optional parameter on M1 should be documented

[Ducasse]

Yes, that is something kinda new, we should document it better

El 28 jun 2022, a las 16:20, Marcus Denker <marcus.denker@inria.fr> escribió:

So.. the problem was related to Variadic functions after all…
(I was already using the version with the latest fix for that)

fcntl is has a third optional parameter, and on X86, it seems there is was no problem to just do a call like that:

fcntl: fd command: cmd struct: struct

^ self 
	ffiCall: #(int fcntl(int fd, int cmd, FLock *struct))
	module: LibC

Which on the M1 lead to an ERRNO 14.

After exploring for far too long what could go wrong, I realised that it was missing the fixedArgumentCount: specification.

fcntl: fd command: cmd struct: struct
^ self
ffiCall: #(int fcntl(int fd, int cmd, FLock *struct))
library: LibC
fixedArgumentCount:2

The M1 changes how var args are passed, they are always passed via the stack, while on x86, it follows normal
functions of using the register if possible.

Marcus

[Rinzwind]

Some background information about this from the article “Addressing Architectural Differences in Your macOS Code” in Apple’s documentation on Apple Silicon (previously posted in PTerm issue #67):

The x86_64 and arm64 architectures have different calling conventions for variadic functions—functions with a variable number of parameters. On x86_64, the compiler treats fixed and variadic parameters the same, placing parameters in registers first and only using the stack when no more registers are available. On arm64, the compiler always places variadic parameters on the stack, regardless of whether registers are available.

[Rinzwind]

As raised in Pharo issue #12576, another question the booklet should cover regarding variadic functions is whether it’s possible, and if so how, to call one using an array for some or all of the variadic arguments.