revert "highmem: catch illegal nesting"
authorAndrew Morton <akpm@linux-foundation.org>
Tue, 11 Sep 2007 22:24:10 +0000 (15:24 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Wed, 12 Sep 2007 00:21:27 +0000 (17:21 -0700)
Revert

commit 656dad312fb41ed95ef08325e9df9bece3aacbbb
Author: Ingo Molnar <mingo@elte.hu>
Date:   Sat Feb 10 01:46:36 2007 -0800

[PATCH] highmem: catch illegal nesting

Catch illegally nested kmap_atomic()s even if the page that is mapped by
the 'inner' instance is from lowmem.

This avoids spuriously zapped kmap-atomic ptes and turns hard to find
crashes into clear asserts at the bug site.

Problem is, a get_zeroed_page(GFP_KERNEL) from interrupt context will trigger
this check if non-irq code on this CPU holds a KM_USER0 mapping.  But that
get_zeroed_page() will never be altering the kmap slot anyway due to the
GFP_KERNEL.

Cc: Christoph Lameter <clameter@sgi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/i386/mm/highmem.c

index ad8d86cc683ee3f6786142c86a95f1fffb7e4f94..1c3bf95f7356826a9243081c8269bd1475df0b1a 100644 (file)
@@ -34,17 +34,16 @@ void *kmap_atomic_prot(struct page *page, enum km_type type, pgprot_t prot)
        /* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
        pagefault_disable();
 
-       idx = type + KM_TYPE_NR*smp_processor_id();
-       BUG_ON(!pte_none(*(kmap_pte-idx)));
-
        if (!PageHighMem(page))
                return page_address(page);
 
+       idx = type + KM_TYPE_NR*smp_processor_id();
        vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
+       BUG_ON(!pte_none(*(kmap_pte-idx)));
        set_pte(kmap_pte-idx, mk_pte(page, prot));
        arch_flush_lazy_mmu_mode();
 
-       return (void*) vaddr;
+       return (void *)vaddr;
 }
 
 void *kmap_atomic(struct page *page, enum km_type type)