arm64/mm: Add warning for outside range requests in vmemmap_populate()
authorAnshuman Khandual <anshuman.khandual@arm.com>
Tue, 5 Jan 2021 11:24:11 +0000 (16:54 +0530)
committerWill Deacon <will@kernel.org>
Wed, 20 Jan 2021 12:27:57 +0000 (12:27 +0000)
vmemmap_populate() does not validate the requested vmemmap address range to
be inside the platform assigned space i.e [VMEMMAP_START..VMEMMAP_END] for
vmemmap. Instead it would just go ahead and create the mapping which might
then overlap with other sections in the kernel virtual address space.

Just adding an warning here for range overrun which would help detect the
problem earlier on, before a potential struct page corruption. This also
makes vmemmap_populate() symmetrical with vmemmap_free() which already has
a similar warning.

Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Anshuman Khandual <anshuman.khandual@arm.com>
Link: https://lore.kernel.org/r/1609845851-25064-1-git-send-email-anshuman.khandual@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
arch/arm64/mm/mmu.c

index ae0c3d0..f938e80 100644 (file)
@@ -1094,6 +1094,7 @@ static void free_empty_tables(unsigned long addr, unsigned long end,
 int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
                struct vmem_altmap *altmap)
 {
+       WARN_ON((start < VMEMMAP_START) || (end > VMEMMAP_END));
        return vmemmap_populate_basepages(start, end, node, altmap);
 }
 #else  /* !ARM64_SWAPPER_USES_SECTION_MAPS */
@@ -1107,6 +1108,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
        pud_t *pudp;
        pmd_t *pmdp;
 
+       WARN_ON((start < VMEMMAP_START) || (end > VMEMMAP_END));
        do {
                next = pmd_addr_end(addr, end);