Hi Mimi,
> Hi Petr,
> On Thu, 2020-08-13 at 20:25 +0200, Petr Vorel wrote:
> > Hi,
> > we talked about Docker based Travis CI builds.
> > Here they are [1]. Working on it revealed few bugs, that's why they're
> > included.
> > I've resent 2 commits ("man:" prefix).
> > I haven't addressed Mimi's suggestion to replace ibmswtpm2 with libtmps/swtpm [2].
> To summarize:
> - Add libtpms/swtpm support
> - Remove debugging from travis/fedora.sh
> - CentOS tss & tss2-devel issues
> - Petr todo's comment:
> * add pgrep and pkill to _require (fix of a different commit)
> * add magic constant 114 to exit codes (in tests/functions.sh)
> * I'd like reuse exit codes in build.sh. I could source tests/functions.sh (they
> look to be general enough, just I didn't like script in root directory depending
> on it, as it's general enough for build without testing) have these constants in
> separate file.
> swtpm: patch was posted, reviewed, and fixed.
Great.
> CentOS: Our internal Travis on ppc64 is installing both tpm2-tss-2.0.0-
> 4.el8.x86_64.rpm and tpm2-tss-devel-2.0.0-4.el8.x86_64.rpm on CentOS.
> In general, the "boot_aggregate.test" is only run if both a software
> TPM and the tsseventextend exist. If either one of them are missing,
> the test is skipped. For CentOS on x86, the "boot_aggregate" would be
> skipped.
Great. BTW it'd be nice to have native non-intel builds instead of
cross-compilation. But, they're constantly broken on Travis CI, that's why I
didn't implement it (I've been testing ppc64le and s390x for several months).
> Left is Petr's "todo's" comment. Petr, would these be additional
> patches on top of the existing ones or were you planning on re-posting
> them? If these are additional patches, I can remove the debugging
> from travis/fedora.sh in patch 9/9. Please let me know how you want to
> go forward.
Please, remove the debugging from travis/fedora.sh, I'll send additional patch.
Kind regards,
Petr
