[PATCH v7 00/70] Introducing the Maple Tree

[Yu Zhao]

On Fri, Apr 15, 2022 at 7:03 PM Liam Howlett <liam.howlett at oracle.com> wrote:
>
> * Yu Zhao <yuzhao at google.com> [220415 03:11]:
> > On Thu, Apr 14, 2022 at 12:19:11PM -0700, Andrew Morton wrote:
> > > On Thu, 14 Apr 2022 17:15:26 +0000 Liam Howlett <liam.howlett at oracle.com> wrote:
> > >
> > > > > Also I noticed, for the end address to walk_page_range(), Matthew used
> > > > > -1 and you used ULONG_MAX in the maple branch; Andrew used TASK_SIZE
> > > > > below. Having a single value throughout would be great.
> > > >
> > > > I think ULONG_MAX would be best, we should probably change the below to
> > > > ULONG_MAX.
> > >
> > > I switched it to ULONG_MAX/
> > >
> > > > I don't see the code below in mglru-mapletree (62dd11ea8d).  Am I on the
> > > > right branch/commit?
> > >
> > > oop, sorry, sleepy guy failed to include all the mglru patches!  It
> > > should be fixed now (4e03b8e70232).
> >
> > Hi Liam,
> >
> > Mind taking a look? Thanks.
> >
> > I used
> >   1fe4e0d45c05 (HEAD) mm/vmscan: remove obsolete comment in get_scan_count
> >
> > On aarch64:
> >   arch/arm64/kernel/elfcore.c:120:2: error: no member named 'mmap' in 'struct mm_struct'
> >   arch/arm64/kernel/elfcore.c:120:2: error: no member named 'vm_next' in 'struct vm_area_struct'
> >   arch/arm64/kernel/elfcore.c:130:2: error: no member named 'mmap' in 'struct mm_struct'
> >   arch/arm64/kernel/elfcore.c:130:2: error: no member named 'vm_next' in 'struct vm_area_struct'
> >   arch/arm64/kernel/elfcore.c:13:23: note: expanded from macro 'for_each_mte_vma'
> >   arch/arm64/kernel/elfcore.c:13:45: note: expanded from macro 'for_each_mte_vma'
> >   arch/arm64/kernel/elfcore.c:85:2: error: no member named 'mmap' in 'struct mm_struct'
> >   arch/arm64/kernel/elfcore.c:85:2: error: no member named 'vm_next' in 'struct vm_area_struct'
> >   arch/arm64/kernel/elfcore.c:95:2: error: no member named 'mmap' in 'struct mm_struct'
> >   arch/arm64/kernel/elfcore.c:95:2: error: no member named 'vm_next' in 'struct vm_area_struct'
>
> This was fixed in linux-next by commit 3a4f7ef4bed5 [1].  Using the same
> patch fixes this issue, although I will clean up the define prior to
> inclusion in the patches.

Thanks. With that commit, I was able to build and test on aarch64:

  ==================================================================
  BUG: KASAN: invalid-access in mas_destroy+0x10a4/0x126c
  Read of size 8 at addr 7bffff8015c1a110 by task CompositorTileW/9966
  Pointer tag: [7b], memory tag: [fe]

  CPU: 1 PID: 9966 Comm: CompositorTileW Not tainted 5.18.0-rc2-mm1-lockdep+ #2
  Call trace:
   dump_backtrace+0x1a0/0x200
   show_stack+0x24/0x30
   dump_stack_lvl+0x7c/0xa0
   print_report+0x15c/0x524
   kasan_report+0x84/0xb4
   kasan_tag_mismatch+0x28/0x3c
   __hwasan_tag_mismatch+0x30/0x60
   mas_destroy+0x10a4/0x126c
   mas_nomem+0x40/0xf4
   mas_store_gfp+0x9c/0xfc
   do_mas_align_munmap+0x344/0x688
   do_mas_munmap+0xf8/0x118
   __vm_munmap+0x154/0x1e0
   __arm64_sys_munmap+0x44/0x54
   el0_svc_common+0xfc/0x1cc
   do_el0_svc_compat+0x38/0x5c
   el0_svc_compat+0x68/0x118
   el0t_32_sync_handler+0xc0/0xf0
   el0t_32_sync+0x190/0x194

  Allocated by task 9966:
   kasan_set_track+0x4c/0x7c
   __kasan_slab_alloc+0x84/0xa8
   kmem_cache_alloc_bulk+0x300/0x408
   mas_alloc_nodes+0x188/0x268
   mas_nomem+0x88/0xf4
   mas_store_gfp+0x9c/0xfc
   do_mas_align_munmap+0x344/0x688
   do_mas_munmap+0xf8/0x118
   __vm_munmap+0x154/0x1e0
   __arm64_sys_munmap+0x44/0x54
   el0_svc_common+0xfc/0x1cc
   do_el0_svc_compat+0x38/0x5c
   el0_svc_compat+0x68/0x118
   el0t_32_sync_handler+0xc0/0xf0
   el0t_32_sync+0x190/0x194

  Freed by task 9966:
   kasan_set_track+0x4c/0x7c
   kasan_set_free_info+0x2c/0x38
   ____kasan_slab_free+0x13c/0x184
   __kasan_slab_free+0x14/0x24
   slab_free_freelist_hook+0x100/0x1ac
   kmem_cache_free_bulk+0x230/0x3b0
   mas_destroy+0x10d4/0x126c
   mas_nomem+0x40/0xf4
   mas_store_gfp+0x9c/0xfc
   do_mas_align_munmap+0x344/0x688
   do_mas_munmap+0xf8/0x118
   __vm_munmap+0x154/0x1e0
   __arm64_sys_munmap+0x44/0x54
   el0_svc_common+0xfc/0x1cc
   do_el0_svc_compat+0x38/0x5c
   el0_svc_compat+0x68/0x118
   el0t_32_sync_handler+0xc0/0xf0
   el0t_32_sync+0x190/0x194

  The buggy address belongs to the object at ffffff8015c1a100
   which belongs to the cache maple_node of size 256
  The buggy address is located 16 bytes inside of
   256-byte region [ffffff8015c1a100, ffffff8015c1a200)

  The buggy address belongs to the physical page:
  page:fffffffe00570600 refcount:1 mapcount:0 mapping:0000000000000000
index:0xa8ffff8015c1ad00 pfn:0x95c18
  head:fffffffe00570600 order:3 compound_mapcount:0 compound_pincount:0
  flags: 0x10200(slab|head|zone=0|kasantag=0x0)
  raw: 0000000000010200 6cffff8080030850 fffffffe003ec608 dbffff8080016280
  raw: a8ffff8015c1ad00 000000000020001e 00000001ffffffff 0000000000000000
  page dumped because: kasan: bad access detected

  Memory state around the buggy address:
   ffffff8015c19f00: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
   ffffff8015c1a000: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
  >ffffff8015c1a100: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
                        ^
   ffffff8015c1a200: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
   ffffff8015c1a300: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
  ==================================================================