Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Subject: [PATCH v1 0/5] Fix NFSv3 READDIRPLUS failures
From:   Chuck Lever <chuck.lever@oracle.com>
To:     linux-nfs@vger.kernel.org
Date:   Sun, 05 Jun 2022 15:58:19 -0400
Message-ID: <165445865736.1664.4497130284832282034.stgit@bazille.1015granger.net>
User-Agent: StGit/1.5
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
Precedence: bulk

While looking into the filecache CPU soft lock-up issue, I ran
across this problem. I thought I could run it down in just an
afternoon (I was wrong) and that this problem probably affects more
users than the soft lock-up (jury's still out).

Anyway, NFSD's new READDIRPLUS dirent encoder blows past the end of
the directory payload xdr_stream when the client requests more than
a page worth of directory entries. I tracked this down to how
xdr_get_next_encode_buffer() computes xdr->end. First patch in this
series is the fix. The remaining patches are clean-ups and
optimizations.

I want to get this into 5.19-rc quickly. I would appreciate getting
at least two R-b's for this series.

---

Chuck Lever (5):
      SUNRPC: Fix the calculation of xdr->end in xdr_get_next_encode_buffer()
      SUNRPC: Optimize xdr_reserve_space()
      SUNRPC: Clean up xdr_commit_encode()
      SUNRPC: Clean up xdr_get_next_encode_buffer()
      SUNRPC: Remove pointer type casts from xdr_get_next_encode_buffer()


 net/sunrpc/xdr.c | 31 ++++++++++++++++++++++---------
 1 file changed, 22 insertions(+), 9 deletions(-)

--
Chuck Lever