On Thu, Jun 19, 2025 at 08:33:10AM +0200, Richard Biener wrote:
> How does this interact with -mincoming-stack-boundary? Is this, and thus
> when we need stack realignment, visible here? Do we know whether we need
> to realign the stack anyway for other reasons?
-mincoming-stack-boundary= like -mpreferred-stack-boundary= argument must
be [3,12] for -m64 and -mx32 and [2,12] for -m32. The stack pointer is always
at least word aligned, this patch doesn't change anything about that.
The patch just restores what we had from 2005 except for avr, pru, m68k and some
options on cris where it keeps the current behavior (at least if my greps were
sufficient, was looking for <= 16 STACK_BOUNDARY on !STACK_ALIGNMENT targets
with UNITS_PER_WORD bigger than that STACK_BOUNDARY / 8).
Jakub
