Add support for musl c

[fealebenpae]

What, How & Why?

As it says on the tin.

The changes are mostly about things that don't exist in musl, or need configure-time detection:

• posix_fallocate
• pthread_getname_np/pthread_setname_np
• backtrace

And various other quirks like the hardcoded messages produced by strerror that are different between glibc and musl.

☑️ ToDos

• 📝 Changelog update
• 🚦 Tests (or not relevant)
• C-API, if public C++ API changed.

[coveralls-official]

Pull Request Test Coverage Report for Build yavor.georgiev_380

• 13 of 13 (100.0%) changed or added relevant lines in 4 files are covered.
• 221 unchanged lines in 19 files lost coverage.
• Overall coverage decreased (-0.06%) to 91.583%

---

Files with Coverage Reduction	New Missed Lines	%
src/realm/aggregate_ops.hpp	1	97.84%
src/realm/util/file.cpp	1	81.69%
src/realm/cluster.cpp	2	80.25%
src/realm/query_engine.hpp	2	94.09%
src/realm/sync/network/websocket.cpp	2	75.36%
src/realm/sync/noinst/server/server_history.cpp	2	67.69%
src/realm/util/fifo_helper.cpp	2	85.71%
src/realm/uuid.cpp	2	97.06%
src/realm/sync/network/network.cpp	3	89.94%
test/test_util_network.cpp	3	97.2%

Totals
Change from base Build 1796:	-0.06%
Covered Lines:	230692
Relevant Lines:	251894

---

💛 - Coveralls

[jedelbo]

@fealebenpae There is a test failing on Evergreen (Sync_BundledRealmFile). Could this be caused by the changes here?

[fealebenpae]

Not from what I can tell - the changes here are mostly about low-level things in Core and shouldn’t be affecting existing targets. I will run tests a couple more times and see if I catch a sporadic failure or not.

[fealebenpae]

I'm pretty sure the sync test failures are unrelated. I haven't been able to consistently trigger them across many CI runs and we have sporadic failures in other branches as well.

[michael-wb]

Looks good - I was wondering if we should add a CMake define for Alpine/Musl c, but I don't think we need it.

[michael-wb]

Of course it does...

[michael-wb]

FYI - The TSAN failure seems to be around the mktime() command: realm-core//test/test_json.cpp:884
Could this have something to do with the different POSIX version?
https://spruce.mongodb.com/task/realm_core_stable_ubuntu2004_tsan_core_tests_patch_e6271d72308b40399890060f58a88cf568c2ee22_6540f3012fbabe7d696a23bf_23_10_31_12_28_50/logs?execution=0&sortBy=STATUS&sortDir=ASC

[fealebenpae]

add a CMake define for Alpine/Musl c

unfortunately there isn't a way I could find to easily detect the C library we're targeting, especially if cross-compilation is involved, and musl deliberately doesn't expose any macros that allow detection like __GLIBC__ for example.

[fealebenpae]

I'll try downgrading the POSIX version and see if that has an impact on the tsan warning.

[ironage]

LGTM

[ironage]

it could be nice to make a REALM_MUSL macro in features.h for readability

[fealebenpae]

I wouldn't do that outside of tests as there isn't a sure-fire way to detect musl based on macros alone - in this particular case we know we only build the tests against glibc and musl on Linux, but if someone were to build Core itself with something like uclibc, for example, it would trigger the above code path if it were in Core itself and not just in the tests we build and run ourselves.

[ironage]

Looks like it took CI 18 minutes to compile with musl c. It's not the bottleneck on CI so doesn't matter too much, but that's around double the time for our other linux builders 😄