| Message ID | 20140726162143.GB20396@n2100.arm.linux.org.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Thanks Russell, the patch looks good. Andrew sent the same patch just a little earlier as well. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sun, Jul 27, 2014 at 9:19 AM, Christoph Hellwig <hch@infradead.org> wrote:
> Thanks Russell, the patch looks good.
No. It needs to use an IS_ERR_OR_NULL() test at least in order to work
correctly.
On Sun, Jul 27, 2014 at 11:15:40AM -0400, Trond Myklebust wrote: > On Sun, Jul 27, 2014 at 9:19 AM, Christoph Hellwig <hch@infradead.org> wrote: > > Thanks Russell, the patch looks good. > > No. It needs to use an IS_ERR_OR_NULL() test at least in order to work > correctly. What's happening on this? Did this get fixed for v3.16? My patch still seems to apply to v3.16 so I guess NFS is broken there...
On Wed, Aug 06, 2014 at 11:34:41PM +0100, Russell King - ARM Linux wrote:
> What's happening on this? Did this get fixed for v3.16?
Trond has Andrey's patch queued up for 3.17 with a Cc to stable, so it
will go into a 3.16 stable kernel as soon as it hits Linus' tree.
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/nfs/nfs3acl.c b/fs/nfs/nfs3acl.c index 8f854dde4150..ada8c66aa9e5 100644 --- a/fs/nfs/nfs3acl.c +++ b/fs/nfs/nfs3acl.c @@ -256,7 +256,7 @@ nfs3_list_one_acl(struct inode *inode, int type, const char *name, void *data, char *p = data + *result; acl = get_acl(inode, type); - if (!acl) + if (IS_ERR(acl)) return 0; posix_acl_release(acl);
Today, I got a nice oops while trying to update my initramfs: Alignment trap: not handling instruction e1901f9f at [<c020b8ec>] Unhandled fault: alignment exception (0x001) at 0xffffffa1 Internal error: : 1 [#1] SMP ARM Modules linked in: bnep rfcomm bluetooth nfsd exportfs hid_cypress brcmfmac brcmutil snd_soc_fsl_spdif imx_pcm_dma imx2_wdt imx_thermal imx_sdma snd_soc_imx_spdif CPU: 2 PID: 1704 Comm: cp Not tainted 3.16.0-rc6+ #1281 task: e9be5580 ti: e3ce6000 task.ti: e3ce6000 PC is at nfs3_list_one_acl+0x38/0xa0 LR is at get_acl+0x4c/0x60 pc : [<c020b8f0>] lr : [<c0144774>] psr: a00d0013 sp : e3ce7f08 ip : e3ce7ee8 fp : e3ce7f2c r10: 00000000 r9 : e3ce6000 r8 : 00000000 r7 : 00000000 r6 : c06d728c r5 : 00000000 r4 : e3ce7f3c r3 : 00000000 r2 : e3ce7ea4 r1 : 00003fe7 r0 : ffffffa1 Flags: NzCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user Control: 10c5387d Table: 33dec04a DAC: 00000015 Process cp (pid: 1704, stack limit = 0xe3ce6248) Stack: (0xe3ce7f08 to 0xe3ce8000) 7f00: 00000000 e3ce7f3c 00000000 e9ff3858 00000000 00000000 7f20: e3ce7f5c e3ce7f30 c020c0f8 c020b8c4 00000000 e3ce7f3c 00000000 00000000 7f40: 00000000 c020c0b4 00000000 e8c05198 e3ce7f84 e3ce7f60 c011a0d0 c020c0c0 7f60: e3d04600 e3d04600 00000000 00000000 c000ec04 00000000 e3ce7fa4 e3ce7f88 7f80: c011aed8 c011a07c 00008000 be9995a8 b6f756db 000000ea 00000000 e3ce7fa8 7fa0: c000ea80 c011aeb0 00008000 be9995a8 00000003 00000000 00000000 00022150 7fc0: 00008000 be9995a8 b6f756db 000000ea be9994d8 b6f10d99 00000003 be9995a8 7fe0: be9993a0 be999390 b6f105db b6ea8ca0 800d0010 00000003 6f74616c 69662d72 Backtrace: [<c020b8b8>] (nfs3_list_one_acl) from [<c020c0f8>] (nfs3_listxattr+0x44/0x80) [<c020c0b4>] (nfs3_listxattr) from [<c011a0d0>] (listxattr+0x60/0x11c) [<c011a070>] (listxattr) from [<c011aed8>] (SyS_flistxattr+0x34/0x58) [<c011aea4>] (SyS_flistxattr) from [<c000ea80>] (ret_fast_syscall+0x0/0x30) Code: 0a000016 f57ff05b f590f000 e1901f9f (e2411001) ---[ end trace 05ecbdab16531f0b ]--- The problem is that get_acl() does *not* return NULL on error, it returns an error code. Hence this: + acl = get_acl(inode, type); + if (!acl) + return 0; introduced by 74adf83f5d77 ("nfs: only show Posix ACLs in listxattr if actually present") ends up breaking when get_acl() fails (as is the case if NFS is used against a server with ACL support disabled.) The original commit was marked for stable, which means that this brokenness is going to spread to all stable kernels. Fixes: 74adf83f5d77 ("nfs: only show Posix ACLs in listxattr if actually present") Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Cc: <stable@vger.kernel.org> Cc: Trond Myklebust <trond.myklebust@primarydata.com> --- I've included <stable@vger.kernel.org> in the Cc list so they can hold off spreading the broken patch further into stable trees until this issue is resolved. fs/nfs/nfs3acl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)