perf/core: Disable page faults when getting phys address

[ Upstream commit d3296fb ]

We hit following warning when running tests on kernel
compiled with CONFIG_DEBUG_ATOMIC_SLEEP=y:

 WARNING: CPU: 19 PID: 4472 at mm/gup.c:2381 __get_user_pages_fast+0x1a4/0x200
 CPU: 19 PID: 4472 Comm: dummy Not tainted 5.6.0-rc6+ #3
 RIP: 0010:__get_user_pages_fast+0x1a4/0x200
 ...
 Call Trace:
  perf_prepare_sample+0xff1/0x1d90
  perf_event_output_forward+0xe8/0x210
  __perf_event_overflow+0x11a/0x310
  __intel_pmu_pebs_event+0x657/0x850
  intel_pmu_drain_pebs_nhm+0x7de/0x11d0
  handle_pmi_common+0x1b2/0x650
  intel_pmu_handle_irq+0x17b/0x370
  perf_event_nmi_handler+0x40/0x60
  nmi_handle+0x192/0x590
  default_do_nmi+0x6d/0x150
  do_nmi+0x2f9/0x3c0
  nmi+0x8e/0xd7

While __get_user_pages_fast() is IRQ-safe, it calls access_ok(),
which warns on:

  WARN_ON_ONCE(!in_task() && !pagefault_disabled())

Peter suggested disabling page faults around __get_user_pages_fast(),
which gets rid of the warning in access_ok() call.

Suggested-by: Peter Zijlstra <[email protected]>
Signed-off-by: Jiri Olsa <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Sasha Levin <[email protected]>

kernel/events/core.c

@@ -6734,9 +6734,12 @@ static u64 perf_virt_to_phys(u64 virt)
  * Try IRQ-safe __get_user_pages_fast first.
  * If failed, leave phys_addr as 0.
  */
- if ((current->mm != NULL) &&
- (__get_user_pages_fast(virt, 1, 0, &p) == 1))
- phys_addr = page_to_phys(p) + virt % PAGE_SIZE;
+ if (current->mm != NULL) {
+ pagefault_disable();
+ if (__get_user_pages_fast(virt, 1, 0, &p) == 1)
+ phys_addr = page_to_phys(p) + virt % PAGE_SIZE;
+ pagefault_enable();
+ }
 
  if (p)
  put_page(p);