[macOS] v3.9+ failing to compile abseil-cpp-20210324.2

[simpleshadow]

Bug Report

Receive the following error upon yarn add mediasoup:

FAILED: subprojects/abseil-cpp-20210324.2/libabsl_base.a.p/absl_base_internal_cycleclock.cc.o 
c++ -Isubprojects/abseil-cpp-20210324.2/libabsl_base.a.p -Isubprojects/abseil-cpp-20210324.2 -I../../subprojects/abseil-cpp-20210324.2 -fcolor-diagnostics -DNDEBUG -std=c++11 -O3 -Wno-sign-compare -march=armv8-a+crypto -MD -MQ subprojects/abseil-cpp-20210324.2/libabsl_base.a.p/absl_base_internal_cycleclock.cc.o -MF subprojects/abseil-cpp-20210324.2/libabsl_base.a.p/absl_base_internal_cycleclock.cc.o.d -o subprojects/abseil-cpp-20210324.2/libabsl_base.a.p/absl_base_internal_cycleclock.cc.o -c ../../subprojects/abseil-cpp-20210324.2/absl/base/internal/cycleclock.cc
error: unknown target CPU 'armv8-a+crypto'
note: valid target CPU values are: nocona, core2, penryn, bonnell, atom, silvermont, slm, goldmont, goldmont-plus, tremont, nehalem, corei7, westmere, sandybridge, corei7-avx, ivybridge, core-avx-i, haswell, core-avx2, broadwell, skylake, skylake-avx512, skx, cascadelake, cooperlake, cannonlake, icelake-client, icelake-server, tigerlake, sapphirerapids, alderlake, knl, knm, k8, athlon64, athlon-fx, opteron, k8-sse3, athlon64-sse3, opteron-sse3, amdfam10, barcelona, btver1, btver2, bdver1, bdver2, bdver3, bdver4, znver1, znver2, znver3, x86-64, x86-64-v2, x86-64-v3, x86-64-v4

v3.8.4 installs so I'm guessing something changed from the v3.7.1 release that added M1 support (#548)?

Your environment

• Operating system: macOS Monterey (Macbook Air M1, 2020)
• Node version: 17.0.1
• npm version: 8.1.1
• gcc/clang version: clang-1300.0.29.3 (Apple)
• mediasoup version: 3.9+

[nazar-pc]

According to abseil/abseil-cpp#970 you might actually try to cross-compile it. Make sure your compiler is for ARM. People had other issues of a similar nature the other way around: https://mediasoup.discourse.group/t/partly-solved-could-not-detect-ninja-v1-8-2-or-newer/3464

[nazar-pc]

Yeah, looks like your compiler is x86-64, you need ARM compiler instead on M1 machine.

[simpleshadow]

roger! thanks @nazar-pc, will have a look.

[olegmelnik]

@simpleshadow Did you resolve your issue? I have the same problem

[satoren]

I have the same problem too with cargo test at the latest v3 branch.

[nazar-pc]

It doesn't help that you have the same issue.

Please do your research and feel free to contribute to documentation on how to fix this unfortunate quirk of Apple's AMD64->AArch64 transition process.

[satoren]

Hmm, I don't know why, but in my environment, c files seem to compile to arm64 and c++ files to x86_64.

deps/libwebrtc/liblibwebrtc.a.p/libwebrtc_rtc_base_rate_statistics.cc.o:    Mach-O 64-bit object x86_64
subprojects/openssl-1.1.1l/libcrypto.a.p/crypto_aes_aes_core.c.o:    Mach-O 64-bit object arm64 :

When executing manually complete command in generated compile_commands.json, that output to arm64 😕

[ibc]

@satoren you are not even using mediasoup 3.9.X which has a completely new build system so your report is not useful at all.

[satoren]

@ibc
The new build system is meson, right? It used. And the same problem also occurred when using an npm package("mediasoup": "3.9.2").

[satoren]

I discovered that the ninja binary is x86_64.
That downloaded from pip by Makefile

The build was successful when replaced ninja binary to aarch64(installed from brew) instead of x86_64.
I don't know how to fix this, but it's probably not just my environment.

[ibc]

@nazar-pc is this something we can fix in Makefile by checking current arch and forcing download of proper ninja version?

[nazar-pc]

I need to research, looks like PIP bug to me, but I'll investigate. Let's reopen this for now as I believe this is something we can act upon.