Skip to contentSkip to conversation

kgdboc,input: Fix regression with keyboard release key and early debugging

PATCH In review Jason Wessel 2 participants 2 messages Dec 10, 2010

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 kgdboc,input: Fix regression with keyboard release key and early debugging

=== NOTES excluded from the real commit ===

Dmitry, I added this commit into kgdb-next and I will send a pull request to Linus Monday, unless someone comes up with an objection or a different implementation.

Thanks,
Jason.

=== END NOTES ===

The commit 111c182340cd22e238ab1cc6564df336c6ebd7cb introduced a regression in early debugging such that you get a kernel oops on continue (with the go command) if you boot a kernel with:

earlyprintk=vga ekgdboc=kbd kgdbwait

The restore input state routine which intended to send release key events for any key that was depressed on entry to the debugger. A simple fix is to not invoke the schedule_work() if the kernel system_state is anything other than SYSTEM_RUNNING.

CC Dmitry Torokhov <[email protected]>
---
 drivers/serial/kgdboc.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
drivers/serial/kgdboc.c +2 -1
index 3374618..25a8bc5 100644
@@ -90,7 +90,8 @@ static DECLARE_WORK(kgdboc_restore_input_work, kgdboc_restore_input_helper);
 
 static void kgdboc_restore_input(void)
 {
-	schedule_work(&kgdboc_restore_input_work);
+	if (likely(system_state == SYSTEM_RUNNING))
+		schedule_work(&kgdboc_restore_input_work);
 }
 
 static int kgdboc_register_kbd(char **cptr)
Signature · 1 line
1.7.0.4
#1 lore raw
Quoted 5 lines · Jason Wessel

=== NOTES excluded from the real commit ===

Dmitry, I added this commit into kgdb-next and I will send a pull request to Linus Monday, unless someone comes up with an objection or a different implementation.

Yes, I think it makes sense. Such an early oopses are unlikely to have input devices fully bound anyway so they'll get "reset" natually.

Thanks.

Signature · 1 line
Dmitry