On 30.07.2025 00:31, Andrew Cooper wrote: > MISRA Rule 5.5 objects to a macro aliasing a function, which is what > pirq_cleanup_check() does. The macro is an overly-complicated way of writing: > > if ( !pirq->evtchn ) > pirq_cleanup_check(pirq, d);
The way you word this does not in any way appreciate the original intention behind introducing this macro: It wasn't to write something in an overly complicated form, but to make sure no call site would mistakenly omit the condition. IOW ... > There are only a handful of users, so expand it inline to be plain regular C. > > Doing this shows one path now needing braces, and one path in > evtchn_bind_pirq() where the expanded form simplies back to no delta, as it > follows an unconditional clear of info->evtchn. ... this one place you don't need to touch is now becoming at least close to a bad precedent. The code changes themselves being all fine, the above is not an objection to the patch. I'd much appreciate if the description was re-worded, though, to take into account the original intention. And obviously with the original intention in mind, whether pleasing Misra this way is actually outweighing the (latent) risk wants (re)considering. Jan
