t/05_ca_hashes_up_to_date.t was relying on time stamps but those do
not work in GHA. This PR now uses git diff which is reliable.
Also it check the wording of the generator utils/create_ca_hashes.sh
also it cleans up the generated crtl files which can cause a problem
when ones run it locally.
The test t/05_ca_hashes_up_to_date.t compared the regenerated etc/ca_hashes.txt
byte-for-byte against the committed file, but two aspects are not stable across
platforms:
- The name column comes from 'openssl x509 -subject', which formats
differently between LibreSSL (macOS) and OpenSSL (Linux).
- The line order comes from 'sort', whose collation differs between
BSD libc and glibc even under an identically-named locale.
Only the SPKI hash (field 1) is deterministic, and it is the sole
column testssl.sh consumes. So:
- The test now compares the set of SPKI hashes (field 1) between the
committed and regenerated files, order-independently, using
'LC_ALL=C sort -u' for a stable, platform-independent collation.
- create_ca_hashes.sh pins the name format ('-nameopt compat') and
sort collation ('LC_ALL=C') so the generated file is stable.
Verified on macOS (LibreSSL) and Ubuntu 24.04 (OpenSSL).
This commit adds support for the two cipher suites in RFC 9150, TLS_SHA256_SHA256 and TLS_SHA384_SHA384. These are authentication and integrity-only cipher suites.
.. mainly copied from testssl.sh. Also it adds a detection for the
strings ssl and https. If those run at non-stanadard ports but nmap
detected it, it'll show up in the output file.
That will be backported to the main program, see #1931 .
According to the POSIX Programmer's Manual, the exit status specified by
the unsigned decimal integer. If n is specified, but its value is not
between 0 and 255 inclusively, the exit status is undefined.
By cross reference the usage between different scripts in this project,
it looks like we could simply remove the `-` before the number.
While we are thankful that Ivan Ristic permitted to use the client
data from SSLlabs, it became of bit outdated now (see #1158). Also
as sslhaf [1] was used, the data comes from HTTP traffic only.
This is a start to address it. It provides data from Android 9
(connecting to the play store, so that it is sure we don't capture
a ClientHello from an application having an own TLS stack.
Also it provides documentation how to grab data yourself, and
provide it back to testssl.sh.
Aim is at least for testssl.sh 3.0 to add Android 8 and OpenSSL 1.1.1 (@drwetter).
My hope others can assist with Safari on OSX 11 and 12. Java 10 and 11,
and a recent Opera and Edge version. (Firefox and Chrome are out of
date too)
Mail clients to follow later.
[1] https://github.com/ssllabs/sslhaf
- Darwin doesn't build with -static (removed; file name suffix changed to "dynamic" in this case)
- Darwin has a different openssldir (/private/etc/ssl)
- script doesn't fail any more at make clean step in case there is no Makefile yet
- Darwin 64 bit compilation needs ./Configure instead of ./config and an explicit reference to darwin64-x86_64-cc
https://api.dev.ssllabs.com/api/v3/getClients incorrectly indicates a highestProtocol of 771 (TLS 1.2) for clients that support TLS 1.3, which leads run_client_simulation() to incorrectly report "no connection" if the client would have actually connected using TLS 1.3.
This has been addressed by manually editing etc/client-simulation.txt to set the highest_protocol to 0x0304 for the clients that support TLS 1.3.
This PR modifies update_client_sim_data.pl to automatically apply the fix for clients that support TLS 1.3 in order to avoid a possible regression when etc/client-simulation.txt is updated.
When performing client simulations in "--ssl-native" mode, provide the client's list of supported curves to "$OPENSSL s_client" in order to make the results even more accurate.
This PR improves client simulation in "--ssl-native" mode:
* It changes ${protos[i]} to list the protocols that should be disabled rather than those that should be enabled, except in the case that the client only supports one protocol.
* It sets the values for ${tlsvers[i]}, which is used in run_client_simulation(), but was not defined.
* It adds a new variable, ${ciphersuites[i]}, that lists the TLSv1.3 cipher suites supported by a client.
Client simulation still produces false results in "--ssl-native" mode, but the results are better than before.
Dirk Wetter