Hey Jorge, thanks for figuring out all this OpenSSL stuff.
Is it fair to say that the bug only really affects resolute, since
OpenSSL 4 in stonking doesn't support engines at all? It would be
helpful to clarify this in the [ Impact ] section (maybe something like:
"wpa_supplicant in Resolute registers an OpenSSL Engine as the RSA
EVP_PKEY_METHOD; Engine is deprecated in OpenSSL 3 and is no longer
supported in OpenSSL 4").
I'm a bit torn as to whether we should make this change in stonking; we
know that LIB_TPM2_OPENSSL_OPENSSL_POST400 will be defined in stonking
builds, so the second patch becomes essentially a no-op. Is there any
reason to include the first patch (5894be2) in stonking? Is there any
user-facing benefit?
You note that this regressed in OpenSSL 3.0.14; is is possible to link
the patch that was responsible for the regression so that we can
understand why it happened? Having a reason for the regression helps me
be confident that we don't need to make changes in Noble or Jammy (where
OpenSSL also supports both Engine and Provider APIs).
It would be good to include an `apt install ...` for all the deps needed in the
test plan:
```
sudo apt install swtpm tpm2-abrmd libtpm2-pkcs11-tools libtpm2-pkcs11-1
libengine-pkcs11-openssl
```
The test plan script fails before it should for me on both stonking and
resolute:
```
wesley@r0:~$ ./tpm2-tls-minimal.sh
[*] OpenSSL: OpenSSL 3.5.5 27 Jan 2026 (Library: OpenSSL 3.5.5 27 Jan 2026)
[*] engine: /usr/lib/x86_64-linux-gnu/engines-3/pkcs11.so | module:
/usr/lib/x86_64-linux-gnu/pkcs11/libtpm2_pkcs11.so
[!] tpm2-abrmd did not become ready (see
/tmp/tpm2-tls-minimal.bAllr5/tabrmd.log)
wesley@r0:~$ cat /tmp/tpm2-tls-minimal.bAllr5/tabrmd.log
File: /tmp/tpm2-tls-minimal.bAllr5/tabrmd.log <EMPTY>
```
Stonking has OpenSSL 4.0, which doesn't support engines at all, so it maybe
makes sense that this fails, and that the bug isn't reproducible. Rather than
testing that the bug is fixed in Stonking, the test plan should make sure that
this functionality in tpm2-pkcs11 hasn't regressed there (or we drop the
stonking upload, RE my questions above).
```
wesley@s0:~$ ./tpm2-tls-minimal.sh
[*] OpenSSL: OpenSSL 4.0.3 29 Sep 2026 (Library: OpenSSL 4.0.3 29 Sep 2026)
[!] engine_pkcs11 not found (package libengine-pkcs11-openssl)
wesley@s0:~$ apt policy libengine-pkcs11-openssl
libengine-pkcs11-openssl:
Installed: 0.4.20-1ubuntu2
Candidate: 0.4.20-1ubuntu2
Version table:
*** 0.4.20-1ubuntu2 500
500 http://archive.ubuntu.com/ubuntu stonking/universe amd64 Packages
100 /var/lib/dpkg/status
```
If you've built PPA test packages with your patches, could you link the
PPA here?
Packaging nits (1 is a blocker, the others are optional):
1. "Origin: backport" should be "Origin: upstream" in the patch you authored, I
don't see any adjustments in the range-diff
2. Ideally the patches would appear in d/p/series in the same order they are
applied upstream, even if they apply cleanly in either order. Upstream's patch
order suggests that 6bde508 depends on 5894be2, so we should preserve this in
the packaging if possible
3. Personally, I prefer to use the full commit hash for the upstream URL
Changelogs & version numbers look good.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2167891
Title:
"Host memory error" when using tpm with wpa_supplicant
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/tpm2-pkcs11/+bug/2167891/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs