I can confirm that with the zero store patch applied libzstd tests run
successfully \o/
It takes a long time though, time reports:
real 151m26.790s
user 0m0.020s
sys 0m0.010s
There are two tests that outputs great amounts of data (~4GB) into a
temporary file, then md5sum it and then compress it. They put a lot of
stress on ext2fs and rumpdisk and make them take a lot of CPU time,
running at almost 100%. Especially the md5sum step.
There is a merge request on Debian's packaging repo [1] that proposes
to apply the hurd-i386 patches to hurd-amd64 (disabling some tests). I
think that's not necessary, unless the run time is not acceptable.
I also sent a PR to upstream [2] to remove the usage of /dev/random
such that when both patches (hurd one and libzstd one) get released
the test won't longer fail.
The patches that [1] is wanting to apply to hurd-amd64 are:
* debian/patches/0013-skip-memory-greedy-tests.patch
* debian/patches/0015-Skip-dev-random-tests-on-hurd.patch
Number 0015 will be no longer necessary once both patches are
released, as I mentioned earlier.
Number 0013, as the name says, skips memory greedy tests that on
hurd-i386 could have failed due to the constrained virtual memory
space, which is not a limitation in hurd-amd64.
I'm not granted access to [1] yet, so I can't comment about our
findings in the MR.
--
[1] https://salsa.debian.org/pkg-rpm-team/libzstd/-/merge_requests/5
[2] https://github.com/facebook/zstd/pull/4212
