Re: [BUG] Bad #define, nonportable C, missing {}

[Andreas Schwab <schwab@suse.de>]

vda <vda@port.imtp.ilyichevsk.odessa.ua> writes:

|> Upon random browsing in the kernel tree I noticed in accel.c:
|>     *a++ = byte_rev[*a]
|> which isn't 100% correct C AFAIK. At least Stroustrup in his C++ book
|> warns that this kind of code has to be avoided.

This is definitely causing undefined behaviour.  AFAIK, gcc 3.1 (current
CVS version) can warn about such errors (-Wsequence-point).

|> ======= bad_c.diff =======
|> diff -u --recursive linux-2.4.13-old/arch/mips/lib/tinycon.c 
|> linux-2.4.13-new/arch/mips/lib/tinycon.c
|> --- linux-2.4.13-old/arch/mips/lib/tinycon.c	Thu Jun 26 17:33:37 1997
|> +++ linux-2.4.13-new/arch/mips/lib/tinycon.c	Wed Nov 21 00:54:05 2001
|> @@ -83,14 +83,18 @@
|>    register int i;
|>  
|>    caddr = vram_addr;
|> -  for(i=0; i<size_x * (size_y-1); i++)
|> -    *(caddr++) = *(caddr + size_x);
|> +  for(i=0; i<size_x * (size_y-1); i++) {
|> +    *caddr = *(caddr + size_x);
|> +    caddr++;
|> +  }

Alternatively you can write:

  for(i=0; i<size_x * (size_y-1); i++, caddr++)
    *caddr = *(caddr + size_x);

or even:

  for(i=0; i<size_x * (size_y-1); i++, caddr++)
    *caddr = caddr[size_x];

Andreas.

-- 
Andreas Schwab                                  "And now for something
Andreas.Schwab@suse.de				completely different."
SuSE Labs, SuSE GmbH, Schanzäckerstr. 10, D-90443 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5