linux-kernel - [PATCH] x86/shstk: Use __force when casting away _

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Tue,  9 Jan 2024 14:46:19 -0800
From: Rick Edgecombe <rick.p.edgecombe@...el.com>
To: lkp@...el.com,
	x86@...nel.org
Cc: bp@...en8.de,
	dave.hansen@...ux.intel.com,
	keescook@...omium.org,
	linux-kernel@...r.kernel.org,
	oe-kbuild-all@...ts.linux.dev,
	rick.p.edgecombe@...el.com,
	yu-cheng.yu@...el.com
Subject: [PATCH] x86/shstk: Use __force when casting away __user

In setup_signal_shadow_stack() the kernel needs to push the restorer
address to the shadow stack. This involves writing the value of the
restorer pointer to the shadow stack. Since the restorer is defined as a
__user in struct k_sigaction, the __user needs to be casted away to read
the value. It is safe to do, because nothing is being dereferenced, and
the de-__user-ed value is not stashed in an accessible local variable
where it might accidentally be used for another purpose.

However, sparse warns about casting away __user. So use __force to
silence sparse and add a comment to explain why it is ok.

Fixes: 05e36022c054 ("x86/shstk: Handle signals for shadow stack")
Reported-by: kernel test robot <lkp@...el.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202401080003.duO4RmjK-lkp@intel.com/
Signed-off-by: Rick Edgecombe <rick.p.edgecombe@...el.com>
---
 arch/x86/kernel/shstk.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/shstk.c b/arch/x86/kernel/shstk.c
index 59e15dd8d0f8..7cc294482011 100644
--- a/arch/x86/kernel/shstk.c
+++ b/arch/x86/kernel/shstk.c
@@ -367,7 +367,11 @@ int setup_signal_shadow_stack(struct ksignal *ksig)

 	/* Push restorer address */
 	ssp -= SS_FRAME_SIZE;
-	err = write_user_shstk_64((u64 __user *)ssp, (u64)restorer);
+	/*
+	 * Use __force because this is just writing the address of the pointer
+	 * to the shadow stack, not dereferencing it.
+	 */
+	err = write_user_shstk_64((u64 __user *)ssp, (u64 __force)restorer);
 	if (unlikely(err))
 		return -EFAULT;

-- 
2.34.1