]> www.infradead.org Git - users/willy/pagecache.git/commit
x86: Clear _PAGE_DIRTY when we clear _PAGE_RW
authorMatthew Wilcox (Oracle) <willy@infradead.org>
Tue, 25 Feb 2025 19:25:12 +0000 (14:25 -0500)
committerMatthew Wilcox (Oracle) <willy@infradead.org>
Tue, 25 Feb 2025 19:47:20 +0000 (14:47 -0500)
commitef84acfdb97cdd3278bc5f17d1bfb0be570ed247
treefc2f6dcf248f1d7b730f2af904603a710551493f
parentc22431a39e97861e8b191d47d653b2c5cad813f8
x86: Clear _PAGE_DIRTY when we clear _PAGE_RW

The bit pattern of _PAGE_DIRTY set and _PAGE_RW clear is used to
mark shadow stacks.  This is currently checked for in mk_pte() but
not pfn_pte().  If we add the check to pfn_pte(), it catches vfree()
calling set_direct_map_invalid_noflush() which calls __change_page_attr()
which loads the old protection bits from the PTE, clears the specified
bits and uses pfn_pte() to construct the new PTE.

We should, therefore, clear the _PAGE_DIRTY bit whenever we clear
_PAGE_RW.  I opted to do it in the callers in case we want to use
__change_page_attr() to create shadow stacks inside the kernel at some
point in the future.  Arguably, we might also want to clear _PAGE_ACCESSED
here.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Rick Edgecombe <rick.p.edgecombe@intel.com>
Reported-by: kernel test robot <oliver.sang@intel.com>
Closes: https://lore.kernel.org/oe-lkp/202502241646.719f4651-lkp@intel.com
arch/x86/mm/pat/set_memory.c