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;
Date:   Fri,  5 Aug 2022 22:21:23 +0000
In-Reply-To: <20220805222126.142525-1-jeffxu@google.com>
Message-Id: <20220805222126.142525-3-jeffxu@google.com>
Mime-Version: 1.0
References: <20220805222126.142525-1-jeffxu@google.com>
Subject: [PATCH v2 2/5] mm/memfd: add MFD_NOEXEC flag to memfd_create
From:   <jeffxu@google.com>
To:     skhan@linuxfoundation.org
Cc:     akpm@linux-foundation.org, dmitry.torokhov@gmail.com,
        dverkamp@chromium.org, hughd@google.com, jeffxu@google.com,
        jorgelo@chromium.org, keescook@chromium.org,
        linux-kernel@vger.kernel.org, linux-kselftest@vger.kernel.org,
        linux-mm@kvack.org, mnissler@chromium.org, jannh@google.com,
        Jeff Xu <jeffxu@chromium.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

From: Daniel Verkamp <dverkamp@chromium.org>

The new MFD_NOEXEC flag allows the creation of a permanently
non-executable memfd. This is accomplished by creating it with a
different set of file mode bits (0666) than the default (0777) and
applying the F_SEAL_EXEC seal at creation time, so there is no window
between memfd creation and seal application.

Unfortunately, the default for memfd must remain executable, since
changing this would be an API break, and some programs depend on being
able to exec code from a memfd directly. However, this new flag will
allow programs to create non-executable memfds, and a distribution may
choose to enforce use of this flag in memfd_create calls via other
security mechanisms.

Co-developed-by: Jeff Xu <jeffxu@chromium.org>
Signed-off-by: Jeff Xu <jeffxu@chromium.org>
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
---
 include/uapi/linux/memfd.h |  1 +
 mm/memfd.c                 | 10 +++++++++-
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/include/uapi/linux/memfd.h b/include/uapi/linux/memfd.h
index 7a8a26751c23..140e125c9f65 100644
--- a/include/uapi/linux/memfd.h
+++ b/include/uapi/linux/memfd.h
@@ -8,6 +8,7 @@
 #define MFD_CLOEXEC		0x0001U
 #define MFD_ALLOW_SEALING	0x0002U
 #define MFD_HUGETLB		0x0004U
+#define MFD_NOEXEC              0x0008U
 
 /*
  * Huge page size encoding when MFD_HUGETLB is specified, and a huge page
diff --git a/mm/memfd.c b/mm/memfd.c
index 4ebeab94aa74..b841514eb0fd 100644
--- a/mm/memfd.c
+++ b/mm/memfd.c
@@ -263,7 +263,7 @@ long memfd_fcntl(struct file *file, unsigned int cmd, unsigned long arg)
 #define MFD_NAME_PREFIX_LEN (sizeof(MFD_NAME_PREFIX) - 1)
 #define MFD_NAME_MAX_LEN (NAME_MAX - MFD_NAME_PREFIX_LEN)
 
-#define MFD_ALL_FLAGS (MFD_CLOEXEC | MFD_ALLOW_SEALING | MFD_HUGETLB)
+#define MFD_ALL_FLAGS (MFD_CLOEXEC | MFD_ALLOW_SEALING | MFD_HUGETLB | MFD_NOEXEC)
 
 SYSCALL_DEFINE2(memfd_create,
 		const char __user *, uname,
@@ -333,6 +333,14 @@ SYSCALL_DEFINE2(memfd_create,
 		*file_seals &= ~F_SEAL_SEAL;
 	}
 
+	if (flags & MFD_NOEXEC) {
+		struct inode *inode = file_inode(file);
+
+		inode->i_mode &= ~0111;
+		file_seals = memfd_file_seals_ptr(file);
+		*file_seals |= F_SEAL_EXEC;
+	}
+
 	fd_install(fd, file);
 	kfree(name);
 	return fd;
-- 
2.37.1.559.g78731f0fdb-goog