Re: [PATCH v2 2/2] mm/uffd: Sanity check write bit for uffd-wp protected ptes

From: Peter Xu <peterx@redhat.com>
To: kernel test robot <lkp@intel.com>
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org,
	oe-kbuild-all@lists.linux.dev,
	Andrea Arcangeli <aarcange@redhat.com>,
	Axel Rasmussen <axelrasmussen@google.com>,
	Ives van Hoorne <ives@codesandbox.io>,
	Nadav Amit <nadav.amit@gmail.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Mike Rapoport <rppt@linux.vnet.ibm.com>
Subject: Re: [PATCH v2 2/2] mm/uffd: Sanity check write bit for uffd-wp protected ptes
Date: Sun, 13 Nov 2022 17:33:52 -0500	[thread overview]
Message-ID: <Y3Fw0DcIr/rXLw3v@x1n> (raw)
In-Reply-To: <202211120515.eqlTdrD3-lkp@intel.com>

On Sat, Nov 12, 2022 at 06:06:22AM +0800, kernel test robot wrote:
> vim +306 arch/x86/include/asm/pgtable.h
> 
>    290	
>    291	#ifdef CONFIG_HAVE_ARCH_USERFAULTFD_WP
>    292	static inline int pte_uffd_wp(pte_t pte)
>    293	{
>    294		bool wp = pte_flags(pte) & _PAGE_UFFD_WP;
>    295	
>    296		/*
>    297		 * Having write bit for wr-protect-marked present ptes is fatal,
>    298		 * because it means the uffd-wp bit will be ignored and write will
>    299		 * just go through.
>    300		 *
>    301		 * Use any chance of pgtable walking to verify this (e.g., when
>    302		 * page swapped out or being migrated for all purposes). It means
>    303		 * something is already wrong.  Tell the admin even before the
>    304		 * process crashes. We also nail it with wrong pgtable setup.
>    305		 */
>  > 306		VM_WARN_ON_ONCE(wp && pte_write(pte));

I just saw the comment in check_pgprot() right below:

	/* mmdebug.h can not be included here because of dependencies */
#ifdef CONFIG_DEBUG_VM
...
#endif

I'll repost, will be the same as v1 code-wise.

-- 
Peter Xu