[printk,v3,06/40] um: kmsg_dump: only dump when no output console available

Message ID 20221107141638.3790965-7-john.ogness@linutronix.de
State New
Headers
Series reduce console_lock scope |

Commit Message

John Ogness Nov. 7, 2022, 2:16 p.m. UTC
  The initial intention of the UML kmsg_dumper is to dump the kernel
buffers to stdout if there is no console available to perform the
regular crash output.

However, if ttynull was registered as a console, no crash output was
seen. Commit e23fe90dec28 ("um: kmsg_dumper: always dump when not tty
console") tried to fix this by performing the kmsg_dump unless the
stdio console was behind /dev/console or enabled. But this allowed
kmsg dumping to occur even if other non-stdio consoles will output
the crash output. Also, a console being the driver behind
/dev/console has nothing to do with a crash scenario.

Restore the initial intention by dumping the kernel buffers to stdout
only if a non-ttynull console is registered and enabled. Also add
detailed comments so that it is clear why these rules are applied.

Signed-off-by: John Ogness <john.ogness@linutronix.de>
---
 arch/um/kernel/kmsg_dump.c | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)
  

Comments

Petr Mladek Nov. 8, 2022, 12:41 p.m. UTC | #1
On Mon 2022-11-07 15:22:04, John Ogness wrote:
> The initial intention of the UML kmsg_dumper is to dump the kernel
> buffers to stdout if there is no console available to perform the
> regular crash output.
> 
> However, if ttynull was registered as a console, no crash output was
> seen. Commit e23fe90dec28 ("um: kmsg_dumper: always dump when not tty
> console") tried to fix this by performing the kmsg_dump unless the
> stdio console was behind /dev/console or enabled. But this allowed
> kmsg dumping to occur even if other non-stdio consoles will output
> the crash output. Also, a console being the driver behind
> /dev/console has nothing to do with a crash scenario.
> 
> Restore the initial intention by dumping the kernel buffers to stdout
> only if a non-ttynull console is registered and enabled. Also add
> detailed comments so that it is clear why these rules are applied.
> 
> Signed-off-by: John Ogness <john.ogness@linutronix.de>

The change makes sense to me:

Reviewed-by: Petr Mladek <pmladek@suse.com>

Best Regards,
Petr
  

Patch

diff --git a/arch/um/kernel/kmsg_dump.c b/arch/um/kernel/kmsg_dump.c
index 0224fcb36e22..40abf1e9ccb1 100644
--- a/arch/um/kernel/kmsg_dump.c
+++ b/arch/um/kernel/kmsg_dump.c
@@ -17,13 +17,22 @@  static void kmsg_dumper_stdout(struct kmsg_dumper *dumper,
 	unsigned long flags;
 	size_t len = 0;
 
-	/* only dump kmsg when no console is available */
+	/*
+	 * If no consoles are available to output crash information, dump
+	 * the kmsg buffer to stdout.
+	 */
+
 	if (!console_trylock())
 		return;
 
 	for_each_console(con) {
-		if(strcmp(con->name, "tty") == 0 &&
-		   (con->flags & (CON_ENABLED | CON_CONSDEV)) != 0) {
+		/*
+		 * The ttynull console and disabled consoles are ignored
+		 * since they cannot output. All other consoles are
+		 * expected to output the crash information.
+		 */
+		if (strcmp(con->name, "ttynull") != 0 &&
+		    (con->flags & CON_ENABLED)) {
 			break;
 		}
 	}