Bernd Edlinger [Fri, 2 Nov 2018 10:46:38 +0000 (11:46 +0100)]
Fix error handling in RAND_DRBG_uninstantiate
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7517)
Bernd Edlinger [Tue, 30 Oct 2018 20:02:22 +0000 (21:02 +0100)]
Fix error handling in drbgtest.c
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7517)
Bernd Edlinger [Tue, 30 Oct 2018 19:57:53 +0000 (20:57 +0100)]
Fix error handling in rand_drbg_new
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7517)
Bernd Edlinger [Mon, 29 Oct 2018 12:48:53 +0000 (13:48 +0100)]
Fix error handling in RAND_DRBG_set
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7517)
Pauli [Mon, 5 Nov 2018 04:30:37 +0000 (14:30 +1000)]
EVP_MAC ctrl numbering duplicate removal.
Both EVP_MAC_CTRL_SET_MD and EVP_MAC_CTRL_SET_CIPHER were numbered 4.
This would preclude any future MAC from using both.
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Paul Yang <yang.yang@baishancloud.com>
(Merged from https://github.com/openssl/openssl/pull/7566)
Richard Levitte [Sat, 3 Nov 2018 17:38:04 +0000 (18:38 +0100)]
Simplify the processing of skipped source directories
We kept a number of arrays of directory names to keep track of exactly
which directories to look for build.info. Some of these had the extra
function to hold the directories to actually build.
With the added SUBDIRS keyword, these arrays are no longer needed.
The logic for skipping certain directories needs to be kept, though.
That is now very much simplified, and is made opportunistic.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7558)
Richard Levitte [Sat, 3 Nov 2018 17:34:09 +0000 (18:34 +0100)]
Collapse different classes of macro databases
We have $config{openssl_algorithm_defines}, $config{openssl_other_defines}
and $config{openssl_thread_defines}. These are treated exactly the same
in include/openssl/opensslconf.h.in, so having them separated into three
different databases isn't necessary, the reason for the separation being
long gone. Therefore, we collapse them into one and the same,
$config{openssl_feature_defines}.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7558)
These will trickle down to the build of object files, so building
foo.o and bar.o will be done with these options: -DFOO=1 -DBAR=0
(exact syntax depending on platform, of course)
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7553)
Benjamin Kaduk [Mon, 22 Oct 2018 16:54:20 +0000 (11:54 -0500)]
Restore sensible "sess_accept" counter tracking
Commit 9ef9088c1585e13b9727796f15f77da64dbbe623 switched the SSL/SSL_CTX
statistics counters to using Thread-Sanitizer-friendly primitives.
However, it erroneously converted an addition of -1
(for s->session_ctx->stats.sess_accept) to an addition of +1, since that
is the only counter API provided by the internal tsan_assist.h header
until the previous commit. This means that for each accepted (initial)
connection, the session_ctx's counter would get doubly incremented, and the
(switched) ctx's counter would also get incremented.
Restore the counter decrement so that each accepted connection increments
exactly one counter exactly once (in net effect).
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7464)
Benjamin Kaduk [Mon, 22 Oct 2018 16:51:35 +0000 (11:51 -0500)]
Add tsan_decr() API, counterpart of tsan_counter()
The existing tsan_counter() API increments a reference counter.
Provide a new API, tsan_decr(), to decrement such a reference counter.
This can be used, for example, when a reference is added to the session_ctx's
sess_accept stats but should more properly be tracked in the regular ctx's
statistics.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7464)
Richard Levitte [Wed, 31 Oct 2018 18:23:44 +0000 (19:23 +0100)]
crypto/engine/eng_devcrypto.c: open /dev/crypto only once
We opened /dev/crypto once for each session, which is quite unnecessary.
With this change, we open /dev/crypto once at engine init, and close
it on unload.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/7506)
Richard Levitte [Sat, 27 Oct 2018 07:15:04 +0000 (09:15 +0200)]
crypto/engine/eng_devcrypto.c: add digest copy
Copying an EVP_MD_CTX, including the implementation local bits, is a
necessary operation. In this case, though, it's the same as
initializing the local bits to be "copied to".
Fixes #7495
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/7506)
Pauli [Thu, 1 Nov 2018 04:25:20 +0000 (14:25 +1000)]
openssl list -mac-algorithms support.
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7541)
Richard Levitte [Thu, 1 Nov 2018 12:55:32 +0000 (13:55 +0100)]
Configure: ensure empty arrays aren't created inadvertently
Just refering to a hash table element as an array reference will
automatically create that element. Avoid that by defaulting to
a separate empty array reference.
Fixes #7543
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7544)
Bernd Edlinger [Sat, 27 Oct 2018 09:31:21 +0000 (11:31 +0200)]
Avoid two memory allocations in each RAND_DRBG_bytes
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7507)
This excluses user from additional PATH adjustments in case NDK has
llvm-ar.
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7443)
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7443)
Bernd Edlinger [Mon, 29 Oct 2018 12:48:53 +0000 (13:48 +0100)]
Fix a race condition in drbg_add
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7523)
Matt Caswell [Thu, 11 Oct 2018 16:01:06 +0000 (17:01 +0100)]
Don't call the client_cert_cb immediately in TLSv1.3
In TLSv1.2 and below a CertificateRequest is sent after the Certificate
from the server. This means that by the time the client_cert_cb is called
on receipt of the CertificateRequest a call to SSL_get_peer_certificate()
will return the server certificate as expected. In TLSv1.3 a
CertificateRequest is sent before a Certificate message so calling
SSL_get_peer_certificate() returns NULL.
To workaround this we delay calling the client_cert_cb until after we
have processed the CertificateVerify message, when we are doing TLSv1.3.
Fixes #7384
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/7413)
Richard Levitte [Wed, 24 Oct 2018 19:40:00 +0000 (21:40 +0200)]
Adapt other EVP code to use EVP_MAC instead of direct implementation calls
The EVP_PKEY methods for CMAC and HMAC needed a rework, although it
wasn't much change apart from name changes.
This also meant that EVP_PKEY_new_CMAC_key() needed an adjustment.
(the possibility to rewrite this function to work with any MAC is yet
to be explored)
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7484)
Chocobo1 [Fri, 19 Oct 2018 14:05:49 +0000 (22:05 +0800)]
Fix MSVC warning C4819
CLA: trivial
Reviewed-by: Tim Hudson <tjh@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7444)
Richard Levitte [Fri, 26 Oct 2018 11:42:40 +0000 (13:42 +0200)]
test/evp_test.c: don't misuse pkey_test_ctrl() in mac_test_run()
pkey_test_ctrl() was designed for parsing values, not for using in
test runs. Relying on its returned value when it returned 1 even for
control errors made it particularly useless for mac_test_run().
Here, it gets replaced with a MAC specific control function, that
parses values the same way but is designed for use in a _run() rather
than a _parse() function.
This uncovers a SipHash test with an invalid control that wasn't
caught properly. After all, that stanza is supposed to test that
invalid control values do generate an error. Now we catch that.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7500)
Richard Levitte [Tue, 23 Oct 2018 08:35:48 +0000 (10:35 +0200)]
Windows build: build foo.d after foo.obj
We made the build of foo.obj depend on foo.d, meaning the latter gets
built first. Unfortunately, the way the compiler works, we are forced
to redirect all output to foo.d, meaning that if the source contains
an error, the build fails without showing those errors.
We therefore remove the dependency and force the build of foo.d to
always happen after build of foo.obj.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7469)
Richard Levitte [Wed, 24 Oct 2018 18:20:00 +0000 (20:20 +0200)]
Add convenience functions EVP_str2ctrl() and EVP_hex2ctrl()
These functions are generalizations of EVP_PKEY_CTX_str2ctrl() and
EVP_PKEY_CTX_hex2ctrl(). They will parse the value, and then pass the
parsed result and length to a callback that knows exactly how to pass
them on to a main _ctrl function, along with a context structure
pointer.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7393)
Richard Levitte [Wed, 24 Oct 2018 16:35:32 +0000 (18:35 +0200)]
Add generic EVP_PKEY_METHOD for EVP_MACs
The MAC EVP_PKEY implementations are currently implemented for each
MAC. However, with the EVP_MAC API, only one such implementation is
needed.
This implementation takes into account the differences between HMAC
and CMAC implementations, and observes that all other current MAC
implementations seem to follow the HMAC model.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7393)
Richard Levitte [Fri, 12 Oct 2018 20:27:18 +0000 (22:27 +0200)]
Add EVP_MAC API
We currently implement EVP MAC methods as EVP_PKEY methods. This
change creates a separate EVP API for MACs, to replace the current
EVP_PKEY ones.
A note about this EVP API and how it interfaces with underlying MAC
implementations:
Other EVP APIs pass the EVP API context down to implementations, and
it can be observed that the implementations use the pointer to their
own private data almost exclusively. The EVP_MAC API deviates from
that pattern by passing the pointer to the implementation's private
data directly, and thereby deny the implementations access to the
EVP_MAC context structure. This change is made to provide a clearer
separation between the EVP library itself and the implementations of
its supported algorithm classes.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7393)
Bernd Edlinger [Fri, 26 Oct 2018 19:06:14 +0000 (21:06 +0200)]
Rework and simplify resource flow in drbg_add
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7504)
Pauli [Tue, 23 Oct 2018 21:42:46 +0000 (07:42 +1000)]
Timing vulnerability in DSA signature generation (CVE-2018-0734).
Avoid a timing attack that leaks information via a side channel that
triggers when a BN is resized. Increasing the size of the BNs
prior to doing anything with them suppresses the attack.
Thanks due to Samuel Weiser for finding and locating this.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/7486)
Pauli [Sun, 28 Oct 2018 20:50:51 +0000 (06:50 +1000)]
DSA mod inverse fix
There is a side channel attack against the division used to calculate one of
the modulo inverses in the DSA algorithm. This change takes advantage of the
primality of the modulo and Fermat's little theorem to calculate the inverse
without leaking information.
Thanks to Samuel Weiser for finding and reporting this.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/7487)
Configure: Improve warning if no random seed source was configured
The new Configure summary box (41349b5e6db) now hides the warning
about the missing seed source (2805ee1e095) too much. To make it
more visible again, add warning markers.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7509)
Fraser Tweedale [Sat, 27 Oct 2018 04:11:41 +0000 (12:11 +0800)]
Support GeneralSubtrees with minimum = 0
The Name Constraints extension contains GeneralSubtree values
indicating included or excluded subtrees. It is defined as:
GeneralSubtree ::= SEQUENCE {
base GeneralName,
minimum [0] BaseDistance DEFAULT 0,
maximum [1] BaseDistance OPTIONAL }
RFC 5280 further specifies:
Within this profile, the minimum and maximum fields are not used with
any name forms, thus, the minimum MUST be zero, and maximum MUST be
absent.
Because the minimum fields has DEFAULT 0, and certificates should be
encoded using DER, the situation where minimum = 0 occurs in a
certificate should not arise. Nevertheless, it does arise. For
example, I have seen certificates issued by Microsoft programs that
contain GeneralSubtree values encoded thus.
Enhance the Name Constraints matching routine to handle the case
where minimum is specified. If present, it must be zero. The
maximum field remains prohibited.
Reviewed-by: Paul Yang <yang.yang@baishancloud.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7039)
Bernd Edlinger [Tue, 23 Oct 2018 14:30:20 +0000 (16:30 +0200)]
Fix a possible crash in rand_drbg_get_entropy
Reviewed-by: Paul Yang <yang.yang@baishancloud.com> Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7474)
In commit 820e414d2830 (pr #5247) the summary output of the
Configure command was optimized towards instructing people
how to create issue reports.
It turned out that the wording of this message can confuse new
OpenSSL users and make them think that they are seeing an error
message. This commit makes the summary output start with a success
to prevent a misunderstanding. Also it gives more hints to new
OpenSSL users.
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7499)
RAND_add()/RAND_seed(): fix failure on short input or low entropy
Commit 5b4cb385c18a (#7382) introduced a bug which had the effect
that RAND_add()/RAND_seed() failed for buffer sizes less than
32 bytes. The reason was that now the added random data was used
exlusively as entropy source for reseeding. When the random input
was too short or contained not enough entropy, the DRBG failed
without querying the available entropy sources.
This commit makes drbg_add() act smarter: it checks the entropy
requirements explicitely. If the random input fails this check,
it won't be added as entropy input, but only as additional data.
More precisely, the behaviour depends on whether an os entropy
source was configured (which is the default on most os):
- If an os entropy source is avaible then we declare the buffer
content as additional data by setting randomness to zero and
trigger a regular reseeding.
- If no os entropy source is available, a reseeding will fail
inevitably. So drbg_add() uses a trick to mix the buffer contents
into the DRBG state without forcing a reseeding: it generates a
dummy random byte, using the buffer content as additional data.
Related-to: #7449 Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7456)
RAND_load_file(): avoid adding small chunks to RAND_add()
Increase the load buffer size such that it exceeds the chunk
size by a comfortable amount. This is done to avoid calling
RAND_add() with a small final chunk. Instead, such a small
final chunk will be added together with the previous chunk
(unless it's the only one).
Related-to: #7449 Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7456)
RAND_load_file(): return error if reseeding failed
The failure of RAND_load_file was only noticed because of the
heap corruption which was reported in #7499 and fixed in commit 5b4cb385c18a. To prevent this in the future, RAND_load_file()
now explicitly checks RAND_status() and reports an error if it
fails.
Related-to: #7449 Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7456)
Richard Levitte [Thu, 25 Oct 2018 13:55:15 +0000 (15:55 +0200)]
Windows: Produce a static version of the public libraries, always
When building shared libraries on Windows, we had a clash between
'libcrypto.lib' the static routine library and 'libcrypto.lib' the
import library.
We now change it so the static versions of our libraries get '_static'
appended to their names. These will never get installed, but can
still be used for our internal purposes, such as internal tests.
When building non-shared, the renaming mechanism doesn't come into
play. In that case, the static libraries 'libcrypto.lib' and
'libssl.lib' are installed, just as always.
Fixes #7492
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7496)
Pauli [Thu, 4 Oct 2018 23:19:30 +0000 (09:19 +1000)]
RSA security bits calculation
NIST has updated their guidelines in appendix D of SP 800-56B rev2 (draft)
providing a formula for the number of security bits it terms of the length
of the RSA key.
This is an implementation of this formula using fixed point arithmetic.
For integers 1 .. 100,000 it rounds down to the next smaller 8 bit strength
270 times. It never errs to the high side. None of the rounded values occur
near any of the commonly selected lengths.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7352)
This bug was introduced by #7382 which enhanced RAND_add() to
accept large buffer sizes. As a consequence, RAND_add() now fails
for buffer sizes less than 32 bytes (i.e. less than 256 bits).
In addition, rand_drbg_get_entropy() forgets to reset the attached
drbg->pool in the case of an error, which leads to the heap corruption.
The problem occurred with RAND_load_file(), which reads the file in
chunks of 1024 bytes each. If the size of the final chunk is less than
32 bytes, then RAND_add() fails, whence RAND_load_file() fails
silently for buffer sizes n = k * 1024 + r with r = 1,...,31.
This commit fixes the heap corruption only. The other issues will
be addressed in a separate pull request.
Thanks to Gisle Vanem for reporting this issue.
Fixes #7449
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7455)
Richard Levitte [Sun, 21 Oct 2018 09:03:02 +0000 (11:03 +0200)]
build file templates: have targets for all shared library names
We only had targets for the "simple" shared library names (libfoo.so
and not libfoo.so.x.y on Unix, import library libfoo.lib but not
libfoo.dll on Windows). This has created some confusion why it wasn't
possible to rebuild the less "simple" name directly (just as an
example, someone who mistook the import library on Windows for a
static library, removed the DLL and then found it was difficult to
rebuild directly), so we change the target to include all possible
names.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7451)
Matt Caswell [Thu, 18 Oct 2018 13:45:59 +0000 (14:45 +0100)]
Don't complain and fail about unknown TLSv1.3 PSK identities in s_server
An unknown PSK identity could be because its actually a session resumption
attempt. Sessions resumptions and external PSKs are indistinguishable so
the callbacks need to fail gracefully if they don't recognise the identity.
Fixes #7433
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7434)
Matt Caswell [Tue, 9 Oct 2018 09:22:06 +0000 (10:22 +0100)]
Buffer a ClientHello with a cookie received via DTLSv1_listen
Previously when a ClientHello arrives with a valid cookie using
DTLSv1_listen() we only "peeked" at the message and left it on the
underlying fd. This works fine for single threaded applications but for
multi-threaded apps this does not work since the fd is typically reused for
the server thread, while a new fd is created and connected for the client.
By "peeking" we leave the message on the server fd, and consequently we
think we've received another valid ClientHello and so we create yet another
fd for the client, and so on until we run out of fds.
In this new approach we remove the ClientHello and buffer it in the SSL
object.
Fixes #6934
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/7375)
Matt Caswell [Mon, 8 Oct 2018 14:46:51 +0000 (15:46 +0100)]
Use the read and write buffers in DTLSv1_listen()
Rather than using init_buf we use the record layer read and write buffers
in DTLSv1_listen(). These seem more appropriate anyway and will help with
the next commit.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/7375)
Andy Polyakov [Sat, 22 Sep 2018 12:39:51 +0000 (14:39 +0200)]
sha/asm/keccak1600-armv8.pl: halve the size of hw-assisted subroutine.
Yes, it's second halving, i.e. it's now 1/4 of original size, or more
specifically inner loop. The challenge with Keccak is that you need
more temporary registers than there are available. By reversing the
order in which columns are assigned in Chi, it's possible to use
three of A[][] registers as temporary prior their assigment.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7294)
Notice that md_block_size=64 for SHA256, which results on the
magic constant kVarianceBlocks = 6.
However, md_block_size=128 for SHA384 leading to kVarianceBlocks = 4.
CLA:trivial
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7342)
Viktor Dukhovni [Mon, 8 Oct 2018 16:05:14 +0000 (12:05 -0400)]
Apply self-imposed path length also to root CAs
Also, some readers of the code find starting the count at 1 for EE
cert confusing (since RFC5280 counts only non-self-issued intermediate
CAs, but we also counted the leaf). Therefore, never count the EE
cert, and adjust the path length comparison accordinly. This may
be more clear to the reader.
Viktor Dukhovni [Fri, 5 Oct 2018 03:53:01 +0000 (23:53 -0400)]
Only CA certificates can be self-issued
At the bottom of https://tools.ietf.org/html/rfc5280#page-12 and
top of https://tools.ietf.org/html/rfc5280#page-13 (last paragraph
of above https://tools.ietf.org/html/rfc5280#section-3.3), we see:
This specification covers two classes of certificates: CA
certificates and end entity certificates. CA certificates may be
further divided into three classes: cross-certificates, self-issued
certificates, and self-signed certificates. Cross-certificates are
CA certificates in which the issuer and subject are different
entities. Cross-certificates describe a trust relationship between
the two CAs. Self-issued certificates are CA certificates in which
the issuer and subject are the same entity. Self-issued certificates
are generated to support changes in policy or operations. Self-
signed certificates are self-issued certificates where the digital
signature may be verified by the public key bound into the
certificate. Self-signed certificates are used to convey a public
key for use to begin certification paths. End entity certificates
are issued to subjects that are not authorized to issue certificates.
that the term "self-issued" is only applicable to CAs, not end-entity
certificates. In https://tools.ietf.org/html/rfc5280#section-4.2.1.9
the description of path length constraints says:
The pathLenConstraint field is meaningful only if the cA boolean is
asserted and the key usage extension, if present, asserts the
keyCertSign bit (Section 4.2.1.3). In this case, it gives the
maximum number of non-self-issued intermediate certificates that may
follow this certificate in a valid certification path. (Note: The
last certificate in the certification path is not an intermediate
certificate, and is not included in this limit. Usually, the last
certificate is an end entity certificate, but it can be a CA
certificate.)
This makes it clear that exclusion of self-issued certificates from
the path length count applies only to some *intermediate* CA
certificates. A leaf certificate whether it has identical issuer
and subject or whether it is a CA or not is never part of the
intermediate certificate count. The handling of all leaf certificates
must be the same, in the case of our code to post-increment the
path count by 1, so that we ultimately reach a non-self-issued
intermediate it will be the first one (not zeroth) in the chain
of intermediates.