From: Andrew Morton <akpm@osdl.org>

prep_zero_page() uses KM_USER0 and hence may not be used from IRQ context, at
least for highmem pages.

Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Christoph Lameter <christoph@lameter.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
---

 mm/page_alloc.c |    5 +++++
 1 files changed, 5 insertions(+)

diff -puN mm/page_alloc.c~mm-prep_zero_page-in-irq-is-a-bug mm/page_alloc.c
--- devel/mm/page_alloc.c~mm-prep_zero_page-in-irq-is-a-bug	2006-02-19 20:23:00.000000000 -0800
+++ devel-akpm/mm/page_alloc.c	2006-02-19 20:24:48.000000000 -0800
@@ -217,6 +217,11 @@ static inline void prep_zero_page(struct
 	int i;
 
 	BUG_ON((gfp_flags & (__GFP_WAIT | __GFP_HIGHMEM)) == __GFP_HIGHMEM);
+	/*
+	 * clear_highpage() will use KM_USER0, so it's a bug to use __GFP_ZERO
+	 * and __GFP_HIGHMEM from hard or soft interrupt context.
+	 */
+	BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
 	for (i = 0; i < (1 << order); i++)
 		clear_highpage(page + i);
 }
_