summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Split out BN_div and BN_mod into ct and nonct versions for Internal use.beck2017-01-2117-58/+92
| | | | ok jsing@
* Make explicit _ct and _nonct versions of bn_mod_exp funcitons thatbeck2017-01-2119-47/+249
| | | | | | | | | | | | matter for constant time, and make the public interface only used external to the library. This moves us to a model where the important things are constant time versions unless you ask for them not to be, rather than the opposite. I'll continue with this method by method. Add regress tests for same. ok jsing@
* whitespacederaadt2017-01-212-5/+5
|
* 1. When shrinking a chunk allocation, compare the size of the currentotto2017-01-211-46/+87
| | | | | | | | | | allocation to the size of the new allocation (instead of the requested size). 2. Previously realloc takes the easy way and always reallocates if C is active. This commit fixes by carefully updating the recorded requested size in all cases, and writing the canary bytes in the proper location after reallocating. 3. Introduce defines to test if MALLOC_MOVE should be done and to compute the new value.
* Specify minimum and maximum protocol version for each method. This isjsing2017-01-216-6/+36
| | | | | | currently unused, but will be in the near future. ok beck@
* Place {DECLARE,IMPLEMENT}_OBJ_BSEARCH{_GLOBAL,}_CMP_FN macros undefjsing2017-01-211-1/+5
| | | | LIBRESSL_INTERNAL.
* Expand DECLARE_OBJ_BSEARCH_CMP_FN and IMPLEMENT_OBJ_BSEARCH_CMP_FN macros.jsing2017-01-216-28/+168
| | | | No change to generated assembly excluding line numbers.
* /usr/bin/unifdef -D MONT_MUL_MOD -D MONT_EXP_WORD -D RECP_MUL_MOD -m bn_exp.cbeck2017-01-211-23/+2
| | | | | with some style cleanup after. no binary change ok jsing@
* Expand DECLARE_OBJ_BSEARCH_CMP_FN and IMPLEMENT_OBJ_BSEARCH_CMP_FN macros.jsing2017-01-212-8/+38
| | | | No change to generated assembly excluding line numbers.
* Expand DECLARE_OBJ_BSEARCH_GLOBAL_CMP_FN macro.jsing2017-01-211-2/+2
| | | | No change in preprocessor output (ignoring whitespace and line numbers).
* Expand IMPLEMENT_OBJ_BSEARCH_GLOBAL_CMP_FN macro.jsing2017-01-211-2/+16
| | | | No change in generated assembly.
* fix bogus commentbeck2017-01-211-2/+2
| | | | ok jsing@
* Make return value of X509_verify_cert be consistent with the error code,beck2017-01-211-2/+10
| | | | | | with the caveat that we force V_OK when a user provided callback has us returning success. ok inoguchi@ jsing@
* rearrange pledge promises into the canonical order; easier to eyeballderaadt2017-01-2039-77/+77
|
* fix pledge for openssl ocsp - we will need tty to ask for a cert pwbeck2017-01-201-2/+2
|
* Rework internal_verify, mostly from OpenSSL. so we can progressbeck2017-01-201-102/+102
| | | | | towards cleaning up the V_OK stuff. ok kinichiro@
* fix openssl ocsp to not report sucess when the ocsp responder rejects usbeck2017-01-191-3/+3
| | | | ok deraadt@ krw@
* unhook ocsp test from the default since it currently requires network access andbeck2017-01-181-2/+1
| | | | therefore appears to break in bluhm's test setup
* Correctly tls_config_set_ca_file() return value (no effective change).jsing2017-01-171-2/+2
| | | | Spotted by inoguchi@
* whitespacederaadt2017-01-131-3/+3
|
* Inline strlen() call to reduce/simplify code.jsing2017-01-121-5/+2
|
* Inline tls_get_new_cb_bio() from the only place that it gets called,jsing2017-01-121-22/+11
| | | | | simplifying the code. Also check the provided read and write callbacks before assigning to the context.
* If tls_set_cbs() fails an error will already be specified, so do notjsing2017-01-122-8/+4
| | | | replace it with a less specific one.
* change two trailing Xr to Fn; ok schwarzejmc2017-01-122-6/+6
|
* Simplify the TLS callback BIO code - a pointer to the tls context can bejsing2017-01-121-66/+7
| | | | | | | stored directly in bio->ptr, rather than allocating and deallocating an intermediate struct. Diff from Marko Kreen <markokr at gmail dot com> - thanks!
* If tls_get_new_cb_bio() fails, an error will already be set.jsing2017-01-121-5/+2
|
* If no callbacks are specified, return after setting an error rather thanjsing2017-01-121-2/+4
| | | | | | continuing on. Also noticed by Marko Kreen.
* Add regress tests for libtls, which currently cover handshakes and closesjsing2017-01-123-1/+356
| | | | using callbacks, file descriptors and sockets.
* Add support for AArch64.patrick2017-01-112-0/+182
|
* Avoid leaking conninfo servername.jsing2017-01-091-1/+3
| | | | Issue found by and fix from Shuo Chen <chenshuo at chenshuo dot com>.
* Provide TLS_INT for consistency with libssl/libcrypto.jsing2017-01-092-2/+4
|
* add missing comma in the NAME sectionschwarze2017-01-071-3/+3
|
* fix a typo in an .Xr reported by jmc@schwarze2017-01-071-3/+3
|
* Add and remove some blank lines, in order to make X509_verify_cert()jsing2017-01-071-6/+4
| | | | (slightly) more readable.
* a little more cleanup;jmc2017-01-075-23/+23
|
* Revert part of r1.54 as there are at least two situations where we are stilljsing2017-01-071-4/+2
| | | | | | | returning ok == 1, with ctx->error not being X509_V_OK. Hopefully we can restore this behaviour once these are ironed out. Discussed with beck@
* correctly mark all documented macros found in <openssl/bn.h>schwarze2017-01-073-9/+11
|
* Use .Fn rather than .Xr for X509_VERIFY_PARAM_lookup(),schwarze2017-01-071-3/+3
| | | | | | | | fixing a dead link reported by jmc@. Only about half of X509_VERIFY_PARAM is documented so far, and the extensible lookup table feels like one of the more arcane features and probably not the next thing to document.
* Document X509_NAME_hash(3), listed in <openssl/x509.h>;schwarze2017-01-071-3/+20
| | | | | | | jmc@ reported that X509_LOOKUP_hash_dir(3) references it. Even though OpenSSL does not document it, given that it is used for file names that users have to create, it is sufficiently exposed to users to be worth documenting.
* Write a new manual page X509_STORE_load_locations(3) from scratch.schwarze2017-01-072-1/+119
| | | | | | | | | | | Not documented by OpenSSL, but listed in <openssl/x509_vfy.h> and referenced from X509_LOOKUP_hash_dir(3), and clearly more important than the latter. Fixes three dead links reported by jmc@. Most of the information from SSL_CTX_load_verify_locations(3) should probably be moved here, but not all, since the SSL page also talks about SSL servers and clients and the like. As i'm not completely sure regarding the boundaries, i'm leaving that as it is for now.
* Remove cross references to the undocumented functions X509_STORE_new(3)schwarze2017-01-062-7/+4
| | | | | | | and X509_STORE_add_lookup(3) reported by jmc@. Even though these functions are public, they seem more useful internally than for application programs, so now is not the time to document them.
* Delete a sentence containing a cross reference to an undocumentedschwarze2017-01-061-6/+2
| | | | | | | function that had the the sole purpose of discouraging its use. Not talking about it at all discourages using it even more. Dangling cross reference reported by jmc@.
* resolve duplication of names and prototypes in manuals related to ex_dataschwarze2017-01-065-67/+23
| | | | and sprinkle cross references instead; more work is obviously needed here
* Replace two dangling .Xrs to sk_*() macros with .Fn; reported by jmc@.schwarze2017-01-061-4/+6
| | | | | | The safestack stuff is the most ill-designed user interface i have seen so far in OpenSSL. It looks positively undocumentable. At least i'm not trying to document it right now.
* Delete a cross reference to the undocumented function X509_check_purpose(3)schwarze2017-01-061-3/+2
| | | | | | | | | that wasn't accompanied by any related information. Reported by jmc@. There are a dozen functions handling X509_PURPOSE objects, all undocumented, a host of defines, and it seems that a callback is required. So this seems complicated, i doubt that is much used in practice, and i'm not diving into it at this point in time.
* Remove bogus cross reference to ui_create(3) reported by jmc@schwarze2017-01-061-6/+8
| | | | | | | | | and refer readers to the header file instead. I'm not convinced customized prompting is such a bright idea, it feels somewhat like overengineering, so i'm not documenting it right now. People who really feel compelled to roll their own prompting can go read the source code.
* Remove dangling .Xrs to PKCS7_final(3) reported by jmc@schwarze2017-01-062-6/+5
| | | | | | | | | and just use .Fn for now. Not counting constructors, destructors, decoders, encoders, and debuggers, six out of 24 public functions operating on PKCS7 objects are currently documented. I'm not documenting the remaining 18 ones at this point in time.
* Remove a dangling .Xr to PKCS7_SIGNER_INFO_sign(3) reported by jmc@schwarze2017-01-061-3/+3
| | | | | | | | | and just use .Fn for now. There are about two dozen interfaces dealing with PKCS7_SIGNER_INFO objects and none but the constructor, destructor, decoder, and encoder are documented so far. It makes no sense to document one random one, and i'm not going to document all of PKCS7_SIGNER_INFO right now.
* Remove a dangling cross reference reported by jmc@.schwarze2017-01-061-5/+8
| | | | | | I'm not convinced documenting EVP_MD_CTX_set_flags(3) would be wise. Instead, refer people to the header file to make it more obvious that they are tinkering with internals when using such flags.
* Add EVP_read_pw_string(3) to NAME and SYNOPSIS,schwarze2017-01-061-14/+29
| | | | | | | | | resolving a dangling cross reference reported by jmc@. Sort NAME and SYNOPSIS to agree with .Dt and DESCRIPTION. Unify parameter names. Delete a sentence about an implementation detail that is no longer true. Mention the length limitation of the *_string() variants.
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-05-17 12:46:54 +0000