Add clang/ARM support for MINIX

[sambuc]

Not to be merged as is, at least I believe the known bugs needs to be sorted out first.

This patch set does not throw away GCC, nor changes the default toolchain for ARM. This will be done when the aforementioned problems are fixed.

[dcvmoole]

Fantastic. A huge step in the right direction for minix, bugs or not :)

For those who may be interested in helping out with debugging efforts: if you happen to know, are any of the test failures occurring on either the actual hardware or qemu, or both? Specifically, it might be easier to do debugging in qemu. Also, I seem to recall that test75 was already having issues before, so that might not have gotten worse at least?

For clarification: given my current project I have -1 spare time for any time unit of -1, so I won't be of much use at least in the short term unfortunately..

[sambuc]

What is mentioned in the README.md is on HW, for reference here are the failing tests:

1. 53: Division by zero does not trigger exceptions
2. 75: ru.tv_secs can't be zero (and is zero)
3. 85: hangs
4. isofs: Fails because of an out of memory condition
5. vnd: crash
6. Running two times the kyua tests in a row, without rebooting in between will lead to a mostly failed second run because of copy-on-write errors.

On emulation, the results are a bit better / different:

1. 53:

Test 53 error line 189; i=0x0000000100000000; j=0x0000000000000000; k=0x0000000000000000

/Volumes/Minix/src-arm/minix/tests/test53.c:31: Subtest 0,  error 7,  errno 9: Bad file descriptor
error line 190; i=0x0000000100000000; j=0x0000000000000000; k=0x0000000000000000
/Volumes/Minix/src-arm/minix/tests/test53.c:31: Subtest 0,  error 7,  errno 9: Bad file descriptor
error line 191; i=0x0000000100000000; j=0x0000000000000000; k=0x0000000000000000
/Volumes/Minix/src-arm/minix/tests/test53.c:31: Subtest 0,  error 7,  errno 9: Bad file descriptor
error line 192; i=0x0000000100000000; j=0x0000000000000000; k=0x0000000000000000
/Volumes/Minix/src-arm/minix/tests/test53.c:31: Subtest 0,  error 7,  errno 9: Bad file descriptor
error line 193; i=0x0000000100000000; j=0x0000000000000000; k=0x0000000000000000
/Volumes/Minix/src-arm/minix/tests/test53.c:31: Subtest 0,  error 7,  errno 9: Bad file descriptor
Too many errors; test aborted

2. 73:

Test 73 VM: pagefault: SIGSEGV 131109 bad addr 0x6374652f; err 0x5 nopage read
rs*F     131109 0x23c34 
PM: coredump signal 11 for 502 / rs
rs*F     131109 0x23c34 
ok

2. 85: hangs

[dcvmoole]

Good to know, thanks! For reference, the test-73 crash on address 0x6374652f very much looks like memory corruption: that value is actually the string "/etc" in 32-bit little-endian format, and as such should obviously never be dereferenced as an address :)

[MerlijnWajer]

What is the status of this PR now? Do all six problems remain? And presumably the OOM in isofs is unrelated to the boot failure in #104? (I recall that was likely a gcc bug?)