Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   "Fabio M. De Francesco" <fmdefrancesco@gmail.com>
To:     Miklos Szeredi <miklos@szeredi.hu>,
        Vivek Goyal <vgoyal@redhat.com>,
        Stefan Hajnoczi <stefanha@redhat.com>,
        Christian Brauner <brauner@kernel.org>,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
        virtualization@lists.linux-foundation.org, bpf@vger.kernel.org
Cc:     "Fabio M. De Francesco" <fmdefrancesco@gmail.com>,
        "Venkataramanan, Anirudh" <anirudh.venkataramanan@intel.com>,
        Ira Weiny <ira.weiny@intel.com>
Subject: [PATCH] fs/fuse: Replace kmap() with kmap_local_page()
Date:   Wed, 12 Oct 2022 13:23:23 +0200
Message-Id: <20221012112323.23283-1-fmdefrancesco@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

The use of kmap() is being deprecated in favor of kmap_local_page().

There are two main problems with kmap(): (1) It comes with an overhead as
the mapping space is restricted and protected by a global lock for
synchronization and (2) it also requires global TLB invalidation when the
kmap’s pool wraps and it might block when the mapping space is fully
utilized until a slot becomes available.

With kmap_local_page() the mappings are per thread, CPU local, can take
page faults, and can be called from any context (including interrupts).
It is faster than kmap() in kernels with HIGHMEM enabled. Furthermore,
the tasks can be preempted and, when they are scheduled to run again, the
kernel virtual addresses are restored and still valid.

Therefore, replace kmap() with kmap_local_page() in fuse_readdir_cached(), 
it being the only call site of kmap() currently left in fs/fuse.

Cc: "Venkataramanan, Anirudh" <anirudh.venkataramanan@intel.com>
Suggested-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
---

These changes are not tested in a 32 bits VM as I use to do with other more
problematic conversions. Mere code inspection makes me reasonably think
that the rules of local mappings are not violated by this conversion.

Furthermore, I have no idea how to test this code. If maintainers think
that tests are absolutely necessary, any hints about how to perform them
would be greatly appreciated.

 fs/fuse/readdir.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/fuse/readdir.c b/fs/fuse/readdir.c
index b4e565711045..9e40c19e90dc 100644
--- a/fs/fuse/readdir.c
+++ b/fs/fuse/readdir.c
@@ -539,9 +539,9 @@ static int fuse_readdir_cached(struct file *file, struct dir_context *ctx)
 	 * Contents of the page are now protected against changing by holding
 	 * the page lock.
 	 */
-	addr = kmap(page);
+	addr = kmap_local_page(page);
 	res = fuse_parse_cache(ff, addr, size, ctx);
-	kunmap(page);
+	kunmap_local(addr);
 	unlock_page(page);
 	put_page(page);
 
-- 
2.37.3