Skip to contentSkip to conversation

vfs: no "(unreachable)" prefix for SYSVIPC maps in /proc/PID/maps

PATCH In review [email protected] 2 participants 3 messages Sep 18, 2009 – Sep 21, 2009

Subsystems (4)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

PATCH 20/21 vfs: no "(unreachable)" prefix for SYSVIPC maps in /proc/PID/maps

From: Miklos Szeredi <[email protected]>

The patch

"vfs: fix d_path() for unreachable paths"

generally changed d_path() to report unreachable paths with a special prefix. This has an effect on /proc/${PID}/maps as well for memory maps set up with shmem_file_setup() or hugetlb_file_setup(). These functions set up unlinked files under a kernel-private vfsmount. Since this vfsmount is unreachable from userspace, these maps will be reported with the "(unreachable)" prefix.

This is undesirable, because it changes the kernel ABI and might break applications for no good reason.

Cc Valdis Kletnieks <[email protected]>
Cc Matthew Wilcox <[email protected]>
Cc Andreas Gruenbacher <[email protected]>
Cc Al Viro <[email protected]>
Cc Christoph Hellwig <[email protected]>

---

 fs/hugetlbfs/inode.c |   17 +++++++++++++++++
 mm/shmem.c           |   17 +++++++++++++++++
 2 files changed, 34 insertions(+)
fs/hugetlbfs/inode.c renamed from fs/hugetlbfs/inode.c~vfs-no-unreachable-prefix-for-sysvipc-maps-in-proc-pid-maps +17 -0
diff -puN fs/hugetlbfs/inode.c~vfs-no-unreachable-prefix-for-sysvipc-maps-in-proc-pid-maps fs/hugetlbfs/inode.c
@@ -906,6 +906,21 @@ static struct file_system_type hugetlbfs
 
 static struct vfsmount *hugetlbfs_vfsmount;
 
+/*
+ * Do a special d_dname function so that these are not prefixed by
+ * "(unreachable)".
+ */
+static char *hugetlb_unlinked_d_dname(struct dentry *dentry, char *buf,
+				      int buflen)
+{
+	return dynamic_dname(dentry, buf, buflen, "/%s (deleted)",
+			     dentry->d_name.name);
+}
+
+static struct dentry_operations hugetlb_unlinked_dentry_operations = {
+	.d_dname = hugetlb_unlinked_d_dname,
+};
+
 static int can_do_hugetlb_shm(void)
 {
 	return capable(CAP_IPC_LOCK) || in_group_p(sysctl_hugetlb_shm_group);
@@ -943,6 +958,8 @@ struct file *hugetlb_file_setup(const ch
 	if (!dentry)
 		goto out_shm_unlock;
 
+	dentry->d_op = &hugetlb_unlinked_dentry_operations;
+
 	error = -ENOSPC;
 	inode = hugetlbfs_get_inode(root->d_sb, current_fsuid(),
 				current_fsgid(), S_IFREG | S_IRWXUGO, 0);
mm/shmem.c renamed from mm/shmem.c~vfs-no-unreachable-prefix-for-sysvipc-maps-in-proc-pid-maps +17 -0
diff -puN mm/shmem.c~vfs-no-unreachable-prefix-for-sysvipc-maps-in-proc-pid-maps mm/shmem.c
@@ -2602,6 +2602,21 @@ int shmem_unuse(swp_entry_t entry, struc
 
 /* common code */
 
+/*
+ * Do a special d_dname function so that these are not prefixed by
+ * "(unreachable)".
+ */
+static char *shmem_unlinked_d_dname(struct dentry *dentry, char *buf,
+				    int buflen)
+{
+	return dynamic_dname(dentry, buf, buflen, "/%s (deleted)",
+			     dentry->d_name.name);
+}
+
+static struct dentry_operations shmem_unlinked_dentry_operations = {
+	.d_dname = shmem_unlinked_d_dname,
+};
+
 /**
  * shmem_file_setup - get an unlinked file living in tmpfs
  * @name: name for dentry (to be seen in /proc/<pid>/maps
@@ -2634,6 +2649,8 @@ struct file *shmem_file_setup(const char
 	if (!dentry)
 		goto put_memory;
 
+	dentry->d_op = &shmem_unlinked_dentry_operations;
+
 	error = -ENFILE;
 	file = get_empty_filp();
 	if (!file)

_

maintainer
#1 lore raw
Quoted 13 lines · On Fri, Sep 18, 2009 at 01:06:02PM -0700, [email protected] wrote:

From: Miklos Szeredi <[email protected]>

The patch

"vfs: fix d_path() for unreachable paths"

generally changed d_path() to report unreachable paths with a special prefix. This has an effect on /proc/${PID}/maps as well for memory maps set up with shmem_file_setup() or hugetlb_file_setup(). These functions set up unlinked files under a kernel-private vfsmount. Since this vfsmount is unreachable from userspace, these maps will be reported with the "(unreachable)" prefix.

Quoted 2 lines

This is undesirable, because it changes the kernel ABI and might break applications for no good reason.

I'd rather see 17, 19 and 20 collapsed...

maintainer
#2 lore raw
Quoted 1 line · Al Viro

I'd rather see 17, 19 and 20 collapsed...

D'oh - s/17/16/ in the above. IOW, d_path() patch + these two fixups to it.