linux-cifs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [Patch] cifs: Deferred close perfomance improvements
@ 2021-09-17 18:33 Rohith Surabattula
  0 siblings, 0 replies; only message in thread
From: Rohith Surabattula @ 2021-09-17 18:33 UTC (permalink / raw)
  To: Steve French, linux-cifs, Shyam Prasad N, ronnie sahlberg, sribhat.msa

[-- Attachment #1: Type: text/plain, Size: 291 bytes --]

Hi Steve/All,

Please find the attached patches related to deferred close fixes for
recently found xfstests failures(generic/478 -- locking test,
generic/461 -- fsstress test) and also performance improvement patch
related to close of deferred handles during rename/unlink.

Regards,
Rohith

[-- Attachment #2: cifs-Not-to-defer-close-on-file-when-lock-is-set.patch --]
[-- Type: application/octet-stream, Size: 1977 bytes --]

From 25e14bd1ebda883e5f9222b010588b5b632275fa Mon Sep 17 00:00:00 2001
From: Rohith Surabattula <rohiths@microsoft.com>
Date: Fri, 17 Sep 2021 17:23:30 +0000
Subject: [PATCH] cifs: Not to defer close on file when lock is set

Close file immediately when lock is set.

Signed-off-by: Rohith Surabattula <rohiths@microsoft.com>
---
 fs/cifs/cifsglob.h | 1 +
 fs/cifs/file.c     | 2 ++
 2 files changed, 3 insertions(+)

diff --git a/fs/cifs/cifsglob.h b/fs/cifs/cifsglob.h
index c6a9542ca281..cf2141483b37 100644
--- a/fs/cifs/cifsglob.h
+++ b/fs/cifs/cifsglob.h
@@ -1403,6 +1403,7 @@ struct cifsInodeInfo {
 #define CIFS_INO_INVALID_MAPPING	  (4) /* pagecache is invalid */
 #define CIFS_INO_LOCK			  (5) /* lock bit for synchronization */
 #define CIFS_INO_MODIFIED_ATTR            (6) /* Indicate change in mtime/ctime */
+#define CIFS_INO_CLOSE_ON_LOCK            (7) /* Not to defer the close when lock is set */
 	unsigned long flags;
 	spinlock_t writers_lock;
 	unsigned int writers;		/* Number of writers on this inode */
diff --git a/fs/cifs/file.c b/fs/cifs/file.c
index bb98fbdd22a9..9d3bc6784771 100644
--- a/fs/cifs/file.c
+++ b/fs/cifs/file.c
@@ -881,6 +881,7 @@ int cifs_close(struct inode *inode, struct file *file)
 		dclose = kmalloc(sizeof(struct cifs_deferred_close), GFP_KERNEL);
 		if ((cinode->oplock == CIFS_CACHE_RHW_FLG) &&
 		    cinode->lease_granted &&
+		    !test_bit(CIFS_INO_CLOSE_ON_LOCK, &cinode->flags) &&
 		    dclose) {
 			if (test_bit(CIFS_INO_MODIFIED_ATTR, &cinode->flags))
 				inode->i_ctime = inode->i_mtime = current_time(inode);
@@ -1861,6 +1862,7 @@ int cifs_lock(struct file *file, int cmd, struct file_lock *flock)
 	cifs_read_flock(flock, &type, &lock, &unlock, &wait_flag,
 			tcon->ses->server);
 	cifs_sb = CIFS_FILE_SB(file);
+	set_bit(CIFS_INO_CLOSE_ON_LOCK, &CIFS_I(d_inode(cfile->dentry))->flags);
 
 	if (cap_unix(tcon->ses) &&
 	    (CIFS_UNIX_FCNTL_CAP & le64_to_cpu(tcon->fsUnixInfo.Capability)) &&
-- 
2.33.0.rc1


[-- Attachment #3: cifs-Deferred-close-performance-improvements.patch --]
[-- Type: application/octet-stream, Size: 3615 bytes --]

From ce8b14eb4621df79b1ea9ba9b06b99704abbbb40 Mon Sep 17 00:00:00 2001
From: Rohith Surabattula <rohiths@microsoft.com>
Date: Fri, 17 Sep 2021 18:14:26 +0000
Subject: [PATCH] cifs: Deferred close performance improvements

During unlink/rename instead of closing all the deferred handles
under tcon, close only handles under the requested dentry.

Signed-off-by: Rohith Surabattula <rohiths@microsoft.com>
---
 fs/cifs/cifsproto.h |  3 +++
 fs/cifs/inode.c     |  6 +++---
 fs/cifs/misc.c      | 37 +++++++++++++++++++++++++++++++++++++
 3 files changed, 43 insertions(+), 3 deletions(-)

diff --git a/fs/cifs/cifsproto.h b/fs/cifs/cifsproto.h
index e0def0f0714b..689d73869a25 100644
--- a/fs/cifs/cifsproto.h
+++ b/fs/cifs/cifsproto.h
@@ -268,6 +268,9 @@ extern void cifs_close_deferred_file(struct cifsInodeInfo *cifs_inode);
 
 extern void cifs_close_all_deferred_files(struct cifs_tcon *cifs_tcon);
 
+extern void cifs_close_deferred_file_under_dentry(struct cifs_tcon *cifs_tcon,
+				const char *path);
+
 extern struct TCP_Server_Info *cifs_get_tcp_session(struct smb3_fs_context *ctx);
 extern void cifs_put_tcp_session(struct TCP_Server_Info *server,
 				 int from_reconnect);
diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c
index 65f8a70cece3..59c2fe634f77 100644
--- a/fs/cifs/inode.c
+++ b/fs/cifs/inode.c
@@ -1625,7 +1625,7 @@ int cifs_unlink(struct inode *dir, struct dentry *dentry)
 		goto unlink_out;
 	}
 
-	cifs_close_deferred_file(CIFS_I(inode));
+	cifs_close_deferred_file_under_dentry(tcon, full_path);
 	if (cap_unix(tcon->ses) && (CIFS_UNIX_POSIX_PATH_OPS_CAP &
 				le64_to_cpu(tcon->fsUnixInfo.Capability))) {
 		rc = CIFSPOSIXDelFile(xid, tcon, full_path,
@@ -2114,9 +2114,9 @@ cifs_rename2(struct user_namespace *mnt_userns, struct inode *source_dir,
 		goto cifs_rename_exit;
 	}
 
-	cifs_close_deferred_file(CIFS_I(d_inode(source_dentry)));
+	cifs_close_deferred_file_under_dentry(tcon, from_name);
 	if (d_inode(target_dentry) != NULL)
-		cifs_close_deferred_file(CIFS_I(d_inode(target_dentry)));
+		cifs_close_deferred_file_under_dentry(tcon, to_name);
 
 	rc = cifs_do_rename(xid, source_dentry, from_name, target_dentry,
 			    to_name);
diff --git a/fs/cifs/misc.c b/fs/cifs/misc.c
index 57e695e3c969..e16f9cbcd97d 100644
--- a/fs/cifs/misc.c
+++ b/fs/cifs/misc.c
@@ -781,6 +781,43 @@ cifs_close_all_deferred_files(struct cifs_tcon *tcon)
 		kfree(tmp_list);
 	}
 }
+void
+cifs_close_deferred_file_under_dentry(struct cifs_tcon *tcon, const char *path)
+{
+	struct cifsFileInfo *cfile;
+	struct list_head *tmp;
+	struct file_list *tmp_list, *tmp_next_list;
+	struct list_head file_head;
+	void *page;
+	const char *full_path;
+
+	INIT_LIST_HEAD(&file_head);
+	page = alloc_dentry_path();
+	spin_lock(&tcon->open_file_lock);
+	list_for_each(tmp, &tcon->openFileList) {
+		cfile = list_entry(tmp, struct cifsFileInfo, tlist);
+		full_path = build_path_from_dentry(cfile->dentry, page);
+		if (strstr(full_path, path)) {
+			if (delayed_work_pending(&cfile->deferred)) {
+				if (cancel_delayed_work(&cfile->deferred)) {
+					tmp_list = kmalloc(sizeof(struct file_list), GFP_ATOMIC);
+					if (tmp_list == NULL)
+						break;
+					tmp_list->cfile = cfile;
+					list_add_tail(&tmp_list->list, &file_head);
+				}
+			}
+		}
+	}
+	spin_unlock(&tcon->open_file_lock);
+
+	list_for_each_entry_safe(tmp_list, tmp_next_list, &file_head, list) {
+		_cifsFileInfo_put(tmp_list->cfile, true, false);
+		list_del(&tmp_list->list);
+		kfree(tmp_list);
+	}
+	free_dentry_path(page);
+}
 
 /* parses DFS refferal V3 structure
  * caller is responsible for freeing target_nodes
-- 
2.33.0.rc1


[-- Attachment #4: cifs-Fix-soft-lockup-during-fsstress.patch --]
[-- Type: application/octet-stream, Size: 1282 bytes --]

From 54760d81624ef2abe662da17e3f2c86bf9e0af49 Mon Sep 17 00:00:00 2001
From: Rohith Surabattula <rohiths@microsoft.com>
Date: Fri, 17 Sep 2021 17:29:42 +0000
Subject: [PATCH] cifs: Fix soft lockup during fsstress

Below traces are observed during fsstress and system got hung.
[  130.698396] watchdog: BUG: soft lockup - CPU#6 stuck for 26s!

Signed-off-by: Rohith Surabattula <rohiths@microsoft.com>
---
 fs/cifs/misc.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/cifs/misc.c b/fs/cifs/misc.c
index 9469f1cf0b46..57e695e3c969 100644
--- a/fs/cifs/misc.c
+++ b/fs/cifs/misc.c
@@ -736,7 +736,7 @@ cifs_close_deferred_file(struct cifsInodeInfo *cifs_inode)
 			if (cancel_delayed_work(&cfile->deferred)) {
 				tmp_list = kmalloc(sizeof(struct file_list), GFP_ATOMIC);
 				if (tmp_list == NULL)
-					continue;
+					break;
 				tmp_list->cfile = cfile;
 				list_add_tail(&tmp_list->list, &file_head);
 			}
@@ -767,7 +767,7 @@ cifs_close_all_deferred_files(struct cifs_tcon *tcon)
 			if (cancel_delayed_work(&cfile->deferred)) {
 				tmp_list = kmalloc(sizeof(struct file_list), GFP_ATOMIC);
 				if (tmp_list == NULL)
-					continue;
+					break;
 				tmp_list->cfile = cfile;
 				list_add_tail(&tmp_list->list, &file_head);
 			}
-- 
2.33.0.rc1


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2021-09-17 18:34 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-09-17 18:33 [Patch] cifs: Deferred close perfomance improvements Rohith Surabattula

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).