This one's more straightforward than the debate around aligned_t because syncvars were already treated as atomics internally. This was just a matter of making the loads and stores explicit.
I also made ARM take the same branch as x86 for these blocks of code since we're currently assuming lock-free atomics there too.
This one's more straightforward than the debate around
aligned_tbecause syncvars were already treated as atomics internally. This was just a matter of making the loads and stores explicit.I also made ARM take the same branch as x86 for these blocks of code since we're currently assuming lock-free atomics there too.