The single stack allocator seems to be an arena by other terms, but the double headed version is pretty neat.
Is the final pattern described actually safe for multi-threaded use? Something like the stack has exactly 1 byte left, T1 allocates the dynamic end with 1 byte (dynamic object just happened to be 1 byte, say) and T2 allocates the fixed size end with 1 byte. Wouldn't they end up overlapping in that case?
carplaybits 1 hours ago [-]
[flagged]
Rendered at 03:28:10 GMT+0000 (Coordinated Universal Time) with Vercel.
Is the final pattern described actually safe for multi-threaded use? Something like the stack has exactly 1 byte left, T1 allocates the dynamic end with 1 byte (dynamic object just happened to be 1 byte, say) and T2 allocates the fixed size end with 1 byte. Wouldn't they end up overlapping in that case?