fuse: separate queue for FORGET requests

Terje Malmedal reports that a fuse filesystem with 32 million inodes
on a machine with lots of memory can go unresponsive for up to 30
minutes when all those inodes are evicted from the icache.

The reason is that FORGET messages, sent when the inode is evicted,
are queued up together with regular filesystem requests, and while the
huge queue of FORGET messages are processed no other filesystem
operation can proceed.

Since a full fuse request structure is allocated for each inode, these
take up quite a bit of memory as well.

To solve these issues, create a slim 'fuse_forget_link' structure
containing just the minimum of information required to send the FORGET
request and chain these on a separate queue.

When userspace is asking for a request make sure that FORGET and
non-FORGET requests are selected fairly: for each 8 non-FORGET allow
16 FORGET requests.  This will make sure FORGETs do not pile up, yet
other requests are also allowed to proceed while the queued FORGETs
are processed.

Reported-by: Terje Malmedal <terje.malmedal@usit.uio.no>
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>

1 files changed, 11 insertions, 19 deletions

diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index cfce3ad86a92..7ba4d351da65 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -71,6 +71,11 @@ struct fuse_mount_data {
         unsigned blksize;
 };
 
+struct fuse_forget_link *fuse_alloc_forget()
+{
+        return kzalloc(sizeof(struct fuse_forget_link), GFP_KERNEL);
+}
+
 static struct inode *fuse_alloc_inode(struct super_block *sb)
 {
         struct inode *inode;
@@ -90,8 +95,8 @@ static struct inode *fuse_alloc_inode(struct super_block *sb)
         INIT_LIST_HEAD(&fi->queued_writes);
         INIT_LIST_HEAD(&fi->writepages);
         init_waitqueue_head(&fi->page_waitq);
-        fi->forget_req = fuse_request_alloc();
-        if (!fi->forget_req) {
+        fi->forget = fuse_alloc_forget();
+        if (!fi->forget) {
                 kmem_cache_free(fuse_inode_cachep, inode);
                 return NULL;
         }
@@ -104,24 +109,10 @@ static void fuse_destroy_inode(struct inode *inode)
         struct fuse_inode *fi = get_fuse_inode(inode);
         BUG_ON(!list_empty(&fi->write_files));
         BUG_ON(!list_empty(&fi->queued_writes));
-        if (fi->forget_req)
-                fuse_request_free(fi->forget_req);
+        kfree(fi->forget);
         kmem_cache_free(fuse_inode_cachep, inode);
 }
 
-void fuse_send_forget(struct fuse_conn *fc, struct fuse_req *req,
-                      u64 nodeid, u64 nlookup)
-{
-        struct fuse_forget_in *inarg = &req->misc.forget_in;
-        inarg->nlookup = nlookup;
-        req->in.h.opcode = FUSE_FORGET;
-        req->in.h.nodeid = nodeid;
-        req->in.numargs = 1;
-        req->in.args[0].size = sizeof(struct fuse_forget_in);
-        req->in.args[0].value = inarg;
-        fuse_request_send_noreply(fc, req);
-}
-
 static void fuse_evict_inode(struct inode *inode)
 {
         truncate_inode_pages(&inode->i_data, 0);
@@ -129,8 +120,8 @@ static void fuse_evict_inode(struct inode *inode)
         if (inode->i_sb->s_flags & MS_ACTIVE) {
                 struct fuse_conn *fc = get_fuse_conn(inode);
                 struct fuse_inode *fi = get_fuse_inode(inode);
-                fuse_send_forget(fc, fi->forget_req, fi->nodeid, fi->nlookup);
-                fi->forget_req = NULL;
+                fuse_queue_forget(fc, fi->forget, fi->nodeid, fi->nlookup);
+                fi->forget = NULL;
         }
 }
 
@@ -534,6 +525,7 @@ void fuse_conn_init(struct fuse_conn *fc)
         INIT_LIST_HEAD(&fc->interrupts);
         INIT_LIST_HEAD(&fc->bg_queue);
         INIT_LIST_HEAD(&fc->entry);
+        fc->forget_list_tail = &fc->forget_list_head;
         atomic_set(&fc->num_waiting, 0);
         fc->max_background = FUSE_DEFAULT_MAX_BACKGROUND;
         fc->congestion_threshold = FUSE_DEFAULT_CONGESTION_THRESHOLD;