Hi
I’m working on the XCore target which does not (nor is there a reason to) handle stack alignment other than 4 byte.
It seems sensible therefore to explicitly set TargetFrameLowering to use:
TransientStackAlignment = 4, StackRealignable= false
“If a constant alignment is specified, the value result of the allocation is guaranteed to be aligned to at least that boundary”
Thus should I expect an error if I compiled the following?
declare void @f1(i32*, i32*)
define void @f2() nounwind {
entry:
%0 = alloca i32, align 1
%1 = alloca i32, align 16
call void @f1(i32* %0, i32* %1)
ret void
}
The output currently seems to ignore the ‘align’ and uses 4 byte alignment on the stack.
Thank you.
Robert