Re: [PATCH] serial: extend compile-test coverage

[Johan Hovold]

On Thu, Apr 22, 2021 at 09:40:38AM +0200, Geert Uytterhoeven wrote:
> Hi Johan,
> 
> On Thu, Apr 22, 2021 at 9:37 AM Johan Hovold <johan@xxxxxxxxxx> wrote:
> > On Thu, Apr 22, 2021 at 09:11:50AM +0200, Geert Uytterhoeven wrote:
> > > On Thu, Apr 22, 2021 at 3:20 AM kernel test robot <lkp@xxxxxxxxx> wrote:
> > > > >> drivers/tty/serial/serial_txx9.c:987:12: warning: no previous prototype for function 'early_serial_txx9_setup' [-Wmissing-prototypes]
> > > >    int __init early_serial_txx9_setup(struct uart_port *port)
> > > >               ^
> > > >    drivers/tty/serial/serial_txx9.c:987:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
> > > >    int __init early_serial_txx9_setup(struct uart_port *port)
> > > >    ^
> > > >    static
> > > >    1 warning generated.
> > >
> > > This function is called from arch/mips/txx9/generic/setup.c, and does
> > > have a forward declaration in arch/mips/include/asm/txx9/generic.h.
> > > Unfortunately the latter cannot be included from the driver, unless
> > > the || COMPILE_TEST is dropped again.
> >
> > Thanks, Geert. I was just about to send a v2 without the txx9 hunk.
> 
> well, I guess apart from this (W=1!) warning, this driver still
> compile-tests fine.
> Do we consider hard-to-fix W=1 warnings to be legitimate blockers
> for enabling compile-testing?

Since the build bots have enabled them (and people have turned it into a
crusade to suppress every W=1 warning by any means) I think it's
reasonable to not introduce new ones knowingly.

And I'm not too motivated right now to try to work around this one
myself. The idea here was just to enable COMPILE_TEST for drivers that
did not have any build-time dependencies.

Johan