11.08.2023 16:39, Jeff Layton пишет:
Given that you're now creating the semaphore before launching the test program, do you need to clean it up if the shell script dies abnormally?
I'd rather say "no" or at least I wasn't planning that. Unlike the current impl in C code, ipcmk creates the sem with random ID, so the clash is no longer a threat. As such we don't care about any previous invocations, were they successful or faulty. Which is yet another advantage of this approach. rm_sem() is here just for "politeness", whereas in prev approach you have to deal with the results of the past invocations.
Seems like a reasonable change overall, and I like not having to micromanage the semaphore inside the C program. The cleanup thing can be fixed after the fact if it's even necessary. Reviwed-by: Jeff Layton <jlayton@xxxxxxxxxx>
Thank you!
