Received-SPF: pass (google.com: domain of linux-kernel+bounces-924-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
From: Petr Tesarik <petrtesarik@huaweicloud.com>
To: Richard Weinberger <richard@nod.at>,
	Anton Ivanov <anton.ivanov@cambridgegreys.com>,
	Johannes Berg <johannes@sipsolutions.net>,
	linux-um@lists.infradead.org (open list:USER-MODE LINUX (UML)),
	linux-kernel@vger.kernel.org (open list)
Cc: Roberto Sassu <roberto.sassu@huaweicloud.com>,
	petr@tesarici.cz,
	Petr Tesarik <petr.tesarik1@huawei-partners.com>
Subject: [PATCH 1/2] um: do not panic on kernel mode faults
Date: Fri, 15 Dec 2023 13:14:30 +0100
Message-Id: <20231215121431.680-2-petrtesarik@huaweicloud.com>
In-Reply-To: <20231215121431.680-1-petrtesarik@huaweicloud.com>
References: <20231215121431.680-1-petrtesarik@huaweicloud.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

From: Petr Tesarik <petr.tesarik1@huawei-partners.com>

Do not call panic() on unrecoverable page faults in kernel mode. Although
such page faults always indicate a bug in the kernel, other architectures
prefer to kill only the current process and continue.

The new behavior is useful for testing intentional kernel mode page faults
with KUnit.

Signed-off-by: Petr Tesarik <petr.tesarik1@huawei-partners.com>
---
 arch/um/kernel/trap.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/arch/um/kernel/trap.c b/arch/um/kernel/trap.c
index 6d8ae86ae978..1124a382fd14 100644
--- a/arch/um/kernel/trap.c
+++ b/arch/um/kernel/trap.c
@@ -17,6 +17,14 @@
 #include <os.h>
 #include <skas.h>
 
+static void page_fault_oops(struct uml_pt_regs *regs, unsigned long address,
+			    unsigned long ip)
+{
+	pr_alert("Kernel mode fault at addr 0x%lx, ip 0x%lx\n", address, ip);
+	show_regs(container_of(regs, struct pt_regs, regs));
+	make_task_dead(SIGKILL);
+}
+
 /*
  * Note this is constrained to return 0, -EFAULT, -EACCES, -ENOMEM by
  * segv().
@@ -249,11 +257,8 @@ unsigned long segv(struct faultinfo fi, unsigned long ip, int is_user,
 	else if (!is_user && arch_fixup(ip, regs))
 		goto out;
 
-	if (!is_user) {
-		show_regs(container_of(regs, struct pt_regs, regs));
-		panic("Kernel mode fault at addr 0x%lx, ip 0x%lx",
-		      address, ip);
-	}
+	if (!is_user)
+		page_fault_oops(regs, address, ip);
 
 	show_segv_info(regs);
 
-- 
2.34.1