Rich Salz [Sat, 22 May 2021 19:57:07 +0000 (15:57 -0400)]
Remove engine_table_select_int
Add missing file/line args and call it engine_table_select
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15419)
Juergen Christ [Thu, 20 May 2021 11:27:43 +0000 (13:27 +0200)]
Fix warning in gf_serialize
Compiling under -Werror fails in gf_serialize:
crypto/ec/curve448/f_generic.c:21:27: error: argument 1 of type 'uint8_t[56]' {aka 'unsigned char[56]'} with mismatched bound [-Werror=array-parameter=]
21 | void gf_serialize(uint8_t serial[SER_BYTES], const gf x, int with_hibit)
| ~~~~~~~~^~~~~~~~~~~~~~~~~
In file included from crypto/ec/curve448/f_generic.c:12:
crypto/ec/curve448/field.h:65:28: note: previously declared as 'uint8_t *' {aka 'unsigned char *'}
void gf_serialize(uint8_t *serial, const gf x, int with_highbit);
~~~~~~~~~^~~~~~
Changed parameter to pointer to fix this warning.
Signed-off-by: Juergen Christ <jchrist@linux.ibm.com> Reviewed-by: Paul Dale <pauli@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15376)
Florian Mickler [Mon, 13 Jan 2020 01:06:49 +0000 (02:06 +0100)]
openssl srp: make index.txt parsing error more verbose
If index.txt exists but has some problems (like for example
consisting of a single \n character or number of fields wrong in one of the lines)
then openssl will just exit. This fixes it by printing an error when
load_index returns null.
Reviewed-by: Paul Dale <pauli@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15360)
Florian Mickler [Mon, 13 Jan 2020 01:05:22 +0000 (02:05 +0100)]
openssl ocsp: make index.txt parsing error more verbose
If index.txt exists but has some problems (like for example consisting of a single \n character in it,
or some field-number error in one of the lines) openssl will just exit without any error message.
Reviewed-by: Paul Dale <pauli@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15360)
Matt Caswell [Fri, 21 May 2021 11:21:32 +0000 (12:21 +0100)]
Don't try the same decoder multiple times
The function collect_decoder decides whether a given decoder should be
tried or not. It loops through all the names for matching keymgmts to
see if any are a match or not. If there is a match then the decoder gets
added. However, each keymgmt may have multiple aliases and a decoder was
being added for each one. For example DHX has 4 alias names, and therefore
4 instances of the DHX decoder were added and being tried.
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15404)
Matt Caswell [Thu, 20 May 2021 13:02:12 +0000 (14:02 +0100)]
Remove some perl 5.14 use from rsaz-avx512.pl
The non-destructive substitution syntax (s///r), was introduced in perl
5.14. We need to support 5.10 and above.
Fixes #15378
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15379)
Shane Lontis [Fri, 14 May 2021 03:08:42 +0000 (13:08 +1000)]
Rename the field 'provctx and data' to 'algctx' inside some objects containing
pointers to provider size algorithm contexts.
Fixes #14284
The gettable_ctx_params methods were confusingly passing a 'provctx' and
a provider context which are completely different objects.
Some objects such as EVP_KDF used 'data' while others such as EVP_MD used 'provctx'.
For libcrypto this 'ctx' is an opaque ptr returned when a providers algorithm
implementation creates an internal context using a new_ctx() method.
Hence the new name 'algctx'.
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15275)
Matt Caswell [Thu, 20 May 2021 10:52:56 +0000 (11:52 +0100)]
Fix a memleak in the FIPS provider
If the DRBG is used within the scope of the FIPS OSSL_provider_init
function then it attempts to register a thread callback via c_thread_start.
However the implementation of c_thread_start assumed that the provider's
provctx was already present. However because OSSL_provider_init is still
running it was actually NULL. This means the thread callback fail to work
correctly and a memory leak resulted.
Instead of having c_thread_start use the provctx as the callback argument
we change the definition of c_thread_start to have an explicit callback
argument to use.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15278)
Richard Levitte [Mon, 17 May 2021 21:10:11 +0000 (23:10 +0200)]
Disable loader_attic by default on VMS
The reason is that it currently doesn't build properly, due to the of
pvkfmt.c, causing multiply defined symbols since libcrypto exports
them as well. At the same time, it can't do without that source file,
or it won't have access to certain internal symbols from there.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15320)
Richard Levitte [Wed, 19 May 2021 08:57:48 +0000 (10:57 +0200)]
Configurations/descrip.mms.tmpl: rework the inclusion hacks
Because VMS C has some trouble with recursive inclusion of header
files, we have had to help it out for object files where there is such
an inclusion structure.
Previously, we did so with temporary logical names that were the same
as the first directory in an inclusion, so for example, to enable this
inclusion (found in ssl/ssl_local.h), we created the logical name
"record" when building any of the object files in the ssl/
subdirectories:
#include "record/record.h"
However, there is another way with the VMS C compiler, to selectively
specify extra include directories in Unix form directly to the
compiler. The logic is that from the directory where the source file
to compile is located, the specified inclusion directory merged with
the inclusion string should be able to access to specified header
file.
So for example, when a file in ssl/record/ is compiled, the following
inclusion is found:
#include "../ssl_local.h"
So far so good, VMS C handles it properly. However, the recursive
inclusion of "record/record.h" fails. However, if the compiler is
helped out a little bit, with the following extra qualifier, then it
works:
/INCLUDE="../"
The reason is that the compiler merges "../" and "record/record.h"
into "../record/record.h", which is the correct path to that header
file from the directory of the source file being compiled.
All that remained was to figure out all places where this trouble may
occur, and specify extra Unix formatted inclusion directories to
specify on per object file basis.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15369)
Pauli [Fri, 21 May 2021 01:19:30 +0000 (11:19 +1000)]
property: convert integers to strings properly.
The int64_t type was converted to int (truncation).
Negative values were not handled at all.
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15396)
Richard Levitte [Thu, 20 May 2021 08:31:21 +0000 (10:31 +0200)]
VMS: don't use app_malloc() in apps/lib/vms_decc_argv.c
The reason being that it would otherwise force test programs to link
with all of libapps.a, which unfortunately causes multiple symbol
definition issues.
The quick and dirty fix is to use OPENSSL_malloc() instead of
app_malloc() in apps/lib/vms_decc_argv.c, and clean up libapps.a
later.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15368)
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15371)
Richard Levitte [Wed, 19 May 2021 16:51:07 +0000 (18:51 +0200)]
PROV: Relegate most of the FIPS provider code to libfips.a
provider/fips/fipsprov.c contains a number of symbols that get used by
anything that's included in libfips.a, at least on Unix.
Unfortunately, there are platforms that do not support resolving
symbols to things that are already included in the end product (module
in this case) being built; they only support resolving symbols with
what comes next in the linking process.
The offending symbols in this case are FIPS_security_check_enabled,
c_thread_start and ossl_fips_intern_provider_init.
We resolve this by placing provider/fips/fipsprov.c in libfips.a along
with everything else there. That takes care of the offending symbols.
What remains is to ensure that there is an entry point in an object
file used directly when linking the module, providers/fips/fips_entry.c
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15370)
Matt Caswell [Tue, 18 May 2021 16:39:56 +0000 (17:39 +0100)]
Cleanup the missing*.txt files
One macro existed that was added since 1.1.1 and was undocumented. This
had been added to missingmacro.txt. This is the wrong approach and so
has been removed from there.
There were some entries in missingcrypto.txt that don't exist as functions
at all. There were also some which were in fact documented.
Additionally 2 entries from missingcrypto.txt have been moved to
missingmacro.txt. These entries existed in 1.1.1 and were undocumented. In
master they have been deprecated and compatibility macros for them
implemented. The replacement functions have been documented.
An entry in missingcrypto111.txt was not in alphabetical order (and was
also) duplicated, but the equivalent entry in missingcrypto.txt was in the
correct place. This has been corrected to make comparisons between the files
easier.
Finally a function has been added to missingcrypto111.txt. This function
did exist in 1.1.1 and was undocumented. Its unclear why this wasn't in
missingcrypto111.txt to start with.
Reviewed-by: Paul Dale <pauli@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15333)
Matt Caswell [Wed, 19 May 2021 10:46:00 +0000 (11:46 +0100)]
Clean up the "fips" option to Configure
Don't die if someone says "fips" instead of "enable-fips"
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15346)
Richard Levitte [Wed, 19 May 2021 07:43:13 +0000 (09:43 +0200)]
VMS: Copy __DECC_INCLUDE_{PROLOGUE,EPILOGUE}.H to more places
Every inclusion directory related to a library we build need these two
files. That signals to any other module using anything from these
libraries what to expect in terms of case sensitivity as well as how
long symbol names are dealt with.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15341)
Richard Levitte [Sat, 15 May 2021 05:45:31 +0000 (07:45 +0200)]
EVP: Modify EVP_PKEY_export() to handle legacy EVP_PKEYs
We use a fake EVP_KEYMGMT import function with the newly modified
EVP_PKEY_ASN1_METHOD export_to function to pass the exported
OSSL_PARAM array directly to the EVP_PKEY_export() callback instead of
exporting to an actual provided key and then getting the OSSL_PARAM
array from there, just to throw away that key again.
Fixes #15290
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15293)
Richard Levitte [Sat, 15 May 2021 05:43:06 +0000 (07:43 +0200)]
Modify EVP_PKEY_ASN1_METHOD's export_to function to take an importer
We previously took an EVP_KEYMGMT pointer, but now found it necessary
to use a different import function in some cases. Since that's the
only thing we use from EVP_KEYMGMT, we might as well pass the import
function directly, allowing for some flexibility in how export_to is
used.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15293)
Matt Caswell [Thu, 20 May 2021 09:47:47 +0000 (10:47 +0100)]
Refer to the migration guide rather than the wiki in our announcements
We now have a migration guide which should be the definitive source of
information for upgrading from a previous version of OpenSSL.
Fixes #15186
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com> Reviewed-by: Tim Hudson <tjh@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15373)
Pauli [Tue, 18 May 2021 08:45:31 +0000 (18:45 +1000)]
app: add a -store_loaders option to list.
Fixes #15307
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15323)
Matt Caswell [Tue, 11 May 2021 15:50:27 +0000 (16:50 +0100)]
Ensure mirroring of properties works for subsequent updates
If the global properties are updated after a provider with a child libctx
has already started we need to make sure those updates are mirrored in
that child.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15242)
Benjamin Kaduk [Fri, 2 Apr 2021 17:04:24 +0000 (10:04 -0700)]
Let SSL_new_session_ticket() enter init immediately
The initial implementation always deferred the generation of the
requested ticket(s) until the next application write, but this
is not a great fit for what it actually does, architecturally wise.
A request to send a session ticket means entering back into the
handshake state machine (or "in init", as it's known in the
implementation). The state machine transition is not something that
only occurs at an application-data write, and in general could occur at
any time. The only constraint is that we can't enter "init" while in
the middle of writing application data. In such cases we will need to
wait until the next TLS record boundary to enter the state machine,
as is currently done.
However, there is no reason why we cannot enter the handshake state
machine immediately in SSL_new_session_ticket() if there are no
application writes pending. Doing so provides a cleaner API surface to
the application, as then calling SSL_do_handshake() suffices to drive
the actual ticket generation. In the previous state of affairs a dummy
zero-length SSL_write() would be needed to trigger the ticket
generation, which is a logical mismatch in the type of operation being
performed.
This commit should only change whether SSL_do_handshake() vs zero-length
SSL_write() is needed to immediately generate a ticket after the
SSL_new_session_ticket() call -- the default behavior is still to defer
the actual write until there is other application data to write, unless
the application requests otherwise.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14817)