# it was measured to be ~6.6x. It's less than previously mentioned 8x,
# because software implementation was optimized.
+while (($output=shift) && ($output!~/^\w[\w\-]*\.\w+$/)) {}
+open STDOUT,">$output";
+
$softonly=0; # allow hardware support
$t0="%r0"; $mask="%r0";
srl %r5,6
ar %r5,%r0
+ larl %r1,OPENSSL_s390xcap_P
+ lg %r0,0(%r1)
+ tmhl %r0,0x4000 # check for message-security assist
+ jz .Lekey_internal
+
lghi %r0,0 # query capability vector
la %r1,16($sp)
.long 0xb92f0042 # kmc %r4,%r2
4: ex $len,0($s1)
j .Lcbc_dec_exit
.size AES_cbc_encrypt,.-AES_cbc_encrypt
+.comm OPENSSL_s390xcap_P,8,8
___
}
$code.=<<___;
$code =~ s/\`([^\`]*)\`/eval $1/gem;
print $code;
+close STDOUT; # force flush