Re: [RFC PATCH v5 2/7] selftests: x86: test_mremap_vdso: skip if vdso is msealed

[Jeff Xu <jeffxu@chromium.org>]

On Thu, Feb 13, 2025 at 11:28 AM Kees Cook <kees@kernel.org> wrote:
>
>
>
> On February 13, 2025 6:14:00 AM PST, Jeff Xu <jeffxu@chromium.org> wrote:
> >On Wed, Feb 12, 2025 at 5:04 AM Thomas Weißschuh
> ><thomas.weissschuh@linutronix.de> wrote:
> >>
> >> On Wed, Feb 12, 2025 at 03:21:50AM +0000, jeffxu@chromium.org wrote:
> >> > From: Jeff Xu <jeffxu@chromium.org>
> >> >
> >> > Add code to detect if the vdso is memory sealed, skip the test
> >> > if it is.
> >> >
> >> > Signed-off-by: Jeff Xu <jeffxu@chromium.org>
> >> > ---
> >> >  .../testing/selftests/x86/test_mremap_vdso.c  | 38 +++++++++++++++++++
> >> >  1 file changed, 38 insertions(+)
> >> >
> >> > diff --git a/tools/testing/selftests/x86/test_mremap_vdso.c b/tools/testing/selftests/x86/test_mremap_vdso.c
> >> > index d53959e03593..c68077c56b22 100644
> >> > --- a/tools/testing/selftests/x86/test_mremap_vdso.c
> >> > +++ b/tools/testing/selftests/x86/test_mremap_vdso.c
> >> > @@ -14,6 +14,7 @@
> >> >  #include <errno.h>
> >> >  #include <unistd.h>
> >> >  #include <string.h>
> >> > +#include <stdbool.h>
> >> >
> >> >  #include <sys/mman.h>
> >> >  #include <sys/auxv.h>
> >> > @@ -55,13 +56,50 @@ static int try_to_remap(void *vdso_addr, unsigned long size)
> >> >
> >> >  }
> >> >
> >> > +#define VDSO_NAME "[vdso]"
> >> > +#define VMFLAGS "VmFlags:"
> >> > +#define MSEAL_FLAGS "sl"
> >> > +#define MAX_LINE_LEN 512
> >> > +
> >> > +bool vdso_sealed(FILE *maps)
> >> > +{
> >> > +     char line[MAX_LINE_LEN];
> >> > +     bool has_vdso = false;
> >> > +
> >> > +     while (fgets(line, sizeof(line), maps)) {
> >> > +             if (strstr(line, VDSO_NAME))
> >> > +                     has_vdso = true;
> >> > +
> >> > +             if (has_vdso && !strncmp(line, VMFLAGS, strlen(VMFLAGS))) {
> >> > +                     if (strstr(line, MSEAL_FLAGS))
> >> > +                             return true;
> >> > +
> >> > +                     return false;
> >>
> >> This only tests that any mapping after the vdso is sealed.
> >
> >The code above begins by searching for the "[vdso]" string, then looks
> >for the first line that starts with "VmFlags:", and looks for the "sl"
> >substring within that line. We're assuming the format of smaps won't
> >change from release to release.
>
> Oh, I missed this in my reviews: nothing _resets_ has_vdso to false, so if any other mapping follows vdso that happens to be sealed, this will return true...
>
It won't return the next mapping's sealing flag.
After finding the "[vdso]", if the next line that contains VMFLAGS
doesn't have the "sl" flag, the function returns false immediately.

> >
> >> There is a real parser for /proc/self/smaps in
> >> tools/testing/selftests/mm/vm_util.[ch], see check_vmflag_io().
> >>
> >This test is in selftest/x86, which makes it hard to include the
> >vm_util from selftest/mm, if that's what you're asking.
>
> Hm, we have done these kinds of inter-tests dependencies before. (E.g. seccomp includes stuff from the clone tests.) I think it should be possible if it can just be a header include. Linking across tests would be more frustrating.
>
I can switch to vm_util, I will need to add a new parsing function in
vm_util, the existing __get_smap_entry() only searches for vm's
starting address, not name.

Thanks
-Jeff


> -Kees

>
> --
> Kees Cook