[PATCH] staging: greybus: use proper return type for wait_for_completion_timeout

[PATCH] staging: greybus: use proper return type for wait_for_completion_timeout

[Nicholas Mc Guire]

wait_for_completion_timeout() returns unsigned long (0 on timeout or
remaining jiffies) not int. 

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
---
Problem located with experimental API conformance checking cocci script

Patch was compile-tested with: x86_64_defconfig + GREYBUS=m

Patch is against 5.1-rc6 (localversion-next is next-20190426)

 drivers/staging/greybus/uart.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/staging/greybus/uart.c b/drivers/staging/greybus/uart.c
index b3bffe9..ff18112 100644
--- a/drivers/staging/greybus/uart.c
+++ b/drivers/staging/greybus/uart.c
@@ -323,7 +323,7 @@ static int send_break(struct gb_tty *gb_tty, u8 state)
 
 static int gb_uart_wait_for_all_credits(struct gb_tty *gb_tty)
 {
-	int ret;
+	unsigned long ret;
 
 	if (gb_tty->credits == GB_UART_FIRMWARE_CREDITS)
 		return 0;
-- 
2.1.4

Re: [PATCH] staging: greybus: use proper return type for wait_for_completion_timeout

[Dan Carpenter]

On Sat, Apr 27, 2019 at 05:27:25AM +0200, Nicholas Mc Guire wrote:
> wait_for_completion_timeout() returns unsigned long (0 on timeout or
> remaining jiffies) not int. 
> 

Yeah, but it's fine though because 10000 / 256 fits into int without a
problem.

I'm not sure this sort of patch is worth it when it's just a style
debate instead of a bugfix.  I'm a little bit torn about this.  In
Smatch, I run into this issue one in a while where Smatch doesn't know
if the timeout is less than int.  Right now I hacked the DB to say that
these functions always return < INT_MAX.

Anyway, for sure the commit message should say that it's just a cleanup
and not a bugfix.

regards,
dan carpenter

Re: [PATCH] staging: greybus: use proper return type for wait_for_completion_timeout

[Nicholas Mc Guire]

On Tue, Apr 30, 2019 at 12:58:21PM +0300, Dan Carpenter wrote:
> On Sat, Apr 27, 2019 at 05:27:25AM +0200, Nicholas Mc Guire wrote:
> > wait_for_completion_timeout() returns unsigned long (0 on timeout or
> > remaining jiffies) not int. 
> > 
> 
> Yeah, but it's fine though because 10000 / 256 fits into int without a
> problem.
> 
> I'm not sure this sort of patch is worth it when it's just a style
> debate instead of a bugfix.  I'm a little bit torn about this.  In
> Smatch, I run into this issue one in a while where Smatch doesn't know
> if the timeout is less than int.  Right now I hacked the DB to say that
> these functions always return < INT_MAX.
> 
> Anyway, for sure the commit message should say that it's just a cleanup
> and not a bugfix.
>
I know its not a functional bug its "only" an API violation - the problem
is more that code is often cut&past and at some point it may be a 
problem or someoe expects a negative return value without that this evef
can occure.

But yes - the commit message should have stated that this non-conformance
in this case has no effect - will resend.

thx!
hofrat