crypto/bn/asm/s390x.S: drop redundant instructions.

diff --git a/crypto/bn/asm/s390x.S b/crypto/bn/asm/s390x.S
index 8f45f5d513cec93cb3213f26fba5a34fd4b6db7f..27ef61bd70d8d240b04fe6b0a20a8975b1346bf9 100755 (executable)
--- a/crypto/bn/asm/s390x.S
+++ b/crypto/bn/asm/s390x.S
@@ -1,4 +1,4 @@
-.ident "s390x.S, version 1.0"
+.ident "s390x.S, version 1.1"
 // ====================================================================
 // Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
 // project.
@@ -24,67 +24,67 @@ bn_mul_add_words:
        bler    %r14            // if (len<=0) return 0;
 
        stmg    %r6,%r10,48(%r15)
+       lghi    %r10,3
        lghi    %r8,0           // carry = 0
-       srag    %r10,%r4,2      // cnt=len/4
-       jz      .Loop1_madd
+       nr      %r10,%r4        // len%4
+       sra     %r4,2           // cnt=len/4
+       jz      .Loop1_madd     // carry is incidentally cleared if branch taken
+       algr    zero,zero       // clear carry
 
 .Loop4_madd:
        lg      %r7,0(%r2,%r3)  // ap[i]
        mlgr    %r6,%r5         // *=w
-       algr    %r7,%r8         // +=carry
+       alcgr   %r7,%r8         // +=carry
        alcgr   %r6,zero
        alg     %r7,0(%r2,%r1)  // +=rp[i]
-       alcgr   %r6,zero
        stg     %r7,0(%r2,%r1)  // rp[i]=
 
        lg      %r9,8(%r2,%r3)
        mlgr    %r8,%r5
-       algr    %r9,%r6
+       alcgr   %r9,%r6
        alcgr   %r8,zero
        alg     %r9,8(%r2,%r1)
-       alcgr   %r8,zero
        stg     %r9,8(%r2,%r1)
 
        lg      %r7,16(%r2,%r3)
        mlgr    %r6,%r5
-       algr    %r7,%r8
+       alcgr   %r7,%r8
        alcgr   %r6,zero
        alg     %r7,16(%r2,%r1)
-       alcgr   %r6,zero
        stg     %r7,16(%r2,%r1)
 
        lg      %r9,24(%r2,%r3)
        mlgr    %r8,%r5
-       algr    %r9,%r6
+       alcgr   %r9,%r6
        alcgr   %r8,zero
        alg     %r9,24(%r2,%r1)
-       alcgr   %r8,zero
        stg     %r9,24(%r2,%r1)
 
        la      %r2,32(%r2)     // i+=4
-       brct    %r10,.Loop4_madd
+       brct    %r4,.Loop4_madd
 
-       lghi    %r10,3
-       nr      %r4,%r10        // cnt=len%4
-       jz      .Lend_madd
+       la      %r10,1(%r10)            // see if len%4 is zero ...
+       brct    %r10,.Loop1_madd        // without touching condition code:-)
+
+.Lend_madd:
+       alcgr   %r8,zero        // collect carry bit
+       lgr     %r2,%r8
+       lmg     %r6,%r10,48(%r15)
+       br      %r14
 
 .Loop1_madd:
        lg      %r7,0(%r2,%r3)  // ap[i]
        mlgr    %r6,%r5         // *=w
-       algr    %r7,%r8         // +=carry
+       alcgr   %r7,%r8         // +=carry
        alcgr   %r6,zero
        alg     %r7,0(%r2,%r1)  // +=rp[i]
-       alcgr   %r6,zero
        stg     %r7,0(%r2,%r1)  // rp[i]=
 
        lgr     %r8,%r6
        la      %r2,8(%r2)      // i++
-       brct    %r4,.Loop1_madd
+       brct    %r10,.Loop1_madd
 
-.Lend_madd:
-       lgr     %r2,%r8
-       lmg     %r6,%r10,48(%r15)
-       br      %r14
+       j       .Lend_madd
 .size  bn_mul_add_words,.-bn_mul_add_words
 
 // BN_ULONG bn_mul_words(BN_ULONG *r2,BN_ULONG *r3,int r4,BN_ULONG r5);
@@ -99,57 +99,57 @@ bn_mul_words:
        bler    %r14            // if (len<=0) return 0;
 
        stmg    %r6,%r10,48(%r15)
+       lghi    %r10,3
        lghi    %r8,0           // carry = 0
-       srag    %r10,%r4,2      // cnt=len/4
-       jz      .Loop1_mul
+       nr      %r10,%r4        // len%4
+       sra     %r4,2           // cnt=len/4
+       jz      .Loop1_mul      // carry is incidentally cleared if branch taken
+       algr    zero,zero       // clear carry
 
 .Loop4_mul:
        lg      %r7,0(%r2,%r3)  // ap[i]
        mlgr    %r6,%r5         // *=w
-       algr    %r7,%r8         // +=carry
-       alcgr   %r6,zero
+       alcgr   %r7,%r8         // +=carry
        stg     %r7,0(%r2,%r1)  // rp[i]=
 
        lg      %r9,8(%r2,%r3)
        mlgr    %r8,%r5
-       algr    %r9,%r6
-       alcgr   %r8,zero
+       alcgr   %r9,%r6
        stg     %r9,8(%r2,%r1)
 
        lg      %r7,16(%r2,%r3)
        mlgr    %r6,%r5
-       algr    %r7,%r8
-       alcgr   %r6,zero
+       alcgr   %r7,%r8
        stg     %r7,16(%r2,%r1)
 
        lg      %r9,24(%r2,%r3)
        mlgr    %r8,%r5
-       algr    %r9,%r6
-       alcgr   %r8,zero
+       alcgr   %r9,%r6
        stg     %r9,24(%r2,%r1)
 
        la      %r2,32(%r2)     // i+=4
-       brct    %r10,.Loop4_mul
+       brct    %r4,.Loop4_mul
 
-       lghi    %r10,3
-       nr      %r4,%r10        // cnt=len%4
-       jz      .Lend_mul
+       la      %r10,1(%r10)            // see if len%4 is zero ...
+       brct    %r10,.Loop1_madd        // without touching condition code:-)
+
+.Lend_mul:
+       alcgr   %r8,zero        // collect carry bit
+       lgr     %r2,%r8
+       lmg     %r6,%r10,48(%r15)
+       br      %r14
 
 .Loop1_mul:
        lg      %r7,0(%r2,%r3)  // ap[i]
        mlgr    %r6,%r5         // *=w
-       algr    %r7,%r8         // +=carry
-       alcgr   %r6,zero
+       alcgr   %r7,%r8         // +=carry
        stg     %r7,0(%r2,%r1)  // rp[i]=
 
        lgr     %r8,%r6
        la      %r2,8(%r2)      // i++
-       brct    %r4,.Loop1_mul
+       brct    %r10,.Loop1_mul
 
-.Lend_mul:
-       lgr     %r2,%r8
-       lmg     %r6,%r10,48(%r15)
-       br      %r14
+       j       .Lend_mul
 .size  bn_mul_words,.-bn_mul_words
 
 // void bn_sqr_words(BN_ULONG *r2,BN_ULONG *r2,int r4)