> I've got a segmentation fault in openssl-0.9.8a.
> ../util/shlib_wrap.sh ./sha512t
> make[2]: *** [test_sha] Illegal instruction

It's not considered to be bug, but a feature. As documented in FAQ.

> OpenSSL self-test report:
> 
> OpenSSL version:  0.9.8a
> Last change:      Remove the functionality of
> SSL_OP_MSIE_SSLV2_RSA_PADDI...
> Options:          -g enable-shared enable-zlib no-gmp no-krb5 no-mdc2
> no-rc5 no-zlib-dynamic
> OS (uname):       Linux b-woody 2.2.20 #1 Thu Feb 26 18:12:28 UTC 2004
> i686 unknown
> OS (config):      i686-whatever-linux2
> Target (default): linux-elf
> Target:           linux-elf
> Compiler:         gcc version 2.95.4 20011002 (Debian prerelease)
> 
> The results of `make report` are attached.
> 
> Adding no-sse2 to config options fixes this problem.

Well, sometimes you have to be more specific and explicitly express 
opinion. One can discuss that automatic detection when no-sse2 is due 
[either at config or run-time] can be improved. In which case it's more 
appropriate to turn to <openssl-dev>, make suggestion, collect opinions 
and reach consensus. Or one can discuss that whole SSE2 support should 
be treated as bug, in which case it's [again] more appropaite to turn to 
openssl-dev to explicitly state something like "it appears to me that 
these sse2 code-pathes create more trouble on elder systems than good on 
new ones and in my opinion it should be disabled everywhere." So I'm 
dismissing the case with "not treated as bug, not for the moment of this 
writing" and see you at <openssl-dev>. A.

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [email protected]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to