2021-09-30 20:07:40

by Trond Myklebust

Subject: [PATCH 1/2] nfsd4: Handle the NFSv4 READDIR 'dircount' hint being zero

From: Trond Myklebust <[email protected]>

RFC3530 notes that the 'dircount' field may be zero, in which case the
recommendation is to ignore it, and only enforce the 'maxcount' field.
In RFC5661, this recommendation to ignore a zero valued field becomes a
requirement.

Fixes: aee377644146 ("nfsd4: fix rd_dircount enforcement")
Cc: <[email protected]>
Signed-off-by: Trond Myklebust <[email protected]>
---
fs/nfsd/nfs4xdr.c | 19 +++++++++++--------
1 file changed, 11 insertions(+), 8 deletions(-)

diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index 7abeccb975b2..cf030ebe2827 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -3544,15 +3544,18 @@ nfsd4_encode_dirent(void *ccdv, const char *name, int namlen,
goto fail;
cd->rd_maxcount -= entry_bytes;
/*
- * RFC 3530 14.2.24 describes rd_dircount as only a "hint", so
- * let's always let through the first entry, at least:
+ * RFC 3530 14.2.24 describes rd_dircount as only a "hint", and
+ * notes that it could be zero. If it is zero, then the server
+ * should enforce only the rd_maxcount value.
*/
- if (!cd->rd_dircount)
- goto fail;
- name_and_cookie = 4 + 4 * XDR_QUADLEN(namlen) + 8;
- if (name_and_cookie > cd->rd_dircount && cd->cookie_offset)
- goto fail;
- cd->rd_dircount -= min(cd->rd_dircount, name_and_cookie);
+ if (cd->rd_dircount) {
+ name_and_cookie = 4 + 4 * XDR_QUADLEN(namlen) + 8;
+ if (name_and_cookie > cd->rd_dircount && cd->cookie_offset)
+ goto fail;
+ cd->rd_dircount -= min(cd->rd_dircount, name_and_cookie);
+ if (!cd->rd_dircount)
+ cd->rd_maxcount = 0;
+ }

cd->cookie_offset = cookie_offset;
skip_entry:
--
2.31.1

2021-10-01 16:10:18

by Chuck Lever III

Subject: Re: [PATCH 1/2] nfsd4: Handle the NFSv4 READDIR 'dircount' hint being zero

> On Sep 30, 2021, at 3:44 PM, [email protected] wrote:
>
> From: Trond Myklebust <[email protected]>
>
> RFC3530 notes that the 'dircount' field may be zero, in which case the
> recommendation is to ignore it, and only enforce the 'maxcount' field.
> In RFC5661, this recommendation to ignore a zero valued field becomes a
> requirement.
>
> Fixes: aee377644146 ("nfsd4: fix rd_dircount enforcement")
> Cc: <[email protected]>
> Signed-off-by: Trond Myklebust <[email protected]>

Pulled into the next 5.15-rc for NFSD. Thanks!


> ---
> fs/nfsd/nfs4xdr.c | 19 +++++++++++--------
> 1 file changed, 11 insertions(+), 8 deletions(-)
>
> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
> index 7abeccb975b2..cf030ebe2827 100644
> --- a/fs/nfsd/nfs4xdr.c
> +++ b/fs/nfsd/nfs4xdr.c
> @@ -3544,15 +3544,18 @@ nfsd4_encode_dirent(void *ccdv, const char *name, int namlen,
> goto fail;
> cd->rd_maxcount -= entry_bytes;
> /*
> - * RFC 3530 14.2.24 describes rd_dircount as only a "hint", so
> - * let's always let through the first entry, at least:
> + * RFC 3530 14.2.24 describes rd_dircount as only a "hint", and
> + * notes that it could be zero. If it is zero, then the server
> + * should enforce only the rd_maxcount value.
> */
> - if (!cd->rd_dircount)
> - goto fail;
> - name_and_cookie = 4 + 4 * XDR_QUADLEN(namlen) + 8;
> - if (name_and_cookie > cd->rd_dircount && cd->cookie_offset)
> - goto fail;
> - cd->rd_dircount -= min(cd->rd_dircount, name_and_cookie);
> + if (cd->rd_dircount) {
> + name_and_cookie = 4 + 4 * XDR_QUADLEN(namlen) + 8;
> + if (name_and_cookie > cd->rd_dircount && cd->cookie_offset)
> + goto fail;
> + cd->rd_dircount -= min(cd->rd_dircount, name_and_cookie);
> + if (!cd->rd_dircount)
> + cd->rd_maxcount = 0;
> + }
>
> cd->cookie_offset = cookie_offset;
> skip_entry:
> --
> 2.31.1
>

--
Chuck Lever