Re: [PATCH v6 18/20] selftests/liveupdate: Add kexec-based selftest for session lifecycle

[David Matlack <dmatlack@google.com>]

On 2025-11-15 06:34 PM, Pasha Tatashin wrote:

> +/* Stage 1: Executed before the kexec reboot. */
> +static void run_stage_1(int luo_fd)
> +{
> +	int session_fd;
> +
> +	ksft_print_msg("[STAGE 1] Starting pre-kexec setup...\n");
> +
> +	ksft_print_msg("[STAGE 1] Creating state file for next stage (2)...\n");
> +	create_state_file(luo_fd, STATE_SESSION_NAME, STATE_MEMFD_TOKEN, 2);
> +
> +	ksft_print_msg("[STAGE 1] Creating session '%s' and preserving memfd...\n",
> +		       TEST_SESSION_NAME);
> +	session_fd = luo_create_session(luo_fd, TEST_SESSION_NAME);
> +	if (session_fd < 0)
> +		fail_exit("luo_create_session for '%s'", TEST_SESSION_NAME);
> +
> +	if (create_and_preserve_memfd(session_fd, TEST_MEMFD_TOKEN,
> +				      TEST_MEMFD_DATA) < 0) {
> +		fail_exit("create_and_preserve_memfd for token %#x",
> +			  TEST_MEMFD_TOKEN);
> +	}
> +
> +	ksft_print_msg("[STAGE 1] Executing kexec...\n");
> +	if (system(KEXEC_SCRIPT) != 0)
> +		fail_exit("kexec script failed");
> +	exit(EXIT_FAILURE);

Can we separate the kexec from the test and allow the user/automation to
trigger it however is appropriate for their system? The current
do_kexec.sh script does not do any sort of graceful shutdown, and I bet
everyone will have different ways of initiating kexec on their systems.

For example, something like this (but sleeping in the child instead of
busy waiting):

diff --git a/tools/testing/selftests/liveupdate/luo_kexec_simple.c b/tools/testing/selftests/liveupdate/luo_kexec_simple.c
index 67ab6ebf9eec..513693bfb77b 100644
--- a/tools/testing/selftests/liveupdate/luo_kexec_simple.c
+++ b/tools/testing/selftests/liveupdate/luo_kexec_simple.c
@@ -24,6 +24,7 @@
 static void run_stage_1(int luo_fd)
 {
 	int session_fd;
+	int ret;
 
 	ksft_print_msg("[STAGE 1] Starting pre-kexec setup...\n");
 
@@ -42,10 +43,17 @@ static void run_stage_1(int luo_fd)
 			  TEST_MEMFD_TOKEN);
 	}
 
-	ksft_print_msg("[STAGE 1] Executing kexec...\n");
-	if (system(KEXEC_SCRIPT) != 0)
-		fail_exit("kexec script failed");
-	exit(EXIT_FAILURE);
+	ksft_print_msg("[STAGE 1] Forking child process to hold session open\n");
+	ret = fork();
+	if (ret < 0)
+		fail_exit("fork() failed");
+	if (!ret)
+		for (;;) {}
+
+	ksft_print_msg("[STAGE 1] Child Process: %d\n", ret);
+	ksft_print_msg("[STAGE 1] Complete!\n");
+	ksft_print_msg("[STAGE 1] Execute kexec to continue\n");
+	exit(0);
 }
 
 /* Stage 2: Executed after the kexec reboot. */

> +int main(int argc, char *argv[])
> +{
> +	int luo_fd;
> +	int state_session_fd;
> +
> +	luo_fd = luo_open_device();
> +	if (luo_fd < 0)
> +		ksft_exit_skip("Failed to open %s. Is the luo module loaded?\n",
> +			       LUO_DEVICE);
> +
> +	/*
> +	 * Determine the stage by attempting to retrieve the state session.
> +	 * If it doesn't exist (ENOENT), we are in Stage 1 (pre-kexec).
> +	 */
> +	state_session_fd = luo_retrieve_session(luo_fd, STATE_SESSION_NAME);

I don't think the test should try to infer the stage from the state of
the system. If a user runs this test, then does the kexec, then runs
this test again and the session can't be retrieved, that should be a
test failure (not just run stage 1 again).

I think it'd be better to require the user to pass in what stage of the
test should be run when invoking the test. e.g.

 $ ./luo_kexec_simple stage_2