[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Tinycc-devel] macOS M1 is broken since last Grischka commit
From: |
grischka |
Subject: |
Re: [Tinycc-devel] macOS M1 is broken since last Grischka commit |
Date: |
Tue, 27 Jul 2021 12:41:46 +0200 |
User-agent: |
Thunderbird 2.0.0.23 (Windows/20090812) |
Christian JULLIEN wrote:
Grischka, you last set of commits breaks macOS M1 port (see make -k test
log in attachment).
Do you test your macOS changes on a real M1 machine? If you like and if
you don't have a M1 machine, I'll be glad to test your patches before
they go to mob.
Better if you create a macos-M1 branch to experiment changes which is
then merged in mod if nothing gets wrong.
Well according to one of your recent commits
#if defined(TCC_TARGET_MACHO) && defined(TCC_TARGET_X86_64)
/* Special case for Rosetta to handle --cpu=x86_64 on macOS */
else if (sizeof(double) == sizeof(long double))
memcpy(ptr, &vtop->c.ld, sizeof(double));
#endif
I was concluding that long doubles are just doubles on macos-x86_64.
If that is not the case (as the test results suggest) then I probably
was confused by what these lines were trying to say.
I still have no idea, actually.
Anyway, obviously setting TCC_USING_DOUBLE_FOR_LDOUBLE for osx-x86_64 was
a mistake but it might still apply to macos-arm64 eventually (according to
https://developer.apple.com/forums/thread/673482).
--- grischka
C.