On 2024-09-02 11:43, Peter Zijlstra wrote:
On Wed, Aug 28, 2024 at 10:37:19AM -0400, Mathieu Desnoyers wrote:To cover scenarios where the scope of the guard differs from the scope of its activation, introduce DEFINE_INACTIVE_GUARD() and activate_guard(). Here is an example use for a conditionally activated guard variable: void func(bool a) { DEFINE_INACTIVE_GUARD(preempt_notrace, myguard); [...] if (a) { might_sleep(); activate_guard(preempt_notrace, myguard)(); } [ protected code ] }So... I more or less proposed this much earlier: https://lore.kernel.org/all/20230919131038.GC39346@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/T/#mb7b84212619ac743dfe4d2cc81decce451586b27 and Linus took objection to similar patterns. But perhaps my naming wasn't right.
Then you suggested something like a "guard_if()": https://lore.kernel.org/lkml/20231120221524.GD8262@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/ which I find really odd because it requires to evaluate the same condition twice within the function if it is used as guard_if expression and needed as expression within the rest of the function flow. I find the original patch with labels and gotos less ugly than the guard_if(). Hence my proposal to optionally separate the definition from the activation, which nicely integrates with the existing code flow. If Linus' objections were mainly about naming, perhaps what I am suggestion here may be more to his liking ? Thanks, Mathieu -- Mathieu Desnoyers EfficiOS Inc. https://www.efficios.com
