sparcv9a-mont was modified to handle 32-bit aligned input, but check

[openssl.git] / crypto / bn / asm / sparcv9a-mont.pl

diff --git a/crypto/bn/asm/sparcv9a-mont.pl b/crypto/bn/asm/sparcv9a-mont.pl
index 8bc64ad5f5628b00196b7c825ffa51cd940e39ba..cecf07c6f438eed5c1fa3ed6c0871e011c3320d1 100755 (executable)
--- a/crypto/bn/asm/sparcv9a-mont.pl
+++ b/crypto/bn/asm/sparcv9a-mont.pl
@@ -18,8 +18,8 @@
 # implementations from compatibility matrix. But the rest, whole Sun
 # UltraSPARC family and brand new Fujitsu's SPARC64 V, all support
 # VIS extension instructions used in this module. This is considered
-# good enough to recommend HAL SPARC64 users [if any] to simply fall
-# down to no-asm configuration.
+# good enough to not care about HAL SPARC64 users [if any] who have
+# integer-only pure SPARCv9 module to "fall down" to.
 
 # USI&II cores currently exhibit uniform 2x improvement [over pre-
 # bn_mul_mont codebase] for all key lengths and benchmarks. On USIII
@@ -31,14 +31,12 @@
 # is pipelined, which in turn *might* be impossible to match... On
 # additional note, SPARC64 V implements FP Multiply-Add instruction,
 # which is perfectly usable in this context... In other words, as far
-# as HAL/Fujitsu SPARC64 family goes, talk to the author:-)
+# as Fujitsu SPARC64 V goes, talk to the author:-)
 
 # The implementation implies following "non-natural" limitations on
 # input arguments:
 # - num may not be less than 4;
 # - num has to be even;
-# - ap, bp, rp, np has to be 64-bit aligned [which is not a problem
-#   as long as BIGNUM.d are malloc-ated];
 # Failure to meet either condition has no fatal effects, simply
 # doesn't give any performance gain.
 
@@ -130,8 +128,6 @@ $code=<<___;
 .align  32
 $fname:
        save    %sp,-$frame-$locals,%sp
-       sethi   %hi(0xffff),$mask
-       or      $mask,%lo(0xffff),$mask
 
        cmp     $num,4
        bl,a,pn %icc,.Lret
@@ -139,14 +135,11 @@ $fname:
        andcc   $num,1,%g0              ! $num has to be even...
        bnz,a,pn %icc,.Lret
        clr     %i0                     ! signal "unsupported input value"
-       or      $bp,$ap,%l0
+
        srl     $num,1,$num
-       or      $rp,$np,%l1
-       or      %l0,%l1,%l0
-       andcc   %l0,7,%g0               ! ...and pointers has to be 8-byte aligned
-       bnz,a,pn %icc,.Lret
-       clr     %i0                     ! signal "unsupported input value"
+       sethi   %hi(0xffff),$mask
        ld      [%i4+0],$n0             ! $n0 reassigned, remember?
+       or      $mask,%lo(0xffff),$mask
        ld      [%i4+4],%o0
        sllx    %o0,32,%o0
        or      %o0,$n0,$n0             ! $n0=n0[1].n0[0]
@@ -183,12 +176,12 @@ $fname:
        add     $ap,$j,%o3
        add     $bp,$i,%o4
 
-       ldx     [$bp+$i],%o0            ! bp[0]
-       ldx     [$ap+$j],%o1            ! ap[0]
-       sllx    %o0,32,%g1
-       sllx    %o1,32,%g5
-       srlx    %o0,32,%o0
-       srlx    %o1,32,%o1
+       ld      [%o3+4],%g1             ! bp[0]
+       ld      [%o3+0],%o0
+       ld      [%o4+4],%g5             ! ap[0]
+       sllx    %g1,32,%g1
+       ld      [%o4+0],%o1
+       sllx    %g5,32,%g5
        or      %g1,%o0,%o0
        or      %g5,%o1,%o1
 
@@ -510,14 +503,15 @@ $fname:
        sub     %g0,$num,$j             ! j=-num
        add     %sp,$bias+$frame+$locals,$tp
 
+       add     $ap,$j,%o3
        add     $bp,$i,%o4
 
-       ldx     [$bp+$i],%o0            ! bp[i]
-       ldx     [$ap+$j],%o1            ! ap[0]
-       sllx    %o0,32,%g1
-       sllx    %o1,32,%g5
-       srlx    %o0,32,%o0
-       srlx    %o1,32,%o1
+       ld      [%o3+4],%g1             ! bp[i]
+       ld      [%o3+0],%o0
+       ld      [%o4+4],%g5             ! ap[0]
+       sllx    %g1,32,%g1
+       ld      [%o4+0],%o1
+       sllx    %g5,32,%g5
        or      %g1,%o0,%o0
        or      %g5,%o1,%o1
 
@@ -818,14 +812,18 @@ $fname:
 
 .align 32,0x1000000
 .Lsub:
-       ldd     [$tp+%o7],%o0
-       ldd     [$np+%o7],%o2
-       subccc  %o1,%o2,%o2
-       subccc  %o0,%o3,%o3
-       std     %o2,[$rp+%o7]
+       ldx     [$tp+%o7],%o0
+       add     $np,%o7,%g1
+       ld      [%g1+0],%o2
+       ld      [%g1+4],%o3
+       srlx    %o0,32,%o1
+       subccc  %o0,%o2,%o2
+       add     $rp,%o7,%g1
+       subccc  %o1,%o3,%o3
+       st      %o2,[%g1+0]
        add     %o7,8,%o7
        brnz,pt %o7,.Lsub
-       nop
+       st      %o3,[%g1+4]
        subccc  $carry,0,$carry
        bcc,pt  %icc,.Lzap
        sub     %g0,$num,%o7            ! n=-num
@@ -834,11 +832,11 @@ $fname:
 .Lcopy:
        ldx     [$tp+%o7],%o0
        srlx    %o0,32,%o1
-       std     %o0,[$rp+%o7]
+       add     $rp,%o7,%g1
+       st      %o0,[%g1+0]
        add     %o7,8,%o7
        brnz,pt %o7,.Lcopy
-       nop
-       ba      .Lzap
+       st      %o1,[%g1+4]
        sub     %g0,$num,%o7            ! n=-num
 
 .align 32
@@ -861,6 +859,7 @@ $fname:
        restore
 .type   $fname,#function
 .size  $fname,(.-$fname)
+.asciz "Montgomery Multipltication for UltraSPARC, CRYPTOGAMS by <appro\@openssl.org>"
 ___
 
 $code =~ s/\`([^\`]*)\`/eval($1)/gem;