Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1666768

Re: [PATCH 25/27] ipc: Convert mqueue fs to fs_context [ver #5]

From David Howells <dhowells@redhat.com>
Newsgroups linux.kernel
Subject Re: [PATCH 25/27] ipc: Convert mqueue fs to fs_context [ver #5]
Date 2017-06-15 16:50 +0200
Message-ID <tSEfT-3Hl-9@gated-at.bofh.it> (permalink)
References <tSzSV-18C-1@gated-at.bofh.it> <tSifn-6QJ-7@gated-at.bofh.it> <tSifp-6QJ-41@gated-at.bofh.it> <tSzSV-18C-1@gated-at.bofh.it>
Organization Red Hat UK Ltd. Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SI4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 3798903

Show all headers | View raw


Al Viro <viro@ZenIV.linux.org.uk> wrote:

> > +	if (ctx->ipc_ns != ns) {
> 
> How could they possibly be equal?  You are setting that ns up here, right?
> How could it be in any process' nsproxy?

Fair point.

> Ugh, again...  Is there any reason for dynamic allocation of that thing in
> this particular case?  AFAICS, these contortions are all due to going through
> vfs_new_fs_context()/put_fs_context().  And it's not as if they had been
> refcounted...

I can statically initialise it as below, but whilst I don't need to call
vfs_new_fs_context() and put_fs_context(), I do have to call the security
hooks (Smack makes no distinction for internal filesystems) to set up the
security context and clean it up, and I do have to have the error handling for
in case kern_mount_data_fc() fails.

So it actually makes both the source and the object file bigger.  Now, I could
hide some of this inside a pair of inline functions, but it doesn't help that
much.

What might be better is to provide a function that wraps the vfs_get_tree()
and kern_mount_data_fc() calls and the cleanup.

David
---
diff --git a/ipc/mqueue.c b/ipc/mqueue.c
index f2e1d1d69961..a18a5f6763f9 100644
--- a/ipc/mqueue.c
+++ b/ipc/mqueue.c
@@ -1431,8 +1431,15 @@ static struct file_system_type mqueue_fs_type = {
 
 int mq_init_ns(struct ipc_namespace *ns)
 {
-	struct mqueue_fs_context *ctx;
-	struct fs_context *fc;
+	struct mqueue_fs_context ctx = {
+		.fc.ops		= &mqueue_fs_context_ops,
+		.fc.fs_type	= &mqueue_fs_type,
+		.fc.cred	= current_cred(),
+		.fc.user_ns	= current_user_ns(),
+		.fc.purpose	= FS_CONTEXT_FOR_NEW,
+		.ipc_ns		= ns,
+	};
+	struct super_block *sb;
 	struct vfsmount *mnt;
 	int ret;
 
@@ -1443,29 +1450,32 @@ int mq_init_ns(struct ipc_namespace *ns)
 	ns->mq_msg_default   = DFLT_MSG;
 	ns->mq_msgsize_default  = DFLT_MSGSIZE;
 
-	fc = vfs_new_fs_context(&mqueue_fs_type, NULL, 0, FS_CONTEXT_FOR_NEW);
-	if (IS_ERR(fc))
-		return PTR_ERR(fc);
-
-	ctx = container_of(fc, struct mqueue_fs_context, fc);
-	put_ipc_ns(ctx->ipc_ns);
-	ctx->ipc_ns = get_ipc_ns(ns);
+	ret = security_fs_context_alloc(&ctx.fc, NULL);
+	if (ret < 0)
+		return ret;
 
-	ret = vfs_get_tree(fc);
+	ret = vfs_get_tree(&ctx.fc);
 	if (ret < 0)
 		goto out_fc;
 
-	mnt = kern_mount_data_fc(fc);
+	mnt = kern_mount_data_fc(&ctx.fc);
 	if (IS_ERR(mnt)) {
 		ret = PTR_ERR(mnt);
-		goto out_fc;
+		goto out_root;
 	}
 
 	ns->mq_mnt = mnt;
 	ret = 0;
 out_fc:
-	put_fs_context(fc);
+	security_fs_context_free(&ctx.fc);
 	return ret;
+
+out_root:
+	sb = ctx.fc.root->d_sb;
+	dput(ctx.fc.root);
+	ctx.fc.root = NULL;
+	deactivate_super(sb);
+	goto out_fc;
 }
 
 void mq_clear_sbinfo(struct ipc_namespace *ns)

Back to linux.kernel | Previous | NextPrevious in thread | Next in thread | Find similar | Unroll thread


Thread

[RFC][PATCH 00/27] VFS: Introduce filesystem context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 11/27] VFS: Remove unused code after filesystem context  changes [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 20/27] NFS: Split nfs_parse_mount_options() [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 03/27] VFS: Make get_mnt_ns() return the namespace [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
    Re: [PATCH 03/27] VFS: Make get_mnt_ns() return the namespace [ver  #5] Al Viro <viro@ZenIV.linux.org.uk> - 2017-06-15 11:20 +0200
  [PATCH 27/27] kernfs, sysfs, cgroup: Support fs_context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
    Re: [PATCH 27/27] kernfs, sysfs, cgroup: Support fs_context [ver #5] Tejun Heo <tj@kernel.org> - 2017-06-14 20:00 +0200
    Re: [PATCH 27/27] ... and the intel_rdt driver David Howells <dhowells@redhat.com> - 2017-06-15 00:40 +0200
  [PATCH 26/27] cpuset: Use fs_context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 16/27] proc: Add fs_context support to procfs [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
    Re: [PATCH 16/27] proc: Add fs_context support to procfs [ver #5] Al Viro <viro@ZenIV.linux.org.uk> - 2017-06-15 12:20 +0200
  [PATCH 01/27] Provide a function to create a NUL-terminated string  from unterminated data [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 09/27] VFS: Add LSM hooks for filesystem context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 25/27] ipc: Convert mqueue fs to fs_context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
    Re: [PATCH 25/27] ipc: Convert mqueue fs to fs_context [ver #5] Al Viro <viro@ZenIV.linux.org.uk> - 2017-06-15 12:10 +0200
      Re: [PATCH 25/27] ipc: Convert mqueue fs to fs_context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-15 16:50 +0200
  [PATCH 23/27] NFS: Do some tidying of the parsing code [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 22/27] NFS: Add a small buffer in nfs_fs_context to avoid  string dup [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 14/27] VFS: Add a sample program for fsopen/fsmount [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 02/27] VFS: Clean up whitespace in fs/namespace.c and  fs/super.c [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 04/27] VFS: Make get_filesystem() return the affected  filesystem [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 18/27] NFS: Constify mount argument match tables [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:20 +0200
  [PATCH 05/27] VFS: Provide empty name qstr [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:30 +0200
  [PATCH 15/27] procfs: Move proc_fill_super() to fs/proc/root.c [ver  #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:30 +0200
  [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem  context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:30 +0200
    Re: [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem  context [ver #5] Randy Dunlap <rdunlap@infradead.org> - 2017-06-14 20:10 +0200
      Updated docs David Howells <dhowells@redhat.com> - 2017-06-15 01:00 +0200
        Re: Updated docs Randy Dunlap <rdunlap@infradead.org> - 2017-06-15 04:00 +0200
    Re: [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem  context [ver #5] Casey Schaufler <casey@schaufler-ca.com> - 2017-06-14 22:10 +0200
      Re: [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 22:50 +0200
        Re: [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem  context [ver #5] Casey Schaufler <casey@schaufler-ca.com> - 2017-06-14 23:00 +0200
          Re: [PATCH 08/27] VFS: Introduce the structs and doc for a  filesystem context [ver #5] Theodore Ts'o <tytso@mit.edu> - 2017-06-17 12:00 +0200
            Re: [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-17 16:20 +0200
              Re: [PATCH 08/27] VFS: Introduce the structs and doc for a  filesystem context [ver #5] Jeff Layton <jlayton@redhat.com> - 2017-06-17 17:00 +0200
                Re: [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem  context [ver #5] Randy Dunlap <rdunlap@infradead.org> - 2017-06-17 17:20 +0200
                Re: [PATCH 08/27] VFS: Introduce the structs and doc for a filesystem context [ver #5] David Howells <dhowells@redhat.com> - 2017-06-19 09:50 +0200
  [PATCH 21/27] NFS: Deindent nfs_fs_context_parse_option() [ver #5] David Howells <dhowells@redhat.com> - 2017-06-14 17:30 +0200
  Re: [PATCH 07/27] VFS: Differentiate mount flags (MS_*) from  internal superblock flags [ver #5] Al Viro <viro@ZenIV.linux.org.uk> - 2017-06-15 11:50 +0200
    Re: [PATCH 07/27] VFS: Differentiate mount flags (MS_*) from  internal superblock flags [ver #5] Christoph Hellwig <hch@infradead.org> - 2017-06-16 11:10 +0200
      Re: [PATCH 07/27] VFS: Differentiate mount flags (MS_*) from internal superblock flags [ver #5] David Howells <dhowells@redhat.com> - 2017-06-16 17:00 +0200
        Re: [PATCH 07/27] VFS: Differentiate mount flags (MS_*) from  internal superblock flags [ver #5] Christoph Hellwig <hch@infradead.org> - 2017-06-16 18:00 +0200
          Re: [PATCH 07/27] VFS: Differentiate mount flags (MS_*) from internal superblock flags [ver #5] David Howells <dhowells@redhat.com> - 2017-06-16 18:00 +0200

csiph-web