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:   Gregory Price <gourry.memverge@gmail.com>
To:     linux-kernel@vger.kernel.org
Cc:     linux-doc@vger.kernel.org, linux-arch@vger.kernel.org,
        oleg@redhat.com, avagin@gmail.com, peterz@infradead.org,
        luto@kernel.org, krisman@collabora.com, tglx@linutronix.de,
        corbet@lwn.net, shuah@kernel.org, catalin.marinas@arm.com,
        arnd@arndb.de, Gregory Price <gregory.price@memverge.com>
Subject: [PATCH v16 2/4] syscall user dispatch: untag selector addresses before access_ok
Date:   Fri,  7 Apr 2023 13:18:32 -0400
Message-Id: <20230407171834.3558-3-gregory.price@memverge.com>
In-Reply-To: <20230407171834.3558-1-gregory.price@memverge.com>
References: <20230407171834.3558-1-gregory.price@memverge.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

This is a preparatory patch for enabling checkpoint/restart of tasks
utilizing syscall user dispatch via ptrace.

To support checkpoint/restart, ptrace must be able to set the selector
of the tracee.  The selector is a user pointer that may be subject to
memory tagging extensions on some architectures (namely ARM MTE).

access_ok will clear memory tags for tagged addresses on tasks where
memory tagging is enabled.  However, to allow ptrace to set a task's
selector when tracer and tracee are not both tagged or untagged,
the selector address must be untagged when calling access_ok.

Since access_ok utilizes current to determine whether or not to untag
an address, an untagged tracer will always fail to restore a tagged
address in a tagged tracee.  This patch will resolve this issue.

The result of this is that a tagged tracer may be capable of setting
an invalid address, which will cause the tracee to SIGSEGV on next
syscall.  This is equivalent to the tracee setting a bad selector
address (such as selector=0x1).  This is preferable to the alternative
of creating a task_access_ok variant, and is consistent with other
operations which change tracee pointers via ptrace.

For more information, see:
https://lore.kernel.org/all/ZCWXE04nLZ4pXEtM@arm.com/

Signed-off-by: Gregory Price <gregory.price@memverge.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Suggested-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
---
 kernel/entry/syscall_user_dispatch.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/kernel/entry/syscall_user_dispatch.c b/kernel/entry/syscall_user_dispatch.c
index 22396b234854..424f24350f8b 100644
--- a/kernel/entry/syscall_user_dispatch.c
+++ b/kernel/entry/syscall_user_dispatch.c
@@ -87,7 +87,14 @@ static int task_set_syscall_user_dispatch(struct task_struct *task, unsigned lon
 		if (offset && offset + len <= offset)
 			return -EINVAL;
 
-		if (selector && !access_ok(selector, sizeof(*selector)))
+		/*
+		 * access_ok will clear memory tags for tagged addresses on tasks where
+		 * memory tagging is enabled.  To enable a tracer to set a tracee's
+		 * selector not in the same tagging state, the selector address must be
+		 * untagged for access_ok, otherwise an untagged tracer will always fail
+		 * to set a tagged tracee's selector.
+		 */
+		if (selector && !access_ok(untagged_addr(selector), sizeof(*selector)))
 			return -EFAULT;
 
 		break;
-- 
2.39.1