[PATCH] staging: panel: fix stackdump

[PATCH] staging: panel: fix stackdump

[Sudip Mukherjee]

if we load the module, unload and then again try to load the module, we
will get a stackdump. In the module_exit function we are unregistering
the device and releasing the parport. So when we reach the detach
function parport is already null and the unregister_reboot_notifier()
is never called. When we again try to load the module it again tries
register_reboot_notifier() and gives us a stackdump as its earlier
registration is still not removed.
Fix this by moving all the unregistering and releasing in the detach
function, which should be the ideal case as the detach will be called if
we try to unregister the driver or if the parport is removed.

Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
---

This was caused by one of my patch :(
Faced this problem while working on the device-model code of
parallelport.

 drivers/staging/panel/panel.c | 44 +++++++++++++++++++++----------------------
 1 file changed, 21 insertions(+), 23 deletions(-)

diff --git a/drivers/staging/panel/panel.c b/drivers/staging/panel/panel.c
index 1d8ed8b..0046ee0 100644
--- a/drivers/staging/panel/panel.c
+++ b/drivers/staging/panel/panel.c
@@ -2250,8 +2250,28 @@ static void panel_detach(struct parport *port)
 		       __func__, port->number, parport);
 		return;
 	}
+	if (scan_timer.function != NULL)
+		del_timer_sync(&scan_timer);
 
-	unregister_reboot_notifier(&panel_notifier);
+	if (pprt != NULL) {
+		if (keypad.enabled) {
+			misc_deregister(&keypad_dev);
+			keypad_initialized = 0;
+		}
+
+		if (lcd.enabled) {
+			panel_lcd_print("\x0cLCD driver " PANEL_VERSION
+					"\nunloaded.\x1b[Lc\x1b[Lb\x1b[L-");
+			misc_deregister(&lcd_dev);
+			lcd.initialized = false;
+		}
+
+		/* TODO: free all input signals */
+		parport_release(pprt);
+		parport_unregister_device(pprt);
+		pprt = NULL;
+		unregister_reboot_notifier(&panel_notifier);
+	}
 }
 
 static struct parport_driver panel_driver = {
@@ -2384,28 +2404,6 @@ static int __init panel_init_module(void)
 
 static void __exit panel_cleanup_module(void)
 {
-
-	if (scan_timer.function != NULL)
-		del_timer_sync(&scan_timer);
-
-	if (pprt != NULL) {
-		if (keypad.enabled) {
-			misc_deregister(&keypad_dev);
-			keypad_initialized = 0;
-		}
-
-		if (lcd.enabled) {
-			panel_lcd_print("\x0cLCD driver " PANEL_VERSION
-					"\nunloaded.\x1b[Lc\x1b[Lb\x1b[L-");
-			misc_deregister(&lcd_dev);
-			lcd.initialized = false;
-		}
-
-		/* TODO: free all input signals */
-		parport_release(pprt);
-		parport_unregister_device(pprt);
-		pprt = NULL;
-	}
 	parport_unregister_driver(&panel_driver);
 }
 
-- 
1.8.1.2

Re: [PATCH] staging: panel: fix stackdump

[Willy Tarreau]

On Mon, May 11, 2015 at 04:42:19PM +0530, Sudip Mukherjee wrote:
> if we load the module, unload and then again try to load the module, we
> will get a stackdump. In the module_exit function we are unregistering
> the device and releasing the parport. So when we reach the detach
> function parport is already null and the unregister_reboot_notifier()
> is never called. When we again try to load the module it again tries
> register_reboot_notifier() and gives us a stackdump as its earlier
> registration is still not removed.
> Fix this by moving all the unregistering and releasing in the detach
> function, which should be the ideal case as the detach will be called if
> we try to unregister the driver or if the parport is removed.
> 
> Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
> ---
> 
> This was caused by one of my patch :(
> Faced this problem while working on the device-model code of
> parallelport.

Introducing regressions (and fixing them) is part of the job :-)

However please mention in the commit message the commit of your other patch
which introduced the regression, and if you noticed it was backported to
stable, please mention it as well by a Cc: stable and the idea of the commit.
That will help everyone involved in the chain.

Thanks,
Willy