openssl.git
17 years agoAddapt VMS scripts to the newer disk layout system ODS-5, which allows more than...
Richard Levitte [Mon, 29 Oct 2001 13:05:28 +0000 (13:05 +0000)]
Addapt VMS scripts to the newer disk layout system ODS-5, which allows more than one period and mixed size characters in file names

17 years agoAnother noemailDN fix.
Dr. Stephen Henson [Sat, 27 Oct 2001 17:53:06 +0000 (17:53 +0000)]
Another noemailDN fix.

17 years agoAllow ca to certify requests containing BMPStrings and UTF8Strings.
Dr. Stephen Henson [Sat, 27 Oct 2001 17:04:47 +0000 (17:04 +0000)]
Allow ca to certify requests containing BMPStrings and UTF8Strings.

17 years agoBugfixes for noemailDN option. Make it use the
Dr. Stephen Henson [Sat, 27 Oct 2001 17:03:20 +0000 (17:03 +0000)]
Bugfixes for noemailDN option. Make it use the
correct name (instead of NULL) if nomailDN is
not set, fix memory leaks and retain DN structure
when deleting emailAddress.

17 years agoStop compiler warnings.
Dr. Stephen Henson [Sat, 27 Oct 2001 00:18:04 +0000 (00:18 +0000)]
Stop compiler warnings.

17 years agoAdd support for Subject Info Acess extension.
Dr. Stephen Henson [Sat, 27 Oct 2001 00:16:53 +0000 (00:16 +0000)]
Add support for Subject Info Acess extension.

17 years agoNote BUF_MEM_grow() consistency fix.
Bodo Möller [Fri, 26 Oct 2001 14:06:33 +0000 (14:06 +0000)]
Note BUF_MEM_grow() consistency fix.

17 years agoConsistency fix in BUF_MEM_grow: Initialise to zero when new memory
Bodo Möller [Fri, 26 Oct 2001 13:12:25 +0000 (13:12 +0000)]
Consistency fix in BUF_MEM_grow: Initialise to zero when new memory
had to be allocated, not just when reusing the existing buffer.

17 years agodisable caching in BIO_gethostbyname
Bodo Möller [Fri, 26 Oct 2001 13:04:23 +0000 (13:04 +0000)]
disable caching in BIO_gethostbyname

17 years agoNew options to allow req to accept UTF8 strings as input.
Dr. Stephen Henson [Fri, 26 Oct 2001 12:40:38 +0000 (12:40 +0000)]
New options to allow req to accept UTF8 strings as input.

17 years agoremove compatibility notes that no longer apply
Ulf Möller [Thu, 25 Oct 2001 17:45:25 +0000 (17:45 +0000)]
remove compatibility notes that no longer apply

17 years agoCorrect some links...
Richard Levitte [Thu, 25 Oct 2001 16:56:06 +0000 (16:56 +0000)]
Correct some links...

17 years agoChange the DES documentation to reflect the current status. Note that
Richard Levitte [Thu, 25 Oct 2001 16:55:17 +0000 (16:55 +0000)]
Change the DES documentation to reflect the current status.  Note that
some password reading functions are really part of the UI
compatibility library...

17 years agoMake sure openssl speed is compilable on systems where fork() doesn't
Richard Levitte [Thu, 25 Oct 2001 16:08:17 +0000 (16:08 +0000)]
Make sure openssl speed is compilable on systems where fork() doesn't
exist.  For now, that's all the ones we "support" except Unix.

17 years agoAdd paralellism to speed - note that this currently causes a weird memory leak.
Ben Laurie [Thu, 25 Oct 2001 14:27:17 +0000 (14:27 +0000)]
Add paralellism to speed - note that this currently causes a weird memory leak.

17 years agoFix warning.
Ben Laurie [Thu, 25 Oct 2001 14:24:59 +0000 (14:24 +0000)]
Fix warning.

17 years agoLike MD_Init, MD now must include a NULL engine pointer in its definition.
Bodo Möller [Thu, 25 Oct 2001 08:53:54 +0000 (08:53 +0000)]
Like MD_Init, MD now must include a NULL engine pointer in its definition.

17 years agoremove redundant definitions that are also in des.h
Bodo Möller [Thu, 25 Oct 2001 08:46:10 +0000 (08:46 +0000)]
remove redundant definitions that are also in des.h

17 years agomention des_old.h
Bodo Möller [Thu, 25 Oct 2001 08:44:10 +0000 (08:44 +0000)]
mention des_old.h

17 years agoAdd '-noemailDN' option to 'openssl ca'. This prevents inclusion of
Bodo Möller [Thu, 25 Oct 2001 08:25:19 +0000 (08:25 +0000)]
Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion of
the e-mail address in the DN (i.e., it will go into a certificate
extension only).  The new configuration file option 'email_in_dn = no'
has the same effect.

Submitted by: Massimiliano Pala madwolf@openca.org

17 years agofilenames are des_old.[ch], not des.comp*
Bodo Möller [Thu, 25 Oct 2001 08:23:13 +0000 (08:23 +0000)]
filenames are des_old.[ch], not des.comp*

17 years agoConsistency with s2_... and s23_... variants (no real functional
Bodo Möller [Thu, 25 Oct 2001 08:17:53 +0000 (08:17 +0000)]
Consistency with s2_... and s23_... variants (no real functional
change)

17 years agoRemove DES_random_seed() but retain des_random_seed() for now. Change
Richard Levitte [Thu, 25 Oct 2001 06:46:22 +0000 (06:46 +0000)]
Remove DES_random_seed() but retain des_random_seed() for now.  Change
the docs to reflect this change and correct libeay.num.

17 years agoAssume TLS 1.0 when ClientHello fragment is too short.
Bodo Möller [Thu, 25 Oct 2001 06:09:51 +0000 (06:09 +0000)]
Assume TLS 1.0 when ClientHello fragment is too short.

17 years agoHave the removal warnings very high up in the source.
Richard Levitte [Thu, 25 Oct 2001 05:37:10 +0000 (05:37 +0000)]
Have the removal warnings very high up in the source.

17 years agoA C file is a C file is a C file!
Richard Levitte [Wed, 24 Oct 2001 21:31:14 +0000 (21:31 +0000)]
A C file is a C file is a C file!

17 years agoDue to an increasing number of clashes between modern OpenSSL and
Richard Levitte [Wed, 24 Oct 2001 21:21:12 +0000 (21:21 +0000)]
Due to an increasing number of clashes between modern OpenSSL and
libdes (which is still used out there) or other des implementations,
the OpenSSL DES functions are renamed to begin with DES_ instead of
des_.  Compatibility routines are provided and declared by including
openssl/des_old.h.  Those declarations are the same as were in des.h
when the OpenSSL project started, which is exactly how libdes looked
at that time, and hopefully still looks today.

The compatibility functions will be removed in some future release, at
the latest in version 1.0.

17 years agoFix SSL handshake functions and SSL_clear() such that SSL_clear()
Bodo Möller [Wed, 24 Oct 2001 19:03:22 +0000 (19:03 +0000)]
Fix SSL handshake functions and SSL_clear() such that SSL_clear()
never resets s->method to s->ctx->method when called from within
one of the SSL handshake functions.

17 years agoDeprecate the macro MAC_OS_pre_X.
Richard Levitte [Wed, 24 Oct 2001 15:32:53 +0000 (15:32 +0000)]
Deprecate the macro MAC_OS_pre_X.

17 years agomake update
Richard Levitte [Wed, 24 Oct 2001 15:32:14 +0000 (15:32 +0000)]
make update

17 years agomake update
Richard Levitte [Wed, 24 Oct 2001 08:27:22 +0000 (08:27 +0000)]
make update

17 years agoCorrection of the id-pda OID's.
Richard Levitte [Tue, 23 Oct 2001 14:30:57 +0000 (14:30 +0000)]
Correction of the id-pda OID's.
Submitted by Frederic.Giudicelli@INTRINsec.com

17 years agoSanitize CHANGES entry.
Dr. Stephen Henson [Tue, 23 Oct 2001 00:54:58 +0000 (00:54 +0000)]
Sanitize CHANGES entry.

17 years agoFix memory leak.
Bodo Möller [Mon, 22 Oct 2001 13:59:36 +0000 (13:59 +0000)]
Fix memory leak.

17 years agoReject certificates with unhandled critical extensions.
Dr. Stephen Henson [Sun, 21 Oct 2001 02:09:15 +0000 (02:09 +0000)]
Reject certificates with unhandled critical extensions.

17 years agoStop spurious "unable to load config info" errors in req
Dr. Stephen Henson [Sun, 21 Oct 2001 01:05:53 +0000 (01:05 +0000)]
Stop spurious "unable to load config info" errors in req

17 years agoCall msg_callback with correct length parameter if ssl3_write_bytes had to
Bodo Möller [Sat, 20 Oct 2001 18:56:01 +0000 (18:56 +0000)]
Call msg_callback with correct length parameter if ssl3_write_bytes had to
be called multiple times

17 years agoNew functions SSL[_CTX]_set_msg_callback().
Bodo Möller [Sat, 20 Oct 2001 17:56:36 +0000 (17:56 +0000)]
New functions SSL[_CTX]_set_msg_callback().
New macros SSL[_CTX]_set_msg_callback_arg().

Message callback imlementation for SSL 3.0/TLS 1.0 (no SSL 2.0 yet).

New '-msg' option for 'openssl s_client' and 'openssl s_server'
that enable a message callback that displays all protocol messages.

In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert if
client_version is smaller than the protocol version in use.
Also change ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0
if the client demanded SSL 3.0 but only TLS 1.0 is enabled; then the
client will at least see that alert.

Fix SSL[_CTX]_ctrl prototype (void * instead of char * for generic
pointer).

Add/update some OpenSSL copyright notices.

17 years agoTypo.
Dr. Stephen Henson [Sat, 20 Oct 2001 16:22:28 +0000 (16:22 +0000)]
Typo.

17 years agoAdd missing EVP_CIPHER_CTX_{init,cleanup}
Dr. Stephen Henson [Sat, 20 Oct 2001 16:18:03 +0000 (16:18 +0000)]
Add missing EVP_CIPHER_CTX_{init,cleanup}

17 years agogcc complained about "write" being shadowed even though the "write"
Bodo Möller [Wed, 17 Oct 2001 20:44:25 +0000 (20:44 +0000)]
gcc complained about "write" being shadowed even though the "write"
variable name occured just in a function *prototype* -- so rename it

17 years agoWrong place...
Richard Levitte [Wed, 17 Oct 2001 17:54:17 +0000 (17:54 +0000)]
Wrong place...

17 years agoThe EVP_*Init_ex() functions take one extra argument. Let's default
Richard Levitte [Wed, 17 Oct 2001 16:03:42 +0000 (16:03 +0000)]
The EVP_*Init_ex() functions take one extra argument.  Let's default
it to NULL.

17 years agodocument SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION
Bodo Möller [Wed, 17 Oct 2001 11:56:26 +0000 (11:56 +0000)]
document SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION

17 years agoUpdate docs.
Dr. Stephen Henson [Wed, 17 Oct 2001 01:50:32 +0000 (01:50 +0000)]
Update docs.

17 years agoModify EVP cipher behaviour in a similar way
Dr. Stephen Henson [Wed, 17 Oct 2001 00:37:12 +0000 (00:37 +0000)]
Modify EVP cipher behaviour in a similar way
to digests to retain compatibility.

17 years agoFlush buffers to prevent mixed output (Adam Back <adam@cypherspace.org>).
Lutz Jänicke [Tue, 16 Oct 2001 14:24:46 +0000 (14:24 +0000)]
Flush buffers to prevent mixed output (Adam Back <adam@cypherspace.org>).

17 years agoAdd per-SSL 'msg_callback' with 'msg_callback_arg'.
Bodo Möller [Tue, 16 Oct 2001 13:09:24 +0000 (13:09 +0000)]
Add per-SSL 'msg_callback' with 'msg_callback_arg'.
Both have per-SSL_CTX defaults.
These new values can be set by calling SSL[_CTX]_[callback_]ctrl
with codes SSL_CTRL_SET_MSG_CALLBACK and SSL_CTRL_SET_MSG_CALLBACK_ARG.

So far, the callback is never actually called.

Also rearrange some SSL_CTX struct members (some exist just in
SSL_CTXs, others are defaults for SSLs and are either copied
during SSL_new, or used if the value in the SSL is not set;
these three classes of members were not in a logical order),
and add some missing assignments to SSL_dup.

17 years agoUpdate docs.
Dr. Stephen Henson [Tue, 16 Oct 2001 02:22:59 +0000 (02:22 +0000)]
Update docs.

17 years agoRetain compatibility of EVP_DigestInit() and EVP_DigestFinal()
Dr. Stephen Henson [Tue, 16 Oct 2001 01:24:29 +0000 (01:24 +0000)]
Retain compatibility of EVP_DigestInit() and EVP_DigestFinal()
with existing code.

Modify library to use digest *_ex() functions.

17 years agoThe message header for fake SSL 3.0/TLS 1.0 client hellos created from
Bodo Möller [Tue, 16 Oct 2001 00:56:04 +0000 (00:56 +0000)]
The message header for fake SSL 3.0/TLS 1.0 client hellos created from
SSL 2.0 client hellos added with the previous commit was totally wrong --
it must start with the message type, not the protocol version.
(Not that this particular header is actually used anywhere ...)

17 years agoFor consistency, set s->init_num in the 'reuse_message' case
Bodo Möller [Mon, 15 Oct 2001 20:16:36 +0000 (20:16 +0000)]
For consistency, set s->init_num in the 'reuse_message' case
(if s23_srvr.c faked the message, s->init_num is 0).

17 years agoChange ssl3_get_message and the functions using it so that complete
Bodo Möller [Mon, 15 Oct 2001 19:49:25 +0000 (19:49 +0000)]
Change ssl3_get_message and the functions using it so that complete
'Handshake' protocol structures are kept in memory, including
'msg_type' and 'length'.

(This is in preparation of future support for callbacks that get to
peek at handshake messages and the like.)

17 years agomake sure .rnd exists
Bodo Möller [Mon, 15 Oct 2001 17:58:00 +0000 (17:58 +0000)]
make sure .rnd exists

17 years agoFix ssl3_get_message handle message fragmentation correctly.
Bodo Möller [Mon, 15 Oct 2001 17:41:41 +0000 (17:41 +0000)]
Fix ssl3_get_message handle message fragmentation correctly.

17 years agothe previous commit accidentily removed 'ret = 1' from the SSL_ST_OK
Bodo Möller [Mon, 15 Oct 2001 17:40:42 +0000 (17:40 +0000)]
the previous commit accidentily removed 'ret = 1' from the SSL_ST_OK
case of ssl3_accept

17 years agoopenbsd-x86 macros
Ulf Möller [Sun, 14 Oct 2001 00:57:30 +0000 (00:57 +0000)]
openbsd-x86 macros

Submitted by: Toomas Kiisk <vix@cyber.ee>

17 years agoUpdate information as a partial response to the post
Lutz Jänicke [Fri, 12 Oct 2001 12:29:16 +0000 (12:29 +0000)]
Update information as a partial response to the post
  From: "Chris D. Peterson" <cpeterson@aventail.com>
  Subject: Implementation Issues with OpenSSL
  To: openssl-users@openssl.org
  Date: Wed, 22 Aug 2001 16:13:17 -0700
The patch included in the original post may improve the internal session
list handling (and is therefore worth a seperate investigation).
No change to the list handling will however solve the problems of incorrect
SSL_SESSION_free() calls. The session list is only one possible point of
failure, dangling pointers would also occur for SSL object currently
using the session. The correct solution is to only use SSL_SESSION_free()
when applicable!

17 years agoIn certain cases, no encoding has been set up for the b64 filter. In
Richard Levitte [Thu, 11 Oct 2001 19:38:40 +0000 (19:38 +0000)]
In certain cases, no encoding has been set up for the b64 filter.  In
such cases, a flush should *not* attempt to finalise the encoding, as
the EVP_ENCODE_CTX structure will only be filled with garbage.  For
the same reason, do the same check when a wpending is performed.

17 years ago'make update'
Richard Levitte [Wed, 10 Oct 2001 21:52:06 +0000 (21:52 +0000)]
'make update'

17 years agoAdd support for md4WithRSAEncryption.
Richard Levitte [Wed, 10 Oct 2001 21:37:45 +0000 (21:37 +0000)]
Add support for md4WithRSAEncryption.

17 years agoFor systems where gcc is used and where we don't know if GNU ld is
Richard Levitte [Wed, 10 Oct 2001 14:46:41 +0000 (14:46 +0000)]
For systems where gcc is used and where we don't know if GNU ld is
used or not, let's ask collect2 which ld it uses and choose to use the
target do-gnu_shared if GNU ld is used.

This solves the reported problems on Solaris systems where GNU cc is
used but GNU ld isn't, and probably on other systems with similar
setups.

17 years ago'make update'
Richard Levitte [Wed, 10 Oct 2001 08:27:52 +0000 (08:27 +0000)]
'make update'

17 years agoIt seems like gcc does canonicalisation of file names. More
Richard Levitte [Wed, 10 Oct 2001 08:27:28 +0000 (08:27 +0000)]
It seems like gcc does canonicalisation of file names.  More
specifically, a starting './' is removed.  makedepend doesn't do this,
resulting in another possible commit war, so let's fix that by doing a
poor mans canonicalisation of file names that gives the same effect as
doing dependencies through gcc.

17 years ago'make update'
Richard Levitte [Wed, 10 Oct 2001 07:56:20 +0000 (07:56 +0000)]
'make update'

17 years agoTo avoid commit wars over dependencies, let's make it so things that
Richard Levitte [Wed, 10 Oct 2001 07:55:02 +0000 (07:55 +0000)]
To avoid commit wars over dependencies, let's make it so things that
depend on the environment, like the presence of the OpenBSD crypto
device or of Kerberos, do not change the dependencies within OpenSSL.

17 years agomakedepend sometimes produces duplicates. Remove them.
Richard Levitte [Wed, 10 Oct 2001 07:44:54 +0000 (07:44 +0000)]
makedepend sometimes produces duplicates.  Remove them.

17 years agoA few more OIDs, contributed by Peter Sylvester <Peter.Sylvester@EdelWeb.fr>
Richard Levitte [Tue, 9 Oct 2001 15:32:23 +0000 (15:32 +0000)]
A few more OIDs, contributed by Peter Sylvester <Peter.Sylvester@EdelWeb.fr>

17 years agoevp_test.c and evptests.txt both need to be linked in the test/ directory
Geoff Thorpe [Tue, 9 Oct 2001 01:38:31 +0000 (01:38 +0000)]
evp_test.c and evptests.txt both need to be linked in the test/ directory
however for different reasons. This separation should prevent the win32
build from interpreting evptests.txt as source code.

17 years agoChange some EVP prototypes to use "cipher" rather than "type" as a variable
Geoff Thorpe [Mon, 8 Oct 2001 17:25:42 +0000 (17:25 +0000)]
Change some EVP prototypes to use "cipher" rather than "type" as a variable
name. The implementations already use this anyway.

17 years agoEVP_EncryptInit_ex() and EVP_DecryptInit_ex() had been defined in evp.h but
Geoff Thorpe [Mon, 8 Oct 2001 17:24:10 +0000 (17:24 +0000)]
EVP_EncryptInit_ex() and EVP_DecryptInit_ex() had been defined in evp.h but
not implemented. (Bug reported by Martin Szotkowski)

This also changes the non-"_ex" versions to defer directly to
EVP_CipherInit_ex() rather than EVP_CipherInit() to avoid an unecessary
level of indirection.

17 years agoAs ENGINE_load_openbsd_dev_crypto() is an API function, it makes sense for
Geoff Thorpe [Mon, 8 Oct 2001 17:08:17 +0000 (17:08 +0000)]
As ENGINE_load_openbsd_dev_crypto() is an API function, it makes sense for
it to be defined on all platforms whether or not it is of any practical
use on them. This also resolves linker problems on "special" platforms,
such as win32.

17 years agoMake sure the "ENGINE_TABLE" cleanup callbacks have correct prototypes.
Geoff Thorpe [Mon, 8 Oct 2001 17:06:52 +0000 (17:06 +0000)]
Make sure the "ENGINE_TABLE" cleanup callbacks have correct prototypes.

17 years agoMissing pointer in the eng_table_register function. Reported by
Geoff Thorpe [Mon, 8 Oct 2001 14:44:38 +0000 (14:44 +0000)]
Missing pointer in the eng_table_register function. Reported by
Martin Szotkowski.

17 years agoSmall documentation fixes (Howard Lum <howard@pumpkin.canada.sun.com>)
Lutz Jänicke [Mon, 8 Oct 2001 08:37:24 +0000 (08:37 +0000)]
Small documentation fixes (Howard Lum <howard@pumpkin.canada.sun.com>)

17 years agoCopy evptests.txt to the right place.
Richard Levitte [Thu, 4 Oct 2001 21:15:03 +0000 (21:15 +0000)]
Copy evptests.txt to the right place.

17 years agoTypo...
Richard Levitte [Thu, 4 Oct 2001 19:25:12 +0000 (19:25 +0000)]
Typo...

17 years agoSSL_add_dir_cert_subjects_to_stack for Win32 finally implemented.
Richard Levitte [Thu, 4 Oct 2001 12:27:39 +0000 (12:27 +0000)]
SSL_add_dir_cert_subjects_to_stack for Win32 finally implemented.
Submitted by Massimo Santin <msantin@santineassociati.com>.

17 years ago'make update'
Richard Levitte [Thu, 4 Oct 2001 07:49:09 +0000 (07:49 +0000)]
'make update'

17 years agoSince ossl_typ.h is an exported header, we sure need to export it on
Richard Levitte [Thu, 4 Oct 2001 07:46:30 +0000 (07:46 +0000)]
Since ossl_typ.h is an exported header, we sure need to export it on
VMS as well :-).

17 years agoBecause there's chances we clash with the system's types.h, rename our
Richard Levitte [Thu, 4 Oct 2001 07:34:45 +0000 (07:34 +0000)]
Because there's chances we clash with the system's types.h, rename our
types.h to ossl_typ.h.
Also, it seems like krb5 was forgotten in some places.

17 years agoBecause there's chances we clash with the system's types.h, rename our
Richard Levitte [Thu, 4 Oct 2001 07:32:46 +0000 (07:32 +0000)]
Because there's chances we clash with the system's types.h, rename our
types.h to ossl_typ.h.

17 years agoUse the maximum block length for the extra size in the encrypt
Dr. Stephen Henson [Wed, 3 Oct 2001 12:47:03 +0000 (12:47 +0000)]
Use the maximum block length for the extra size in the encrypt
BIO buffer instead of hard coding it as 8.

17 years agoMake EVP_DecryptUpdate work again.
Dr. Stephen Henson [Tue, 2 Oct 2001 16:19:49 +0000 (16:19 +0000)]
Make EVP_DecryptUpdate work again.

17 years agosch isn't an array, how did this pass through gcc?
Richard Levitte [Tue, 2 Oct 2001 11:49:55 +0000 (11:49 +0000)]
sch isn't an array, how did this pass through gcc?

17 years agoA lot of things are undeclared unless x509.h is included.
Richard Levitte [Tue, 2 Oct 2001 11:06:42 +0000 (11:06 +0000)]
A lot of things are undeclared unless x509.h is included.

17 years agoHmm, everything "open" isn't necessarely "openssl" :-).
Richard Levitte [Tue, 2 Oct 2001 10:03:15 +0000 (10:03 +0000)]
Hmm, everything "open" isn't necessarely "openssl" :-).
*sigh* habit...

17 years agoWoopsie...
Richard Levitte [Mon, 1 Oct 2001 17:20:28 +0000 (17:20 +0000)]
Woopsie...

17 years ago'make update'
Richard Levitte [Mon, 1 Oct 2001 17:16:24 +0000 (17:16 +0000)]
'make update'

17 years agosk_ENGINE_CLEANUP_ITEM_pop_free() is duplicated in ENGINE_cleanup().
Richard Levitte [Mon, 1 Oct 2001 17:15:28 +0000 (17:15 +0000)]
sk_ENGINE_CLEANUP_ITEM_pop_free() is duplicated in ENGINE_cleanup().
Let's use sk_ENGINE_CLEANUP_ITEM_pop_free() instead.

17 years agoAddapt seldom compiled code to new semantics of the key schedule (not
Richard Levitte [Mon, 1 Oct 2001 17:10:10 +0000 (17:10 +0000)]
Addapt seldom compiled code to new semantics of the key schedule (not
a pointer any more).

17 years agoSome new symbols have very long names...
Richard Levitte [Mon, 1 Oct 2001 17:09:17 +0000 (17:09 +0000)]
Some new symbols have very long names...

17 years agoo_time.c contains symbols with dollar signs in them, so we must tell
Richard Levitte [Mon, 1 Oct 2001 17:08:18 +0000 (17:08 +0000)]
o_time.c contains symbols with dollar signs in them, so we must tell
the compiler not to warn about that.

17 years agoThe cleanup stack in ENGINE changed slightly, so this "make update" is
Geoff Thorpe [Mon, 1 Oct 2001 16:39:58 +0000 (16:39 +0000)]
The cleanup stack in ENGINE changed slightly, so this "make update" is
needed.

17 years agoThe STACK macros take care of casting to and from the designated item type
Geoff Thorpe [Mon, 1 Oct 2001 16:26:00 +0000 (16:26 +0000)]
The STACK macros take care of casting to and from the designated item type
of the stack, and the (void *) type used in the underlying sk_***
functions.  However, declaring a STACK_OF(type) where type is a *function*
type implicitly involves casts between function pointers and data pointers.
That's a no-no. This changes the ENGINE_CLEANUP handling to use a regular
data type in the stack.

17 years agoMake an (overdue) note about the recent ENGINE restructuring. Apart from
Geoff Thorpe [Mon, 1 Oct 2001 15:56:25 +0000 (15:56 +0000)]
Make an (overdue) note about the recent ENGINE restructuring. Apart from
a few items however, most of the details are deferred to the
crypto/engine/README file.

17 years agoMake "openssl engine -c" list any supported digests as well as supported
Geoff Thorpe [Mon, 1 Oct 2001 15:41:31 +0000 (15:41 +0000)]
Make "openssl engine -c" list any supported digests as well as supported
ciphers.

17 years agoTypos (Chris Pepper <pepper@mail.reppep.com>)
Lutz Jänicke [Mon, 1 Oct 2001 14:43:47 +0000 (14:43 +0000)]
Typos (Chris Pepper <pepper@mail.reppep.com>)

17 years agoChange HZ in speed to rely on sysconf() if the clock tick is available
Richard Levitte [Fri, 28 Sep 2001 10:34:48 +0000 (10:34 +0000)]
Change HZ in speed to rely on sysconf() if the clock tick is available
that way.  Synchronise s_time with these changes.

17 years agoENGINE_register_all_complete() will register all implementations of all
Geoff Thorpe [Fri, 28 Sep 2001 02:25:14 +0000 (02:25 +0000)]
ENGINE_register_all_complete() will register all implementations of all
algorithms present in all loaded ENGINEs. The result is that if any of
those ENGINEs successfully initialises, and the ENGINE_TABLE_FLAG_NOINIT
flag isn't set, then they will always be used (and cached as defaults) in
preference to software implementations. Ie. accidental auto-detection of
acceleration hardware :-)

This change stops all implementations being automatically registered in
"openssl" sub-commands, so that the "setup_engine()" handler in apps.c
controls which ENGINEs are registered for use. A special case has been
added that will revert to this "auto-detect" logic, ie. if the "-engine"
switch is used as;
   -engine auto