Re: [PATCH] MAINTAINERS: update phylink/sfp keyword matching

From: Russell King - ARM Linux admin
Date: Wed Aug 05 2020 - 14:23:25 EST

On Wed, Aug 05, 2020 at 11:11:28AM -0700, Linus Torvalds wrote:
> On Wed, Aug 5, 2020 at 7:34 AM Russell King <rmk+kernel@xxxxxxxxxxxxxxx> wrote:
> >
> > Is this something you're willing to merge directly please?
> Done.
> That said:
> > -K: phylink
> > +K: phylink\.h|struct\s+phylink|\.phylink|>phylink_|phylink_(autoneg|clear|connect|create|destroy|disconnect|ethtool|helper|mac|mii|of|set|start|stop|test|validate)
> That's a very awkward pattern. I wonder if there could be better ways
> to express this (ie "only apply this pattern to these files" kind of
> thing)

Yes, it's extremely awkward - I spent much of the morning with perl
testing it out on the drivers/ subtree.

> Isn't the 'F' pattern already complete enough that maybe the K pattern
> isn't even worth it?

Unfortunately not; I used not to have a K: line, which presented the
problem that we had users of phylink added to the kernel that were not
being reviewed. So, the suggestion was to add a K: line.

However, I'm now being spammed by syzbot (I've received multiple emails
about the same problem) because, rather than MAINTAINERS being applied
to just patches, it is now being applied to entire source files. This
means that the previous "K: phylink" entry matches not just on patches
(which can be easily ignored) but entire files, such as
net/bluetooth/hci_event.c which happens to contain "phylink" in a
function name.

So, when syzbot identifies there is a problem in
net/bluetooth/hci_event.c, it sends me a report, despite it having
no relevance for me.

RMK's Patch system:
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!