Re: [PATCH v2] selftests: mm: fix map_hugetlb failure on 64K page size systems

[Nico Pache]

Hi Andrew,

No, I think it's always been broken-- I don't think the test was
written with 512M huge page sizes in mind.

-- Nico

On Sat, Jan 20, 2024 at 9:39 PM Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Fri, 19 Jan 2024 06:14:29 -0700 Nico Pache <npache@xxxxxxxxxx> wrote:
>
> > On systems with 64k page size and 512M huge page sizes, the allocation
> > and test succeeds but errors out at the munmap. As the comment states,
> > munmap will failure if its not HUGEPAGE aligned. This is due to the
> > length of the mapping being 1/2 the size of the hugepage causing the
> > munmap to not be hugepage aligned. Fix this by making the mapping length
> > the full hugepage if the hugepage is larger than the length of the
> > mapping.
>
> Is
>
> Fixes: fa7b9a805c79 ("tools/selftest/vm: allow choosing mem size and page size in map_hugetlb")
>
> a suitable Fixes: target for this?
>
> > --- a/tools/testing/selftests/mm/map_hugetlb.c
> > +++ b/tools/testing/selftests/mm/map_hugetlb.c
> > @@ -15,6 +15,7 @@
> >  #include <unistd.h>
> >  #include <sys/mman.h>
> >  #include <fcntl.h>
> > +#include "vm_util.h"
> >
> >  #define LENGTH (256UL*1024*1024)
> >  #define PROTECTION (PROT_READ | PROT_WRITE)
> > @@ -58,10 +59,16 @@ int main(int argc, char **argv)
> >  {
> >       void *addr;
> >       int ret;
> > +     size_t hugepage_size;
> >       size_t length = LENGTH;
> >       int flags = FLAGS;
> >       int shift = 0;
> >
> > +     hugepage_size = default_huge_page_size();
> > +     /* munmap with fail if the length is not page aligned */
> > +     if (hugepage_size > length)
> > +             length = hugepage_size;
> > +
> >       if (argc > 1)
> >               length = atol(argv[1]) << 20;
> >       if (argc > 2) {
> > --
> > 2.43.0
>