Re: [PATCH v5 6/8] PCI/AER: Introduce ratelimit for error logs

[Bjorn Helgaas <helgaas@kernel.org>]

On Mon, Mar 31, 2025 at 05:30:50PM -0700, Jon Pan-Doh wrote:
> On Mon, Mar 31, 2025 at 11:48 AM Bjorn Helgaas <helgaas@kernel.org> wrote:
> > I found the __ratelimit() return values a little confusing (1 == print
> > the message, 0 == don't print), so this is appealing because it's less
> > confusing by itself.
> >
> > But I think we should name this "aer_ratelimit()" and return the
> > result of __ratelimit() without inverting it so it works the same way
> > as __ratelimit() and similar wrappers like ata_ratelimit(),
> > net_ratelimit(), drbd_ratelimit().
> 
> Ack. Caught between readability and consistency :).
> 
> > On Thu, Mar 20, 2025 at 06:58:04PM -0700, Jon Pan-Doh wrote:
> > > --- a/drivers/pci/pci.h
> > > +++ b/drivers/pci/pci.h
> > > @@ -533,6 +533,7 @@ static inline bool pci_dev_test_and_set_removed(struct pci_dev *dev)
> > >
> > >  struct aer_err_info {
> > >       struct pci_dev *dev[AER_MAX_MULTI_ERR_DEVICES];
> > > +     bool ratelimited[AER_MAX_MULTI_ERR_DEVICES];
> 
> s/ratelimited/ratelimit here as well? Should it store aer_ratelimit()
> or !aer_ratelimit()?

I'm in favor of avoiding negation when possible, so I would name it
"ratelimit" with the semantic of "1 == print", even though that seems
a little backwards to me.  But I think it will make sense to people
who read ratelimiting in other areas.