x86-32: make sure virt_addr_valid() returns false for fixmap addresses
authorJeremy Fitzhardinge <jeremy@goop.org>
Fri, 6 Mar 2009 18:09:26 +0000 (10:09 -0800)
committerIngo Molnar <mingo@elte.hu>
Sun, 8 Mar 2009 19:03:52 +0000 (20:03 +0100)
I found that virt_addr_valid() was returning true for fixmap addresses.

I'm not sure whether pfn_valid() is supposed to include this test,
but there's no harm in being explicit.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Jiri Slaby <jirislaby@gmail.com>
Cc: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <49B166D6.2080505@goop.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/x86/mm/ioremap.c

index 62773ab..62def57 100644 (file)
@@ -87,6 +87,8 @@ bool __virt_addr_valid(unsigned long x)
                return false;
        if (__vmalloc_start_set && is_vmalloc_addr((void *) x))
                return false;
+       if (x >= FIXADDR_START)
+               return false;
        return pfn_valid((x - PAGE_OFFSET) >> PAGE_SHIFT);
 }
 EXPORT_SYMBOL(__virt_addr_valid);