On Fri, Jan 31, 2025 at 6:55 AM Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx> wrote: > So if I follow correctly. We count irqs for any error type and > then mask whatever was set on one that triggered this rate_limit check? > That last one isn't reported other than via a log message. > > Imagine that is a totally unrelated error to the earlier ones, > now RASDaemon has no info on it at all as the tracepoint never > fired. To me that's a very different situation to it knowing there > were 10 errors of the type vs more. > > I'd like to see that final trace point and also to see a tracepoint > that lets rasdaemon etc know you cut off errors after this point > + rasdaemon support for using that. Your understanding is correct. I get your point and will try to address this when I send out the IRQ ratelimiting series. Thanks, Jon
