On Thu, Jan 24, 2008 at 01:18:48AM +0100, Roel Kluin wrote:
> Joe Perches wrote:
> > On Wed, 2008-01-23 at 23:37 +0100, Roel Kluin wrote:
> >> Signed-off-by: Roel Kluin <[EMAIL PROTECTED]>
> >> ---
> >> diff --git a/arch/powerpc/boot/4xx.c b/arch/powerpc/boot/4xx.c
> >> index ebf9e21..dcfb459 100644
> >> --- a/arch/powerpc/boot/4xx.c
> >> +++ b/arch/powerpc/boot/4xx.c
> >> @@ -104,7 +104,7 @@ void ibm4xx_denali_fixup_memsize(void)
> >> val = DDR_GET_VAL(val, DDR_CS_MAP, DDR_CS_MAP_SHIFT);
> >> cs = 0;
> >> while (val) {
> >> - if (val && 0x1)
> >> + if (val & 0x1)
> >> cs++;
> >> val = val >> 1;
> >
> > I think this pattern should be added to checkpatch
>
> I agree but...
>
> > Signed-off-by: Joe Perches <[EMAIL PROTECTED]>
> >
> > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> > index 579f50f..147e573 100755
> > --- a/scripts/checkpatch.pl
> > +++ b/scripts/checkpatch.pl
> > @@ -1337,6 +1337,11 @@ sub process {
> > }
> > }
> >
> > +# Check for bitwise tests written as boolean
> > + if ($line =~ /\&\&\s*0[xX]/) {
> > + WARN("boolean test with hexadecimal, perhaps just 1
> > \&?\n" . $herecurr);
> > + }
> > +
>
> when you use git-grep -n "\(&&\|||\)${s}0x\([A-Z0-9]*\|[a-z0-9]*\)",
> (with s="[[:space:]]*") there will be false positives like
>
> if (relocation < -0x20000 || 0x1fffc < relocation)
> if (ARCH_SUN4C_SUN4 && addr < 0xe0000000 && 0x20000000 - len < addr) {
> (and more)
>
> However, if you search with
> git-grep -n " \(&&\|||\)${s}0x\([A-Z0-9]*\|[a-z0-9]*\)$s\()\|&&\|||\)"
> you won't get these false positives, but you do get the one I fixed.
>
> Also there is the same logic flipped (though it did not give matches at
> this time):
> git-grep -n " \(&&\|||\|(\)${s}0x\([A-Z0-9]*\|[a-z0-9]*\)$s\(&&\|||\)"
>
> so i'd propose to change that to
> --
> Catch boolean tests with hexadecimals, based on suggestion by Joe Perches
>
> Signed-off-by: Roel Kluin <[EMAIL PROTECTED]>
> ---
> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> index 579f50f..8ac7c7b 100755
> --- a/scripts/checkpatch.pl
> +++ b/scripts/checkpatch.pl
> @@ -1337,6 +1337,12 @@ sub process {
> }
> }
>
> +# Check for bitwise tests written as boolean
> + if ($line =~ /(?:(?:\(|\&\&|\|\|)\s*0[xX]\s*(?:&&|\|\|)|
> + (?:\&\&|\|\|)\s*0[xX]\s*(?:\)|&&|\|\|))/) {
> + WARN("boolean test with hexadecimal, perhaps just 1 \&
> or \|?\n" . $herecurr);
> + }
> +
> # if and else should not have general statements after it
> if ($line =~ /^.\s*(?:}\s*)?else\b(.*)/ &&
> $1 !~ /^\s*(?:\sif|{|\\|$)/) {
That one doesn't even match the original example. Seems to be missing
some number matching. The concept seems sound though. Basically
looking for numbers which are definatly adjacent to a boolean or a brace
on both sides.
/me goes play with some examples to see if the falsies are better than
the reports.
-apw
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
