Re: Rep:Re: Rep:Re: Rep:Re: [f-cpu] Stack handling

[Cedric BAIL <cedric.bail@free.fr>]

> For calling convention I've just realesed a draft about it in my folder

I have read it, it's a good resume, and I have only one thing to say. Invert
return address and stack pointer. Two reasons for this choice : on MIPS
the return address is at the end of the register bank, so why not use the same
idee. And if we need frame pointer, or some thing like that we will creat
separate the pointer with an jump destination, and I think that's not clean.

An other question, why didn't you speak about the relocation problem (big
problem on F-CPU) ?
 
> on seul.org (in thomas\call_draft.html) it's not yet finished and I'd 
> love your feedback and comment.
> But he explain that we only need to define a call convention to 
> inter-language call so this CC don't have to be time critical, but he 
> need to be simple, so the return address was stored in a register.

I agree.
 
> But each compiler could choose to implement a stack for return address
> for all is internal call if he want (it seems better for security, but
> if programmer make good job...)

Programmer make good job ;-) Prefer to say compiler do the must do the best work
they can, because programmer didn't care about security and speed in most case.
 
> This talk about two different stack was hardware related it was a 
> software discution and it was dependent of each compiler.

Cedric
*************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/