x86 calling conventions
=======================

x86-64
------

Arguments:
Integer: rdi, rsi, rdx, rcx, r8, r9
Float: xmm0-xmm7

Return:
Integer: rax
Float: xmm0, xmm1

Callee saved:
rbp, rbx, r12, r13, r14, r15

Notes:
For variadic arguments, al contains the number of vector registers
used
rdx is the "second return register"
Vector type argument  (__m128, __m256 or __m512) goes in Float
registers (xmm, ymm or zmm respectively)

i386
----

Arguments:
__mm64: mm0-mm2
__mm128: xmm0-xmm2

Return:
Integer: eax
Float: st0
__mm64: mm0
__mm128: xmm0

Callee saved:
ebp, ebx, esi, edi

Linux syscall convention
------------------------

Arch   | Syscall | Arguments:                        | return |
       | number  |  1  |  2  |  3  |  4  |  5  |  6  | value  |
-------|---------|-----|-----|-----|-----|-----|-----|--------|
x86-64 | rax     | rdi | rsi | rdx | r10 | r8  | r9  | rax    |
i386   | eax     | ebx | ecx | edx | esi | edi | ebp | eax    |

Notes:
x86-64 lose the content of rcx and r11, as they are overwritten with
the syscall instruction