https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121213
Jeffrey A. Law <law at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2025-08-10
Assignee|unassigned at gcc dot gnu.org |austinkylelaw at gmail
dot com
Status|UNCONFIRMED |ASSIGNED
Ever confirmed|0 |1
--- Comment #2 from Jeffrey A. Law <law at gcc dot gnu.org> ---
So what Austin is doing is:
1. Change the predicate on the two amoswapp insns so that the source operand
can be reg_or_0_operand. That allows (const_int 0) to be recognized in that
operand position.
2. Change that operand's constraint to be "rJ" so that register allocation
knows (const_int 0) is OK.
3. Change the constraint of operand 0 to "=r", dropping the earlyclobber tag
which isn't needed anymore.
I think he's got that coded up. He still needs to build a new test for the
regression suite, then run the regression suite. But he's getting close.
