From: Andrea Arcangeli <andrea@suse.de>
Subject: Tell the end user they should not worry about GFP_ATOMIC failures
Patch-mainline: no
References: SUSE48965

x

Signed-off-by: Andrea Arcangeli <andrea@suse.de>

---
 mm/page_alloc.c |    8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2087,7 +2087,13 @@ rebalance:
 		va_end(args);
 	}
 
-	pr_warn("%s: page allocation failure: order:%d, mode:0x%x\n",
+	if (!(gfp_mask & __GFP_WAIT)) {
+		pr_info("The following is only an harmless informational message.\n");
+		pr_info("Unless you get a _continuous_flood_ of these messages it means\n");
+		pr_info("everything is working fine. Allocations from irqs cannot be\n");
+		pr_info("perfectly reliable and the kernel is designed to handle that.\n");
+	}
+	pr_info("%s: page allocation failure. order:%d, mode:0x%x\n",
 		current->comm, order, gfp_mask);

 	dump_stack();