Revert "kmemleak: allow to coexist with fault injection"

[ Upstream commit df9576d ]

When running ltp's oom test with kmemleak enabled, the below warning was
triggerred since kernel detects __GFP_NOFAIL & ~__GFP_DIRECT_RECLAIM is
passed in:

  WARNING: CPU: 105 PID: 2138 at mm/page_alloc.c:4608 __alloc_pages_nodemask+0x1c31/0x1d50
  Modules linked in: loop dax_pmem dax_pmem_core ip_tables x_tables xfs virtio_net net_failover virtio_blk failover ata_generic virtio_pci virtio_ring virtio libata
  CPU: 105 PID: 2138 Comm: oom01 Not tainted 5.2.0-next-20190710+ #7
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.10.2-0-g5f4c7b1-prebuilt.qemu-project.org 04/01/2014
  RIP: 0010:__alloc_pages_nodemask+0x1c31/0x1d50
  ...
   kmemleak_alloc+0x4e/0xb0
   kmem_cache_alloc+0x2a7/0x3e0
   mempool_alloc_slab+0x2d/0x40
   mempool_alloc+0x118/0x2b0
   bio_alloc_bioset+0x19d/0x350
   get_swap_bio+0x80/0x230
   __swap_writepage+0x5ff/0xb20

The mempool_alloc_slab() clears __GFP_DIRECT_RECLAIM, however kmemleak
has __GFP_NOFAIL set all the time due to d9570ee ("kmemleak:
allow to coexist with fault injection").  But, it doesn't make any sense
to have __GFP_NOFAIL and ~__GFP_DIRECT_RECLAIM specified at the same
time.

According to the discussion on the mailing list, the commit should be
reverted for short term solution.  Catalin Marinas would follow up with
a better solution for longer term.

The failure rate of kmemleak metadata allocation may increase in some
circumstances, but this should be expected side effect.

Link: http://lkml.kernel.org/r/[email protected]
Fixes: d9570ee ("kmemleak: allow to coexist with fault injection")
Signed-off-by: Yang Shi <[email protected]>
Suggested-by: Catalin Marinas <[email protected]>
Acked-by: Michal Hocko <[email protected]>
Cc: Dmitry Vyukov <[email protected]>
Cc: David Rientjes <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Qian Cai <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Signed-off-by: Sasha Levin <[email protected]>

mm/kmemleak.c

@@ -126,7 +126,7 @@
 /* GFP bitmask for kmemleak internal allocations */
 #define gfp_kmemleak_mask(gfp)	(((gfp) & (GFP_KERNEL | GFP_ATOMIC)) | \
 				 __GFP_NORETRY | __GFP_NOMEMALLOC | \
-				 __GFP_NOWARN | __GFP_NOFAIL)
+				 __GFP_NOWARN)
 
 /* scanning area inside a memory block */
 struct kmemleak_scan_area {