Re: [Patch] Signedness issue in drivers/net/3c515.c

Re: [Patch] Signedness issue in drivers/net/3c515.c

[Andrew Morton]

On Sat, 19 Aug 2006 19:37:57 +0200
Eric Sesterhenn <snakebyte@gmx.de> wrote:

> while playing with gcc 4.1 -Wextra warnings, I came across this one:
> 
> drivers/net/3c515.c:1027: warning: comparison of unsigned expression >= 0 is always true
> 
> Since i is unsigned the >= 0 check in the for loop is always true,
> so we might spin there forever unless the if condition triggers.
> Since i is only used in this loop, this patch changes it to
> an integer.
> 
> Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
> 
> --- linux-2.6.18-rc4/drivers/net/3c515.c.orig	2006-08-19 19:35:04.000000000 +0200
> +++ linux-2.6.18-rc4/drivers/net/3c515.c	2006-08-19 19:35:14.000000000 +0200
> @@ -1003,7 +1003,8 @@ static int corkscrew_start_xmit(struct s
>  		/* Calculate the next Tx descriptor entry. */
>  		int entry = vp->cur_tx % TX_RING_SIZE;
>  		struct boom_tx_desc *prev_entry;
> -		unsigned long flags, i;
> +		unsigned long flags;
> +		int i;
>  
>  		if (vp->tx_full)	/* No room to transmit with */
>  			return 1;

Which affects this loop:

	/* Wait for the stall to complete. */
	for (i = 20; i >= 0; i--)
		if ((inw(ioaddr + EL3_STATUS) & CmdInProgress) == 0) 
			break;

Your fix will convert this indefinit wait into a bounded one.  It might
cause the driver to malfunction.

Given that our pool of 3c515 testers is less than enormous, a more prudent
change might be to remove `i' and simply formalise the existing behaviour
into a while(1) loop.

Re: [Patch] Signedness issue in drivers/net/3c515.c

[Alan Cox]

Ar Llu, 2006-08-21 am 14:05 -0700, ysgrifennodd Andrew Morton:
> 	/* Wait for the stall to complete. */
> 	for (i = 20; i >= 0; i--)
> 		if ((inw(ioaddr + EL3_STATUS) & CmdInProgress) == 0) 
> 			break;
> 
> Your fix will convert this indefinit wait into a bounded one.  It might
> cause the driver to malfunction.

The change is correct. The docs guarantee it can't take that long.