Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20221206105737.69478-1-david@redhat.com>
In-Reply-To: <20221206105737.69478-1-david@redhat.com>
From:   Yang Shi <shy828301@gmail.com>
Date:   Wed, 7 Dec 2022 14:40:34 -0800
Message-ID: <CAHbLzkra0mjn4utzYPNp+-e77W3tb6yJiJMhkK9Kg=Ra_3rQnQ@mail.gmail.com>
Subject: Re: [PATCH v2] mm/swap: fix SWP_PFN_BITS with CONFIG_PHYS_ADDR_T_64BIT
 on 32bit
To:     David Hildenbrand <david@redhat.com>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        Andrew Morton <akpm@linux-foundation.org>,
        Peter Xu <peterx@redhat.com>, Hugh Dickins <hughd@google.com>,
        Andrea Arcangeli <aarcange@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Dec 6, 2022 at 2:57 AM David Hildenbrand <david@redhat.com> wrote:
>
> We use "unsigned long" to store a PFN in the kernel and phys_addr_t to
> store a physical address.
>
> On a 64bit system, both are 64bit wide. However, on a 32bit system, the
> latter might be 64bit wide. This is, for example, the case on x86 with
> PAE: phys_addr_t and PTEs are 64bit wide, while "unsigned long" only
> spans 32bit.
>
> The current definition of SWP_PFN_BITS without MAX_PHYSMEM_BITS misses
> that case, and assumes that the maximum PFN is limited by an 32bit
> phys_addr_t. This implies, that SWP_PFN_BITS will currently only be able to
> cover 4 GiB - 1 on any 32bit system with 4k page size, which is wrong.

Thanks for debugging this. IIUC this means even swap is actually
broken on x86_32 + PAE?

>
> Let's rely on the number of bits in phys_addr_t instead, but make sure to
> not exceed the maximum swap offset, to not make the  BUILD_BUG_ON() in
> is_pfn_swap_entry() unhappy. Note that swp_entry_t is effectively an
> unsigned long and the maximum swap offset shares that value with the
> swap type.
>
> For example, on an 8 GiB x86 PAE system with a kernel config based on
> Debian 11.5 (-> CONFIG_FLATMEM=y, CONFIG_X86_PAE=y), we will currently fail
> removing migration entries (remove_migration_ptes()), because
> mm/page_vma_mapped.c:check_pte() will fail to identify a PFN match as
> swp_offset_pfn() wrongly masks off PFN bits. For example,
> split_huge_page_to_list()->...->remap_page() will leave migration
> entries in place and continue to unlock the page.
>
> Later, when we stumble over these migration entries (e.g., via
> /proc/self/pagemap), pfn_swap_entry_to_page() will BUG_ON() because
> these migration entries shouldn't exist anymore and the page was
> unlocked.
>
> [   33.067591] kernel BUG at include/linux/swapops.h:497!
> [   33.067597] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
> [   33.067602] CPU: 3 PID: 742 Comm: cow Tainted: G            E      6.1.0-rc8+ #16
> [   33.067605] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-1.fc36 04/01/2014
> [   33.067606] EIP: pagemap_pmd_range+0x644/0x650
> [   33.067612] Code: 00 00 00 00 66 90 89 ce b9 00 f0 ff ff e9 ff fb ff ff 89 d8 31 db e8 48 c6 52 00 e9 23 fb ff ff e8 61 83 56 00 e9 b6 fe ff ff <0f> 0b bf 00 f0 ff ff e9 38 fa ff ff 3e 8d 74 26 00 55 89 e5 57 31
> [   33.067615] EAX: ee394000 EBX: 00000002 ECX: ee394000 EDX: 00000000
> [   33.067617] ESI: c1b0ded4 EDI: 00024a00 EBP: c1b0ddb4 ESP: c1b0dd68
> [   33.067619] DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068 EFLAGS: 00010246
> [   33.067624] CR0: 80050033 CR2: b7a00000 CR3: 01bbbd20 CR4: 00350ef0
> [   33.067625] Call Trace:
> [   33.067628]  ? madvise_free_pte_range+0x720/0x720
> [   33.067632]  ? smaps_pte_range+0x4b0/0x4b0
> [   33.067634]  walk_pgd_range+0x325/0x720
> [   33.067637]  ? mt_find+0x1d6/0x3a0
> [   33.067641]  ? mt_find+0x1d6/0x3a0
> [   33.067643]  __walk_page_range+0x164/0x170
> [   33.067646]  walk_page_range+0xf9/0x170
> [   33.067648]  ? __kmem_cache_alloc_node+0x2a8/0x340
> [   33.067653]  pagemap_read+0x124/0x280
> [   33.067658]  ? default_llseek+0x101/0x160
> [   33.067662]  ? smaps_account+0x1d0/0x1d0
> [   33.067664]  vfs_read+0x90/0x290
> [   33.067667]  ? do_madvise.part.0+0x24b/0x390
> [   33.067669]  ? debug_smp_processor_id+0x12/0x20
> [   33.067673]  ksys_pread64+0x58/0x90
> [   33.067675]  __ia32_sys_ia32_pread64+0x1b/0x20
> [   33.067680]  __do_fast_syscall_32+0x4c/0xc0
> [   33.067683]  do_fast_syscall_32+0x29/0x60
> [   33.067686]  do_SYSENTER_32+0x15/0x20
> [   33.067689]  entry_SYSENTER_32+0x98/0xf1
>
> Decrease the indentation level of SWP_PFN_BITS and SWP_PFN_MASK to keep
> it readable and consistent.
>
> Fixes: 0d206b5d2e0d ("mm/swap: add swp_offset_pfn() to fetch PFN from swap entry")
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Peter Xu <peterx@redhat.com>
> Cc: Yang Shi <shy828301@gmail.com>
> Cc: Hugh Dickins <hughd@google.com>
> Cc: Andrea Arcangeli <aarcange@redhat.com>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>
> v1 -> v2:
> * Rely on sizeof(phys_addr_t) and min_t() instead.
> * Survives my various cross compilations and testing on x86 PAE.

Reviewed-by: Yang Shi <shy828301@gmail.com>

>
> ---
>  include/linux/swapops.h | 8 +++++---
>  1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/include/linux/swapops.h b/include/linux/swapops.h
> index 86b95ccb81bb..424d586ed1ca 100644
> --- a/include/linux/swapops.h
> +++ b/include/linux/swapops.h
> @@ -33,11 +33,13 @@
>   * can use the extra bits to store other information besides PFN.
>   */
>  #ifdef MAX_PHYSMEM_BITS
> -#define SWP_PFN_BITS                   (MAX_PHYSMEM_BITS - PAGE_SHIFT)
> +#define SWP_PFN_BITS           (MAX_PHYSMEM_BITS - PAGE_SHIFT)
>  #else  /* MAX_PHYSMEM_BITS */
> -#define SWP_PFN_BITS                   (BITS_PER_LONG - PAGE_SHIFT)
> +#define SWP_PFN_BITS           min_t(phys_addr_t, \
> +                                     sizeof(phys_addr_t) * 8 - PAGE_SHIFT, \
> +                                     SWP_TYPE_SHIFT)
>  #endif /* MAX_PHYSMEM_BITS */
> -#define SWP_PFN_MASK                   (BIT(SWP_PFN_BITS) - 1)
> +#define SWP_PFN_MASK           (BIT(SWP_PFN_BITS) - 1)
>
>  /**
>   * Migration swap entry specific bitfield definitions.  Layout:
>
> base-commit: 76dcd734eca23168cb008912c0f69ff408905235
> --
> 2.38.1
>