25 Aug
2023
25 Aug
'23
10:18 p.m.
But will that still work with the new implementation? Otherwise it is unclear if supporting one known game using SRW locks in an essentially broken way at the cost of stopping supporting another broken case worth it.
I don't think we know that any application uses unaligned SRW locks, to be clear. But regardless, I haven't read the patch, but I can only presume that it doesn't do anything more problematic than interlocked operations on the pointer. Which, if I recall correctly, while slow on x86, should at least be valid. It may be broken on other architectures, though...