linux-kernel - Re: [PATCH 2/2] PCI: mvebu: switch to using gpiod API

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 16 Sep 2022 09:22:59 +0200
From:   Bartosz Golaszewski <brgl@...ev.pl>
To:     Kent Gibson <warthog618@...il.com>
Cc:     Linus Walleij <linus.walleij@...aro.org>,
        Dmitry Torokhov <dmitry.torokhov@...il.com>,
        Pali Rohár <pali@...nel.org>,
        Shawn Guo <shawn.guo@...aro.org>,
        Lorenzo Pieralisi <lpieralisi@...nel.org>,
        Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        Rob Herring <robh@...nel.org>,
        Krzysztof Wilczyński <kw@...ux.com>,
        linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] PCI: mvebu: switch to using gpiod API

On Thu, Sep 15, 2022 at 11:30 AM Kent Gibson <warthog618@...il.com> wrote:
>
> On Thu, Sep 15, 2022 at 10:51:02AM +0200, Linus Walleij wrote:
> > On Thu, Sep 15, 2022 at 4:23 AM Kent Gibson <warthog618@...il.com> wrote:
> >
> > > After sleeping on this, I'm even more in disagreement with renaming
> > > value to state.
> >
> > OK let's not do that then.
> >
> > > OTOH, I totally agree with the addition of GPIOD_ACTIVE/INACTIVE to be
> > > used for the logical cases, and even a script to apply it globally.
> > > Ideally that script would change both the calls to the logical functions
> > > to use ACTIVE/INACTIVE, and the physical to HIGH/LOW.
> >
> > OK we have consensus on this.
> >
> > > Introducing enums for those, and changing the function signatures to
> > > use those rather than int makes sense to me too.
> >
> > Either they can be enum or defined to bool true/false. Not really
> > sure what is best. But intuitively enum "feels better" for me.
> >
>
> Enums work for me - especially if the goal is to differentiate
> logical from physical - there should be a distinct enum for each.
>

We won't even have to change the function signatures if we go with
enums - they already take an int and I'm in general against putting
enum types into function signatures in C as they give you a false
sense of a strong type.

Bart

> > > Though I'm not sure
> > > why that has to wait until after all users are changed to the new macros.
> > > Would that generate lint warnings?
> >
> > I rather want it all to happen at once. One preparatory commit
> > adding the new types and one sed script to refactor the whole
> > lot. Not gradual switchover.
> >
> > The reason is purely administrative: we have too many refactorings
> > lagging behind, mainly the GPIO descriptors that have been
> > lagging behind for what is it? 5 years? 10? GPIO irqchips also dragged
> > out for way too long. We can't keep doing things gradually like
> > this, it takes too much time and effort.
> >
> > I don't want any more "in-transition-indefinitely" stuff in the GPIO
> > subsystem if I can avoid it.
> >
> > Therefore I would advice to switch it all over at the end of a merge
> > window and be done with it.
> >
>
> Agreed - do it all at once.  My question was specific to the change of the
> function signatures to using enums - what is to prevent us doing that
> before running the sed script, and have the script switch usage over to
> the enums?
>
> Cheers,
> Kent.