scx_mitosis: Use RAII style for releasing resources #2798
Conversation
Much of the logic was pretty confusing with many gotos and various release_cpumask/group/rcu lock, etc. This creates a few macros for common types and uses the gcc/clang __attribute__(cleanup, ...) to do the proper destruction of the resource once the variable goes out of scope.
|
Can we copy the guard() macros from the kernel and use them? They're pretty much the same things and it'd be nice to have some commonality. |
|
I wasn't aware of the guard() macros, let me look those up |
| cpu = bpf_cpumask_any_distribute(cast_mask(tctx->cpumask)); | ||
| else | ||
| cpu = prev_cpu; | ||
| return bpf_cpumask_any_distribute(cast_mask(tctx->cpumask)); |
There was a problem hiding this comment.
[nit] I know the tctx->cpumask part of the condition above this line is redundant with the check before it & probably just there for the verifier. But it still looks weird to see the check in that order bc the short circuit implications of:
!bpf_cpumask_test_cpu(prev_cpu, cast_mask(tctx->cpumask)) && tctx->cpumask
vs
tctx->cpumask && !bpf_cpumask_test_cpu(prev_cpu, cast_mask(tctx->cpumask))
| s32 cpu; | ||
| DECLARE_IDLE_CPUMASK(idle_smtmask) = scx_bpf_get_idle_smtmask(); | ||
|
|
||
| if (!(task_cpumask = (struct cpumask *)tctx->cpumask) || |
There was a problem hiding this comment.
Know this is outside the diff, but the bpf_cpumask is larger than the task_cpumask. I think we'll be fine using it as a pointer as long as we never e.g. copy size of bpf_cpumask into a cpumask
https://elixir.bootlin.com/linux/v6.16.8/source/kernel/bpf/cpumask.c#L25
4 participants
Much of the logic was pretty confusing with many gotos and various release_cpumask/group/rcu lock, etc. This creates a few macros for common types and uses the gcc/clang attribute(cleanup, ...) to do the proper destruction of the resource once the variable goes out of scope.