Re: [PATCH v2] scripts: checkpatch.pl: add 2 new checks on memset calls

[Aya Mahfouz]

On Tue, Feb 24, 2015 at 07:10:52PM -0800, Joe Perches wrote:
> On Wed, 2015-02-25 at 04:40 +0200, Aya Mahfouz wrote:
> > This patch adds 2 new checks on memset calls in the file
> > checkpatch.pl as follows:
> > 
> > replace memset by eth_zero_addr if the second argument is
> > an address of zeros (0x00). eth_zero_addr is a wrapper function
> > for memset that takes an address array to set as zero. The size
> > address has to be ETH_ALEN.
> 
> []
> > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> []
> > @@ -4901,10 +4901,22 @@ sub process {
> >  
> >  			if ($ms_size =~ /^(0x|)0$/i) {
> >  				ERROR("MEMSET",
> > -				      "memset to 0's uses 0 as the 2nd argument, not the 3rd\n" . "$here\n$stat\n");
> > +				      "memset to 0's uses 0 as the 2nd argument, not the 3rd\n" . "$here\n$line\n");
> >  			} elsif ($ms_size =~ /^(0x|)1$/i) {
> >  				WARN("MEMSET",
> > -				     "single byte memset is suspicious. Swapped 2nd/3rd argument?\n" . "$here\n$stat\n");
> > +				     "single byte memset is suspicious. Swapped 2nd/3rd argument?\n" . "$here\n$line\n");
> > +			} elsif ($ms_size =~ /^ETH_ALEN/i) {
> > +				if ($ms_val =~ /^0x00/i && WARN("PREFER_ETH_ZERO_ADDR",
> 
> This isn't right.  Look again at what I suggested.
> 
> This would match 0x00ff and wouldn't match 0
> 
> > +					 "Prefer eth_zero_addr() over memset() if the second address is 0x00\n" . $herecurr) &&
> > +				    $fix) {
> > +
> > +					$fixed[$fixlinenr] =~ s/\bmemset\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\,\s*ETH_ALEN\s*\)/eth_zero_addr($ms_addr)/;
> > +				} elsif ($ms_val =~ /^0xff/i && WARN("PREFER_ETH_BROADCAST_ADDR",
> > +					 "Prefer eth_broadcast_addr() over memset() if the second address is 0xff\n" . $herecurr) &&
> > +				    $fix) {
> > +
> > +					$fixed[$fixlinenr] =~ s/\bmemset\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\,\s*ETH_ALEN\s*\)/eth_broadcast_addr($ms_addr)/;
> > +				}
> >  			}
> >  		}
> >  
> 
> 
>

ok, I didn't see your suggestion, sorry. Can you look at the following
modification before sending the third patch? I don't use $stat because
I get false positives with it.

		#check for misused memsets
		if ($^V && $^V ge 5.10.0 &&
		    $line =~ /^\+(?:.*?)\bmemset\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\,\s*$FuncArg\s*\)/s) {

			my $ms_addr = $2;
			my $ms_val = $7;
			my $ms_size = $12;

			if ($ms_val =~ /^(?:0|0x0+)$/i &&
			    $ms_size =~ /^ETH_ALEN$/ &&
			    WARN("PREFER_ETH_ADDR_FUNC",
				 "Prefer eth_zero_addr() over memset() if the second address is 0x00\n" . $herecurr) &&
			    $fix) {
				$fixed[$fixlinenr] =~ s/\bmemcpy\s*\(\s*\Q$ms_addr\E\s*,\s*\Q$ms_val\E\s*\,\s*ETH_ALEN\s*\)/eth_zero_addr($ms_addr)/;
			} elsif ($ms_val =~ /^(?:0xff|255)$/i &&
				 $ms_size =~ /^ETH_ALEN$/ &&
				 WARN("PREFER_ETH_ADDR_FUNC",
				      "Prefer eth_broadcast_addr() over memset() if the second address is 0xff\n" . $herecurr) &&
				 $fix) {
				$fixed[$fixlinenr] =~ s/\bmemcpy\s*\(\s*\Q$ms_addr\E\s*,\s*\Q$ms_val\E\s*\,\s*ETH_ALEN\s*\)/eth_broadcast_addr($ms_addr)/;
			} elsif ($ms_size =~ /^(0x|)0$/i) {
				ERROR("MEMSET",
				      "memset to 0's uses 0 as the 2nd argument, not the 3rd\n" . "$here\n$line\n");
			} elsif ($ms_size =~ /^(0x|)1$/i) {
				WARN("MEMSET",
				     "single byte memset is suspicious. Swapped 2nd/3rd argument?\n" . "$here\n$line\n");
			}
		}

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/