]> git.baikalelectronics.ru Git - kernel.git/commitdiff
kfence, x86: fix preemptible warning on KPTI-enabled systems
authorMarco Elver <elver@google.com>
Fri, 9 Apr 2021 20:27:44 +0000 (13:27 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 9 Apr 2021 21:54:23 +0000 (14:54 -0700)
On systems with KPTI enabled, we can currently observe the following
warning:

  BUG: using smp_processor_id() in preemptible
  caller is invalidate_user_asid+0x13/0x50
  CPU: 6 PID: 1075 Comm: dmesg Not tainted 5.12.0-rc4-gda4a2b1a5479-kfence_1+ #1
  Hardware name: Hewlett-Packard HP Pro 3500 Series/2ABF, BIOS 8.11 10/24/2012
  Call Trace:
   dump_stack+0x7f/0xad
   check_preemption_disabled+0xc8/0xd0
   invalidate_user_asid+0x13/0x50
   flush_tlb_one_kernel+0x5/0x20
   kfence_protect+0x56/0x80
   ...

While it normally makes sense to require preemption to be off, so that
the expected CPU's TLB is flushed and not another, in our case it really
is best-effort (see comments in kfence_protect_page()).

Avoid the warning by disabling preemption around flush_tlb_one_kernel().

Link: https://lore.kernel.org/lkml/YGIDBAboELGgMgXy@elver.google.com/
Link: https://lkml.kernel.org/r/20210330065737.652669-1-elver@google.com
Signed-off-by: Marco Elver <elver@google.com>
Reported-by: Tomi Sarvela <tomi.p.sarvela@intel.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Jann Horn <jannh@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/x86/include/asm/kfence.h

index 97bbb4a9083a76cce625d5255d2e25ce86aeb2a5..05b48b33baf0b811b2788750d4539f18c9bd2255 100644 (file)
@@ -56,8 +56,13 @@ static inline bool kfence_protect_page(unsigned long addr, bool protect)
        else
                set_pte(pte, __pte(pte_val(*pte) | _PAGE_PRESENT));
 
-       /* Flush this CPU's TLB. */
+       /*
+        * Flush this CPU's TLB, assuming whoever did the allocation/free is
+        * likely to continue running on this CPU.
+        */
+       preempt_disable();
        flush_tlb_one_kernel(addr);
+       preempt_enable();
        return true;
 }