Andy Polyakov <[EMAIL PROTECTED]> said:
>> > BN_div_recp failed!
>> > a=21CC27865629353755638C3726DF4C2D1C976729D1CD3C3FFC01039CE01B6687959E2BB84BAB54
>> > D798D9873CAED7007AB955B025F799BDE5AE4C84D79DE7B35E7ED2A43
>> > b=1
>>
>> > d:\net\openssl.095\bin>openssl.exe version -a
>> > OpenSSL 0.9.5beta2-dev 28 Feb 2000
>>
>> Could it be that there were some traces of an earlier version left?
>> That bug _should_ be fixed in beta2.
>I actually tried the buggy version and couldn't reproduce the problem
>either. A.
I've rebuilt the 27-02-2000 beta3 with okay result. (I always do a
'make clean' after applying a new snapshot). The error, I forgot to
say, only occurs in a DOS-box under Win-NT4 (SP3). Under plain
DOS there seems to be no problem (I ran divtest for ~1hour).
>> -DBN_DEBUG
> ^^^^^^^^^ But I wonder how did you get through the compilation? There
>is a typo in crypto/bn/bn_asm.c in 0.9.5beta2 triggered by -DBN_DEBUG.
>If you've never discovered the typo, then it means only one thing, you
>link with assembler module...
Thats right, I do.
> Which assumes bn(64,32) option...
Why it that? What's the CFLAGS and other configure options for if
the PerlAsm modules have to assume anything.
> Well,
>shouldn't really matter... At least I can't imagine it would... But can
>you reproduce the problem if you compile the toolkit with BN_LLONG? And
>post the whole config line:-)
I didn't use ./configure, but my single big homemade makefile with
the CFLAGS I posted. Here's my build-options now:
d:\net\openssl.095\bin>openssl.exe version -a
OpenSSL 0.9.5beta3-dev 27 Feb 2000
built on: Feb 29 2000
platform: DOS (djgpp)
options: bn(32,32) md2(int) rc4(ptr,int) des(ptr,cisc,16,long) idea(int)
^^ blowfish(ptr2)
compiler: gcc.exe -I. -Icrypto -I/net/watt/inc -D__WATT32__ -DL_ENDIAN
-D_ANSI_SOURCE -DBF_PTR2 -DBN_LLONG -DHAVE_LONG_LONG=1
-DDES_UNROLL -DDES_PTR -Wall -O2 -mpentium -malign-loops=2
-malign-jumps=2 -malign-functions=2 -fno-strength-reduce -DCRYPTO_MDEBUG
-DCRYPTO_MDEBUG_ALL -DREF_CHECK -DBN_DEBUG -DBN_MUL_DEBUG
-DMAC_DEBUG -DCIPHER_DEBUG -DERR_DEBUG -DABORT_DEBUG
-DPKT_DEBUG -DTLS_DEBUG -DDES_DEBUG -DDES_OFB_DEBUG
-DSSL_DEBUG -DRSA_DEBUG -DIDEA_DEBUG -DMD5_ASM=1 -DSHA1_ASM=1
-DRMD160_ASM=1
Notice how BN_LLONG doesn't give me "bn(64,32)". I assume it's because
I don't have a correct ./opensslconf.h.
But, most important, the divtest program doesn't give any errors now ?!!
BTW. One more compile error; defining BN_LLONG requires
-DHAVE_LONG_LONG=1 or parse error is produced in
./crypto/bio/b_print.c (line 128).
Gisle V.
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
