aes-ppc.pl, sha512-ppc.pl: comply even with Embedded ABI specification

(most restrictive about r2 and r13 usage).

diff --git a/crypto/aes/asm/aes-ppc.pl b/crypto/aes/asm/aes-ppc.pl
index 7c52cbe5f9fa9ce44da7e31cca5c59f2be57c902..e10ce880c23c65923a4e7e1e303aaf81b32ab624 100644 (file)
--- a/crypto/aes/asm/aes-ppc.pl
+++ b/crypto/aes/asm/aes-ppc.pl
@@ -68,7 +68,7 @@ $key="r5";
 $Tbl0="r3";
 $Tbl1="r6";
 $Tbl2="r7";
-$Tbl3="r2";
+$Tbl3=$out;    # stay away from "r2"; $out is offloaded to stack
 
 $s0="r8";
 $s1="r9";
@@ -76,7 +76,7 @@ $s2="r10";
 $s3="r11";
 
 $t0="r12";
-$t1="r13";
+$t1="r0";      # stay away from "r13";
 $t2="r14";
 $t3="r15";
 
@@ -100,9 +100,6 @@ $acc13="r29";
 $acc14="r30";
 $acc15="r31";
 
-# stay away from TLS pointer
-if ($SIZE_T==8)        { die if ($t1 ne "r13");  $t1="r0";             }
-else           { die if ($Tbl3 ne "r2"); $Tbl3=$t0; $t0="r0";  }
 $mask80=$Tbl2;
 $mask1b=$Tbl3;
 
@@ -337,8 +334,7 @@ $code.=<<___;
        $STU    $sp,-$FRAME($sp)
        mflr    r0
 
-       $PUSH   $toc,`$FRAME-$SIZE_T*20`($sp)
-       $PUSH   r13,`$FRAME-$SIZE_T*19`($sp)
+       $PUSH   $out,`$FRAME-$SIZE_T*19`($sp)
        $PUSH   r14,`$FRAME-$SIZE_T*18`($sp)
        $PUSH   r15,`$FRAME-$SIZE_T*17`($sp)
        $PUSH   r16,`$FRAME-$SIZE_T*16`($sp)
@@ -371,6 +367,7 @@ Lenc_unaligned_ok:
        lwz     $s3,12($inp)
        bl      LAES_Te
        bl      Lppc_AES_encrypt_compact
+       $POP    $out,`$FRAME-$SIZE_T*19`($sp)
        stw     $s0,0($out)
        stw     $s1,4($out)
        stw     $s2,8($out)
@@ -417,6 +414,7 @@ Lenc_xpage:
 
        bl      LAES_Te
        bl      Lppc_AES_encrypt_compact
+       $POP    $out,`$FRAME-$SIZE_T*19`($sp)
 
        extrwi  $acc00,$s0,8,0
        extrwi  $acc01,$s0,8,8
@@ -449,8 +447,6 @@ Lenc_xpage:
 
 Lenc_done:
        $POP    r0,`$FRAME+$LRSAVE`($sp)
-       $POP    $toc,`$FRAME-$SIZE_T*20`($sp)
-       $POP    r13,`$FRAME-$SIZE_T*19`($sp)
        $POP    r14,`$FRAME-$SIZE_T*18`($sp)
        $POP    r15,`$FRAME-$SIZE_T*17`($sp)
        $POP    r16,`$FRAME-$SIZE_T*16`($sp)
@@ -771,8 +767,7 @@ Lenc_compact_done:
        $STU    $sp,-$FRAME($sp)
        mflr    r0
 
-       $PUSH   $toc,`$FRAME-$SIZE_T*20`($sp)
-       $PUSH   r13,`$FRAME-$SIZE_T*19`($sp)
+       $PUSH   $out,`$FRAME-$SIZE_T*19`($sp)
        $PUSH   r14,`$FRAME-$SIZE_T*18`($sp)
        $PUSH   r15,`$FRAME-$SIZE_T*17`($sp)
        $PUSH   r16,`$FRAME-$SIZE_T*16`($sp)
@@ -805,6 +800,7 @@ Ldec_unaligned_ok:
        lwz     $s3,12($inp)
        bl      LAES_Td
        bl      Lppc_AES_decrypt_compact
+       $POP    $out,`$FRAME-$SIZE_T*19`($sp)
        stw     $s0,0($out)
        stw     $s1,4($out)
        stw     $s2,8($out)
@@ -851,6 +847,7 @@ Ldec_xpage:
 
        bl      LAES_Td
        bl      Lppc_AES_decrypt_compact
+       $POP    $out,`$FRAME-$SIZE_T*19`($sp)
 
        extrwi  $acc00,$s0,8,0
        extrwi  $acc01,$s0,8,8
@@ -883,8 +880,6 @@ Ldec_xpage:
 
 Ldec_done:
        $POP    r0,`$FRAME+$LRSAVE`($sp)
-       $POP    $toc,`$FRAME-$SIZE_T*20`($sp)
-       $POP    r13,`$FRAME-$SIZE_T*19`($sp)
        $POP    r14,`$FRAME-$SIZE_T*18`($sp)
        $POP    r15,`$FRAME-$SIZE_T*17`($sp)
        $POP    r16,`$FRAME-$SIZE_T*16`($sp)

diff --git a/crypto/sha/asm/sha512-ppc.pl b/crypto/sha/asm/sha512-ppc.pl
index 6b44a68e599e5fcc59876e9be9af73cb1f54094d..2445ae79c8eadc4ad53db33484ffcf8a94831c51 100755 (executable)
--- a/crypto/sha/asm/sha512-ppc.pl
+++ b/crypto/sha/asm/sha512-ppc.pl
@@ -110,7 +110,7 @@ $B  ="r9";
 $C  ="r10";
 $D  ="r11";
 $E  ="r12";
-$F  ="r13";    $F="r2" if ($SIZE_T==8);# reassigned to exempt TLS pointer
+$F  =$t1;      $t1 = "r0";     # stay away from "r13";
 $G  ="r14";
 $H  ="r15";
 
@@ -123,19 +123,18 @@ $inp="r31";       # reassigned $inp! aliases with @X[15]
 sub ROUND_00_15 {
 my ($i,$a,$b,$c,$d,$e,$f,$g,$h)=@_;
 $code.=<<___;
-       $LD     $T,`$i*$SZ`($Tbl)
        $ROR    $a0,$e,$Sigma1[0]
        $ROR    $a1,$e,$Sigma1[1]
        and     $t0,$f,$e
-       andc    $t1,$g,$e
-       add     $T,$T,$h
        xor     $a0,$a0,$a1
+       add     $h,$h,$t1
+       andc    $t1,$g,$e
        $ROR    $a1,$a1,`$Sigma1[2]-$Sigma1[1]`
        or      $t0,$t0,$t1             ; Ch(e,f,g)
-       add     $T,$T,@X[$i]
+       add     $h,$h,@X[$i%16]
        xor     $a0,$a0,$a1             ; Sigma1(e)
-       add     $T,$T,$t0
-       add     $T,$T,$a0
+       add     $h,$h,$t0
+       add     $h,$h,$a0
 
        $ROR    $a0,$a,$Sigma0[0]
        $ROR    $a1,$a,$Sigma0[1]
@@ -146,9 +145,14 @@ $code.=<<___;
        xor     $t0,$t0,$t1
        and     $t1,$b,$c
        xor     $a0,$a0,$a1             ; Sigma0(a)
-       add     $d,$d,$T
+       add     $d,$d,$h
        xor     $t0,$t0,$t1             ; Maj(a,b,c)
-       add     $h,$T,$a0
+___
+$code.=<<___ if ($i<15);
+       $LD     $t1,`($i+1)*$SZ`($Tbl)
+___
+$code.=<<___;
+       add     $h,$h,$a0
        add     $h,$h,$t0
 
 ___
@@ -169,10 +173,11 @@ $code.=<<___;
        add     @X[$i],@X[$i],@X[($i+9)%16]
        xor     $a0,$a0,$a1             ; sigma0(X[(i+1)&0x0f])
        xor     $t0,$t0,$t1             ; sigma1(X[(i+14)&0x0f])
+       $LD     $t1,`$i*$SZ`($Tbl)
        add     @X[$i],@X[$i],$a0
        add     @X[$i],@X[$i],$t0
 ___
-&ROUND_00_15($i,$a,$b,$c,$d,$e,$f,$g,$h);
+&ROUND_00_15($i+16,$a,$b,$c,$d,$e,$f,$g,$h);
 }
 
 $code=<<___;
@@ -188,8 +193,6 @@ $func:
 
        $PUSH   $ctx,`$FRAME-$SIZE_T*22`($sp)
 
-       $PUSH   $toc,`$FRAME-$SIZE_T*20`($sp)
-       $PUSH   r13,`$FRAME-$SIZE_T*19`($sp)
        $PUSH   r14,`$FRAME-$SIZE_T*18`($sp)
        $PUSH   r15,`$FRAME-$SIZE_T*17`($sp)
        $PUSH   r16,`$FRAME-$SIZE_T*16`($sp)
@@ -283,8 +286,6 @@ Lmemcpy:
 
 Ldone:
        $POP    r0,`$FRAME+$LRSAVE`($sp)
-       $POP    $toc,`$FRAME-$SIZE_T*20`($sp)
-       $POP    r13,`$FRAME-$SIZE_T*19`($sp)
        $POP    r14,`$FRAME-$SIZE_T*18`($sp)
        $POP    r15,`$FRAME-$SIZE_T*17`($sp)
        $POP    r16,`$FRAME-$SIZE_T*16`($sp)
@@ -312,6 +313,7 @@ Ldone:
 
 .align 4
 Lsha2_block_private:
+       $LD     $t1,0($Tbl)
 ___
 for($i=0;$i<16;$i++) {
 $code.=<<___ if ($SZ==4);
@@ -328,8 +330,8 @@ ___
        unshift(@V,pop(@V));
 }
 $code.=<<___;
-       li      $T,`$rounds/16-1`
-       mtctr   $T
+       li      $t0,`$rounds/16-1`
+       mtctr   $t0
 .align 4
 Lrounds:
        addi    $Tbl,$Tbl,`16*$SZ`