summaryrefslogtreecommitdiff
path: root/src/lib (follow)
Commit message (Collapse)AuthorAgeFilesLines
...
* Reimplement {EVP_CIPHER,EVP_MD,OBJ_NAME}_do_all{,_sorted}(3)tb2024-01-133-175/+161
| | | | | | | | | | | | | | | | | | | | | This implements the do_all API by simple loops over the tables of digests and ciphers. Since some ciphers are only available on some platforms, we need to skip them if necessary. We use loops in each of the functions rather the convoluted way of reducing some of the loops to others. Since the tables are sorted, as ensured by regress, both do_all() and do_all_sorted() walk the lists in order. In particular, we no longer need to allocate to be able to sort hash tables by name on the fly in a void function that may end up doing nothing because allocation failed. We still need to do an unchecked OPENSSL_init_crypto() call. But that's what prayer and clean living are there for (as beck put it). The OBJ_NAME API is completely misnamed. It has little to do with objects and a lot to do with EVP. Therefore we implement what will remain from its saner replacement in the evp directory, i.e., evp_names.c. ok jsing
* Add a table of digest names, digests and aliasestb2024-01-131-2/+475
| | | | | | | | This is the corresponding commit for digests and their aliases. It only adds a table to be used in upcoming commits. What was said about ciphers applies mutatis mutandis to digests. ok jsing
* Add a table of cipher names, ciphers and aliasestb2024-01-132-1/+1090
| | | | | | | | | | | | | | | | | | | | | | | | | | This arranges the data provided by dynamic library initialization in a static table and will help avoid gross code with missing error checking and other defects on every use of the library. This table isn't pretty due to various naming inconsistecies accumulated over the decades. It will significantly simplify the implementation of API such as EVP_get_cipherbyname() and EVP_CIPHER_do_all(). All the table does is map strings to ciphers, typically used on the openssl(1) command line or in code it's the mechanism that underlies the map from NID_chacha20 to the data returned by EVP_chacha20(). It's of course more complicated because it just had to be stupid. This is one of the places where the use of bsearch() is justified. The price to pay for the simplification is that adding custom aliases and custom ciphers to this table will no longer be supported. It is one significant user of the LHASH madness. That's just another piece of the awful "toolkit aspect"-guided misdesign that contributes to making this codebase so terrible. A corresponding table for the digests will be added in the next commit. ok jsing
* Remove X509_STORE_CTX_purpose_inherit(3) documentationtb2024-01-121-131/+41
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | This abomination of an API will be removed. Remove the hairy details of its internals and make the documentation of X509_STORE_CTX_set_trust(3) and X509_STORE_CTX_set_purpose(3) independent of it. Neither of these two remaining APIs can be recommended. Once set, trust and purpose are sticky. Setting the trust to a different (valid) value will indicate success but leave the value unchanged. I suppose it means the new trust value was successfully ignored. Also, setting the trust to X509_TRUST_DEFAULT can succeed or fail depending on which OpenSSL derivative you use. Setting the purpose will also set the trust (unless it is already set). Setting some purposes may or may not fail depending on the OpenSSL lib. The only way you have a chance of knowing what will be set is by calling only one of these functions directly after X509_STORE_CTX_init(). This isn't really safe either because in some versions the user can override the values stored in a global table by writing directly to it. The actual contributions here are rather minimal. State more explicitly that 0 is invalid (but results in success being returned), document the error values to be accurate across implementations and call out some of the nonsense in a CAVEATS section. Many thanks to schwarze for the very helpful review with lots of input. ok schwarze
* X509_TRUST: tidy up includestb2024-01-101-2/+4
|
* X509_TRUST: group together all trust_*() functionstb2024-01-101-33/+30
| | | | | Now they are next to the trstandard[] table and listed in the order they appear in the table.
* X509_TRUST: hoist trust_compat() to the other end of the filetb2024-01-101-12/+11
|
* X509_TRUST: start shuffling some code aroundtb2024-01-101-32/+31
| | | | | Hoist obj_trust() to the top and move the static default_trust() next to its setter.
* Rework X509_STORE_CTX_set_{purpose,trust}()tb2024-01-101-36/+35
| | | | | | | | | | | | | Split the two codepaths in x509_vfy_purpose_inherit() into its two callers. What remains is gross, but at least a reader has a chance of following all this nonsense without leaving a significant amount of hair behind. In short, purpose and trust are only overridden if they're not already set. Otherwise silently ignore valid purpose and trust identifiers that were passed in and succeed. Error on almost all invalid trust or purpose ids, except 0, because... well... who knows, really? ok jsing
* Inline rsa_is_pss() and rsa_pkey_is_pss()tb2024-01-101-11/+9
| | | | | | It's more explicit and not that much longer. ok jsing
* Drop an unnecessary casttb2024-01-101-2/+2
| | | | from jsing
* Fix print_fp()tb2024-01-101-7/+5
| | | | | | | | | | | The callback-based printing needs to die. But first BIO_set() will die. We have a FILE *. We have fprintf(). No need to use a static BIO to dump error codes to said stream. This basically undoes an unrelated change of "Move crpytlib.h prior bio.h" from 19 years ago (OpenSSL 25a66ee3). Except we don't cast and check len. ok jsing (who had a nearly identical diff)
* Fix copy-paste error that broke openssl-ruby and openssl regresstb2024-01-091-2/+2
| | | | Noticed by anton
* Disable X509_STORE_CTX_purpose_inherit()tb2024-01-081-23/+22
| | | | | | | | | | | | | | | | | | | | | | Nothing uses this function, except two internal callers. So split its guts temporarily into a helper function and disable the gross general case. The internal helper can be simplified by observing that def_purpose == 0: Overriding 0 by 0 doesn't do anything, so drop that bit. Rename ptmp into purp, and inline X509_PURPOSE_get_by_id(), i.e., make appropriate checks and subtract X509_PURPOSE_MIN. The fallback to X509_PURPOSE_get_by_id(0) will always fail since X509_PURPOSE_MIN == 1. So ditch that call. In particular, X509_STORE_CTX_set_purpose(ctx, X509_PURPOSE_ANY) fails in current because of this. That's nonsense. So remove the purp->trust == X509_TRUST_DEFAULT check as only change of behavior. This matches what OpenSSL do nowadays. They now set def_purpose = purpose if purpose != 0 and def_purpose == 0, so in all real-world uses of this function they will just fetch the same purpose again and do not check for default trust the second time around. Finally, X509_TRUST_get_by_id() is only used to ensure that a non-zero (or overridden) trust is between X509_TRUST_MIN and X509_TRUST_MAX. So expand that into its explicit form. ok jsing
* Inline X509_{TRUST,PUPROSE}_set() in their only callerstb2024-01-081-3/+16
| | | | | | They are now unused and will join the exodus to the attic in the next bump. ok jsing
* const correct aesni_{128,256}_cbc_hmac_sha1_ciphertb2024-01-081-3/+3
|
* Fix logic error (&& -> ||)tb2024-01-081-2/+2
| | | | CID 477172
* Garbage collect call to X509_TRUST_cleanup()tb2024-01-071-2/+1
| | | | Since x509_trs.c r1.33, this is a noop.
* Minor cleanup in X509_STORE_CTX_purpose_inherit()tb2024-01-071-9/+10
| | | | | Make a few checks against 0 explicit to reduce noise in an upcoming diff and tiny KNF tweaks.
* purpose/trust: Improve comments about COUNT/MAX confusiontb2024-01-072-4/+10
|
* const-correct r4_hmac_md5_ciphertb2024-01-071-2/+2
|
* Convert the remaining legacy ciphers to C99 initializerstb2024-01-075-76/+90
| | | | | | No change in the generated aarch64 assembly apart from line number changes. ok jsing
* Improve EVP_CIPHER_{get,set}_asn1_iv()tb2024-01-071-25/+26
| | | | | | | | | Use iv_len for the variables storing the IV length, formerly l and j. Remove use of the unnecessary variable i and unindent the whole mess. Some return values are fishy. That will be addressed in subsequent commits. ok jsing
* Remove X509_TRUST extensibilitytb2024-01-071-102/+10
| | | | | | | | This is pretty much identical to the X509_PURPOSE case: remove the stack used for extending and overriding the trust table and make X509_TRUST_add() always fail. Simplify some other bits accordingly. ok jsing
* Zap some more CRL method thingstb2024-01-061-10/+1
|
* Remove X509_PURPOSE_cleanup() call in OPENSSL_cleanup()tb2024-01-061-2/+1
| | | | | Since x509_purp.c r1.34 this is a noop since there is nothing to clean up anymore. Remove the last caller.
* Remove X509_CRL_METHOD internalstb2024-01-063-92/+25
| | | | | | | | | | | | | | | Another complication of dubious value that nobody's ever used. crl_init(), crl_free() and the meth_data are dead weight, as are their accessors. Inline def_crl_verify() in X509_CRL_verify() so that the latter becomes the trivial wrapper of ASN1_item_verify() that one would expect it to be. It is quite unclear what kind of customization would make sense here... def_crl_lookup() is renamed into crl_lookup() and its two callers, X509_CRL_lookup_by_{serial,cert}(), are moved below it so that we don't need a prototype. ok jsing
* Remove X509_PURPOSE extensibilitytb2024-01-061-112/+10
| | | | | | | | | | | | | | | | | | | Another bit of global state without lock protection. The by now familiar complications of a stack to make this user configurable, which, of course, no one ever did. The table is not currently const, and the API exposes its entries directly, so anyone can modify it. This fits very well with the safety guarantees of Rust's 'static lifetime, which is how rust-openssl exposes it (for no good reason). Remove the stack and make the X509_PURPOSE_add() API always fail. Simplify the other bits accordingly. In addition, this API inflicts the charming difference between purpose identifiers and purpose indexes (the former minus one) onto the user. Neither of the two obvious solutions to avoid this trap seems to have crossed the implementer's mind. ok jsing
* EVP_PKEY_asn1_find_str() tweakstb2024-01-051-4/+3
| | | | | Switch i to a size_t and improve a flag check. Part of an earlier diff that was ok jsing but were lost when I reworked the diff.
* evp_key.c: Remove more unnecessary parenthesestb2024-01-051-7/+6
|
* Minor tweaks in EVP_read_pw_string_min()tb2024-01-051-4/+4
| | | | | Remove unnecessary parentheses and use a better place to break an overlong line.
* Plug a leak in EVP_read_pw_string_min()tb2024-01-051-9/+14
| | | | | | | Use an error exit that frees the ui in case the UI_add_* fail. Also add a few empty lines for readability. ok joshua
* Disable EVP_PKEY_meth_* extensibilitytb2024-01-041-46/+7
| | | | | | | | | | This removes the global pkey_app_methods stack that was never cleaned up and makes EVP_PKEY_meth_add0() always fail and push an error on the stack. EVP_PKEY_meth_find() can now walk the list of PKEY_METHODs forward and things become a bit cleaner. It's still all way more complicated than it needs to be... ok jsing
* Remove last external call to EVP_PKEY_meth_find()tb2024-01-041-5/+6
| | | | | | | | | In order to determine whether GOST is properly enabled, libssl has various weird dances. In this specific case, it calls EVP_PKEY_meth_find() to see whether the relevant cipher is around. Check the same thing with an #ifdef instead. ok jsing
* Remove unused app_data from EVP_CIPHERtb2024-01-0415-84/+15
| | | | | | | | The EVP_CIPHER structs are static const data that the library returns when you call EVP_aes_128_cbc(), for example. It makes no sense whatsoever to hang user data off such a struct, but it's been there since forever. ok jsing
* Clean up EVP_PKEY_asn1_get0_info() a bittb2024-01-041-14/+16
| | | | | Use better variable names without silly p prefix and use explicit checks against NULL.
* Clean up EVP_PKEY_asn1_find_str()tb2024-01-041-9/+18
| | | | | | | | | Use slightly better argument and variable names, do not pointlessly try to match a string of negative length < -1, use a size_t for the strlen() and preserve the logic that allows lookup by a string fragment rather than a full string. ok jsing
* Simplify EVP_PKEY_asn1_find()tb2024-01-041-33/+9
| | | | | | | | | EVP_PKEY_asn1_find() finds the EVP_PKEY_ASN1_METHOD underlying the method or alias with nid (or, rather, pkey_id) passed in. Now that we have the base method stored in a pointer, we can return that method after a simple lookup of said nid (or, rather, pkey_id). ok jsing
* Replace .pkey_base_id with a .base_method pointertb2024-01-0411-32/+32
| | | | | | | | | | | | | | | | Every EVP_PKEY_ASN1_METHOD is either an ASN.1 method or an alias. As such it resolves to an underlying ASN.1 method (in one step). This information can be stored in a base_method pointer in allusion to the pkey_base_id, which is the name for the nid (aka pkey_id aka type) of the underlying method. For an ASN.1 method, the base method is itself, so the base method is set as a pointer to itself. For an alias it is of course a pointer to the underlying method. Then obviously ameth->pkey_base_id is the same as ameth->base_method->pkey_id, so rework all ASN.1 methods to follow that. ok jsing
* Neuter the remainder of the ameth libtb2024-01-041-65/+13
| | | | | | | | | | | The few pieces of the ameth lib that will stay in libcrypto were moved to p_lib.c recently. The functions that still are in ameth_lib.c will be removed in the next major bump. With disabled EVP_PKEY_asn1_add{0,_alias}() API they are completely useless now and they are getting in the way of more ameth surgery. Rip out their guts and turn them into stubs that do nothing but push an error onto the stack. ok jsing
* Split ameth arrays into individual methodstb2024-01-044-149/+150
| | | | | | | | | | | | For some reason DSA, GOST, and RSA had their ASN.1 methods stored in an array. This is clumsy and the only benefit is that one saves a few externs in p_lib.c. They were also arranged by ascending NID because of bsearch() madness. Split them up and arrange the methods by name, which is much saner and simpler. ok jsing
* Improve length checks for oiv and ivtb2024-01-041-4/+5
| | | | | | | | | | There are two unsigned char arrays of size EVP_MAX_IV_LENGTH to store the IVs of block ciphers. In most modes, only iv is used, but in some modes iv is modified and oiv is used to store the original IV. At the moment nothing enforces that they are of the same length. Therefore make sure the correct one or both are checked before writing to or reading from them. ok miod
* Improve order in ancient CMS helperstb2024-01-031-25/+25
| | | | | | | First came EVP_CIPHER_param_to_asn1() which wraps EVP_CIPHER_set_asn1_iv() which was implemented last. Then came EVP_CIPHER_asn1_to_param() wrapping EVP_CIPHER_get_asn1_iv(). Move each param function below the iv function it wraps.
* Move a t to the right place in a commenttb2024-01-021-2/+2
|
* Match struct order for the EVP_CIPHER_CTX accessorstb2024-01-021-69/+73
| | | | | | This isn't great since the struct is ordered in about the silliest way imaginable, but it is better than it was before. Bringing order into this mess is harder than solving a Rubik's cube.
* Move down EVP_CIPHER_CTX accessors expose EVP_CIPHER internalstb2024-01-021-38/+42
| | | | | These confusingly named getters were added "for convenience" in 1.1. They fit best next to the EVP_CIPHER API.
* Move the trivial EVP_CIPHER getters downtb2024-01-021-31/+35
| | | | | They are now below the CMS ASN.1 IV stuff, but above the EVP_CIPHER_meth* API, which are setters, in a way.
* Simplify EVP_CIPHER_{asn1_to_param,parma_to_asn1}()tb2024-01-021-17/+13
| | | | There's no need for a ret variable and else if/else
* Move the EVP_CIPHER API that only exists for CMS/legacy a bit downtb2024-01-021-121/+125
|
* sm4: more NULL misspellingstb2024-01-021-4/+4
|
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-20 02:52:23 +0000