From: hooanon05@yahoo.co.jp Subject: Re: Problems with default ACLs over NFS Date: Sat, 28 Mar 2009 07:50:55 +0900 Message-ID: <18513.1238194255@jrobl> References: <20090323222427.00bb24e7@werewolf.home> <20090326170513.4e7d377a.akpm@linux-foundation.org> Cc: "\"=?ISO-8859-1?Q? \" J. A. _Magall=F3n \" ?=\"" , LKML , linux-nfs@vger.kernel.org, Trond.Myklebust@netapp.com To: Andrew Morton Return-path: In-Reply-To: <20090326170513.4e7d377a.akpm@linux-foundation.org> Sender: linux-kernel-owner@vger.kernel.org List-ID: > On Mon, 23 Mar 2009 22:24:27 +0100 "J.A. Magall__n" wrote: > > > Hi all... > > > > I have a strange problem with 2.6.29-rc8-git3. > > It looks like a kernel bug... ? > > > > I have a box with NFS autofs-mounted home directories. Server is a > > Solaris 10 system. This worked in previous kernel versions, so I > > don't blame Solaris, at this moment. > > > > With this acls set in my home: ::: > > I get this strange message in the linux client: > > > > cicely:~> mkdir kk > > mkdir: cannot create directory `kk': Invalid argument I met a similar problem. In linux-2.6.29 with CONFIG_NFS_V3_ACL enabled, when I executed "cp -p /local/ext2/file /nfsv3/file", I got EINVAL too. Strace-ing this cp, I found setxsttr() return EINVAL. Before 2.6.29, it was EOPNOTSUPPORTED. Tracing some kernel functions, I observed several thigs. - xdr_buf_subsegment() returned -1 due to 'len' is not 0. - xdr_encode_word() call from nfsacl_encode() got this -1, and nfsacl_encode/nfs3_xdr_setaclargs/nfs3_proc_setacls/nfs3_proc_setacl/nfs3_setxattr returned -EINVAL. Finally I am afraid this commit might be related to this issue. commit ae46141ff08f1965b17c531b571953c39ce8b9e2 Author: Trond Myklebust Date: Tue Mar 10 20:33:18 2009 -0400 NFSv3: Fix posix ACL code Fix a memory leak due to allocation in the XDR layer. In cases where the RPC call needs to be retransmitted, we end up allocating new pages without clearing the old ones. Fix this by moving the allocation into nfs3_proc_setacls(). Also fix an issue discovered by Kevin Rudd, whereby the amount of memory reserved for the acls in the xdr_buf->head was miscalculated, and causing corruption. J. R. Okajima