Latest update.

crypto/chacha/asm/chacha-armv4.pl

@@ -34,9 +34,10 @@
 #	but then Snapdragon S4 and Cortex-A8 results get
 #	20-25% worse;
 
-$flavour = shift;
-if ($flavour=~/\w[\w\-]*\.\w+$/) { $output=$flavour; undef $flavour; }
-else { while (($output=shift) && ($output!~/\w[\w\-]*\.\w+$/)) {} }
+# $output is the last argument if it looks like a file (it has an extension)
+# $flavour is the first argument if it doesn't look like a file
+$output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
+$flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : undef;
 
 if ($flavour && $flavour ne "void") {
     $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
@@ -44,9 +45,10 @@ if ($flavour && $flavour ne "void") {
     ( $xlate="${dir}../../perlasm/arm-xlate.pl" and -f $xlate) or
     die "can't locate arm-xlate.pl";
 
-    open STDOUT,"| \"$^X\" $xlate $flavour $output";
+    open STDOUT,"| \"$^X\" $xlate $flavour $output"
+        or die "can't call $xlate: $!";
 } else {
-    open STDOUT,">$output";
+    $output and open STDOUT,">$output";
 }
 
 sub AUTOLOAD()		# thunk [simplified] x86-style perlasm

crypto/chacha/asm/chacha-armv8.pl

@@ -44,15 +44,18 @@
 #
 # (*)	slower than 4+1:-(
 
-$flavour=shift;
-$output=shift;
+# $output is the last argument if it looks like a file (it has an extension)
+# $flavour is the first argument if it doesn't look like a file
+$output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
+$flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : undef;
 
 $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
 ( $xlate="${dir}arm-xlate.pl" and -f $xlate ) or
 ( $xlate="${dir}../../perlasm/arm-xlate.pl" and -f $xlate) or
 die "can't locate arm-xlate.pl";
 
-open OUT,"| \"$^X\" $xlate $flavour $output";
+open OUT,"| \"$^X\" $xlate $flavour \"$output\""
+    or die "can't call $xlate: $!";
 *STDOUT=*OUT;
 
 sub AUTOLOAD()		# thunk [simplified] x86-style perlasm

crypto/chacha/asm/chacha-c64xplus.pl

@@ -24,8 +24,7 @@
 # dependent on input length. This module on the other hand is free
 # from such limitation.
 
-$output=pop;
-open STDOUT,">$output";
+$output=pop and open STDOUT,">$output";
 
 ($OUT,$INP,$LEN,$KEYB,$COUNTERA)=("A4","B4","A6","B6","A8");
 ($KEYA,$COUNTERB,$STEP)=("A7","B7","A3");

crypto/chacha/asm/chacha-ia64.pl

@@ -15,8 +15,7 @@
 # pass runs slower than expected... Overall result is 15.6 cpb, two
 # cycles more than theoretical estimate.
 
-$output = pop;
-open STDOUT, ">$output" if $output;
+$output = pop and open STDOUT, ">$output";
 
 my @k = map("r$_",(16..31));
 my @x = map("r$_",(38..53));

crypto/chacha/asm/chacha-ppc.pl

@@ -44,7 +44,10 @@
 #	instructions, which is why switch to vector-only code pays
 #	off that much;
 
-$flavour = shift;
+# $output is the last argument if it looks like a file (it has an extension)
+# $flavour is the first argument if it doesn't look like a file
+$output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
+$flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : undef;
 
 if ($flavour =~ /64/) {
 	$SIZE_T	=8;
@@ -69,7 +72,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
 ( $xlate="${dir}../../perlasm/ppc-xlate.pl" and -f $xlate) or
 die "can't locate ppc-xlate.pl";
 
-open STDOUT,"| $^X $xlate $flavour ".shift || die "can't call $xlate: $!";
+open STDOUT,"| $^X $xlate $flavour \"$output\""
+    or die "can't call $xlate: $!";
 
 $LOCALS=6*$SIZE_T;
 $FRAME=$LOCALS+64+18*$SIZE_T;	# 64 is for local variables

crypto/chacha/asm/chacha-s390x.pl

@@ -42,7 +42,10 @@ use FindBin qw($Bin);
 use lib "$Bin/../..";
 use perlasm::s390x qw(:DEFAULT :VX :LD AUTOLOAD LABEL INCLUDE);
 
-my $flavour = shift;
+# $output is the last argument if it looks like a file (it has an extension)
+# $flavour is the first argument if it doesn't look like a file
+my $output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
+my $flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : undef;
 
 my ($z,$SIZE_T);
 if ($flavour =~ /3[12]/) {
@@ -53,9 +56,6 @@ if ($flavour =~ /3[12]/) {
 	$SIZE_T=8;
 }
 
-my $output;
-while (($output=shift) && ($output!~/\w[\w\-]*\.\w+$/)) {}
-
 my $sp="%r15";
 my $stdframe=16*$SIZE_T+4*8;
 

crypto/chacha/asm/chacha-x86.pl

@@ -40,8 +40,7 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
 push(@INC,"${dir}","${dir}../../perlasm");
 require "x86asm.pl";
 
-$output=pop;
-open STDOUT,">$output";
+$output = pop and open STDOUT,">$output";
 
 &asm_init($ARGV[0],$ARGV[$#ARGV] eq "386");
 

crypto/chacha/asm/chacha-x86_64.pl

@@ -58,9 +58,10 @@
 # (vi)	even though Skylake-X can execute AVX512F code and deliver 0.57
 #	cpb in single thread, the corresponding capability is suppressed;
 
-$flavour = shift;
-$output  = shift;
-if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+# $output is the last argument if it looks like a file (it has an extension)
+# $flavour is the first argument if it doesn't look like a file
+$output = $#ARGV >= 0 && $ARGV[$#ARGV] =~ m|\.\w+$| ? pop : undef;
+$flavour = $#ARGV >= 0 && $ARGV[0] !~ m|\.| ? shift : undef;
 
 $win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
 
@@ -89,7 +90,8 @@ if (!$avx && `$ENV{CC} -v 2>&1` =~ /((?:^clang|LLVM) version|.*based on LLVM) ([
 	$avx = ($2>=3.0) + ($2>3.0);
 }
 
-open OUT,"| \"$^X\" \"$xlate\" $flavour \"$output\"";
+open OUT,"| \"$^X\" \"$xlate\" $flavour \"$output\""
+    or die "can't call $xlate: $!";
 *STDOUT=*OUT;
 
 # input parameter block