Re: [PATCH] userfaultfd: selftests: make __{s,u}64 format specifiers portable

[Peter Xu]

On Wed, Dec 02, 2020 at 03:52:56PM -0800, Axel Rasmussen wrote:
> On Wed, Dec 2, 2020 at 3:10 PM Peter Xu <peterx@xxxxxxxxxx> wrote:
> >
> > Hi, Axel,
> >
> > Looks mostly good to me, but a few nitpickings below.
> >
> > On Wed, Dec 02, 2020 at 01:15:42PM -0800, Axel Rasmussen wrote:
> >
> > [...]
> >
> > > +static void uffd_error(const char *message, __s64 code)
> > > +{
> > > +     fprintf(stderr, "%s: %" PRId64 "\n", message, (int64_t)code);
> > > +     exit(1);
> > > +}
> >
> > IMHO a macro that can take arbitrary parameters would be nicer, but if it
> > satisfy our need, definitely ok too.
> >
> > [...]
> >
> > > @@ -340,7 +348,8 @@ static void wp_range(int ufd, __u64 start, __u64 len, bool wp)
> > >       prms.mode = wp ? UFFDIO_WRITEPROTECT_MODE_WP : 0;
> > >
> > >       if (ioctl(ufd, UFFDIO_WRITEPROTECT, &prms)) {
> > > -             fprintf(stderr, "clear WP failed for address 0x%Lx\n", start);
> > > +             fprintf(stderr, "clear WP failed for address 0x%" PRIx64 "\n",
> > > +                     (uint64_t)start);
> > >               exit(1);
> >
> > Is it intended to not use uffd_error() here?
> 
> Yes, this is intentional. This particular case prints the value in
> hexadecimal, rather than decimal.
> 
> (Agree that uffd_error() could be made more general to cover cases
> like this. I opted for the simplest thing which covers all but two
> cases - this one, and one where we "return 1;" instead of "exit(1);" -
> but I don't feel strongly.)

Actually it's as simple as:

#define  uffd_error(...)  do {                  \
        fprintf(stderr, __VA_ARGS__);           \
        fprintf(stderr, "\n");                  \
        exit(1);                                \
    } while (0)

But it's okay, I think.

> 
> >
> > >       }
> > >  }
> >
> > [...]
> >
> > > @@ -979,26 +981,20 @@ static int __uffdio_zeropage(int ufd, unsigned long offset, bool retry)
> > >       if (ret) {
> > >               /* real retval in ufdio_zeropage.zeropage */
> > >               if (has_zeropage) {
> > > -                     if (uffdio_zeropage.zeropage == -EEXIST) {
> > > -                             fprintf(stderr, "UFFDIO_ZEROPAGE -EEXIST\n");
> > > -                             exit(1);
> > > -                     } else {
> > > -                             fprintf(stderr, "UFFDIO_ZEROPAGE error %Ld\n",
> > > -                                     uffdio_zeropage.zeropage);
> > > -                             exit(1);
> > > -                     }
> > > +                     uffd_error(uffdio_zeropage.zeropage == -EEXIST ?
> > > +                                              "UFFDIO_ZEROPAGE -EEXIST" :
> > > +                                              "UFFDIO_ZEROPAGE error",
> >
> > Nit: The indents here are a bit odd..
> 
> This is what clang-format yields. Are you thinking it would be better
> to line everything up with the ( in uffd_error( ?
> 
> Or, perhaps this case is a good reason to make uffd_error() a variadic
> macro so we can insert "-EEXIST" || "error" with a "%s".

Yes.  It fixes a build warning, so I think current patch is fine too.

No matter whether you'd like a v2, please feel free to take:

Acked-by: Peter Xu <peterx@xxxxxxxxxx>

Thanks,

-- 
Peter Xu