percpu: fix how @gfp is interpreted by the percpu allocator

When @gfp is specified, the percpu allocator is interested in whether
it contains all of GFP_KERNEL or not.  If it does, the normal
allocation path is taken; otherwise, the atomic allocation path.
Unfortunately, pcpu_alloc() was incorrectly testing for whether @gfp
contains any part of GFP_KERNEL.

Fix it by testing "(gfp & GFP_KERNEL) != GFP_KERNEL" instead of
"!(gfp & GFP_KERNEL)" to decide whether the allocation should be
atomic or not.

Signed-off-by: Tejun Heo <tj@kernel.org>

mm/percpu.c

@@ -876,7 +876,7 @@ static void __percpu *pcpu_alloc(size_t size, size_t align, bool reserved,
 	static int warn_limit = 10;
 	struct pcpu_chunk *chunk;
 	const char *err;
-	bool is_atomic = !(gfp & GFP_KERNEL);
+	bool is_atomic = (gfp & GFP_KERNEL) != GFP_KERNEL;
 	int occ_pages = 0;
 	int slot, off, new_alloc, cpu, ret;
 	unsigned long flags;