Now that we have the foundation for the "unified" build scheme in
place, we add build.info files. They have been generated from the
Makefiles in the same directories. Things that are platform specific
will appear in later commits.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The "unified" build scheme revolves around small information files,
build.info, which each describe their own bit of everything that needs
to be built, using a mini-language described in Configurations/README.
The information in build.info file contain references to source files
and final result. Object files are not mentioned at all, they are
simply from source files. Because of this, all the *_obj items in
Configurations/*.conf are renamed to *_asm_src and the files listed
in the values are change from object files to their corresponding
source files. For the sake of the other build schemes, Configure
generates corresponding *_obj entries in %target.
Furthermore, the "unified" build scheme supports having a build
directory tree separate from the source directry tree.
All paths in a build.info file is assumed to be relative to its
location, either within the source tree or within the build tree.
Reviewed-by: Andy Polyakov <appro@openssl.org>
When auxiliary data contains only reject entries, continue to trust
self-signed objects just as when no auxiliary data is present.
This makes it possible to reject specific uses without changing
what's accepted (and thus overring the underlying EKU).
Added new supported certs and doubled test count from 38 to 76.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
This includes basic constraints, key usages, issuer EKUs and auxiliary
trust OIDs (given a trust suitably related to the intended purpose).
Added tests and updated documentation.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
d2i_ECPrivateKey always caculates the public key so there is
no need to caculate it again in eckey_priv_decode().
Reviewed-by: Rich Salz <rsalz@openssl.org>
New functions to return internal pointer for order and cofactor. This
avoids the need to allocate a new BIGNUM which to copy the value to.
Simplify code to use new functions.
Reviewed-by: Rich Salz <rsalz@openssl.org>
There was an unused macro in ssl_locl.h that used an internal
type, so I removed it.
Move bio_st from bio.h to ossl_type.h
Reviewed-by: Andy Polyakov <appro@openssl.org>
As a side-effect of opaque x509, ex_flags were looked up too early,
before additional policy cache updates.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
The lflags configuration had a weird syntax with a % as separator. If
it was present, whatever came before ended up as PEX_LIBS in Makefile
(usually, this is LDFLAGS), while whatever came after ended up as
EX_LIBS.
This change splits that item into lflags and ex_libs, making their use
more explicit.
Also, PEX_LIBS in all the Makefiles are renamed to LDFLAGS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
By default X509_check_trust() trusts self-signed certificates from
the trust store that have no explicit local trust/reject oids
encapsulated as a "TRUSTED CERTIFICATE" object. (See the -addtrust
and -trustout options of x509(1)).
This commit adds a flag that makes it possible to distinguish between
that implicit trust, and explicit auxiliary settings.
With flags |= X509_TRUST_NO_SS_COMPAT, a certificate is only trusted
via explicit trust settings.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
The use of the uninitialized buffer in the RNG has no real security
benefits and is only a nuisance when using memory sanitizers.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
During precomputation if the group given is well known then we memcpy a
well known precomputation. However we go the wrong label in the code and
don't store the data properly. Consequently if we call have_precompute_mult
the data isn't there and we return 0.
RT#3600
Reviewed-by: Richard Levitte <levitte@openssl.org>
The function DH_check_pub_key() was missing some return value checks in
some calls to BN functions.
RT#4278
Reviewed-by: Andy Polyakov <appro@openssl.org>
These tests are not built, and only usable as hand-tests so not
worth moving into our test framework.
This closes https://github.com/openssl/openssl/pull/561 and RT 4252
Reviewed-by: Richard Levitte <levitte@openssl.org>
Historically OpenSSL only ever generated DH parameters based on "safe"
primes. More recently (in version 1.0.2) support was provided for
generating X9.42 style parameter files such as those required for RFC
5114 support. The primes used in such files may not be "safe". Where an
application is using DH configured with parameters based on primes that
are not "safe" then an attacker could use this fact to find a peer's
private DH exponent. This attack requires that the attacker complete
multiple handshakes in which the peer uses the same DH exponent.
A simple mitigation is to ensure that y^q (mod p) == 1
CVE-2016-0701
Issue reported by Antonio Sanso.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Many options for supporting optimizations for legacy crypto on legacy
platforms have been removed. This simplifies the source code and
does not really penalize anyone.
DES_PTR (always on)
DES_RISC1, DES_RISC2 (always off)
DES_INT (always 'unsigned int')
DES_UNROLL (always on)
BF_PTR (always on) BF_PTR2 (removed)
MD2_CHAR, MD2_LONG (always 'unsigned char')
IDEA_SHORT, IDEA_LONG (always 'unsigned int')
RC2_SHORT, RC2_LONG (always 'unsigned int')
RC4_LONG (only int and char (for assembler) are supported)
RC4_CHUNK (always long), RC_CHUNK_LL (removed)
RC4_INDEX (always on)
And also make D_ENCRYPT macro more clear (@appro)
This is done in consultation with Andy.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This was done by the following
find . -name '*.[ch]' | /tmp/pl
where /tmp/pl is the following three-line script:
print unless $. == 1 && m@/\* .*\.[ch] \*/@;
close ARGV if eof; # Close file to reset $.
And then some hand-editing of other files.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Now that we're using templates, we should warn people not to edit the
resulting file. We do it through util/dofile.pl, which is enhanced
with an option to tell what file it was called from. We also change
the calls so the template files are on the command line instead of
being redirected through standard input. That way, we can display
something like this (example taken from include/openssl/opensslconf.h):
/* WARNING: do not edit! */
/* Generated by Configure from include/openssl/opensslconf.h.in */
Reviewed-by: Rich Salz <rsalz@openssl.org>
Move opensslconf.h.in to include/openssl.
Split off DES,BN,RC4 stuff into separate header file
templates in crypto/include/internal/*_conf.h.in
Reviewed-by: Richard Levitte <levitte@openssl.org>
This is an internal facility, never documented, not for
public consumption. Move it into ssl (where it's only used
for DTLS).
I also made the typedef's for pqueue and pitem follow our style: they
name structures, not pointers.
Reviewed-by: Richard Levitte <levitte@openssl.org>
When experimental-store is enabled, it does not compile due to the
change to opaque data structures.
Change CRYPTO_add() to EVP_PKEY_up_ref() as needed.
Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
RT: #4263, GH: #579
The turn has come to have crypto/opensslconf.h.in get run through
util/dofile.pl. The consequence is that a large number of variables
get moved to the %config table.
Also, the string variables $openssl_*, which were populated with cpp
lines, all being of the form "#define SOMETHING", were converted into
ARRAY refs in %config values, containing just the list of macros to be
defined.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Returning untrusted is enough for for full chains that end in
self-signed roots, because when explicit trust is specified it
suppresses the default blanket trust of self-signed objects.
But for partial chains, this is not enough, because absent a similar
trust-self-signed policy, non matching EKUs are indistinguishable
from lack of EKU constraints.
Therefore, failure to match any trusted purpose must trigger an
explicit reject.
Reviewed-by: Richard Levitte <levitte@openssl.org>
When DANE-EE(3) matches or either of DANE-EE/PKIX-EE fails, we don't
build a chain at all, but rather succeed or fail with just the leaf
certificate. In either case also check for Suite-B violations.
As unlikely as it may seem that anyone would enable both DANE and
Suite-B, we should do what the application asks.
Took the opportunity to eliminate the "cb" variables in x509_vfy.c,
just call ctx->verify_cb(ok, ctx)
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Rename 'update' to 'generate'. Rather than recurse, just explicitly
call the three generate targets directly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The GOST engine is now out of date and is removed by this commit. An up
to date GOST engine is now being maintained in an external repository.
See:
https://wiki.openssl.org/index.php/Binaries
Reviewed-by: Richard Levitte <levitte@openssl.org>
It seems risky in the context of cross-signed certificates when the
same certificate might have multiple potential issuers. Also rarely
used, since chains in OpenSSL typically only employ self-signed
trust-anchors, whose self-signatures are not checked, while untrusted
certificates are generally ephemeral.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Remove lint, tags, dclean, tests.
This is prep for a new makedepend scheme.
This is temporary pending unified makefile, and might help it.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Author: Remi Gacogne <rgacogne-github@coredump.fr>
GH334: Add an OCSP_SINGLERESP_get0_id() accessor to the OCSP_CERTID of
a OCSP_SINGLERESP. It is possible to do it the other way around using
OCSP_resp_find(), but this is more efficient when you have a tree indexed
by OCSP_CERTID, like haproxy does. (This is also RT4251)
Author: Marek Klein <kleinmrk@gmail.com>
GH556: OCSP_resp_get_produced_at() accessor to the producedAt of a
OCSP_BASICRESP
GH555: TS_STATUS_INFO_get_status(), TS_STATUS_INFO_get_text() and
TS_STATUS_INFO_get_failure_info() accessors for a TS_STATUS_INFO
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
- bugfix: should not treat '--' as invalid domain substring.
- '-' should not be the first letter of a domain
Signed-off-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The entropy-gathering daemon is used only on a small number of machines.
Provide a configure knob so that EGD support can be disabled by default
but re-enabled on those systems that do need it.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Add CRYPTO_EX_DATA add EndC_KEY_[gs]et_method, From Roumen Petrov.
Had to add various exdata calls to init/copy/free the exdata.
Had to remove const from some EC functions because exdata isn't
const-correct. :(
Also remove EC_EXTRA_DATA and use a union to hold the possible
pre-computed values and an enum to tell which value is in the
union. (Rich Salz)
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
For some strange reason opensslconf.h was only defining DES_LONG
when included via des.h, but that's exceedingly fragile (as a
result of include guards the include via des.h might not actually
process the content again).
Ripped out the nesting constraint, now always define OSSL_DES_LONG
if not already defined. Note, this could just be DES_LONG, but
trying to avoid exposing DES_LONG in places where it has never been
seen before, so it is up to des.h to actually define DES_LONG as
OSSL_DES_LONG.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Create Makefile's from Makefile.in
Rename Makefile.org to Makefile.in
Rename Makefiles to Makefile.in
Address review feedback from Viktor and Richard
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
They all stop including evp_locl.h, so we also take care of their
adaptation to opaque EVP_CIPHER_CTX, as was promised in an earlier
commit.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We follow the method used for EVP_MD.
Also, move all the internal EVP_CIPHER building macros from evp_locl.h
to evp_int.h. This will benefit our builtin EVP_CIPHERs.
Reviewed-by: Rich Salz <rsalz@openssl.org>
EVP_CIPHER_CTX_flags was returning the flags of its associated
EVP_CIPHER. However, EVP_CIPHER_CTX has flags of its own, so this
function is quite confusing and therefore error prone.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Note: there's a larger number of implementations in crypto/evp/ that
aren't affected because they include evp_locl.h. They will be handled
in a separate commit.
Reviewed-by: Rich Salz <rsalz@openssl.org>
New functions:
- EVP_CIPHER_CTX_encrypting()
- EVP_CIPHER_CTX_iv()
- EVP_CIPHER_CTX_iv_noconst()
- EVP_CIPHER_CTX_original_iv()
- EVP_CIPHER_CTX_buf_noconst()
- EVP_CIPHER_CTX_num()
- EVP_CIPHER_CTX_set_num()
- EVP_CIPHER_CTX_cipher_data()
- EVP_CIPHER_CTX_new_cipher_data()
Note that the accessors / writers for iv, buf and num may go away, as
those rather belong in the implementation's own structure (cipher_data)
when the implementation would affect them (that would be the case when
they are flagged EVP_CIPH_CUSTOM_IV or EVP_CIPH_FLAG_CUSTOM_CIPHER).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Following the method used for EVP_MD_CTX and HMAC_CTX,
EVP_CIPHER_CTX_init and EVP_CIPHER_CTX_cleanup are joined together
into one function, EVP_CIPHER_CTX_reset, with EVP_CIPHER_CTX_init kept
as an alias.
EVP_CIPHER_CTX_cleanup fills no purpose of its own any more and is
therefore removed.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Make LHASH_OF use static inline functions.
Add new lh_get_down_load and lh_set_down_load functions and their
typesafe inline equivalents.
Make lh_error a function instead of a macro.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Make CRYPTO_mem_leaks() and CRYPTO_mem_leaks_fp() return a status value.
Update documentation. Don't abort() if there are leaks.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
In order for mkdep to find #ifdef'd functions, they must be
wrapped (in the header file) with
#ifndef OPENSSL_NO_...
So do that for various CRYPTO_mem_debug... things.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Provide backwards-compatiblity for functions, macros and include
files if OPENSSL_API_COMPAT is either not defined or defined less
than the version number of the release in which the feature was
deprecated.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Only two macros CRYPTO_MDEBUG and CRYPTO_MDEBUG_ABORT to control this.
If CRYPTO_MDEBUG is not set, #ifdef out the whole debug machinery.
(Thanks to Jakob Bohm for the suggestion!)
Make the "change wrapper functions" be the only paradigm.
Wrote documentation!
Format the 'set func' functions so their paramlists are legible.
Format some multi-line comments.
Remove ability to get/set the "memory debug" functions at runtme.
Remove MemCheck_* and CRYPTO_malloc_debug_init macros.
Add CRYPTO_mem_debug(int flag) function.
Add test/memleaktest.
Rename CRYPTO_malloc_init to OPENSSL_malloc_init; remove needless calls.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Applications wishing to include their own stacks now just need to include
DEFINE_STACK_OF(foo)
in a header file.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Also tweak some of the code in demos/bio, to enable interactive
testing of BIO_s_accept's use of SSL_dup. Changed the sconnect
client to authenticate the server, which now exercises the new
SSL_set1_host() function.
Reviewed-by: Richard Levitte <levitte@openssl.org>
If there's a failure allocating md_data, the destination pctx will have
a shared pointer with the source EVP_MD_CTX, which will lead to problems
when either the source or the destination is freed.
Reviewed-by: Stephen Henson <steve@openssl.org>
If DSA parameters are absent return -1 (for unknown) in DSA_security_bits.
If parameters are absent when a certificate is set in an SSL/SSL_CTX
structure this will reject the certificate by default. This will cause DSA
certificates which omit parameters to be rejected but that is never (?)
done in practice.
Thanks to Brian 'geeknik' Carpenter for reporting this issue.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
When it's the last item that is removed int_thread_hash == hash and we would
still call int_thread_release(&hash) while hash is already freed. So
int_thread_release would compare that dangling pointer to NULL which is
undefined behaviour. Instead do already what int_thread_release() would do,
and make the call do nothing instead.
Reviewed-by: Rich Salz <rsalz@openssl.org>
RT: #4155, MR: #1519
Rename to OPENSSL_mem_debug_{push,pop}.
Remove simple calls; keep only calls used in recursive functions.
Ensure we always push, to simplify so that we can always pop
Reviewed-by: Richard Levitte <levitte@openssl.org>
Just like *_clear_free routines. Previously undocumented, used
a half-dozen times within OpenSSL source.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Some URLs in the source code ended up getting mangled by indent. This fixes
it. Based on a patch supplied by Arnaud Lacombe <al@aerilon.ca>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Rename BUF_{strdup,strlcat,strlcpy,memdup,strndup,strnlen}
to OPENSSL_{strdup,strlcat,strlcpy,memdup,strndup,strnlen}
Add #define's for the old names.
Add CRYPTO_{memdup,strndup}, called by OPENSSL_{memdup,strndup} macros.
Reviewed-by: Tim Hudson <tjh@openssl.org>
New functions EC_POINT_point2buf and EC_KEY_key2buf which encode
a point and allocate a buffer in one call.
New function EC_KEY_oct2key() which sets public key in an EC_KEY
structure from an encoded point.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Solaris builds were failing during async compilation because the .o files
created from compiling the corresponding .c files held in async/arch were
ending up in the top level async directory. Consequently the link fails
because it can't find the .o files.
Thanks to Richard Levitte for pointing me in the right direction on this.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Make EVP_PKEY_copy_parameters() work if the destination has no type
(e.g. if obtained from EVP_PKEY_new()) or the underlying key is NULL.
This is useful where we want to copy the parameters from an existing
key to a new key.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Commit 2b0180c37f attempted to do this but
only hit one of many BN_mod_exp codepaths. Fix remaining variants and add
a test for each method.
Thanks to Hanno Boeck for reporting this issue.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Rename ENGINE _EC_KEY functions to _EC.
Add support for EC_KEY_METHOD in ENGINE_set_default et al. Copy
ec_meth.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add set_group, set_public and set_private methods. An EC_KEY_METHOD can use
these to perform any appropriate operation when the key components are set,
such as caching data in some more convenient ENGINE specific format or
returning an error if the parameters are invalid or the operation is
not supported.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Rename ecdh_compute_key into ossl_ecdh_compute_key and modify it
to use EC error codes. Remove superfluous old ECDH functions.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add keygen to EC_KEY_METHOD. Redirect EC_KEY_generate_key through
method and set the current EC key generation function as the default.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add EC_KEY_METHOD. This is part of the EC revision and will make EC behave
more like other algorithms. Specifically:
EC_KEY_METHOD is part of EC_KEY.
It is part of ENGINE.
Default or key specific implementations can be provided to redirect some
or all operations.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Looking over names, it seems like we usually use names ending with
_new and _free as object constructors and destructors. Also, since
EVP_MD_CTX_init is now used to reset a EVP_MD_CTX, it might as well be
named accordingly.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The idea is that with EVP_MD_CTX_create() and EVP_MD_CTX_destroy(),
EVP_MD_CTX_cleanup and EVP_MD_CTX_init is not used the same as before.
Instead, we need a single function that can be used to reinitialise an
existing EVP_MD_CTX that's been created with EVP_MD_CTX_create()
previously. Combining EVP_MD_CTX_cleanup and EVP_MD_CTX_init into
that one function is the answer.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This moves the definition to crypto/hmac/hmac_lcl.h. Constructor and
destructor added, and the typedef moved to include/openssl/ossl_typ.h.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This moves the definition to crypto/include/internal/evp_int.h and
defines all the necessary method creators, destructors, writers and
accessors. The name standard for the latter is inspired from the
corresponding functions to manipulate UI methods.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This change required some special treatment, as HMAC is intertwined
with EVP_MD. For now, all local HMAC_CTX variables MUST be
initialised with HMAC_CTX_EMPTY, or whatever happens to be on the
stack will be mistaken for actual pointers to EVP_MD_CTX. This will
change as soon as HMAC_CTX becomes opaque.
Also, since HMAC_CTX_init() can fail now, its return type changes from
void to int, and it will return 0 on failure, 1 on success.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This moves the definitionto crypto/evp/evp_locl.h, along with a few
associated accessor macros. A few accessor/writer functions added.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Remove RSA_FLAG_SIGN_VER: this was origininally used to retain binary
compatibility after RSA_METHOD was extended to include rsa_sign and
rsa_verify fields. It is no longer needed.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Remove sign/verify and required_pkey_type fields of EVP_MD: these are a
legacy from when digests were linked to public key types. All signing is
now handled by the corresponding EVP_PKEY_METHOD.
Only allow supported digest types in RSA EVP_PKEY_METHOD: other algorithms
already block unsupported types.
Remove now obsolete EVP_dss1() and EVP_ecdsa().
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add CRYPTO_free_ex_index (for shared libraries)
Unify and complete the documentation for all "ex_data" API's and objects.
Replace xxx_get_ex_new_index functions with a macro.
Added an exdata test.
Renamed the ex_data internal datatypes.
Reviewed-by: Matt Caswell <matt@openssl.org>
The SRP_create_verifier_BN function goes to the |err| label if the |salt|
value passed to it is NULL. It is then deref'd.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The function rsa_ossl_mod_exp uses the function BN_with_flags to create a
temporary copy (local_r1) of a BIGNUM (r1) with modified flags. This
temporary copy shares some state with the original r1. If the state of r1
gets updated then local_r1's state will be stale. This was occurring in the
function so that when local_r1 was freed a call to bn_check_top was made
which failed an assert due to the stale state. To resolve this we must free
local_r1 immediately after we have finished using it and not wait until the
end of the function.
This problem prompted a review of all BN_with_flag usage within the
codebase. All other usage appears to be correct, although often not
obviously so. This commit refactors things to make it much clearer for
these other uses.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Add a ctrl to EVP_md5_sha1() to handle the additional operations needed
to handle SSL v3 client authentication and finished message.
Reviewed-by: Tim Hudson <tjh@openssl.org>
This patch contains the necessary changes to provide GOST 2012
ciphersuites in TLS. It requires the use of an external GOST 2012 engine.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
In the async code for MacOS/X define _XOPEN_SOURCE (if not already
defined) as early as possible. We must do this before including
any header files, because on MacOS/X <stlib.h> includes <signal.h>
which includes <ucontext.h>. If we delay defining _XOPEN_SOURCE
and include <ucontext.h> after various system headers are included,
we are very likely to end up with the wrong (truncated) definition
of ucontext_t.
Also, better error handling and some code cleanup in POSIX fibre
construction and destruction. We make sure that async_fibre_makecontext()
always initializes the fibre to a state that can be freed.
For all implementations, check for error returns from
async_fibre_makecontext().
Reviewed-by: Matt Caswell <matt@openssl.org>
Implements Thread Local Storage in the windows async port. This also has
some knock on effects to the posix and null implementations.
Reviewed-by: Rich Salz <rsalz@openssl.org>
In theory the pthreads approach for Thread Local Storage should be more
portable.
This also changes some APIs in order to accommodate this change. In
particular ASYNC_init_pool is renamed ASYNC_init_thread and
ASYNC_free_pool is renamed ASYNC_cleanup_thread. Also introduced ASYNC_init
and ASYNC_cleanup.
Reviewed-by: Rich Salz <rsalz@openssl.org>
A lot of the pool handling code was in the arch specific files, but was
actually boiler plate and the same across the implementations. This commit
moves as much code as possible out of the arch specific files.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We were using _pipe to create a pipe on windows. This uses the "int" type
for its file descriptor for compatibility. However most windows functions
expect to use a "HANDLE". Probably we could get away with just casting but
it seems more robust to use the proper type and main stream windows
functions.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The rand code can aquire locks and then attempt crypto operations. This
can end up in a deadlock if we are using an async engine, because control
returns back to the user code whilst still holding the lock. We need to
force synchronous operation for these sections of code.
Reviewed-by: Rich Salz <rsalz@openssl.org>
There are potential deadlock situations that can occur if code executing
within the context of a job aquires a lock, and then pauses the job. This
adds an ability to temporarily block pauses from occuring whilst performing
work and holding a lock.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Even with _XOPEN_SOURCE defined OS-X still displays warnings that
makecontext and friends are deprecated. This isn't a problem until you
try and build with --strict-warnings, and the build fails. This change
suppresses the warnings. We know they are deprecated but there is no
alternative!
Reviewed-by: Rich Salz <rsalz@openssl.org>
async_fibre_makecontext was initialise the fibre first and then calling
getcontext(). It should be the other way around because the getcontext
call may overwrite some of the things we just initialised. This didn't
cause an issue on Linux and so the problem went unnoticed. On OS-X it
causes a crash.
Reviewed-by: Rich Salz <rsalz@openssl.org>
_longjmp/_setjmp do not manipulate the signal mask whilst
longjmp/setjmp may do. Online sources suggest this could result
in a significant speed up in the context switching.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If config'd without -d (--debug), asynctest was crashing with:
*** longjmp causes uninitialized stack frame ***
This is because gcc will add certain checks for some functions
(including longjmp). The checks assume you can only longjmp down the
stack not up. However, if we are actually jumping to a different
fibre then it can appear as if we are going up the stack when we are
not really. This change disables the check.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add some clarifications to the async documentation. Also changed
ASYNC_pause_job() so that it returns success if you are not within the
context of a job. This is so that engines can be used either asynchronously
or synchronously and can treat an error from ASYNC_pause_job() as a real
error.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Don't recreate a new ASYNC_CTX every time we call ASYNC_start_job() - the
same one can be used for the life of the thread. Instead we only free it
up when we call ASYNC_free_pool().
Reviewed-by: Rich Salz <rsalz@openssl.org>
The ASYNC null implementation has not kept pace with the rest of the async
development and so was failing to compile.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Initial API implemented for notifying applications that an ASYNC_JOB
has completed. Currently only s_server is using this. The Dummy Async
engine "cheats" in that it notifies that it has completed *before* it
pauses the job. A normal async engine would not do that.
Only the posix version of this has been implemented so far, so it will
probably fail to compile on Windows at the moment.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It is expensive to create the ASYNC_JOB objects due to the "makecontext"
call. This change adds support for pools of ASYNC_JOB objects so that we
don't have to create a new ASYNC_JOB every time we want to use one.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Where we can we should use longjmp and setjmp in preference to swapcontext/
setcontext as they seem to be more performant.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Removed the function ASYNC_job_is_waiting() as it was redundant. The only
time user code has a handle on a job is when one is waiting, so all they
need to do is check whether the job is NULL. Also did some cleanups to
make sure the job really is NULL after it has been freed!
Reviewed-by: Rich Salz <rsalz@openssl.org>
Create a "null" async implementation for platforms that lack support. This
just does nothing when called and therefore performs synchronously.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The following entry points have been made async aware:
SSL_accept
SSL_read
SSL_write
Also added is a new mode - SSL_MODE_ASYNC. Calling the above functions with
the async mode enabled will initiate a new async job. If an async pause is
encountered whilst executing the job (such as for example if using SHA1/RSA
with the Dummy Async engine), then the above functions return with
SSL_WANT_ASYNC. Calling the functions again (with exactly the same args
as per non-blocking IO), will resume the job where it left off.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Provides support for running asynchronous jobs. Currently this is completely
stand alone. Future commits will integrate this into libssl and s_server/
s_client. An asynchronous capable engine will be required to see any benefit
from this capability.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We use the sysconf function to provide details about the page size in the
secure memory code. This function can return -1 on error so we should check
for this before proceeding.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
The problem remained unnoticed so far, because it's never called by default.
You have to craft OPENSSL_ppccap environment variable to trigger the problem.
Reviewed-by: Richard Levitte <levitte@openssl.org>
