Make conf_diagnostics apply also to the SSL conf errors

[t8m]

Checklist

• documentation is added or updated
• tests are added or updated

[t8m]

This is a draft PR - the CI failures demonstrate that conf_diagnostics work and make SSL_CTX_new to fail.

[t-j-h]

Looks good to me in terms of improving what we currently have ...

[beldmit]

I'd like to have some syntax check for the config file as a part of the openssl command line util but it can be a separate PR

[mattcaswell]

The CI failure looks relevant

[t8m]

The CI failure looks relevant

The question is whether the failing test really makes sense. It basically loads a config into a single libctx concurrently from multiple threads. What is the semantics of such call? If the same file is being loaded then it could somewhat work, although I would not be surprised if there are some strange circumstances where it would not. However if a different file would be loaded concurrently into the same libctx, what the application can expect then? Random results.

I am very much inclined to drop the testcase.

[openssl-machine]

This pull request is ready to merge

[nhorman]

@t8m the fixups appear to be causing some conflicts during merge. I can merge them if you like, but if you can I would prefer you take a look at them

[t8m]

I've merged this yesterday. Thank you for the reviews.

[xnox]

make test TESTS=test_sysdefault
...
make[2]: Entering directory '/home/xnox/upstream/openssl'
( SRCTOP=. \
  BLDTOP=. \
  PERL="/usr/bin/perl" \
  FIPSKEY="f4556650ac31d35461610bac4ed81b1a181b2d8a43ea2854cbae22ca74560813" \
  EXE_EXT= \
  /usr/bin/perl ./test/run_tests.pl test_sysdefault )
00-prep_fipsmodule_cnf.t .. ok   
All tests successful.
Files=1, Tests=1,  0 wallclock secs ( 0.01 usr  0.00 sys +  0.27 cusr  0.05 csys =  0.33 CPU)
Result: PASS
90-test_sysdefault.t .. 
    # ERROR: (ptr) 'ctx = SSL_CTX_new(TLS_method()) != NULL' failed @ test/sysdefaulttest.c:25
    # 0x0
    # 80CB81B9097E0000:error:0A0001A3:SSL routines:SSL_CTX_new_ex:error in system default config:ssl/ssl_lib.c:4100:
    # OPENSSL_TEST_RAND_SEED=1715381166
    not ok 1 - test_func
# ------------------------------------------------------------------------------
../../util/wrap.pl ../../test/sysdefaulttest => 1
not ok 1 - sysdefaulttest
90-test_sysdefault.t .. 1/? ----------------------------------------------------
#   Failed test 'sysdefaulttest'
#   at test/recipes/90-test_sysdefault.t line 23.
90-test_sysdefault.t .. Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/1 subtests 

Test Summary Report
-------------------
90-test_sysdefault.t (Wstat: 256 (exited 1) Tests: 1 Failed: 1)
  Failed test:  1
  Non-zero exit status: 1
Files=1, Tests=1,  0 wallclock secs ( 0.01 usr  0.01 sys +  0.17 cusr  0.04 csys =  0.23 CPU)
Result: FAIL
make[2]: *** [Makefile:3949: run_tests] Error 1
make[2]: Leaving directory '/home/xnox/upstream/openssl'
make[1]: *** [Makefile:3946: _tests] Error 2
make[1]: Leaving directory '/home/xnox/upstream/openssl'
make: *** [Makefile:3944: tests] Error 2

Fails for me, and git bisect blames:

21819f78b057c254254646a7854bfad0cd40ed83 is the first bad commit
commit 21819f78b057c254254646a7854bfad0cd40ed83
Date:   Fri Apr 26 17:23:13 2024 +0200

    Make conf_diagnostics apply also to the SSL conf errors
    
    Reviewed-by: Paul Dale <ppzgs1@>
    Reviewed-by: Neil Horman <nhorman@>
    (Merged from https://github.com/openssl/openssl/pull/24275)

 crypto/conf/conf_mod.c      |  1 +
 crypto/context.c            | 17 +++++++++++++++++
 crypto/err/openssl.txt      |  1 +
 include/openssl/crypto.h.in |  2 ++
 include/openssl/sslerr.h    |  1 +
 ssl/ssl_err.c               |  2 ++
 ssl/ssl_lib.c               |  5 ++++-
 ssl/ssl_local.h             |  2 +-
 ssl/ssl_mcnf.c              | 13 +++++++++----
 test/sysdefaulttest.c       | 18 ++++++++----------
 util/libcrypto.num          |  2 ++
 11 files changed, 48 insertions(+), 16 deletions(-)

This is on Ubuntu 24.04 host x86-64 with gcc-13.

Is my host system config somehow bad? or is the test case looking for something incorrect? or is Ubuntu 24.04 missbuilding things?

[t8m]

@xnox please open a new issue for this. Is this a clean build without any patches on top?