From: Andreas Gruenbacher <agruen@suse.de>
To: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
Neil Brown <neilb@cse.unsw.edu.au>, Olaf Kirch <okir@suse.de>,
"Andries E. Brouwer" <Andries.Brouwer@cwi.nl>,
Andrew Morton <akpm@osdl.org>
Subject: Re: [patch 12/13] ACL umask handling workaround in nfs client
Date: Tue, 22 Feb 2005 17:47:24 +0100 [thread overview]
Message-ID: <1109090843.6102.443.camel@winden.suse.de> (raw)
In-Reply-To: <1108490682.10073.57.camel@lade.trondhjem.org>
[-- Attachment #1: Type: text/plain, Size: 1826 bytes --]
On Tue, 2005-02-15 at 19:04, Trond Myklebust wrote:
> lau den 22.01.2005 Klokka 21:34 (+0100) skreiv Andreas Gruenbacher:
> > vanlig tekstdokument vedlegg (patches.suse)
> > NFSv3 has no concept of a umask on the server side: The client applies
> > the umask locally, and sends the effective permissions to the server.
> > This behavior is wrong when files are created in a directory that has
> > a default ACL. In this case, the umask is supposed to be ignored, and
> > only the default ACL determines the file's effective permissions.
> >
> > Usually its the server's task to conditionally apply the umask. But
> > since the server knows nothing about the umask, we have to do it on the
> > client side. This patch tries to fetch the parent directory's default
> > ACL before creating a new file, computes the appropriate create mode to
> > send to the server, and finally sets the new file's access and default
> > acl appropriately.
>
> Firstly, this sort of code belongs in the NFSv3-specific code. POSIX
> acls have no business whatsoever in the generic NFS code.
See attached patch.
NOTE:
During testing I noticed that without
nfsacl-cache-acls-on-the-nfs-client-side.patch, no directories or
devices can be created. It's probably a problem with
nfs_set_default_acl(). I'll have to debug this tomorrow.
> Secondly, what is the point of doing all this *after* you have created
> the file with the wrong permissions? How are you avoiding races?
Well, everything but the umask is always correct; that is guaranteed by
the server. The initial create sets permissions that may be more
restrictive than necessary, and then the SETACL RPC sets up the final,
correct permissions. I don't believe that a race-free solution is
possible.
Cheers,
--
Andreas Gruenbacher <agruen@suse.de>
SUSE Labs, SUSE LINUX GMBH
[-- Attachment #2: nfsacl-acl-umask-handling-workaround-in-nfs-client-fix2.patch --]
[-- Type: text/x-patch, Size: 4260 bytes --]
Index: linux-2.6.11-rc3/fs/nfs/dir.c
===================================================================
--- linux-2.6.11-rc3.orig/fs/nfs/dir.c
+++ linux-2.6.11-rc3/fs/nfs/dir.c
@@ -42,12 +42,15 @@ static int nfs_opendir(struct inode *, s
static int nfs_readdir(struct file *, void *, filldir_t);
static struct dentry *nfs_lookup(struct inode *, struct dentry *, struct nameidata *);
static int nfs_create(struct inode *, struct dentry *, int, struct nameidata *);
+static int nfs3_create(struct inode *, struct dentry *, int, struct nameidata *);
static int nfs_mkdir(struct inode *, struct dentry *, int);
+static int nfs3_mkdir(struct inode *, struct dentry *, int);
static int nfs_rmdir(struct inode *, struct dentry *);
static int nfs_unlink(struct inode *, struct dentry *);
static int nfs_symlink(struct inode *, struct dentry *, const char *);
static int nfs_link(struct dentry *, struct inode *, struct dentry *);
static int nfs_mknod(struct inode *, struct dentry *, int, dev_t);
+static int nfs3_mknod(struct inode *, struct dentry *, int, dev_t);
static int nfs_rename(struct inode *, struct dentry *,
struct inode *, struct dentry *);
static int nfs_fsync_dir(struct file *, struct dentry *, int);
@@ -77,14 +80,14 @@ struct inode_operations nfs_dir_inode_op
#ifdef CONFIG_NFS_V3
struct inode_operations nfs3_dir_inode_operations = {
- .create = nfs_create,
+ .create = nfs3_create,
.lookup = nfs_lookup,
.link = nfs_link,
.unlink = nfs_unlink,
.symlink = nfs_symlink,
- .mkdir = nfs_mkdir,
+ .mkdir = nfs3_mkdir,
.rmdir = nfs_rmdir,
- .mknod = nfs_mknod,
+ .mknod = nfs3_mknod,
.rename = nfs_rename,
.permission = nfs_permission,
.getattr = nfs_getattr,
@@ -994,16 +997,14 @@ out_err:
return error;
}
-static int nfs_set_default_acl(struct inode *dir, struct inode *inode,
- mode_t mode)
+#ifdef CONFIG_NFS_V3
+static int nfs3_set_default_acl(struct inode *dir, struct inode *inode,
+ mode_t mode)
{
#ifdef CONFIG_NFS_ACL
struct posix_acl *dfacl, *acl;
int error = 0;
- if (NFS_PROTO(inode)->version != 3 ||
- !NFS_PROTO(dir)->getacl || !NFS_PROTO(inode)->setacls)
- return 0;
dfacl = NFS_PROTO(dir)->getacl(dir, ACL_TYPE_DEFAULT);
if (IS_ERR(dfacl)) {
error = PTR_ERR(dfacl);
@@ -1028,6 +1029,7 @@ out:
return 0;
#endif
}
+#endif
/*
* Following a failed create operation, we drop the dentry rather
@@ -1060,7 +1062,7 @@ static int nfs_create(struct inode *dir,
d_instantiate(dentry, inode);
nfs_renew_times(dentry);
nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
- error = nfs_set_default_acl(dir, inode, mode);
+ error = 0;
} else {
error = PTR_ERR(inode);
d_drop(dentry);
@@ -1069,6 +1071,22 @@ static int nfs_create(struct inode *dir,
return error;
}
+#ifdef CONFIG_NFS_V3
+static int nfs3_create(struct inode *dir, struct dentry *dentry, int mode,
+ struct nameidata *nd)
+{
+ int error;
+
+ lock_kernel();
+ error = nfs_create(dir, dentry, mode, nd);
+ if (!error)
+ error = nfs3_set_default_acl(dir, dentry->d_inode, mode);
+ unlock_kernel();
+
+ return error;
+}
+#endif
+
/*
* See comments for nfs_proc_create regarding failed operations.
*/
@@ -1098,9 +1116,21 @@ nfs_mknod(struct inode *dir, struct dent
error = nfs_instantiate(dentry, &fhandle, &fattr);
else
d_drop(dentry);
+ unlock_kernel();
+ return error;
+}
+
+static int nfs3_mknod(struct inode *dir, struct dentry *dentry, int mode,
+ dev_t rdev)
+{
+ int error;
+
+ lock_kernel();
+ error = nfs_mknod(dir, dentry, mode, rdev);
if (!error)
- error = nfs_set_default_acl(dir, dentry->d_inode, mode);
+ error = nfs3_set_default_acl(dir, dentry->d_inode, mode);
unlock_kernel();
+
return error;
}
@@ -1138,9 +1168,20 @@ static int nfs_mkdir(struct inode *dir,
error = nfs_instantiate(dentry, &fhandle, &fattr);
else
d_drop(dentry);
+ unlock_kernel();
+ return error;
+}
+
+static int nfs3_mkdir(struct inode *dir, struct dentry *dentry, int mode)
+{
+ int error;
+
+ lock_kernel();
+ error = nfs_mkdir(dir, dentry, mode);
if (!error)
- error = nfs_set_default_acl(dir, dentry->d_inode, mode);
+ error = nfs3_set_default_acl(dir, dentry->d_inode, mode);
unlock_kernel();
+
return error;
}
next prev parent reply other threads:[~2005-02-22 16:48 UTC|newest]
Thread overview: 85+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-01-22 20:34 [patch 0/13] NFSACL protocol extension for NFSv3 Andreas Gruenbacher
2005-01-22 20:34 ` [patch 1/13] Qsort Andreas Gruenbacher
2005-01-22 21:00 ` vlobanov
2005-01-23 2:03 ` Felipe Alfaro Solana
2005-01-23 2:39 ` Andi Kleen
2005-01-23 3:02 ` Jesper Juhl
2005-01-23 4:46 ` Andi Kleen
2005-01-23 5:05 ` Jesper Juhl
2005-01-23 10:37 ` Rafael J. Wysocki
2005-01-24 4:29 ` Horst von Brand
2005-01-24 15:45 ` Alan Cox
2005-01-24 17:10 ` H. Peter Anvin
2005-01-25 0:43 ` Horst von Brand
2005-01-25 4:06 ` Eric St-Laurent
2005-01-24 22:04 ` Mike Waychison
2005-01-25 6:51 ` Andi Kleen
2005-01-25 10:12 ` Andreas Gruenbacher
2005-01-25 12:00 ` Andi Kleen
2005-01-25 12:05 ` Olaf Kirch
2005-01-25 16:52 ` Trond Myklebust
2005-01-25 16:53 ` Andreas Gruenbacher
2005-01-25 17:03 ` Trond Myklebust
2005-01-25 17:16 ` Andreas Gruenbacher
2005-01-25 17:37 ` Trond Myklebust
2005-01-25 18:12 ` Andreas Gruenbacher
2005-01-25 19:33 ` Trond Myklebust
2005-01-25 19:49 ` Andreas Gruenbacher
2005-01-23 4:29 ` Matt Mackall
2005-01-24 0:21 ` Nathan Scott
2005-01-24 2:57 ` Matt Mackall
2005-01-24 4:02 ` Horst von Brand
2005-01-24 21:57 ` Matt Mackall
2005-01-23 4:58 ` Felipe Alfaro Solana
2005-01-24 21:20 ` Matt Mackall
2005-01-24 21:50 ` vlobanov
2005-01-23 4:22 ` Matt Mackall
2005-01-23 5:44 ` Willy Tarreau
2005-01-23 21:24 ` Richard Henderson
[not found] ` <1106431568.4153.154.camel@laptopd505.fenrus.org>
2005-01-22 22:10 ` Andreas Gruenbacher
2005-01-22 23:28 ` Matt Mackall
2005-01-23 0:21 ` Matt Mackall
2005-01-23 5:08 ` Andreas Gruenbacher
2005-01-23 5:32 ` Matt Mackall
2005-01-23 12:22 ` Andreas Gruenbacher
2005-01-23 16:49 ` Matt Mackall
2005-01-24 3:48 ` Horst von Brand
2005-01-24 20:15 ` [PATCH] lib/qsort Matt Mackall
2005-01-24 23:09 ` Andrew Morton
2005-01-24 23:30 ` Matt Mackall
2005-01-25 4:11 ` Matt Mackall
2005-01-22 20:34 ` [patch 2/13] Return -ENOSYS for RPC programs that are unavailable Andreas Gruenbacher
2005-02-15 17:04 ` Trond Myklebust
2005-02-16 15:32 ` Andreas Gruenbacher
2005-01-22 20:34 ` [patch 3/13] Add missing -EOPNOTSUPP => NFS3ERR_NOTSUPP mapping in nfsd Andreas Gruenbacher
2005-01-22 20:34 ` [patch 4/13] Allow multiple programs to listen on the same port Andreas Gruenbacher
2005-01-22 20:34 ` [patch 5/13] Allow multiple programs to share the same transport Andreas Gruenbacher
2005-01-22 20:34 ` [patch 6/13] Lazy RPC receive buffer allocation Andreas Gruenbacher
2005-01-22 20:34 ` [patch 7/13] Encode and decode arbitrary XDR arrays Andreas Gruenbacher
2005-02-15 19:17 ` Trond Myklebust
2005-02-16 16:08 ` Andreas Gruenbacher
2005-02-17 14:12 ` Adrian Bunk
2005-01-22 20:34 ` [patch 8/13] Add noacl nfs mount option Andreas Gruenbacher
2005-02-15 17:24 ` Trond Myklebust
2005-02-16 16:10 ` Andreas Gruenbacher
2005-01-22 20:34 ` [patch 9/13] Infrastructure and server side of nfsacl Andreas Gruenbacher
2005-01-22 20:34 ` [patch 10/13] Solaris nfsacl workaround Andreas Gruenbacher
2005-02-15 17:29 ` Trond Myklebust
2005-02-15 20:35 ` Olivier Galibert
2005-02-15 22:43 ` Trond Myklebust
2005-02-15 23:02 ` Olivier Galibert
2005-02-15 23:37 ` Trond Myklebust
2005-02-15 23:43 ` Olivier Galibert
2005-02-16 16:17 ` Andreas Gruenbacher
2005-02-16 17:05 ` Trond Myklebust
2005-02-16 17:39 ` Andreas Gruenbacher
2005-01-22 20:34 ` [patch 11/13] Client side of nfsacl Andreas Gruenbacher
2005-02-15 17:49 ` Trond Myklebust
2005-02-22 13:41 ` Andreas Gruenbacher
2005-02-22 14:13 ` Trond Myklebust
2005-01-22 20:34 ` [patch 12/13] ACL umask handling workaround in nfs client Andreas Gruenbacher
2005-01-25 1:20 ` Andreas Gruenbacher
2005-02-15 18:04 ` Trond Myklebust
2005-02-22 16:47 ` Andreas Gruenbacher [this message]
2005-02-22 17:43 ` Trond Myklebust
2005-01-22 20:34 ` [patch 13/13] Cache acls on the nfs client side Andreas Gruenbacher
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=1109090843.6102.443.camel@winden.suse.de \
--to=agruen@suse.de \
--cc=Andries.Brouwer@cwi.nl \
--cc=akpm@osdl.org \
--cc=linux-kernel@vger.kernel.org \
--cc=neilb@cse.unsw.edu.au \
--cc=okir@suse.de \
--cc=trond.myklebust@fys.uio.no \
/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
Powered by JetHome