From: hubcap@kernel.org
To: linux-fsdevel@vger.kernel.org, christoph@lameter.com
Cc: Martin Brandenburg <martin@omnibond.com>,
Mike Marshall <hubcap@omnibond.com>
Subject: [PATCH 01/22] orangefs: implement xattr cache
Date: Thu, 18 Apr 2019 14:40:53 -0400 [thread overview]
Message-ID: <20190418184113.9152-2-hubcap@kernel.org> (raw)
In-Reply-To: <20190418184113.9152-1-hubcap@kernel.org>
From: Martin Brandenburg <martin@omnibond.com>
This uses the same timeout as the getattr cache. This substantially
increases performance when writing files with smaller buffer sizes.
When writing, the size is (often) changed, which causes a call to
notify_change which calls security_inode_need_killpriv which needs a
getxattr. Caching it reduces traffic to the server.
Signed-off-by: Martin Brandenburg <martin@omnibond.com>
Signed-off-by: Mike Marshall <hubcap@omnibond.com>
---
fs/orangefs/inode.c | 1 +
fs/orangefs/orangefs-kernel.h | 11 ++++
fs/orangefs/super.c | 10 ++++
fs/orangefs/xattr.c | 106 +++++++++++++++++++++++++++++++++-
4 files changed, 127 insertions(+), 1 deletion(-)
diff --git a/fs/orangefs/inode.c b/fs/orangefs/inode.c
index c3334eca18c7..b47765ea6870 100644
--- a/fs/orangefs/inode.c
+++ b/fs/orangefs/inode.c
@@ -364,6 +364,7 @@ static int orangefs_set_inode(struct inode *inode, void *data)
struct orangefs_object_kref *ref = (struct orangefs_object_kref *) data;
ORANGEFS_I(inode)->refn.fs_id = ref->fs_id;
ORANGEFS_I(inode)->refn.khandle = ref->khandle;
+ hash_init(ORANGEFS_I(inode)->xattr_cache);
return 0;
}
diff --git a/fs/orangefs/orangefs-kernel.h b/fs/orangefs/orangefs-kernel.h
index 17b24ad6b264..eba9136207f9 100644
--- a/fs/orangefs/orangefs-kernel.h
+++ b/fs/orangefs/orangefs-kernel.h
@@ -51,6 +51,7 @@
#include <linux/rwsem.h>
#include <linux/xattr.h>
#include <linux/exportfs.h>
+#include <linux/hashtable.h>
#include <asm/unaligned.h>
@@ -193,6 +194,8 @@ struct orangefs_inode_s {
unsigned long getattr_time;
u32 getattr_mask;
+
+ DECLARE_HASHTABLE(xattr_cache, 4);
};
/* per superblock private orangefs info */
@@ -217,6 +220,14 @@ struct orangefs_stats {
unsigned long writes;
};
+struct orangefs_cached_xattr {
+ struct hlist_node node;
+ char key[ORANGEFS_MAX_XATTR_NAMELEN];
+ char val[ORANGEFS_MAX_XATTR_VALUELEN];
+ ssize_t length;
+ unsigned long timeout;
+};
+
extern struct orangefs_stats orangefs_stats;
/*
diff --git a/fs/orangefs/super.c b/fs/orangefs/super.c
index dfaee90d30bd..31db6ac73de1 100644
--- a/fs/orangefs/super.c
+++ b/fs/orangefs/super.c
@@ -10,6 +10,7 @@
#include "orangefs-bufmap.h"
#include <linux/parser.h>
+#include <linux/hashtable.h>
/* a cache for orangefs-inode objects (i.e. orangefs inode private data) */
static struct kmem_cache *orangefs_inode_cache;
@@ -128,6 +129,15 @@ static void orangefs_i_callback(struct rcu_head *head)
{
struct inode *inode = container_of(head, struct inode, i_rcu);
struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
+ struct orangefs_cached_xattr *cx;
+ struct hlist_node *tmp;
+ int i;
+
+ hash_for_each_safe(orangefs_inode->xattr_cache, i, tmp, cx, node) {
+ hlist_del(&cx->node);
+ kfree(cx);
+ }
+
kmem_cache_free(orangefs_inode_cache, orangefs_inode);
}
diff --git a/fs/orangefs/xattr.c b/fs/orangefs/xattr.c
index 03bcb871544d..bdc285aea360 100644
--- a/fs/orangefs/xattr.c
+++ b/fs/orangefs/xattr.c
@@ -1,6 +1,7 @@
// SPDX-License-Identifier: GPL-2.0
/*
* (C) 2001 Clemson University and The University of Chicago
+ * Copyright 2018 Omnibond Systems, L.L.C.
*
* See COPYING in top-level directory.
*/
@@ -14,7 +15,7 @@
#include "orangefs-bufmap.h"
#include <linux/posix_acl_xattr.h>
#include <linux/xattr.h>
-
+#include <linux/hashtable.h>
#define SYSTEM_ORANGEFS_KEY "system.pvfs2."
#define SYSTEM_ORANGEFS_KEY_LEN 13
@@ -50,6 +51,35 @@ static inline int convert_to_internal_xattr_flags(int setxattr_flags)
return internal_flag;
}
+static unsigned int xattr_key(const char *key)
+{
+ unsigned int i = 0;
+ while (key)
+ i += *key++;
+ return i % 16;
+}
+
+static struct orangefs_cached_xattr *find_cached_xattr(struct inode *inode,
+ const char *key)
+{
+ struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
+ struct orangefs_cached_xattr *cx;
+ struct hlist_head *h;
+ struct hlist_node *tmp;
+ h = &orangefs_inode->xattr_cache[xattr_key(key)];
+ if (hlist_empty(h))
+ return NULL;
+ hlist_for_each_entry_safe(cx, tmp, h, node) {
+/* if (!time_before(jiffies, cx->timeout)) {
+ hlist_del(&cx->node);
+ kfree(cx);
+ continue;
+ }*/
+ if (!strcmp(cx->key, key))
+ return cx;
+ }
+ return NULL;
+}
/*
* Tries to get a specified key's attributes of a given
@@ -65,6 +95,7 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
{
struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
struct orangefs_kernel_op_s *new_op = NULL;
+ struct orangefs_cached_xattr *cx;
ssize_t ret = -ENOMEM;
ssize_t length = 0;
int fsuid;
@@ -93,6 +124,27 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
down_read(&orangefs_inode->xattr_sem);
+ cx = find_cached_xattr(inode, name);
+ if (cx && time_before(jiffies, cx->timeout)) {
+ if (cx->length == -1) {
+ ret = -ENODATA;
+ goto out_unlock;
+ } else {
+ if (size == 0) {
+ ret = cx->length;
+ goto out_unlock;
+ }
+ if (cx->length > size) {
+ ret = -ERANGE;
+ goto out_unlock;
+ }
+ memcpy(buffer, cx->val, cx->length);
+ memset(buffer + cx->length, 0, size - cx->length);
+ ret = cx->length;
+ goto out_unlock;
+ }
+ }
+
new_op = op_alloc(ORANGEFS_VFS_OP_GETXATTR);
if (!new_op)
goto out_unlock;
@@ -117,6 +169,15 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
" does not exist!\n",
get_khandle_from_ino(inode),
(char *)new_op->upcall.req.getxattr.key);
+ cx = kmalloc(sizeof *cx, GFP_KERNEL);
+ if (cx) {
+ strcpy(cx->key, name);
+ cx->length = -1;
+ cx->timeout = jiffies +
+ orangefs_getattr_timeout_msecs*HZ/1000;
+ hash_add(orangefs_inode->xattr_cache, &cx->node,
+ xattr_key(cx->key));
+ }
}
goto out_release_op;
}
@@ -156,6 +217,23 @@ ssize_t orangefs_inode_getxattr(struct inode *inode, const char *name,
ret = length;
+ if (cx) {
+ strcpy(cx->key, name);
+ memcpy(cx->val, buffer, length);
+ cx->length = length;
+ cx->timeout = jiffies + HZ;
+ } else {
+ cx = kmalloc(sizeof *cx, GFP_KERNEL);
+ if (cx) {
+ strcpy(cx->key, name);
+ memcpy(cx->val, buffer, length);
+ cx->length = length;
+ cx->timeout = jiffies + HZ;
+ hash_add(orangefs_inode->xattr_cache, &cx->node,
+ xattr_key(cx->key));
+ }
+ }
+
out_release_op:
op_release(new_op);
out_unlock:
@@ -168,6 +246,9 @@ static int orangefs_inode_removexattr(struct inode *inode, const char *name,
{
struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
struct orangefs_kernel_op_s *new_op = NULL;
+ struct orangefs_cached_xattr *cx;
+ struct hlist_head *h;
+ struct hlist_node *tmp;
int ret = -ENOMEM;
if (strlen(name) >= ORANGEFS_MAX_XATTR_NAMELEN)
@@ -209,6 +290,16 @@ static int orangefs_inode_removexattr(struct inode *inode, const char *name,
"orangefs_inode_removexattr: returning %d\n", ret);
op_release(new_op);
+
+ h = &orangefs_inode->xattr_cache[xattr_key(name)];
+ hlist_for_each_entry_safe(cx, tmp, h, node) {
+ if (!strcmp(cx->key, name)) {
+ hlist_del(&cx->node);
+ kfree(cx);
+ break;
+ }
+ }
+
out_unlock:
up_write(&orangefs_inode->xattr_sem);
return ret;
@@ -226,6 +317,9 @@ int orangefs_inode_setxattr(struct inode *inode, const char *name,
struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
struct orangefs_kernel_op_s *new_op;
int internal_flag = 0;
+ struct orangefs_cached_xattr *cx;
+ struct hlist_head *h;
+ struct hlist_node *tmp;
int ret = -ENOMEM;
gossip_debug(GOSSIP_XATTR_DEBUG,
@@ -287,6 +381,16 @@ int orangefs_inode_setxattr(struct inode *inode, const char *name,
/* when request is serviced properly, free req op struct */
op_release(new_op);
+
+ h = &orangefs_inode->xattr_cache[xattr_key(name)];
+ hlist_for_each_entry_safe(cx, tmp, h, node) {
+ if (!strcmp(cx->key, name)) {
+ hlist_del(&cx->node);
+ kfree(cx);
+ break;
+ }
+ }
+
out_unlock:
up_write(&orangefs_inode->xattr_sem);
return ret;
--
2.20.1
next prev parent reply other threads:[~2019-04-18 18:42 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-04-18 18:40 [RFC PATCH 00/22] Orangefs Through the Pagecache hubcap
2019-04-18 18:40 ` hubcap [this message]
2019-04-18 18:40 ` [PATCH 02/22] orangefs: do not invalidate attributes on inode create hubcap
2019-04-18 18:40 ` [PATCH 03/22] orangefs: simplify orangefs_inode_getattr interface hubcap
2019-04-18 18:40 ` [PATCH 04/22] orangefs: update attributes rather than relying on server hubcap
2019-04-18 18:40 ` [PATCH 05/22] orangefs: hold i_lock during inode_getattr hubcap
2019-04-18 18:40 ` [PATCH 06/22] orangefs: set up and use backing_dev_info hubcap
2019-04-18 18:40 ` [PATCH 07/22] orangefs: let setattr write to cached inode hubcap
2019-04-18 18:41 ` [PATCH 08/22] orangefs: reorganize setattr functions to track attribute changes hubcap
2019-04-18 18:41 ` [PATCH 09/22] orangefs: remove orangefs_readpages hubcap
2019-04-18 18:41 ` [PATCH 10/22] orangefs: service ops done for writeback are not killable hubcap
2019-04-18 18:41 ` [PATCH 11/22] orangefs: migrate to generic_file_read_iter hubcap
2019-04-18 18:41 ` [PATCH 12/22] orangefs: implement writepage hubcap
2019-04-18 18:41 ` [PATCH 13/22] orangefs: do not return successful read when the client-core disappeared hubcap
2019-04-18 18:41 ` [PATCH 14/22] orangefs: move do_readv_writev to direct_IO hubcap
2019-04-18 18:41 ` [PATCH 15/22] orangefs: skip inode writeout if nothing to write hubcap
2019-04-18 18:41 ` [PATCH 16/22] orangefs: avoid fsync service operation on flush hubcap
2019-04-18 18:41 ` [PATCH 17/22] orangefs: write range tracking hubcap
2019-04-18 18:41 ` [PATCH 18/22] orangefs: implement writepages hubcap
2019-04-18 18:41 ` [PATCH 19/22] orangefs: add orangefs_revalidate_mapping hubcap
2019-04-18 18:41 ` [PATCH 20/22] orangefs: remember count when reading hubcap
2019-04-18 18:41 ` [PATCH 21/22] orangefs: pass slot index back to readpage hubcap
2019-04-18 18:41 ` [PATCH 22/22] orangefs: copy Orangefs-sized blocks into the pagecache if possible hubcap
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190418184113.9152-2-hubcap@kernel.org \
--to=hubcap@kernel.org \
--cc=christoph@lameter.com \
--cc=hubcap@omnibond.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=martin@omnibond.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).