Groups > linux.kernel > #1287116 > unrolled thread

[PATCHSET v2] ->follow_link() without dropping from RCU mode

Started by	Al Viro <viro@ZenIV.linux.org.uk>
First post	2015-12-09 06:40 +0100
Last post	2015-12-09 23:00 +0100
Articles	16 — 4 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

  [PATCHSET v2] ->follow_link() without dropping from RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 02/11] logfs: don't duplicate page_symlink_inode_operations Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 08/11] teach page_get_link() to work in RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 10/11] teach proc_self_get_link()/proc_thread_self_get_link() to work in RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 11/11] teach nfs_get_link() to work in RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 06/11] don't put symlink bodies in pagecache into highmem Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 01/11] switch befs long symlinks to page_symlink_operations Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 03/11] udf: don't duplicate page_symlink_inode_operations Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 09/11] teach shmem_get_link() to work in RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    [PATCH v2 04/11] ufs: get rid of ->setattr() for symlinks Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 06:40 +0100
    Re: [PATCHSET v2] ->follow_link() without dropping from RCU mode Linus Torvalds <torvalds@linux-foundation.org> - 2015-12-09 18:30 +0100
      Re: [PATCHSET v2] ->follow_link() without dropping from RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-09 19:30 +0100
        Re: [PATCHSET v2] ->follow_link() without dropping from RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-10 01:20 +0100
          Re: [PATCHSET v2] ->follow_link() without dropping from RCU mode Al Viro <viro@ZenIV.linux.org.uk> - 2015-12-10 03:50 +0100
            Re: [PATCHSET v2] ->follow_link() without dropping from RCU mode Rasmus Villemoes <linux@rasmusvillemoes.dk> - 2015-12-11 08:50 +0100
    Re: [PATCHSET v2] ->follow_link() without dropping from RCU mode NeilBrown <neilb@suse.com> - 2015-12-09 23:00 +0100

#1287116 — [PATCHSET v2] ->follow_link() without dropping from RCU mode

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCHSET v2] ->follow_link() without dropping from RCU mode
Message-ID	<qDFDP-485-3@gated-at.bofh.it>

On Tue, Nov 17, 2015 at 10:57:52PM +0000, Al Viro wrote:

> 	Right now we stay in RCU mode for fast symlink traversal.
> However, anything trickier drops out of RCU mode - back in 4.2
> the symlink-related pile had grown too large to add this on top
> of everything else.  Below is an attempt to do that now.
> 
> Those who prefer to use git for review can find that series in
> git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git work.symlinks

Compared to the previous iteration:
	* rebased to 4.4-rc4.
	* NFS (in case of pagecache hit, of course) also stays in RCU mode.
	* renamed inode_nohigh() as suggested by dchinner, moved to
fs/inode.c - less header PITA that way.

What's there:
01/11: switch befs long symlinks to page_symlink_operations
02/11: logfs: don't duplicate page_symlink_inode_operations
03/11: udf: don't duplicate page_symlink_inode_operations
04/11: ufs: get rid of ->setattr() for symlinks
        Simplifying things a bit by switching them to page_symlink_operations
where possible.

05/11: namei: page_getlink() and page_follow_link_light() are the same thing
        Get rid of some code duplication

06/11: don't put symlink bodies in pagecache into highmem
        Bugfix for a long-standing mess.  For pagecache-based symlinks we end
up with kmap() of the body for the duration of traversal.  Which could take
a long time if we walk into e.g. slow automount along the way.  It's DoSable,
actually.  Not hard to fix - there's no reason to use GFP_HIGHUSER_MOVABLE
for those guys; GFP_USER serves just as well.

07/11: replace ->follow_link() with new method that could stay
        Meat of the series - we switch to a new method (->get_link()) that
differs from ->follow_link() in getting dentry and inode separately; it
can be called in RCU mode (with NULL dentry) and it should bail out with
ERR_PTR(-ECHILD) if it needs non-RCU.  All instances converted, most of them
by making them bail out immediately in RCU mode.  Some are trivialy
RCU-safe, though; those do not bail out at all.

08/11: teach page_get_link() to work in RCU mode
09/11: teach shmem_get_link() to work in RCU mode
10/11: teach proc_self_get_link()/proc_thread_self_get_link()
11/11: teach nfs_get_link() to work in RCU mode
        Teaching more instances to stay in RCU mode if they can.  (8) and
(11) are similar to what Neil had done back in March, except for the lack of
kmap mess to deal with - that's one of the benefits of having (6) done...

	Missing, probably worth adding: XFS, overlayfs.  Maybe gfs2 - if it's
at all possible (not sure if it is).  Maybe autofs4 as well.

	Procfs magical symlinks definitely are *not* worth trying to handle
in RCU mode - ->d_revalidate() on them is such that it'll kick us out of
RCU mode before we even see that it's a symlink, let alone try to follow it.

Any help with review and testing would be welcome.  FWIW, it seems to survive
the beating here.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1287117 — [PATCH v2 02/11] logfs: don't duplicate page_symlink_inode_operations

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 02/11] logfs: don't duplicate page_symlink_inode_operations
Message-ID	<qDFDQ-485-7@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/logfs/dir.c   | 8 +-------
 fs/logfs/inode.c | 2 +-
 fs/logfs/logfs.h | 1 -
 3 files changed, 2 insertions(+), 9 deletions(-)

diff --git a/fs/logfs/dir.c b/fs/logfs/dir.c
index f9b45d4..99944a4 100644
--- a/fs/logfs/dir.c
+++ b/fs/logfs/dir.c
@@ -528,7 +528,7 @@ static int logfs_symlink(struct inode *dir, struct dentry *dentry,
 	if (IS_ERR(inode))
 		return PTR_ERR(inode);
 
-	inode->i_op = &logfs_symlink_iops;
+	inode->i_op = &page_symlink_inode_operations;
 	inode->i_mapping->a_ops = &logfs_reg_aops;
 
 	return __logfs_create(dir, dentry, inode, target, destlen);
@@ -776,12 +776,6 @@ fail:
 	return -EIO;
 }
 
-const struct inode_operations logfs_symlink_iops = {
-	.readlink	= generic_readlink,
-	.follow_link	= page_follow_link_light,
-	.put_link	= page_put_link,
-};
-
 const struct inode_operations logfs_dir_iops = {
 	.create		= logfs_create,
 	.link		= logfs_link,
diff --git a/fs/logfs/inode.c b/fs/logfs/inode.c
index af49e2d..06baa92 100644
--- a/fs/logfs/inode.c
+++ b/fs/logfs/inode.c
@@ -64,7 +64,7 @@ static void logfs_inode_setops(struct inode *inode)
 		inode->i_mapping->a_ops = &logfs_reg_aops;
 		break;
 	case S_IFLNK:
-		inode->i_op = &logfs_symlink_iops;
+		inode->i_op = &page_symlink_inode_operations;
 		inode->i_mapping->a_ops = &logfs_reg_aops;
 		break;
 	case S_IFSOCK:	/* fall through */
diff --git a/fs/logfs/logfs.h b/fs/logfs/logfs.h
index 5f09376..209a26d 100644
--- a/fs/logfs/logfs.h
+++ b/fs/logfs/logfs.h
@@ -495,7 +495,6 @@ static inline int logfs_get_sb_mtd(struct logfs_super *s, int mtdnr)
 #endif
 
 /* dir.c */
-extern const struct inode_operations logfs_symlink_iops;
 extern const struct inode_operations logfs_dir_iops;
 extern const struct file_operations logfs_dir_fops;
 int logfs_replay_journal(struct super_block *sb);
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287118 — [PATCH v2 08/11] teach page_get_link() to work in RCU mode

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 08/11] teach page_get_link() to work in RCU mode
Message-ID	<qDFDQ-485-11@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

more or less along the lines of Neil's patchset, sans the insanity
around kmap().

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/namei.c | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

diff --git a/fs/namei.c b/fs/namei.c
index 1da3064..8f51788 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -4533,12 +4533,19 @@ const char *page_get_link(struct dentry *dentry, struct inode *inode,
 	struct page *page;
 	struct address_space *mapping = inode->i_mapping;
 
-	if (!dentry)
-		return ERR_PTR(-ECHILD);
-
-	page = read_mapping_page(mapping, 0, NULL);
-	if (IS_ERR(page))
-		return (char*)page;
+	if (!dentry) {
+		page = find_get_page(mapping, 0);
+		if (!page)
+			return ERR_PTR(-ECHILD);
+		if (!PageUptodate(page)) {
+			put_page(page);
+			return ERR_PTR(-ECHILD);
+		}
+	} else {
+		page = read_mapping_page(mapping, 0, NULL);
+		if (IS_ERR(page))
+			return (char*)page;
+	}
 	*cookie = page;
 	BUG_ON(mapping_gfp_mask(mapping) & __GFP_HIGHMEM);
 	kaddr = page_address(page);
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287119 — [PATCH v2 10/11] teach proc_self_get_link()/proc_thread_self_get_link() to work in RCU mode

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 10/11] teach proc_self_get_link()/proc_thread_self_get_link() to work in RCU mode
Message-ID	<qDFDQ-485-13@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/proc/self.c        | 8 +++-----
 fs/proc/thread_self.c | 9 ++++-----
 2 files changed, 7 insertions(+), 10 deletions(-)

diff --git a/fs/proc/self.c b/fs/proc/self.c
index 9dd0ae6..7a8b19e 100644
--- a/fs/proc/self.c
+++ b/fs/proc/self.c
@@ -25,14 +25,12 @@ static const char *proc_self_get_link(struct dentry *dentry,
 	pid_t tgid = task_tgid_nr_ns(current, ns);
 	char *name;
 
-	if (!dentry)
-		return ERR_PTR(-ECHILD);
 	if (!tgid)
 		return ERR_PTR(-ENOENT);
 	/* 11 for max length of signed int in decimal + NULL term */
-	name = kmalloc(12, GFP_KERNEL);
-	if (!name)
-		return ERR_PTR(-ENOMEM);
+	name = kmalloc(12, dentry ? GFP_KERNEL : GFP_ATOMIC);
+	if (unlikely(!name))
+		return dentry ? ERR_PTR(-ENOMEM) : ERR_PTR(-ECHILD);
 	sprintf(name, "%d", tgid);
 	return *cookie = name;
 }
diff --git a/fs/proc/thread_self.c b/fs/proc/thread_self.c
index 50eef6f..03eaa84 100644
--- a/fs/proc/thread_self.c
+++ b/fs/proc/thread_self.c
@@ -27,13 +27,12 @@ static const char *proc_thread_self_get_link(struct dentry *dentry,
 	pid_t pid = task_pid_nr_ns(current, ns);
 	char *name;
 
-	if (!dentry)
-		return ERR_PTR(-ECHILD);
 	if (!pid)
 		return ERR_PTR(-ENOENT);
-	name = kmalloc(PROC_NUMBUF + 6 + PROC_NUMBUF, GFP_KERNEL);
-	if (!name)
-		return ERR_PTR(-ENOMEM);
+	name = kmalloc(PROC_NUMBUF + 6 + PROC_NUMBUF,
+				dentry ? GFP_KERNEL : GFP_ATOMIC);
+	if (unlikely(!name))
+		return dentry ? ERR_PTR(-ENOMEM) : ERR_PTR(-ECHILD);
 	sprintf(name, "%d/task/%d", tgid, pid);
 	return *cookie = name;
 }
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287120 — [PATCH v2 11/11] teach nfs_get_link() to work in RCU mode

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 11/11] teach nfs_get_link() to work in RCU mode
Message-ID	<qDFDQ-485-5@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

based upon the corresponding patch from Neil's March patchset,
again with kmap-related horrors removed.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/nfs/inode.c         | 21 +++++++++++++++++++++
 fs/nfs/symlink.c       | 30 ++++++++++++++++++++----------
 include/linux/nfs_fs.h |  1 +
 3 files changed, 42 insertions(+), 10 deletions(-)

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index ae9aa0b..aa828e8 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -1087,6 +1087,27 @@ static bool nfs_mapping_need_revalidate_inode(struct inode *inode)
 		|| NFS_STALE(inode);
 }
 
+int nfs_revalidate_mapping_rcu(struct inode *inode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+	unsigned long *bitlock = &nfsi->flags;
+	int ret = 0;
+
+	if (IS_SWAPFILE(inode))
+		goto out;
+	if (nfs_mapping_need_revalidate_inode(inode)) {
+		ret = -ECHILD;
+		goto out;
+	}
+	spin_lock(&inode->i_lock);
+	if (test_bit(NFS_INO_INVALIDATING, bitlock) ||
+	    (nfsi->cache_validity & NFS_INO_INVALID_DATA))
+		ret = -ECHILD;
+	spin_unlock(&inode->i_lock);
+out:
+	return ret;
+}
+
 /**
  * __nfs_revalidate_mapping - Revalidate the pagecache
  * @inode - pointer to host inode
diff --git a/fs/nfs/symlink.c b/fs/nfs/symlink.c
index 8ade8a8..95c69af 100644
--- a/fs/nfs/symlink.c
+++ b/fs/nfs/symlink.c
@@ -48,16 +48,26 @@ static const char *nfs_get_link(struct dentry *dentry,
 	struct page *page;
 	void *err;
 
-	if (!dentry)
-		return ERR_PTR(-ECHILD);
-
-	err = ERR_PTR(nfs_revalidate_mapping(inode, inode->i_mapping));
-	if (err)
-		return err;
-	page = read_cache_page(&inode->i_data, 0,
-				(filler_t *)nfs_symlink_filler, inode);
-	if (IS_ERR(page))
-		return ERR_CAST(page);
+	if (!dentry) {
+		err = ERR_PTR(nfs_revalidate_mapping_rcu(inode));
+		if (err)
+			return err;
+		page = find_get_page(inode->i_mapping, 0);
+		if (!page)
+			return ERR_PTR(-ECHILD);
+		if (!PageUptodate(page)) {
+			put_page(page);
+			return ERR_PTR(-ECHILD);
+		}
+	} else {
+		err = ERR_PTR(nfs_revalidate_mapping(inode, inode->i_mapping));
+		if (err)
+			return err;
+		page = read_cache_page(&inode->i_data, 0,
+					(filler_t *)nfs_symlink_filler, inode);
+		if (IS_ERR(page))
+			return ERR_CAST(page);
+	}
 	*cookie = page;
 	return page_address(page);
 }
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index c0e9614..37a3d29 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -359,6 +359,7 @@ extern int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode);
 extern int nfs_revalidate_inode_rcu(struct nfs_server *server, struct inode *inode);
 extern int __nfs_revalidate_inode(struct nfs_server *, struct inode *);
 extern int nfs_revalidate_mapping(struct inode *inode, struct address_space *mapping);
+extern int nfs_revalidate_mapping_rcu(struct inode *inode);
 extern int nfs_revalidate_mapping_protected(struct inode *inode, struct address_space *mapping);
 extern int nfs_setattr(struct dentry *, struct iattr *);
 extern void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr, struct nfs_fattr *);
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287121 — [PATCH v2 06/11] don't put symlink bodies in pagecache into highmem

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 06/11] don't put symlink bodies in pagecache into highmem
Message-ID	<qDFDQ-485-9@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

kmap() in page_follow_link_light() needed to go - allowing to hold
an arbitrary number of kmaps for long is a great way to deadlocking
the system.

new helper (inode_nohighmem(inode)) needs to be used for pagecache
symlinks inodes; done for all in-tree cases.  page_follow_link_light()
instrumented to yell about anything missed.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 Documentation/filesystems/porting |  5 +++++
 fs/affs/inode.c                   |  1 +
 fs/affs/namei.c                   |  1 +
 fs/affs/symlink.c                 |  4 +---
 fs/afs/inode.c                    |  1 +
 fs/befs/linuxvfs.c                |  5 ++---
 fs/btrfs/inode.c                  |  2 ++
 fs/coda/cnode.c                   |  2 ++
 fs/coda/symlink.c                 |  4 +---
 fs/cramfs/inode.c                 |  1 +
 fs/efs/inode.c                    |  1 +
 fs/efs/symlink.c                  |  4 +---
 fs/exofs/inode.c                  |  1 +
 fs/exofs/namei.c                  |  1 +
 fs/ext2/inode.c                   |  1 +
 fs/ext2/namei.c                   |  1 +
 fs/ext4/inode.c                   |  1 +
 fs/ext4/namei.c                   |  1 +
 fs/ext4/symlink.c                 | 10 +++-------
 fs/f2fs/inode.c                   |  1 +
 fs/f2fs/namei.c                   |  5 ++---
 fs/freevxfs/vxfs_inode.c          |  1 +
 fs/hfsplus/inode.c                |  2 ++
 fs/hpfs/inode.c                   |  1 +
 fs/hpfs/namei.c                   |  5 ++---
 fs/hugetlbfs/inode.c              |  1 +
 fs/inode.c                        |  6 ++++++
 fs/isofs/inode.c                  |  1 +
 fs/isofs/rock.c                   |  4 +---
 fs/jfs/inode.c                    |  1 +
 fs/jfs/namei.c                    |  1 +
 fs/logfs/dir.c                    |  1 +
 fs/logfs/inode.c                  |  1 +
 fs/minix/inode.c                  |  1 +
 fs/namei.c                        |  9 +++------
 fs/ncpfs/inode.c                  |  1 +
 fs/nfs/inode.c                    |  5 +++--
 fs/nfs/symlink.c                  |  2 +-
 fs/nilfs2/inode.c                 |  1 +
 fs/nilfs2/namei.c                 |  1 +
 fs/ocfs2/inode.c                  |  1 +
 fs/ocfs2/namei.c                  |  1 +
 fs/qnx4/inode.c                   |  1 +
 fs/qnx6/inode.c                   |  1 +
 fs/ramfs/inode.c                  |  1 +
 fs/reiserfs/inode.c               |  1 +
 fs/reiserfs/namei.c               |  1 +
 fs/romfs/super.c                  |  1 +
 fs/squashfs/inode.c               |  2 ++
 fs/sysv/inode.c                   |  1 +
 fs/udf/inode.c                    |  1 +
 fs/udf/namei.c                    |  1 +
 fs/udf/symlink.c                  |  4 +---
 fs/ufs/inode.c                    |  1 +
 fs/ufs/namei.c                    |  1 +
 include/linux/fs.h                |  1 +
 mm/shmem.c                        |  9 +++------
 57 files changed, 81 insertions(+), 46 deletions(-)

diff --git a/Documentation/filesystems/porting b/Documentation/filesystems/porting
index f24d1b8..3eb7c35 100644
--- a/Documentation/filesystems/porting
+++ b/Documentation/filesystems/porting
@@ -504,3 +504,8 @@ in your dentry operations instead.
 [mandatory]
 	__fd_install() & fd_install() can now sleep. Callers should not
 	hold a spinlock	or other resources that do not allow a schedule.
+--
+[mandatory]
+	any symlink that might use page_follow_link_light/page_put_link() must
+	have inode_nohighmem(inode) called before anything might start playing with
+	its pagecache.
diff --git a/fs/affs/inode.c b/fs/affs/inode.c
index 1734950..0fdb0f5 100644
--- a/fs/affs/inode.c
+++ b/fs/affs/inode.c
@@ -140,6 +140,7 @@ struct inode *affs_iget(struct super_block *sb, unsigned long ino)
 		break;
 	case ST_SOFTLINK:
 		inode->i_mode |= S_IFLNK;
+		inode_nohighmem(inode);
 		inode->i_op = &affs_symlink_inode_operations;
 		inode->i_data.a_ops = &affs_symlink_aops;
 		break;
diff --git a/fs/affs/namei.c b/fs/affs/namei.c
index 181e05b..00d3002 100644
--- a/fs/affs/namei.c
+++ b/fs/affs/namei.c
@@ -344,6 +344,7 @@ affs_symlink(struct inode *dir, struct dentry *dentry, const char *symname)
 		return -ENOSPC;
 
 	inode->i_op = &affs_symlink_inode_operations;
+	inode_nohighmem(inode);
 	inode->i_data.a_ops = &affs_symlink_aops;
 	inode->i_mode = S_IFLNK | 0777;
 	mode_to_prot(inode);
diff --git a/fs/affs/symlink.c b/fs/affs/symlink.c
index ea5b69a..e3f9dc3 100644
--- a/fs/affs/symlink.c
+++ b/fs/affs/symlink.c
@@ -14,7 +14,7 @@ static int affs_symlink_readpage(struct file *file, struct page *page)
 {
 	struct buffer_head *bh;
 	struct inode *inode = page->mapping->host;
-	char *link = kmap(page);
+	char *link = page_address(page);
 	struct slink_front *lf;
 	int			 i, j;
 	char			 c;
@@ -57,12 +57,10 @@ static int affs_symlink_readpage(struct file *file, struct page *page)
 	link[i] = '\0';
 	affs_brelse(bh);
 	SetPageUptodate(page);
-	kunmap(page);
 	unlock_page(page);
 	return 0;
 fail:
 	SetPageError(page);
-	kunmap(page);
 	unlock_page(page);
 	return -EIO;
 }
diff --git a/fs/afs/inode.c b/fs/afs/inode.c
index e06f5a2..86cc726 100644
--- a/fs/afs/inode.c
+++ b/fs/afs/inode.c
@@ -56,6 +56,7 @@ static int afs_inode_map_status(struct afs_vnode *vnode, struct key *key)
 	case AFS_FTYPE_SYMLINK:
 		inode->i_mode	= S_IFLNK | vnode->status.mode;
 		inode->i_op	= &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		break;
 	default:
 		printk("kAFS: AFS vnode with undefined type\n");
diff --git a/fs/befs/linuxvfs.c b/fs/befs/linuxvfs.c
index 1c8b0dc..25250fa 100644
--- a/fs/befs/linuxvfs.c
+++ b/fs/befs/linuxvfs.c
@@ -397,6 +397,7 @@ static struct inode *befs_iget(struct super_block *sb, unsigned long ino)
 	} else if (S_ISLNK(inode->i_mode)) {
 		if (befs_ino->i_flags & BEFS_LONG_SYMLINK) {
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 			inode->i_mapping->a_ops = &befs_symlink_aops;
 		} else {
 			inode->i_link = befs_ino->i_data.symlink;
@@ -469,7 +470,7 @@ static int befs_symlink_readpage(struct file *unused, struct page *page)
 	struct befs_inode_info *befs_ino = BEFS_I(inode);
 	befs_data_stream *data = &befs_ino->i_data.ds;
 	befs_off_t len = data->size;
-	char *link = kmap(page);
+	char *link = page_address(page);
 
 	if (len == 0 || len > PAGE_SIZE) {
 		befs_error(sb, "Long symlink with illegal length");
@@ -483,12 +484,10 @@ static int befs_symlink_readpage(struct file *unused, struct page *page)
 	}
 	link[len - 1] = '\0';
 	SetPageUptodate(page);
-	kunmap(page);
 	unlock_page(page);
 	return 0;
 fail:
 	SetPageError(page);
-	kunmap(page);
 	unlock_page(page);
 	return -EIO;
 }
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index a70c579..70f98bf 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -3774,6 +3774,7 @@ cache_acl:
 		break;
 	case S_IFLNK:
 		inode->i_op = &btrfs_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &btrfs_symlink_aops;
 		break;
 	default:
@@ -9705,6 +9706,7 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 	btrfs_free_path(path);
 
 	inode->i_op = &btrfs_symlink_inode_operations;
+	inode_nohighmem(inode);
 	inode->i_mapping->a_ops = &btrfs_symlink_aops;
 	inode_set_bytes(inode, name_len);
 	btrfs_i_size_write(inode, name_len);
diff --git a/fs/coda/cnode.c b/fs/coda/cnode.c
index 7740b1c..dd6a79e 100644
--- a/fs/coda/cnode.c
+++ b/fs/coda/cnode.c
@@ -8,6 +8,7 @@
 
 #include <linux/coda.h>
 #include <linux/coda_psdev.h>
+#include <linux/pagemap.h>
 #include "coda_linux.h"
 
 static inline int coda_fideq(struct CodaFid *fid1, struct CodaFid *fid2)
@@ -35,6 +36,7 @@ static void coda_fill_inode(struct inode *inode, struct coda_vattr *attr)
                 inode->i_fop = &coda_dir_operations;
         } else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &coda_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_data.a_ops = &coda_symlink_aops;
 		inode->i_mapping = &inode->i_data;
 	} else
diff --git a/fs/coda/symlink.c b/fs/coda/symlink.c
index ab94ef6..03736e2 100644
--- a/fs/coda/symlink.c
+++ b/fs/coda/symlink.c
@@ -26,7 +26,7 @@ static int coda_symlink_filler(struct file *file, struct page *page)
 	int error;
 	struct coda_inode_info *cii;
 	unsigned int len = PAGE_SIZE;
-	char *p = kmap(page);
+	char *p = page_address(page);
 
 	cii = ITOC(inode);
 
@@ -34,13 +34,11 @@ static int coda_symlink_filler(struct file *file, struct page *page)
 	if (error)
 		goto fail;
 	SetPageUptodate(page);
-	kunmap(page);
 	unlock_page(page);
 	return 0;
 
 fail:
 	SetPageError(page);
-	kunmap(page);
 	unlock_page(page);
 	return error;
 }
diff --git a/fs/cramfs/inode.c b/fs/cramfs/inode.c
index 355c522..b862bc2 100644
--- a/fs/cramfs/inode.c
+++ b/fs/cramfs/inode.c
@@ -100,6 +100,7 @@ static struct inode *get_cramfs_inode(struct super_block *sb,
 		break;
 	case S_IFLNK:
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_data.a_ops = &cramfs_aops;
 		break;
 	default:
diff --git a/fs/efs/inode.c b/fs/efs/inode.c
index 079d203..cdf0872 100644
--- a/fs/efs/inode.c
+++ b/fs/efs/inode.c
@@ -151,6 +151,7 @@ struct inode *efs_iget(struct super_block *super, unsigned long ino)
 			break;
 		case S_IFLNK:
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 			inode->i_data.a_ops = &efs_symlink_aops;
 			break;
 		case S_IFCHR:
diff --git a/fs/efs/symlink.c b/fs/efs/symlink.c
index 75117d0..4870cc8 100644
--- a/fs/efs/symlink.c
+++ b/fs/efs/symlink.c
@@ -13,7 +13,7 @@
 
 static int efs_symlink_readpage(struct file *file, struct page *page)
 {
-	char *link = kmap(page);
+	char *link = page_address(page);
 	struct buffer_head * bh;
 	struct inode * inode = page->mapping->host;
 	efs_block_t size = inode->i_size;
@@ -39,12 +39,10 @@ static int efs_symlink_readpage(struct file *file, struct page *page)
 	}
 	link[size] = '\0';
 	SetPageUptodate(page);
-	kunmap(page);
 	unlock_page(page);
 	return 0;
 fail:
 	SetPageError(page);
-	kunmap(page);
 	unlock_page(page);
 	return err;
 }
diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c
index 73c64da..d8e9c181 100644
--- a/fs/exofs/inode.c
+++ b/fs/exofs/inode.c
@@ -1227,6 +1227,7 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino)
 			inode->i_link = (char *)oi->i_data;
 		} else {
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 			inode->i_mapping->a_ops = &exofs_aops;
 		}
 	} else {
diff --git a/fs/exofs/namei.c b/fs/exofs/namei.c
index 994e078..c20d77d 100644
--- a/fs/exofs/namei.c
+++ b/fs/exofs/namei.c
@@ -111,6 +111,7 @@ static int exofs_symlink(struct inode *dir, struct dentry *dentry,
 	if (l > sizeof(oi->i_data)) {
 		/* slow symlink */
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &exofs_aops;
 		memset(oi->i_data, 0, sizeof(oi->i_data));
 
diff --git a/fs/ext2/inode.c b/fs/ext2/inode.c
index 0aa9bf6..338eefd 100644
--- a/fs/ext2/inode.c
+++ b/fs/ext2/inode.c
@@ -1420,6 +1420,7 @@ struct inode *ext2_iget (struct super_block *sb, unsigned long ino)
 				sizeof(ei->i_data) - 1);
 		} else {
 			inode->i_op = &ext2_symlink_inode_operations;
+			inode_nohighmem(inode);
 			if (test_opt(inode->i_sb, NOBH))
 				inode->i_mapping->a_ops = &ext2_nobh_aops;
 			else
diff --git a/fs/ext2/namei.c b/fs/ext2/namei.c
index 3267a80d..7a2be8f 100644
--- a/fs/ext2/namei.c
+++ b/fs/ext2/namei.c
@@ -183,6 +183,7 @@ static int ext2_symlink (struct inode * dir, struct dentry * dentry,
 	if (l > sizeof (EXT2_I(inode)->i_data)) {
 		/* slow symlink */
 		inode->i_op = &ext2_symlink_inode_operations;
+		inode_nohighmem(inode);
 		if (test_opt(inode->i_sb, NOBH))
 			inode->i_mapping->a_ops = &ext2_nobh_aops;
 		else
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index ea433a7..b3bd912 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -4283,6 +4283,7 @@ struct inode *ext4_iget(struct super_block *sb, unsigned long ino)
 			inode->i_op = &ext4_symlink_inode_operations;
 			ext4_set_aops(inode);
 		}
+		inode_nohighmem(inode);
 	} else if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode) ||
 	      S_ISFIFO(inode->i_mode) || S_ISSOCK(inode->i_mode)) {
 		inode->i_op = &ext4_special_inode_operations;
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
index a969ab3..f27e0c2 100644
--- a/fs/ext4/namei.c
+++ b/fs/ext4/namei.c
@@ -3132,6 +3132,7 @@ static int ext4_symlink(struct inode *dir,
 	if ((disk_link.len > EXT4_N_BLOCKS * 4)) {
 		if (!encryption_required)
 			inode->i_op = &ext4_symlink_inode_operations;
+		inode_nohighmem(inode);
 		ext4_set_aops(inode);
 		/*
 		 * We cannot call page_symlink() with transaction started
diff --git a/fs/ext4/symlink.c b/fs/ext4/symlink.c
index abe2401..0e6dc44 100644
--- a/fs/ext4/symlink.c
+++ b/fs/ext4/symlink.c
@@ -45,7 +45,7 @@ static const char *ext4_encrypted_follow_link(struct dentry *dentry, void **cook
 		cpage = read_mapping_page(inode->i_mapping, 0, NULL);
 		if (IS_ERR(cpage))
 			return ERR_CAST(cpage);
-		caddr = kmap(cpage);
+		caddr = page_address(cpage);
 		caddr[size] = 0;
 	}
 
@@ -75,16 +75,12 @@ static const char *ext4_encrypted_follow_link(struct dentry *dentry, void **cook
 	/* Null-terminate the name */
 	if (res <= plen)
 		paddr[res] = '\0';
-	if (cpage) {
-		kunmap(cpage);
+	if (cpage)
 		page_cache_release(cpage);
-	}
 	return *cookie = paddr;
 errout:
-	if (cpage) {
-		kunmap(cpage);
+	if (cpage)
 		page_cache_release(cpage);
-	}
 	kfree(paddr);
 	return ERR_PTR(res);
 }
diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c
index 97e20de..5528801 100644
--- a/fs/f2fs/inode.c
+++ b/fs/f2fs/inode.c
@@ -202,6 +202,7 @@ make_now:
 			inode->i_op = &f2fs_encrypted_symlink_inode_operations;
 		else
 			inode->i_op = &f2fs_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &f2fs_dblock_aops;
 	} else if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode) ||
 			S_ISFIFO(inode->i_mode) || S_ISSOCK(inode->i_mode)) {
diff --git a/fs/f2fs/namei.c b/fs/f2fs/namei.c
index 2c32110..484df68 100644
--- a/fs/f2fs/namei.c
+++ b/fs/f2fs/namei.c
@@ -351,6 +351,7 @@ static int f2fs_symlink(struct inode *dir, struct dentry *dentry,
 		inode->i_op = &f2fs_encrypted_symlink_inode_operations;
 	else
 		inode->i_op = &f2fs_symlink_inode_operations;
+	inode_nohighmem(inode);
 	inode->i_mapping->a_ops = &f2fs_dblock_aops;
 
 	f2fs_lock_op(sbi);
@@ -942,7 +943,7 @@ static const char *f2fs_encrypted_follow_link(struct dentry *dentry, void **cook
 	cpage = read_mapping_page(inode->i_mapping, 0, NULL);
 	if (IS_ERR(cpage))
 		return ERR_CAST(cpage);
-	caddr = kmap(cpage);
+	caddr = page_address(cpage);
 	caddr[size] = 0;
 
 	/* Symlink is encrypted */
@@ -982,13 +983,11 @@ static const char *f2fs_encrypted_follow_link(struct dentry *dentry, void **cook
 	/* Null-terminate the name */
 	paddr[res] = '\0';
 
-	kunmap(cpage);
 	page_cache_release(cpage);
 	return *cookie = paddr;
 errout:
 	kfree(cstr.name);
 	f2fs_fname_crypto_free_buffer(&pstr);
-	kunmap(cpage);
 	page_cache_release(cpage);
 	return ERR_PTR(res);
 }
diff --git a/fs/freevxfs/vxfs_inode.c b/fs/freevxfs/vxfs_inode.c
index ef73ed6..3e2ccad 100644
--- a/fs/freevxfs/vxfs_inode.c
+++ b/fs/freevxfs/vxfs_inode.c
@@ -326,6 +326,7 @@ vxfs_iget(struct super_block *sbp, ino_t ino)
 	} else if (S_ISLNK(ip->i_mode)) {
 		if (!VXFS_ISIMMED(vip)) {
 			ip->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(ip);
 			ip->i_mapping->a_ops = &vxfs_aops;
 		} else {
 			ip->i_op = &simple_symlink_inode_operations;
diff --git a/fs/hfsplus/inode.c b/fs/hfsplus/inode.c
index 6dd107d..19b33f8 100644
--- a/fs/hfsplus/inode.c
+++ b/fs/hfsplus/inode.c
@@ -403,6 +403,7 @@ struct inode *hfsplus_new_inode(struct super_block *sb, umode_t mode)
 	} else if (S_ISLNK(inode->i_mode)) {
 		sbi->file_count++;
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &hfsplus_aops;
 		hip->clump_blocks = 1;
 	} else
@@ -526,6 +527,7 @@ int hfsplus_cat_read_inode(struct inode *inode, struct hfs_find_data *fd)
 			inode->i_mapping->a_ops = &hfsplus_aops;
 		} else if (S_ISLNK(inode->i_mode)) {
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 			inode->i_mapping->a_ops = &hfsplus_aops;
 		} else {
 			init_special_inode(inode, inode->i_mode,
diff --git a/fs/hpfs/inode.c b/fs/hpfs/inode.c
index 933c737..1f3c6d7 100644
--- a/fs/hpfs/inode.c
+++ b/fs/hpfs/inode.c
@@ -77,6 +77,7 @@ void hpfs_read_inode(struct inode *i)
 			kfree(ea);
 			i->i_mode = S_IFLNK | 0777;
 			i->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(i);
 			i->i_data.a_ops = &hpfs_symlink_aops;
 			set_nlink(i, 1);
 			i->i_size = ea_size;
diff --git a/fs/hpfs/namei.c b/fs/hpfs/namei.c
index ae4d5a1..506765a 100644
--- a/fs/hpfs/namei.c
+++ b/fs/hpfs/namei.c
@@ -332,6 +332,7 @@ static int hpfs_symlink(struct inode *dir, struct dentry *dentry, const char *sy
 	result->i_blocks = 1;
 	set_nlink(result, 1);
 	result->i_size = strlen(symlink);
+	inode_nohighmem(result);
 	result->i_op = &page_symlink_inode_operations;
 	result->i_data.a_ops = &hpfs_symlink_aops;
 
@@ -500,7 +501,7 @@ out:
 
 static int hpfs_symlink_readpage(struct file *file, struct page *page)
 {
-	char *link = kmap(page);
+	char *link = page_address(page);
 	struct inode *i = page->mapping->host;
 	struct fnode *fnode;
 	struct buffer_head *bh;
@@ -516,14 +517,12 @@ static int hpfs_symlink_readpage(struct file *file, struct page *page)
 		goto fail;
 	hpfs_unlock(i->i_sb);
 	SetPageUptodate(page);
-	kunmap(page);
 	unlock_page(page);
 	return 0;
 
 fail:
 	hpfs_unlock(i->i_sb);
 	SetPageError(page);
-	kunmap(page);
 	unlock_page(page);
 	return err;
 }
diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
index de4bdfa..d8f51ee 100644
--- a/fs/hugetlbfs/inode.c
+++ b/fs/hugetlbfs/inode.c
@@ -760,6 +760,7 @@ static struct inode *hugetlbfs_get_inode(struct super_block *sb,
 			break;
 		case S_IFLNK:
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 			break;
 		}
 		lockdep_annotate_inode_mutex_key(inode);
diff --git a/fs/inode.c b/fs/inode.c
index 1be5f90..5bb85a0 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -2028,3 +2028,9 @@ void inode_set_flags(struct inode *inode, unsigned int flags,
 				  new_flags) != old_flags));
 }
 EXPORT_SYMBOL(inode_set_flags);
+
+void inode_nohighmem(struct inode *inode)
+{
+	mapping_set_gfp_mask(inode->i_mapping, GFP_USER);
+}
+EXPORT_SYMBOL(inode_nohighmem);
diff --git a/fs/isofs/inode.c b/fs/isofs/inode.c
index d67a16f..61abdc4 100644
--- a/fs/isofs/inode.c
+++ b/fs/isofs/inode.c
@@ -1417,6 +1417,7 @@ static int isofs_read_inode(struct inode *inode, int relocated)
 		inode->i_fop = &isofs_dir_operations;
 	} else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_data.a_ops = &isofs_symlink_aops;
 	} else
 		/* XXX - parse_rock_ridge_inode() had already set i_rdev. */
diff --git a/fs/isofs/rock.c b/fs/isofs/rock.c
index 735d752..5384ceb 100644
--- a/fs/isofs/rock.c
+++ b/fs/isofs/rock.c
@@ -687,7 +687,7 @@ static int rock_ridge_symlink_readpage(struct file *file, struct page *page)
 	struct inode *inode = page->mapping->host;
 	struct iso_inode_info *ei = ISOFS_I(inode);
 	struct isofs_sb_info *sbi = ISOFS_SB(inode->i_sb);
-	char *link = kmap(page);
+	char *link = page_address(page);
 	unsigned long bufsize = ISOFS_BUFFER_SIZE(inode);
 	struct buffer_head *bh;
 	char *rpnt = link;
@@ -774,7 +774,6 @@ repeat:
 	brelse(bh);
 	*rpnt = '\0';
 	SetPageUptodate(page);
-	kunmap(page);
 	unlock_page(page);
 	return 0;
 
@@ -791,7 +790,6 @@ fail:
 	brelse(bh);
 error:
 	SetPageError(page);
-	kunmap(page);
 	unlock_page(page);
 	return -EIO;
 }
diff --git a/fs/jfs/inode.c b/fs/jfs/inode.c
index 41aa3ca..9d9bae6 100644
--- a/fs/jfs/inode.c
+++ b/fs/jfs/inode.c
@@ -60,6 +60,7 @@ struct inode *jfs_iget(struct super_block *sb, unsigned long ino)
 	} else if (S_ISLNK(inode->i_mode)) {
 		if (inode->i_size >= IDATASIZE) {
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 			inode->i_mapping->a_ops = &jfs_aops;
 		} else {
 			inode->i_op = &jfs_fast_symlink_inode_operations;
diff --git a/fs/jfs/namei.c b/fs/jfs/namei.c
index 9d7551f..701f893 100644
--- a/fs/jfs/namei.c
+++ b/fs/jfs/namei.c
@@ -983,6 +983,7 @@ static int jfs_symlink(struct inode *dip, struct dentry *dentry,
 		jfs_info("jfs_symlink: allocate extent ip:0x%p", ip);
 
 		ip->i_op = &jfs_symlink_inode_operations;
+		inode_nohighmem(ip);
 		ip->i_mapping->a_ops = &jfs_aops;
 
 		/*
diff --git a/fs/logfs/dir.c b/fs/logfs/dir.c
index 99944a4..542468e 100644
--- a/fs/logfs/dir.c
+++ b/fs/logfs/dir.c
@@ -529,6 +529,7 @@ static int logfs_symlink(struct inode *dir, struct dentry *dentry,
 		return PTR_ERR(inode);
 
 	inode->i_op = &page_symlink_inode_operations;
+	inode_nohighmem(inode);
 	inode->i_mapping->a_ops = &logfs_reg_aops;
 
 	return __logfs_create(dir, dentry, inode, target, destlen);
diff --git a/fs/logfs/inode.c b/fs/logfs/inode.c
index 06baa92..0fce46d 100644
--- a/fs/logfs/inode.c
+++ b/fs/logfs/inode.c
@@ -65,6 +65,7 @@ static void logfs_inode_setops(struct inode *inode)
 		break;
 	case S_IFLNK:
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &logfs_reg_aops;
 		break;
 	case S_IFSOCK:	/* fall through */
diff --git a/fs/minix/inode.c b/fs/minix/inode.c
index 086cd0a..67a23bf 100644
--- a/fs/minix/inode.c
+++ b/fs/minix/inode.c
@@ -452,6 +452,7 @@ void minix_set_inode(struct inode *inode, dev_t rdev)
 		inode->i_mapping->a_ops = &minix_aops;
 	} else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &minix_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &minix_aops;
 	} else
 		init_special_inode(inode, inode->i_mode, rdev);
diff --git a/fs/namei.c b/fs/namei.c
index 4bae5cb..2808958 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -4527,7 +4527,8 @@ static const char *page_getlink(struct dentry * dentry, void **cookie)
 	if (IS_ERR(page))
 		return (char*)page;
 	*cookie = page;
-	kaddr = kmap(page);
+	BUG_ON(mapping_gfp_mask(mapping) & __GFP_HIGHMEM);
+	kaddr = page_address(page);
 	nd_terminate_link(kaddr, dentry->d_inode->i_size, PAGE_SIZE - 1);
 	return kaddr;
 }
@@ -4541,7 +4542,6 @@ EXPORT_SYMBOL(page_follow_link_light);
 void page_put_link(struct inode *unused, void *cookie)
 {
 	struct page *page = cookie;
-	kunmap(page);
 	page_cache_release(page);
 }
 EXPORT_SYMBOL(page_put_link);
@@ -4565,7 +4565,6 @@ int __page_symlink(struct inode *inode, const char *symname, int len, int nofs)
 	struct page *page;
 	void *fsdata;
 	int err;
-	char *kaddr;
 	unsigned int flags = AOP_FLAG_UNINTERRUPTIBLE;
 	if (nofs)
 		flags |= AOP_FLAG_NOFS;
@@ -4576,9 +4575,7 @@ retry:
 	if (err)
 		goto fail;
 
-	kaddr = kmap_atomic(page);
-	memcpy(kaddr, symname, len-1);
-	kunmap_atomic(kaddr);
+	memcpy(page_address(page), symname, len-1);
 
 	err = pagecache_write_end(NULL, mapping, 0, len-1, len-1,
 							page, fsdata);
diff --git a/fs/ncpfs/inode.c b/fs/ncpfs/inode.c
index 9605a2f..bb856f7 100644
--- a/fs/ncpfs/inode.c
+++ b/fs/ncpfs/inode.c
@@ -283,6 +283,7 @@ ncp_iget(struct super_block *sb, struct ncp_entry_info *info)
 #if defined(CONFIG_NCPFS_EXTRAS) || defined(CONFIG_NCPFS_NFS_NS)
 		} else if (S_ISLNK(inode->i_mode)) {
 			inode->i_op = &ncp_symlink_inode_operations;
+			inode_nohighmem(inode);
 			inode->i_data.a_ops = &ncp_symlink_aops;
 #endif
 		} else {
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 31b0a52..ae9aa0b 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -408,9 +408,10 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr, st
 				inode->i_fop = NULL;
 				inode->i_flags |= S_AUTOMOUNT;
 			}
-		} else if (S_ISLNK(inode->i_mode))
+		} else if (S_ISLNK(inode->i_mode)) {
 			inode->i_op = &nfs_symlink_inode_operations;
-		else
+			inode_nohighmem(inode);
+		} else
 			init_special_inode(inode, inode->i_mode, fattr->rdev);
 
 		memset(&inode->i_atime, 0, sizeof(inode->i_atime));
diff --git a/fs/nfs/symlink.c b/fs/nfs/symlink.c
index b6de433..abd93bf 100644
--- a/fs/nfs/symlink.c
+++ b/fs/nfs/symlink.c
@@ -56,7 +56,7 @@ static const char *nfs_follow_link(struct dentry *dentry, void **cookie)
 	if (IS_ERR(page))
 		return ERR_CAST(page);
 	*cookie = page;
-	return kmap(page);
+	return page_address(page);
 }
 
 /*
diff --git a/fs/nilfs2/inode.c b/fs/nilfs2/inode.c
index ac2f649..10b2252 100644
--- a/fs/nilfs2/inode.c
+++ b/fs/nilfs2/inode.c
@@ -510,6 +510,7 @@ static int __nilfs_read_inode(struct super_block *sb,
 		inode->i_mapping->a_ops = &nilfs_aops;
 	} else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &nilfs_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &nilfs_aops;
 	} else {
 		inode->i_op = &nilfs_special_inode_operations;
diff --git a/fs/nilfs2/namei.c b/fs/nilfs2/namei.c
index c9a1a49..90b3ba9 100644
--- a/fs/nilfs2/namei.c
+++ b/fs/nilfs2/namei.c
@@ -161,6 +161,7 @@ static int nilfs_symlink(struct inode *dir, struct dentry *dentry,
 
 	/* slow symlink */
 	inode->i_op = &nilfs_symlink_inode_operations;
+	inode_nohighmem(inode);
 	inode->i_mapping->a_ops = &nilfs_aops;
 	err = page_symlink(inode, symname, l);
 	if (err)
diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c
index 8f87e05..97a563b 100644
--- a/fs/ocfs2/inode.c
+++ b/fs/ocfs2/inode.c
@@ -361,6 +361,7 @@ void ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe,
 		    break;
 	    case S_IFLNK:
 		    inode->i_op = &ocfs2_symlink_inode_operations;
+		    inode_nohighmem(inode);
 		    i_size_write(inode, le64_to_cpu(fe->i_size));
 		    break;
 	    default:
diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c
index a03f6f4..2efe8af 100644
--- a/fs/ocfs2/namei.c
+++ b/fs/ocfs2/namei.c
@@ -1960,6 +1960,7 @@ static int ocfs2_symlink(struct inode *dir,
 	inode->i_rdev = 0;
 	newsize = l - 1;
 	inode->i_op = &ocfs2_symlink_inode_operations;
+	inode_nohighmem(inode);
 	if (l > ocfs2_fast_symlink_chars(sb)) {
 		u32 offset = 0;
 
diff --git a/fs/qnx4/inode.c b/fs/qnx4/inode.c
index c4bcb77..f37b3de 100644
--- a/fs/qnx4/inode.c
+++ b/fs/qnx4/inode.c
@@ -316,6 +316,7 @@ struct inode *qnx4_iget(struct super_block *sb, unsigned long ino)
 		inode->i_fop = &qnx4_dir_operations;
 	} else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &qnx4_aops;
 		qnx4_i(inode)->mmu_private = inode->i_size;
 	} else {
diff --git a/fs/qnx6/inode.c b/fs/qnx6/inode.c
index 32d2e1a..9728b54 100644
--- a/fs/qnx6/inode.c
+++ b/fs/qnx6/inode.c
@@ -582,6 +582,7 @@ struct inode *qnx6_iget(struct super_block *sb, unsigned ino)
 		inode->i_mapping->a_ops = &qnx6_aops;
 	} else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &qnx6_aops;
 	} else
 		init_special_inode(inode, inode->i_mode, 0);
diff --git a/fs/ramfs/inode.c b/fs/ramfs/inode.c
index 889d558..38981b0 100644
--- a/fs/ramfs/inode.c
+++ b/fs/ramfs/inode.c
@@ -79,6 +79,7 @@ struct inode *ramfs_get_inode(struct super_block *sb,
 			break;
 		case S_IFLNK:
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 			break;
 		}
 	}
diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c
index 3d8e7e6..ae9e5b3 100644
--- a/fs/reiserfs/inode.c
+++ b/fs/reiserfs/inode.c
@@ -1361,6 +1361,7 @@ static void init_inode(struct inode *inode, struct treepath *path)
 		inode->i_fop = &reiserfs_dir_operations;
 	} else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &reiserfs_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &reiserfs_address_space_operations;
 	} else {
 		inode->i_blocks = 0;
diff --git a/fs/reiserfs/namei.c b/fs/reiserfs/namei.c
index 47f9698..4fc2326 100644
--- a/fs/reiserfs/namei.c
+++ b/fs/reiserfs/namei.c
@@ -1170,6 +1170,7 @@ static int reiserfs_symlink(struct inode *parent_dir,
 	reiserfs_update_inode_transaction(parent_dir);
 
 	inode->i_op = &reiserfs_symlink_inode_operations;
+	inode_nohighmem(inode);
 	inode->i_mapping->a_ops = &reiserfs_address_space_operations;
 
 	retval = reiserfs_add_entry(&th, parent_dir, dentry->d_name.name,
diff --git a/fs/romfs/super.c b/fs/romfs/super.c
index 268733c..bb894e7 100644
--- a/fs/romfs/super.c
+++ b/fs/romfs/super.c
@@ -360,6 +360,7 @@ static struct inode *romfs_iget(struct super_block *sb, unsigned long pos)
 		break;
 	case ROMFH_SYM:
 		i->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(i);
 		i->i_data.a_ops = &romfs_aops;
 		mode |= S_IRWXUGO;
 		break;
diff --git a/fs/squashfs/inode.c b/fs/squashfs/inode.c
index a1ce5ce..0927b1e 100644
--- a/fs/squashfs/inode.c
+++ b/fs/squashfs/inode.c
@@ -41,6 +41,7 @@
 #include <linux/fs.h>
 #include <linux/vfs.h>
 #include <linux/xattr.h>
+#include <linux/pagemap.h>
 
 #include "squashfs_fs.h"
 #include "squashfs_fs_sb.h"
@@ -291,6 +292,7 @@ int squashfs_read_inode(struct inode *inode, long long ino)
 		set_nlink(inode, le32_to_cpu(sqsh_ino->nlink));
 		inode->i_size = le32_to_cpu(sqsh_ino->symlink_size);
 		inode->i_op = &squashfs_symlink_inode_ops;
+		inode_nohighmem(inode);
 		inode->i_data.a_ops = &squashfs_symlink_aops;
 		inode->i_mode |= S_IFLNK;
 		squashfs_i(inode)->start = block;
diff --git a/fs/sysv/inode.c b/fs/sysv/inode.c
index 02fa1dc..ef8bcdb 100644
--- a/fs/sysv/inode.c
+++ b/fs/sysv/inode.c
@@ -163,6 +163,7 @@ void sysv_set_inode(struct inode *inode, dev_t rdev)
 		inode->i_mapping->a_ops = &sysv_aops;
 	} else if (S_ISLNK(inode->i_mode)) {
 		inode->i_op = &sysv_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &sysv_aops;
 	} else
 		init_special_inode(inode, inode->i_mode, rdev);
diff --git a/fs/udf/inode.c b/fs/udf/inode.c
index 8675c2b..0557463 100644
--- a/fs/udf/inode.c
+++ b/fs/udf/inode.c
@@ -1541,6 +1541,7 @@ reread:
 	case ICBTAG_FILE_TYPE_SYMLINK:
 		inode->i_data.a_ops = &udf_symlink_aops;
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mode = S_IFLNK | S_IRWXUGO;
 		break;
 	case ICBTAG_FILE_TYPE_MAIN:
diff --git a/fs/udf/namei.c b/fs/udf/namei.c
index d0e6de1..42eafb9 100644
--- a/fs/udf/namei.c
+++ b/fs/udf/namei.c
@@ -922,6 +922,7 @@ static int udf_symlink(struct inode *dir, struct dentry *dentry,
 
 	inode->i_data.a_ops = &udf_symlink_aops;
 	inode->i_op = &page_symlink_inode_operations;
+	inode_nohighmem(inode);
 
 	if (iinfo->i_alloc_type != ICBTAG_FLAG_AD_IN_ICB) {
 		struct kernel_lb_addr eloc;
diff --git a/fs/udf/symlink.c b/fs/udf/symlink.c
index 862535b..8d61977 100644
--- a/fs/udf/symlink.c
+++ b/fs/udf/symlink.c
@@ -107,7 +107,7 @@ static int udf_symlink_filler(struct file *file, struct page *page)
 	struct buffer_head *bh = NULL;
 	unsigned char *symlink;
 	int err;
-	unsigned char *p = kmap(page);
+	unsigned char *p = page_address(page);
 	struct udf_inode_info *iinfo;
 	uint32_t pos;
 
@@ -141,7 +141,6 @@ static int udf_symlink_filler(struct file *file, struct page *page)
 
 	up_read(&iinfo->i_data_sem);
 	SetPageUptodate(page);
-	kunmap(page);
 	unlock_page(page);
 	return 0;
 
@@ -149,7 +148,6 @@ out_unlock_inode:
 	up_read(&iinfo->i_data_sem);
 	SetPageError(page);
 out_unmap:
-	kunmap(page);
 	unlock_page(page);
 	return err;
 }
diff --git a/fs/ufs/inode.c b/fs/ufs/inode.c
index 737160a..d897e16 100644
--- a/fs/ufs/inode.c
+++ b/fs/ufs/inode.c
@@ -533,6 +533,7 @@ static void ufs_set_inode_ops(struct inode *inode)
 		} else {
 			inode->i_mapping->a_ops = &ufs_aops;
 			inode->i_op = &page_symlink_inode_operations;
+			inode_nohighmem(inode);
 		}
 	} else
 		init_special_inode(inode, inode->i_mode,
diff --git a/fs/ufs/namei.c b/fs/ufs/namei.c
index 24b0cbd..acf4a3b 100644
--- a/fs/ufs/namei.c
+++ b/fs/ufs/namei.c
@@ -124,6 +124,7 @@ static int ufs_symlink (struct inode * dir, struct dentry * dentry,
 	if (l > UFS_SB(sb)->s_uspi->s_maxsymlinklen) {
 		/* slow symlink */
 		inode->i_op = &page_symlink_inode_operations;
+		inode_nohighmem(inode);
 		inode->i_mapping->a_ops = &ufs_aops;
 		err = page_symlink(inode, symname, l);
 		if (err)
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 3aa5142..dfeda44 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -3025,5 +3025,6 @@ static inline bool dir_relax(struct inode *inode)
 }
 
 extern bool path_noexec(const struct path *path);
+extern void inode_nohighmem(struct inode *inode);
 
 #endif /* _LINUX_FS_H */
diff --git a/mm/shmem.c b/mm/shmem.c
index 9187eee..64bf5ac 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -2444,7 +2444,6 @@ static int shmem_symlink(struct inode *dir, struct dentry *dentry, const char *s
 	int len;
 	struct inode *inode;
 	struct page *page;
-	char *kaddr;
 	struct shmem_inode_info *info;
 
 	len = strlen(symname) + 1;
@@ -2483,9 +2482,8 @@ static int shmem_symlink(struct inode *dir, struct dentry *dentry, const char *s
 		}
 		inode->i_mapping->a_ops = &shmem_aops;
 		inode->i_op = &shmem_symlink_inode_operations;
-		kaddr = kmap_atomic(page);
-		memcpy(kaddr, symname, len);
-		kunmap_atomic(kaddr);
+		inode_nohighmem(inode);
+		memcpy(page_address(page), symname, len);
 		SetPageUptodate(page);
 		set_page_dirty(page);
 		unlock_page(page);
@@ -2506,13 +2504,12 @@ static const char *shmem_follow_link(struct dentry *dentry, void **cookie)
 		return ERR_PTR(error);
 	unlock_page(page);
 	*cookie = page;
-	return kmap(page);
+	return page_address(page);
 }
 
 static void shmem_put_link(struct inode *unused, void *cookie)
 {
 	struct page *page = cookie;
-	kunmap(page);
 	mark_page_accessed(page);
 	page_cache_release(page);
 }
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287122 — [PATCH v2 01/11] switch befs long symlinks to page_symlink_operations

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 01/11] switch befs long symlinks to page_symlink_operations
Message-ID	<qDFDQ-485-19@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

just give them the right ->readpage()...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/befs/linuxvfs.c | 41 ++++++++++++++++++++++-------------------
 1 file changed, 22 insertions(+), 19 deletions(-)

diff --git a/fs/befs/linuxvfs.c b/fs/befs/linuxvfs.c
index 46aedac..1c8b0dc 100644
--- a/fs/befs/linuxvfs.c
+++ b/fs/befs/linuxvfs.c
@@ -42,7 +42,7 @@ static struct inode *befs_iget(struct super_block *, unsigned long);
 static struct inode *befs_alloc_inode(struct super_block *sb);
 static void befs_destroy_inode(struct inode *inode);
 static void befs_destroy_inodecache(void);
-static const char *befs_follow_link(struct dentry *, void **);
+static int befs_symlink_readpage(struct file *, struct page *);
 static int befs_utf2nls(struct super_block *sb, const char *in, int in_len,
 			char **out, int *out_len);
 static int befs_nls2utf(struct super_block *sb, const char *in, int in_len,
@@ -79,10 +79,8 @@ static const struct address_space_operations befs_aops = {
 	.bmap		= befs_bmap,
 };
 
-static const struct inode_operations befs_symlink_inode_operations = {
-	.readlink	= generic_readlink,
-	.follow_link	= befs_follow_link,
-	.put_link	= kfree_put_link,
+static const struct address_space_operations befs_symlink_aops = {
+	.readpage	= befs_symlink_readpage,
 };
 
 /* 
@@ -398,7 +396,8 @@ static struct inode *befs_iget(struct super_block *sb, unsigned long ino)
 		inode->i_fop = &befs_dir_operations;
 	} else if (S_ISLNK(inode->i_mode)) {
 		if (befs_ino->i_flags & BEFS_LONG_SYMLINK) {
-			inode->i_op = &befs_symlink_inode_operations;
+			inode->i_op = &page_symlink_inode_operations;
+			inode->i_mapping->a_ops = &befs_symlink_aops;
 		} else {
 			inode->i_link = befs_ino->i_data.symlink;
 			inode->i_op = &simple_symlink_inode_operations;
@@ -463,31 +462,35 @@ befs_destroy_inodecache(void)
  * The data stream become link name. Unless the LONG_SYMLINK
  * flag is set.
  */
-static const char *
-befs_follow_link(struct dentry *dentry, void **cookie)
+static int befs_symlink_readpage(struct file *unused, struct page *page)
 {
-	struct super_block *sb = dentry->d_sb;
-	struct befs_inode_info *befs_ino = BEFS_I(d_inode(dentry));
+	struct inode *inode = page->mapping->host;
+	struct super_block *sb = inode->i_sb;
+	struct befs_inode_info *befs_ino = BEFS_I(inode);
 	befs_data_stream *data = &befs_ino->i_data.ds;
 	befs_off_t len = data->size;
-	char *link;
+	char *link = kmap(page);
 
-	if (len == 0) {
+	if (len == 0 || len > PAGE_SIZE) {
 		befs_error(sb, "Long symlink with illegal length");
-		return ERR_PTR(-EIO);
+		goto fail;
 	}
 	befs_debug(sb, "Follow long symlink");
 
-	link = kmalloc(len, GFP_NOFS);
-	if (!link)
-		return ERR_PTR(-ENOMEM);
 	if (befs_read_lsymlink(sb, data, link, len) != len) {
-		kfree(link);
 		befs_error(sb, "Failed to read entire long symlink");
-		return ERR_PTR(-EIO);
+		goto fail;
 	}
 	link[len - 1] = '\0';
-	return *cookie = link;
+	SetPageUptodate(page);
+	kunmap(page);
+	unlock_page(page);
+	return 0;
+fail:
+	SetPageError(page);
+	kunmap(page);
+	unlock_page(page);
+	return -EIO;
 }
 
 /*
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287124 — [PATCH v2 03/11] udf: don't duplicate page_symlink_inode_operations

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 03/11] udf: don't duplicate page_symlink_inode_operations
Message-ID	<qDFDQ-485-27@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/udf/inode.c   | 2 +-
 fs/udf/namei.c   | 7 +------
 fs/udf/udfdecl.h | 1 -
 3 files changed, 2 insertions(+), 8 deletions(-)

diff --git a/fs/udf/inode.c b/fs/udf/inode.c
index 8d0b3ad..8675c2b 100644
--- a/fs/udf/inode.c
+++ b/fs/udf/inode.c
@@ -1540,7 +1540,7 @@ reread:
 		break;
 	case ICBTAG_FILE_TYPE_SYMLINK:
 		inode->i_data.a_ops = &udf_symlink_aops;
-		inode->i_op = &udf_symlink_inode_operations;
+		inode->i_op = &page_symlink_inode_operations;
 		inode->i_mode = S_IFLNK | S_IRWXUGO;
 		break;
 	case ICBTAG_FILE_TYPE_MAIN:
diff --git a/fs/udf/namei.c b/fs/udf/namei.c
index c97b5a8..d0e6de1 100644
--- a/fs/udf/namei.c
+++ b/fs/udf/namei.c
@@ -921,7 +921,7 @@ static int udf_symlink(struct inode *dir, struct dentry *dentry,
 	}
 
 	inode->i_data.a_ops = &udf_symlink_aops;
-	inode->i_op = &udf_symlink_inode_operations;
+	inode->i_op = &page_symlink_inode_operations;
 
 	if (iinfo->i_alloc_type != ICBTAG_FLAG_AD_IN_ICB) {
 		struct kernel_lb_addr eloc;
@@ -1344,8 +1344,3 @@ const struct inode_operations udf_dir_inode_operations = {
 	.rename				= udf_rename,
 	.tmpfile			= udf_tmpfile,
 };
-const struct inode_operations udf_symlink_inode_operations = {
-	.readlink	= generic_readlink,
-	.follow_link	= page_follow_link_light,
-	.put_link	= page_put_link,
-};
diff --git a/fs/udf/udfdecl.h b/fs/udf/udfdecl.h
index 47bb3f5..ce169b4 100644
--- a/fs/udf/udfdecl.h
+++ b/fs/udf/udfdecl.h
@@ -85,7 +85,6 @@ extern const struct inode_operations udf_dir_inode_operations;
 extern const struct file_operations udf_dir_operations;
 extern const struct inode_operations udf_file_inode_operations;
 extern const struct file_operations udf_file_operations;
-extern const struct inode_operations udf_symlink_inode_operations;
 extern const struct address_space_operations udf_aops;
 extern const struct address_space_operations udf_adinicb_aops;
 extern const struct address_space_operations udf_symlink_aops;
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287125 — [PATCH v2 09/11] teach shmem_get_link() to work in RCU mode

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 09/11] teach shmem_get_link() to work in RCU mode
Message-ID	<qDFDQ-485-23@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 mm/shmem.c | 20 ++++++++++++++------
 1 file changed, 14 insertions(+), 6 deletions(-)

diff --git a/mm/shmem.c b/mm/shmem.c
index 684dbc3..0605716 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -2501,12 +2501,20 @@ static const char *shmem_get_link(struct dentry *dentry,
 {
 	struct page *page = NULL;
 	int error;
-	if (!dentry)
-		return ERR_PTR(-ECHILD);
-	error = shmem_getpage(inode, 0, &page, SGP_READ, NULL);
-	if (error)
-		return ERR_PTR(error);
-	unlock_page(page);
+	if (!dentry) {
+		page = find_get_page(inode->i_mapping, 0);
+		if (!page)
+			return ERR_PTR(-ECHILD);
+		if (!PageUptodate(page)) {
+			put_page(page);
+			return ERR_PTR(-ECHILD);
+		}
+	} else {
+		error = shmem_getpage(inode, 0, &page, SGP_READ, NULL);
+		if (error)
+			return ERR_PTR(error);
+		unlock_page(page);
+	}
 	*cookie = page;
 	return page_address(page);
 }
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287126 — [PATCH v2 04/11] ufs: get rid of ->setattr() for symlinks

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 06:40 +0100
Subject	[PATCH v2 04/11] ufs: get rid of ->setattr() for symlinks
Message-ID	<qDFDR-485-37@gated-at.bofh.it>
In reply to	#1287116

From: Al Viro <viro@zeniv.linux.org.uk>

It was to needed for a couple of months in 2010, until UFS
quota support got dropped.  Since then it's equivalent to
simple_setattr() (i.e. the default) for everything except the
regular files.  And dropping it there allows to convert all
UFS symlinks to {page,simple}_symlink_inode_operations, getting
rid of fs/ufs/symlink.c completely.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/ufs/Makefile  |  2 +-
 fs/ufs/inode.c   |  4 ++--
 fs/ufs/namei.c   |  4 ++--
 fs/ufs/symlink.c | 42 ------------------------------------------
 fs/ufs/ufs.h     |  4 ----
 5 files changed, 5 insertions(+), 51 deletions(-)
 delete mode 100644 fs/ufs/symlink.c

diff --git a/fs/ufs/Makefile b/fs/ufs/Makefile
index 392db25..ec4a6b4 100644
--- a/fs/ufs/Makefile
+++ b/fs/ufs/Makefile
@@ -5,5 +5,5 @@
 obj-$(CONFIG_UFS_FS) += ufs.o
 
 ufs-objs := balloc.o cylinder.o dir.o file.o ialloc.o inode.o \
-	    namei.o super.o symlink.o util.o
+	    namei.o super.o util.o
 ccflags-$(CONFIG_UFS_DEBUG)    += -DDEBUG
diff --git a/fs/ufs/inode.c b/fs/ufs/inode.c
index a064cf44..737160a 100644
--- a/fs/ufs/inode.c
+++ b/fs/ufs/inode.c
@@ -528,11 +528,11 @@ static void ufs_set_inode_ops(struct inode *inode)
 		inode->i_mapping->a_ops = &ufs_aops;
 	} else if (S_ISLNK(inode->i_mode)) {
 		if (!inode->i_blocks) {
-			inode->i_op = &ufs_fast_symlink_inode_operations;
 			inode->i_link = (char *)UFS_I(inode)->i_u1.i_symlink;
+			inode->i_op = &simple_symlink_inode_operations;
 		} else {
-			inode->i_op = &ufs_symlink_inode_operations;
 			inode->i_mapping->a_ops = &ufs_aops;
+			inode->i_op = &page_symlink_inode_operations;
 		}
 	} else
 		init_special_inode(inode, inode->i_mode,
diff --git a/fs/ufs/namei.c b/fs/ufs/namei.c
index 4796655..24b0cbd 100644
--- a/fs/ufs/namei.c
+++ b/fs/ufs/namei.c
@@ -123,14 +123,14 @@ static int ufs_symlink (struct inode * dir, struct dentry * dentry,
 
 	if (l > UFS_SB(sb)->s_uspi->s_maxsymlinklen) {
 		/* slow symlink */
-		inode->i_op = &ufs_symlink_inode_operations;
+		inode->i_op = &page_symlink_inode_operations;
 		inode->i_mapping->a_ops = &ufs_aops;
 		err = page_symlink(inode, symname, l);
 		if (err)
 			goto out_fail;
 	} else {
 		/* fast symlink */
-		inode->i_op = &ufs_fast_symlink_inode_operations;
+		inode->i_op = &simple_symlink_inode_operations;
 		inode->i_link = (char *)UFS_I(inode)->i_u1.i_symlink;
 		memcpy(inode->i_link, symname, l);
 		inode->i_size = l-1;
diff --git a/fs/ufs/symlink.c b/fs/ufs/symlink.c
deleted file mode 100644
index 874480b..0000000
--- a/fs/ufs/symlink.c
+++ /dev/null
@@ -1,42 +0,0 @@
-/*
- *  linux/fs/ufs/symlink.c
- *
- * Only fast symlinks left here - the rest is done by generic code. AV, 1999
- *
- * Copyright (C) 1998
- * Daniel Pirkl <daniel.pirkl@emai.cz>
- * Charles University, Faculty of Mathematics and Physics
- *
- *  from
- *
- *  linux/fs/ext2/symlink.c
- *
- * Copyright (C) 1992, 1993, 1994, 1995
- * Remy Card (card@masi.ibp.fr)
- * Laboratoire MASI - Institut Blaise Pascal
- * Universite Pierre et Marie Curie (Paris VI)
- *
- *  from
- *
- *  linux/fs/minix/symlink.c
- *
- *  Copyright (C) 1991, 1992  Linus Torvalds
- *
- *  ext2 symlink handling code
- */
-
-#include "ufs_fs.h"
-#include "ufs.h"
-
-const struct inode_operations ufs_fast_symlink_inode_operations = {
-	.readlink	= generic_readlink,
-	.follow_link	= simple_follow_link,
-	.setattr	= ufs_setattr,
-};
-
-const struct inode_operations ufs_symlink_inode_operations = {
-	.readlink	= generic_readlink,
-	.follow_link	= page_follow_link_light,
-	.put_link	= page_put_link,
-	.setattr	= ufs_setattr,
-};
diff --git a/fs/ufs/ufs.h b/fs/ufs/ufs.h
index 7da4aca..c87f4c3 100644
--- a/fs/ufs/ufs.h
+++ b/fs/ufs/ufs.h
@@ -136,10 +136,6 @@ extern __printf(3, 4)
 void ufs_panic(struct super_block *, const char *, const char *, ...);
 void ufs_mark_sb_dirty(struct super_block *sb);
 
-/* symlink.c */
-extern const struct inode_operations ufs_fast_symlink_inode_operations;
-extern const struct inode_operations ufs_symlink_inode_operations;
-
 static inline struct ufs_sb_info *UFS_SB(struct super_block *sb)
 {
 	return sb->s_fs_info;
-- 
2.1.4

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

[toc] | [prev] | [next] | [standalone]

#1287677

From	Linus Torvalds <torvalds@linux-foundation.org>
Date	2015-12-09 18:30 +0100
Message-ID	<qDQIW-2Tz-25@gated-at.bofh.it>
In reply to	#1287116

On Tue, Dec 8, 2015 at 9:32 PM, Al Viro <viro@zeniv.linux.org.uk> wrote:
>
> Any help with review and testing would be welcome.  FWIW, it seems to survive
> the beating here.

It looks ok to me, but I *hate* that "cookie" thing we have. And I
don't actually see any reason for it.

Every single user except for a couple seem to just put the resulting
symlink address into the cookie (and then put_link() will kfree it or
put_page() it). And the calling convention would be *so* much clearer
if we just made things be

   char *get_link(dentry, inode);
   put_link(dentry, inode, char *);

and get rid of that cookie entirely.

The only users of the cookie seem to be

 - we currently use it for the page pointer vs the page data address.

   That's because of the HIGHMEM crap that you removed. Without
highmem, the page data address specifies the page fine without any
cookie.

 - overlayfs, which uses it to create a temporary data storage.

   Why? Mainly because it wants to hide the cookie in there. D'oh!

 - proc, which uses it to save the proc_dir_entry

   Why? The reason seems to be that we don't call put_link() with the
dentry that get_link got, so it can't look it up the sane way.

In other words, as far as I can tell, the cookie is just garbage now.

But maybe I missed something.

Other than that, I like the series. And I realize that technically the
"cookie removal" would be a separate thing, but since you're changing
the calling convention from "follow_link()" to "get_link()" in this
series, I _really_ think that as part of that calling convention
change we should just get rid of the cookie, rather than have
*another* calling convention change later.

Hmm?

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

[toc] | [prev] | [next] | [standalone]

#1287727

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-09 19:30 +0100
Message-ID	<qDRF0-3u6-3@gated-at.bofh.it>
In reply to	#1287677

On Wed, Dec 09, 2015 at 09:24:32AM -0800, Linus Torvalds wrote:

>    That's because of the HIGHMEM crap that you removed. Without
> highmem, the page data address specifies the page fine without any
> cookie.
> 
>  - overlayfs, which uses it to create a temporary data storage.
> 
>    Why? Mainly because it wants to hide the cookie in there. D'oh!

Not really - it also wants to stash a way to underlying inode in there.
See below...

>  - proc, which uses it to save the proc_dir_entry
> 
>    Why? The reason seems to be that we don't call put_link() with the
> dentry that get_link got, so it can't look it up the sane way.

What's more, that dentry might very well have gone negative by that
point.  Think what happens if, during the symlink traversal, we run
into the hard "restart from scratch in non-RCU mode".  We'll need to
do ->put_link() on everything we have in stack.  Regardless of what
might've happened to dentries/inodes of symlinks involved - all we
are really promised is that RCU-delayed parts of their destruction hadn't
been entered yet.

Note, BTW, that *all* ->put_link() instances ignore the inode argument
and I seriously considered dropping it completely.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1288071

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-10 01:20 +0100
Message-ID	<qDX7I-73a-7@gated-at.bofh.it>
In reply to	#1287727

On Wed, Dec 09, 2015 at 06:23:09PM +0000, Al Viro wrote:

> What's more, that dentry might very well have gone negative by that
> point.  Think what happens if, during the symlink traversal, we run
> into the hard "restart from scratch in non-RCU mode".  We'll need to
> do ->put_link() on everything we have in stack.  Regardless of what
> might've happened to dentries/inodes of symlinks involved - all we
> are really promised is that RCU-delayed parts of their destruction hadn't
> been entered yet.
> 
> Note, BTW, that *all* ->put_link() instances ignore the inode argument
> and I seriously considered dropping it completely.

PS: I toyed with the idea of replacing cookie a struct callback_head, to make
it really obvious what's going on - we are leaving a closure to be evaluated
once the caller is done with our return value.  That way ->put_link() as a
method would've been gone and we would've had something like
	init_callback(destructor, page_put_link, page);
instead of
	*cookie = page;
plus
	.put_link = page_put_link,
we have right now, etc.

The reason not to go that way (and not all that strong, at that) is that
we get an extra pointer in struct saved, i.e. slightly heavier nameidata
(there are two struct saved embedded into it).  That, or play silly buggers
with
	union {
		struct callback_head callback;
		struct inode *inode;
	};
in struct saved - we need to stash a pointer to inode between pick_link()
and get_link().

Hell knows, still might make sense to go there...  What we want is to set
a closure to be evaluated once the caller of ->get_link() is done with
the string ->get_link() returns.  We end up picking the function to be
called at saved->inode->i_op->put_link and store its argument in
saved->cookie.  Might as well just use the representation we use for such
stuff (rcu_head, etc.) and be done with that...
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1288149

From	Al Viro <viro@ZenIV.linux.org.uk>
Date	2015-12-10 03:50 +0100
Message-ID	<qDZsR-8qg-3@gated-at.bofh.it>
In reply to	#1288071

On Thu, Dec 10, 2015 at 12:10:12AM +0000, Al Viro wrote:

> PS: I toyed with the idea of replacing cookie a struct callback_head, to make

... except that it would have to be something like work_struct, since
callback_head gets embedded into the argument and here we don't want to
do that.  _And_ we've no use for ->entry in work_struct, which makes it
a bad fit as well.  IOW, please ignore that idiocy...
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1289275

From	Rasmus Villemoes <linux@rasmusvillemoes.dk>
Date	2015-12-11 08:50 +0100
Message-ID	<qEqCJ-167-7@gated-at.bofh.it>
In reply to	#1288149

On Fri, Dec 11 2015, Al Viro <viro@ZenIV.linux.org.uk> wrote:

> I would really love to be able to say
> 	set_delayed_call(done, kfree, p);
> but as it is I had to keep a wrapper - void kfree_link(void *).  The problem
> is, you can't assign void f(const void *) to void (*p)(void *) - mismatch of
> qualifiers in the arguments makes the latter not assignment-compatible with
> the former.  If there's a clever trick allowing to sidestep that, I'd be
> very happy; I don't know one.  Any ideas not starting with "use C11" (or,
> worse yet, "use such and such C++ misfeature with arseloads of RTL required
> in order to implement it") would be welcome...

I _think_ this satisfies these very reasonable criteria. What you're
looking for is presumably __attribute__((__transparent_union__)). At
least this compiles without warnings at -Wall -Wextra and gives the
expected disassembly, and the gcc docs mention transparent_union at
least back to 4.0.4.

#include <stddef.h>

struct delayed_call {
	void (*fn)(void *);
	void *arg;
};
union delayed_call_fn {
	void (*fn)(void *);
	void (*kfree_like)(const void *);
} __attribute__((__transparent_union__));

void
set_delayed_call(struct delayed_call *call, union delayed_call_fn u, void *arg)
{
	call->fn = u.fn;
	call->arg = arg;
}

void some_cb(void *);
void kfree(const void *);
extern struct delayed_call done;

void test(void)
{
	set_delayed_call(&done, some_cb, NULL);
	set_delayed_call(&done, kfree, NULL);
}

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

[toc] | [prev] | [next] | [standalone]

#1287912

From	NeilBrown <neilb@suse.com>
Date	2015-12-09 23:00 +0100
Message-ID	<qDUWf-5u5-33@gated-at.bofh.it>
In reply to	#1287116

[Multipart message — attachments visible in raw view] — view raw

Hi Al,
 thanks a lot for this.  I particularly like that you renamed
 "follow_link" which didn't really describe what the function did any
 more.

 Apart from the changelog comment in 4/11 which starts
   "It was to needed for ..."

 I can't fault anything.

Thanks,
NeilBrown

[toc] | [prev] | [standalone]

csiph-web

[PATCHSET v2] ->follow_link() without dropping from RCU mode

Contents

#1287116 — [PATCHSET v2] ->follow_link() without dropping from RCU mode

#1287117 — [PATCH v2 02/11] logfs: don't duplicate page_symlink_inode_operations

#1287118 — [PATCH v2 08/11] teach page_get_link() to work in RCU mode

#1287119 — [PATCH v2 10/11] teach proc_self_get_link()/proc_thread_self_get_link() to work in RCU mode

#1287120 — [PATCH v2 11/11] teach nfs_get_link() to work in RCU mode

#1287121 — [PATCH v2 06/11] don't put symlink bodies in pagecache into highmem

#1287122 — [PATCH v2 01/11] switch befs long symlinks to page_symlink_operations

#1287124 — [PATCH v2 03/11] udf: don't duplicate page_symlink_inode_operations

#1287125 — [PATCH v2 09/11] teach shmem_get_link() to work in RCU mode

#1287126 — [PATCH v2 04/11] ufs: get rid of ->setattr() for symlinks

#1287677

#1287727

#1288071

#1288149

#1289275

#1287912