diff mbox series

[5/9] nfsd: NFSv3 should allow zero length writes

Message ID 20211217215046.40316-6-trondmy@kernel.org (mailing list archive)
State New, archived
Headers show
Series Assorted patches for knfsd | expand

Commit Message

Trond Myklebust Dec. 17, 2021, 9:50 p.m. UTC
From: Trond Myklebust <trond.myklebust@hammerspace.com>

According to RFC1813: "If count is 0, the WRITE will succeed
and return a count of 0, barring errors due to permissions checking."

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfsd/vfs.c    | 3 +++
 net/sunrpc/svc.c | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

Comments

Bruce Fields Dec. 17, 2021, 10:23 p.m. UTC | #1
On Fri, Dec 17, 2021 at 5:07 PM <trondmy@kernel.org> wrote:
>
> From: Trond Myklebust <trond.myklebust@hammerspace.com>
>
> According to RFC1813: "If count is 0, the WRITE will succeed
> and return a count of 0, barring errors due to permissions checking."

Yes, I'm surprised we're not already doing this right.

I wonder how far back this bug goes.

The svc.c code is from 8154ef2776aa "NFSD: Clean up legacy NFS WRITE
argument XDR decoders", but the behavior might predate that code.  The
nfsd_vfs_write() logic, I'm not sure I understand.

We have a pynfs test for the v4 case (WRT4), but I guess we must have
nothing testing the v3 case.

--b.

>
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> ---
>  fs/nfsd/vfs.c    | 3 +++
>  net/sunrpc/svc.c | 2 +-
>  2 files changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
> index 4d57befdac6e..38fdfcbb079e 100644
> --- a/fs/nfsd/vfs.c
> +++ b/fs/nfsd/vfs.c
> @@ -969,6 +969,9 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
>
>         trace_nfsd_write_opened(rqstp, fhp, offset, *cnt);
>
> +       if (!*cnt)
> +               return nfs_ok;
> +
>         if (sb->s_export_op)
>                 exp_op_flags = sb->s_export_op->flags;
>
> diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
> index a3bbe5ce4570..d1ccf37a4588 100644
> --- a/net/sunrpc/svc.c
> +++ b/net/sunrpc/svc.c
> @@ -1692,7 +1692,7 @@ unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, struct page **pages,
>          * entirely in rq_arg.pages. In this case, @first is empty.
>          */
>         i = 0;
> -       if (first->iov_len) {
> +       if (first->iov_len || !total) {
>                 vec[i].iov_base = first->iov_base;
>                 vec[i].iov_len = min_t(size_t, total, first->iov_len);
>                 total -= vec[i].iov_len;
> --
> 2.33.1
>
Chuck Lever III Dec. 18, 2021, 6:41 p.m. UTC | #2
> On Dec 17, 2021, at 5:23 PM, Bruce Fields <bfields@redhat.com> wrote:
> 
> On Fri, Dec 17, 2021 at 5:07 PM <trondmy@kernel.org> wrote:
>> 
>> From: Trond Myklebust <trond.myklebust@hammerspace.com>
>> 
>> According to RFC1813: "If count is 0, the WRITE will succeed
>> and return a count of 0, barring errors due to permissions checking."
> 
> Yes, I'm surprised we're not already doing this right.
> 
> I wonder how far back this bug goes.
> 
> The svc.c code is from 8154ef2776aa "NFSD: Clean up legacy NFS WRITE
> argument XDR decoders", but the behavior might predate that code.  The
> nfsd_vfs_write() logic, I'm not sure I understand.

The new check in nfsd_vfs_write() might circumvent the VFS
layer's security checks, so I agree, that is a little suspect.


> We have a pynfs test for the v4 case (WRT4), but I guess we must have
> nothing testing the v3 case.

I guess cthon04 doesn't check this case.

But it seems to me WRT4 should already tickle any problems
with nfsd_vfs_write(), shouldn't it?


> --b.
> 
>> 
>> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
>> ---
>> fs/nfsd/vfs.c    | 3 +++
>> net/sunrpc/svc.c | 2 +-
>> 2 files changed, 4 insertions(+), 1 deletion(-)
>> 
>> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
>> index 4d57befdac6e..38fdfcbb079e 100644
>> --- a/fs/nfsd/vfs.c
>> +++ b/fs/nfsd/vfs.c
>> @@ -969,6 +969,9 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
>> 
>>        trace_nfsd_write_opened(rqstp, fhp, offset, *cnt);
>> 
>> +       if (!*cnt)
>> +               return nfs_ok;
>> +
>>        if (sb->s_export_op)
>>                exp_op_flags = sb->s_export_op->flags;
>> 
>> diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
>> index a3bbe5ce4570..d1ccf37a4588 100644
>> --- a/net/sunrpc/svc.c
>> +++ b/net/sunrpc/svc.c
>> @@ -1692,7 +1692,7 @@ unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, struct page **pages,
>>         * entirely in rq_arg.pages. In this case, @first is empty.
>>         */
>>        i = 0;
>> -       if (first->iov_len) {
>> +       if (first->iov_len || !total) {
>>                vec[i].iov_base = first->iov_base;
>>                vec[i].iov_len = min_t(size_t, total, first->iov_len);
>>                total -= vec[i].iov_len;
>> --
>> 2.33.1
>> 
> 

--
Chuck Lever
Bruce Fields Dec. 19, 2021, 10:25 p.m. UTC | #3
On Sat, Dec 18, 2021 at 1:42 PM Chuck Lever III <chuck.lever@oracle.com> wrote:
> But it seems to me WRT4 should already tickle any problems
> with nfsd_vfs_write(), shouldn't it?

I was just taking Trond's word that this is NFSv3-specific, but it's
not clear to me why from the code, and I know that WRT4 is passing.
Something's weird.  I'm travelling or I'd test it.

--b.
diff mbox series

Patch

diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 4d57befdac6e..38fdfcbb079e 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -969,6 +969,9 @@  nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct nfsd_file *nf,
 
 	trace_nfsd_write_opened(rqstp, fhp, offset, *cnt);
 
+	if (!*cnt)
+		return nfs_ok;
+
 	if (sb->s_export_op)
 		exp_op_flags = sb->s_export_op->flags;
 
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index a3bbe5ce4570..d1ccf37a4588 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -1692,7 +1692,7 @@  unsigned int svc_fill_write_vector(struct svc_rqst *rqstp, struct page **pages,
 	 * entirely in rq_arg.pages. In this case, @first is empty.
 	 */
 	i = 0;
-	if (first->iov_len) {
+	if (first->iov_len || !total) {
 		vec[i].iov_base = first->iov_base;
 		vec[i].iov_len = min_t(size_t, total, first->iov_len);
 		total -= vec[i].iov_len;