8250: Eliminate compile warning of 8250.c

The patch based on linux-2.6 upstream and elimiate this compile warning info:

drivers/serial/8250.c: In function ‘serial8250_shutdown’:
drivers/serial/8250.c:1612: warning: ‘i’ may be used uninitialized in this function

---
 drivers/serial/8250.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

index 303272a..c78eee6 100644
@@ -1609,7 +1609,7 @@ static int serial_link_irq_chain(struct uart_8250_port *up)
 
 static void serial_unlink_irq_chain(struct uart_8250_port *up)
 {
-	struct irq_info *i;
+	struct irq_info *i = NULL;
 	struct hlist_node *n;
 	struct hlist_head *h;
 

1.5.4.5

--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

(Something wrong with encoding of the first mail...)

The patch based on linux-2.6 upstream and elimiate this compile warning info:

drivers/serial/8250.c: In function 'serial8250_shutdown':
drivers/serial/8250.c:1612: warning: 'i' may be used uninitialized in this function

---
 drivers/serial/8250.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

index 303272a..c78eee6 100644
@@ -1609,7 +1609,7 @@ static int serial_link_irq_chain(struct uart_8250_port *up)
 
 static void serial_unlink_irq_chain(struct uart_8250_port *up)
 {
-	struct irq_info *i;
+	struct irq_info *i = NULL;
 	struct hlist_node *n;
 	struct hlist_head *h;
 

1.5.4.5

NAK - again.

There is no real bug here and you risk hiding future ones. It also doesn't produce a warning with a modern gcc.

Alan

Can't understand initializing got "risk hiding future ones". As I know, didn't initialize the variable got risk hiding future random problems. If you think i=NULL is not enough, how about this? I don't use this version just because I think BUG_ON(n==null) cover this. And I don't think leave it to compiler is more proper here.

--
From: Sheng Yang <[email protected]>
Date: Tue, 25 Nov 2008 18:25:26 +0800
Subject: [PATCH 1/1] 8250: Eliminate compile warning of 8250.c

The patch based on linux-2.6 upstream and elimiate this compile warning info:

drivers/serial/8250.c: In function 'serial8250_shutdown':
drivers/serial/8250.c:1612: warning: 'i' may be used uninitialized in this function

---
 drivers/serial/8250.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

index 303272a..c3120be 100644
@@ -1609,7 +1609,7 @@ static int serial_link_irq_chain(struct uart_8250_port 

*up)

static void serial_unlink_irq_chain(struct uart_8250_port *up) {
- struct irq_info *i;
+ struct irq_info *i = NULL;
struct hlist_node *n;
struct hlist_head *h;

@@ -1624,6 +1624,7 @@ static void serial_unlink_irq_chain(struct uart_8250_port *up)
}

BUG_ON(n == NULL);
+ BUG_ON(i == NULL);
BUG_ON(i->head == NULL);

if (list_empty(i->head))

1.5.4.5

Because if you set it to NULL and later delete a line which does the intended assignment you will no longer get a warning.

The BUG_ON covers it, the current gcc gets this right and works it out.

NAK again

The logic here is strange...

If you used old compiler, you would get a warning, and you thought that's ensured we won't delete a intended assignment by mistake.

If you used new compiler, you wouldn't get a warning, and you think this time the compiler get it right.

So, what result did you expect? A warning to notice that we didn't delete a intended assignment, or a "right" result?

OK. if you worry about "delete a line which does the intended assignment", the updated version with BUG_ON(i == NULL) can help.

regards
Yang, Sheng

Why

No.. if you hid the warning by hacking around the compiler mistake with extra bogus assignments you would hide a future mistake.

The compiler does now get it right - I checked.

Alan

OK, I got it now.

Comparing the result of deleting a meaningful line, this modification result in no more warning. And you also think one additional BUG_ON just for fixing compiler's mistake, so it's unacceptable.

So leave the patch as it was. Seems many people need to upgrade compiler to avoid this thing. :)

regards
Yang, Sheng

The new compiler gets it right. If you delete the assignment with the new compiler, *then* you get a warning.

Compile-time error detection is better than runtime.

OG.