Re: [PATCH 2/4] kfence: add function to mask address bits

From: Alexander Potapenko <glider@google.com>
To: Sven Schnelle <svens@linux.ibm.com>
Cc: Heiko Carstens <hca@linux.ibm.com>,
	Marco Elver <elver@google.com>,
	 Vasily Gorbik <gor@linux.ibm.com>,
	Christian Borntraeger <borntraeger@de.ibm.com>,
	 kasan-dev <kasan-dev@googlegroups.com>,
	 Linux Memory Management List <linux-mm@kvack.org>,
	LKML <linux-kernel@vger.kernel.org>,
	 linux-s390 <linux-s390@vger.kernel.org>
Subject: Re: [PATCH 2/4] kfence: add function to mask address bits
Date: Thu, 29 Jul 2021 15:59:21 +0200	[thread overview]
Message-ID: <CAG_fn=XHr2j+xVaxjxqSUKU7ddDoekvxVoac9sSJ+Yk3voRUnA@mail.gmail.com> (raw)
In-Reply-To: <yt9dtukdteoj.fsf@linux.ibm.com>

On Thu, Jul 29, 2021 at 3:47 PM Sven Schnelle <svens@linux.ibm.com> wrote:
>
> Alexander Potapenko <glider@google.com> writes:
>
> > On Wed, Jul 28, 2021 at 9:03 PM Heiko Carstens <hca@linux.ibm.com> wrote:
> >>
> >> From: Sven Schnelle <svens@linux.ibm.com>
> >>
> >> s390 only reports the page address during a translation fault.
> >> To make the kfence unit tests pass, add a function that might
> >> be implemented by architectures to mask out address bits.
> >>
> >> Signed-off-by: Sven Schnelle <svens@linux.ibm.com>
> >> Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
> >> ---
> >>  mm/kfence/kfence_test.c | 13 ++++++++++++-
> >>  1 file changed, 12 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/mm/kfence/kfence_test.c b/mm/kfence/kfence_test.c
> >> index 942cbc16ad26..eb6307c199ea 100644
> >> --- a/mm/kfence/kfence_test.c
> >> +++ b/mm/kfence/kfence_test.c
> >> @@ -23,8 +23,15 @@
> >>  #include <linux/tracepoint.h>
> >>  #include <trace/events/printk.h>
> >>
> >> +#include <asm/kfence.h>
> >> +
> >>  #include "kfence.h"
> >>
> >> +/* May be overridden by <asm/kfence.h>. */
> >> +#ifndef arch_kfence_test_address
> >> +#define arch_kfence_test_address(addr) (addr)
> >> +#endif
> >> +
> >>  /* Report as observed from console. */
> >>  static struct {
> >>         spinlock_t lock;
> >> @@ -82,6 +89,7 @@ static const char *get_access_type(const struct expect_report *r)
> >>  /* Check observed report matches information in @r. */
> >>  static bool report_matches(const struct expect_report *r)
> >>  {
> >> +       unsigned long addr = (unsigned long)r->addr;
> >>         bool ret = false;
> >>         unsigned long flags;
> >>         typeof(observed.lines) expect;
> >> @@ -131,22 +139,25 @@ static bool report_matches(const struct expect_report *r)
> >>         switch (r->type) {
> >>         case KFENCE_ERROR_OOB:
> >>                 cur += scnprintf(cur, end - cur, "Out-of-bounds %s at", get_access_type(r));
> >> +               addr = arch_kfence_test_address(addr);
> >
> > Can we normalize addr once before (or after) this switch?
> >
>

> I don't think so. When reporing corrupted memory or an invalid free the
> address is not generated by hardware but kfence itself, and therefore we
> would strip valid bits.

Ah, sorry, I missed that.