEPAGES
instead of nfs_add_stats.

However nfs_inc_stats does not do the same thing as nfs_add_stats with
value 1 because these functions work on distinct stats:
nfs_inc_stats increments stats from "enum nfs_stat_eventcounters" (in
server->io_stats->events) and nfs_add_stats those from "enum
nfs_stat_bytecounters" (in server->io_stats->bytes).

Signed-off-by: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Fixes: 5a254d08b086 ("nfs: replace nfs_add_stats with nfs_inc_stats...")
Cc: stable@vger.kernel.org # 3.19+
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
fs/nfs/read.c
fs/nfs/write.c

index 568ecf0a880f1145aedb69a1d93241b5eba300e0..848d8b1db4ceca7095dbfd68ec610283b20052a8 100644 (file)
@@ -284,7 +284,7 @@ int nfs_readpage(struct file *file, struct page *page)
        dprintk("NFS: nfs_readpage (%p %ld@%lu)\n",
                page, PAGE_CACHE_SIZE, page_file_index(page));
        nfs_inc_stats(inode, NFSIOS_VFSREADPAGE);
-       nfs_inc_stats(inode, NFSIOS_READPAGES);
+       nfs_add_stats(inode, NFSIOS_READPAGES, 1);
 
        /*
         * Try to flush any pending writes to the file..
index 7f933a39f3854f5b3ae4e8a64a23fc7b67763470..36d0c0a6e68aef76b1915be9322838b35cf951c1 100644 (file)
@@ -580,7 +580,7 @@ static int nfs_do_writepage(struct page *page, struct writeback_control *wbc, st
        int ret;
 
        nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
-       nfs_inc_stats(inode, NFSIOS_WRITEPAGES);
+       nfs_add_stats(inode, NFSIOS_WRITEPAGES, 1);
 
        nfs_pageio_cond_complete(pgio, page_file_index(page));
        ret = nfs_page_async_flush(pgio, page, wbc->sync_mode == WB_SYNC_NONE);