On 11/6/25 7:22 PM, Andrew MacLeod wrote:
I can add a check in the inferred range manager for atomic loads to
resolve this PR.
The IL sequence tends to look like:
_1 = &this_2(D)->b;
__atomic_load_8 (_1, 0);
Just want to make sure I get this right since memory operations are not
my strong suit.
The first argument to the atomic load is non-null (so _1), as well as
the base of the RHS of the address expression that defines _1 are non-
zero?. (this_2)
The attached patch scavenges a little code from
fold_using_range::range_of_address that I think works... but perhaps
there is a more efficient way to do this? Is this likely to work OK and
be safe? or are there additional checks I need to be doing?
And I suppose there is an entire range of atomic operations this applies
to? certainly atomic_store should qualify...
Bootstraps on x86_64-pc-linux-gnu with no regressions, but I'm not sure
that is a really extensive test for this..
I think my question would be do we have to handle this specially at all.
As you later noted, we may get what we want just by getting the
non-null on the builtins set properly.
jeff
