Re: [PATCH 3/3] PCI: imx6: Make fallthrough comments more consistent

[Andrey Smirnov]

On Mon, Dec 17, 2018 at 5:58 AM Bjorn Helgaas <helgaas@xxxxxxxxxx> wrote:
>
> On Sun, Dec 16, 2018 at 03:09:16PM -0800, Andrey Smirnov wrote:
> > Convert all fallthrough comments to say "fall through", as well as
> > modify their placement to the point where the "break" would normally
> > be.
> >
> > Cc: Bjorn Helgaas <bhelgaas@xxxxxxxxxx>
> > Cc: Fabio Estevam <fabio.estevam@xxxxxxx>
> > Cc: Chris Healy <cphealy@xxxxxxxxx>
> > Cc: Lucas Stach <l.stach@xxxxxxxxxxxxxx>
> > Cc: Leonard Crestez <leonard.crestez@xxxxxxx>
> > Cc: "A.s. Dong" <aisheng.dong@xxxxxxx>
> > Cc: Richard Zhu <hongxing.zhu@xxxxxxx>
> > Cc: linux-imx@xxxxxxx
> > Cc: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
> > Cc: linux-kernel@xxxxxxxxxxxxxxx
> > Cc: linux-pci@xxxxxxxxxxxxxxx
> > Suggested-by: Bjorn Helgaas <helgaas@xxxxxxxxxx>
>
> I didn't make it very clear, but my suggestion was really to remove
> the annotation completely; see below.
>
> > Signed-off-by: Andrey Smirnov <andrew.smirnov@xxxxxxxxx>
> > ---
> >  drivers/pci/controller/dwc/pci-imx6.c | 14 +++++++++-----
> >  1 file changed, 9 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/pci/controller/dwc/pci-imx6.c b/drivers/pci/controller/dwc/pci-imx6.c
> > index 59658577e81d..a0510e185d44 100644
> > --- a/drivers/pci/controller/dwc/pci-imx6.c
> > +++ b/drivers/pci/controller/dwc/pci-imx6.c
> > @@ -362,7 +362,8 @@ static void imx6_pcie_assert_core_reset(struct imx6_pcie *imx6_pcie)
> >
> >       switch (imx6_pcie->variant) {
> >       case IMX7D:
> > -     case IMX8MQ: /* FALLTHROUGH */
> > +             /* fall through */
> > +     case IMX8MQ:
> >               reset_control_assert(imx6_pcie->pciephy_reset);
> >               reset_control_assert(imx6_pcie->apps_reset);
> >               break;
>
> IMO this use of "fall through" is superfluous and unusual in the Linux
> source.
>
> A "fall through" comment would be useful if the IMX7D case had
> executable code but no "break".  Then the comment shows that the
> intent was to execute *both* the IMX7D code and the IMX8MQ code and
> the lack of a "break" was intentional.
>
> In this case, the intent is to treat IMX7D and IMX8MQ the same, and
> there's no executable code specifically for the IMX7D.  I think it's
> easiest to read that when the list of identical cases is all together
> without the comment in the middle, i.e., as
>
> >       case IMX7D:
> >       case IMX8MQ:
> >               reset_control_assert(imx6_pcie->pciephy_reset);
>
> rather than this:
>
> >       case IMX7D:
> >               /* fall through */
> >       case IMX8MQ:
> >               reset_control_assert(imx6_pcie->pciephy_reset);

OK, understood, will remove in next version.

Thanks,
Andrey Smirnov