Re: [syzbot] [fs?] KASAN: slab-use-after-free Read in invalidate_bh_lru

From: Matthew Wilcox
Date: Fri Sep 13 2024 - 10:00:01 EST

Next message: Christophe Leroy: "Re: [PATCH] crypto: Removing CRYPTO_AES_GCM_P10."
Previous message: Jeff Layton: "[PATCH v7 11/11] tmpfs: add support for multigrain timestamps"
In reply to: Hillf Danton: "Re: [syzbot] [fs?] KASAN: slab-use-after-free Read in invalidate_bh_lru"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Fri, Sep 13, 2024 at 08:31:04PM +0800, Hillf Danton wrote:
> On Thu, 12 Sep 2024 02:19:26 -0700
> > ==================================================================
> > BUG: KASAN: slab-use-after-free in instrument_atomic_read include/linux/instrumented.h:68 [inline]
> > BUG: KASAN: slab-use-after-free in atomic_read include/linux/atomic/atomic-instrumented.h:32 [inline]
> > BUG: KASAN: slab-use-after-free in __brelse fs/buffer.c:1235 [inline]
> > BUG: KASAN: slab-use-after-free in brelse include/linux/buffer_head.h:325 [inline]
> > BUG: KASAN: slab-use-after-free in __invalidate_bh_lrus fs/buffer.c:1508 [inline]
> > BUG: KASAN: slab-use-after-free in invalidate_bh_lru+0xa8/0x1b0 fs/buffer.c:1521
> > Read of size 4 at addr ffff88801c989a58 by task udevd/5114
> >
> A worse case than the report looks like
>
> buff head is freed and reused with bh lru left intact

How is that supposed to happen? We take a reference to the BH when
it goes on the LRU list, so it shouldn't be freed.

We free BHs in free_buffer_head(). That's called from two places; the
failure path of folio_alloc_buffers() and try_to_free_buffers().
try_to_free_buffers() calls drop_buffers() first, which checks
buffer_busy(), which will fail if bh->b_count is non-zero.

You can't even get to the BHs in folio_alloc_buffers() because they're
not attached to anything until the function returns.

So I have no idea what the problem is here. And also I don't think it's
my responsibility to track it down. I'm not the buffer_head maintainer.

> flush bh lru
> __brelse()
> if (atomic_read(&bh->b_count)) {
> put_bh(bh);
> return;
> }
>
> and bh->b_count gets corrupted by put_bh().
>
> > CPU: 0 UID: 0 PID: 5114 Comm: udevd Not tainted 6.11.0-rc6-syzkaller-00308-gb31c44928842 #0
> > Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2~bpo12+1 04/01/2014
> > Call Trace:
> > <TASK>
> > __dump_stack lib/dump_stack.c:93 [inline]
> > dump_stack_lvl+0x241/0x360 lib/dump_stack.c:119
> > print_address_description mm/kasan/report.c:377 [inline]
> > print_report+0x169/0x550 mm/kasan/report.c:488
> > kasan_report+0x143/0x180 mm/kasan/report.c:601
> > kasan_check_range+0x282/0x290 mm/kasan/generic.c:189
> > instrument_atomic_read include/linux/instrumented.h:68 [inline]
> > atomic_read include/linux/atomic/atomic-instrumented.h:32 [inline]
> > __brelse fs/buffer.c:1235 [inline]
> > brelse include/linux/buffer_head.h:325 [inline]
> > __invalidate_bh_lrus fs/buffer.c:1508 [inline]
> > invalidate_bh_lru+0xa8/0x1b0 fs/buffer.c:1521
> > csd_do_func kernel/smp.c:134 [inline]
> > smp_call_function_many_cond+0x15d7/0x29d0 kernel/smp.c:847
> > on_each_cpu_cond_mask+0x3f/0x80 kernel/smp.c:1023
> > kill_bdev block/bdev.c:89 [inline]
> > blkdev_flush_mapping+0xfe/0x250 block/bdev.c:664
> > blkdev_put_whole block/bdev.c:671 [inline]
> > bdev_release+0x466/0x700 block/bdev.c:1096
> > blkdev_release+0x15/0x20 block/fops.c:638
> > __fput+0x24a/0x8a0 fs/file_table.c:422
> > __do_sys_close fs/open.c:1566 [inline]
> > __se_sys_close fs/open.c:1551 [inline]
> > __x64_sys_close+0x7f/0x110 fs/open.c:1551
> > do_syscall_x64 arch/x86/entry/common.c:52 [inline]
> > do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83
> > entry_SYSCALL_64_after_hwframe+0x77/0x7f
> > RIP: 0033:0x7f8226f170a8
> > Code: 48 8b 05 83 9d 0d 00 64 c7 00 16 00 00 00 83 c8 ff 48 83 c4 20 5b c3 64 8b 04 25 18 00 00 00 85 c0 75 20 b8 03 00 00 00 0f 05 <48> 3d 00 f0 ff ff 76 5b 48 8b 15 51 9d 0d 00 f7 d8 64 89 02 48 83
> > RSP: 002b:00007ffe0319be58 EFLAGS: 00000246 ORIG_RAX: 0000000000000003
> > RAX: ffffffffffffffda RBX: 00007f8226dee0e0 RCX: 00007f8226f170a8
> > RDX: 000056579a6766d5 RSI: 00007ffe0319b658 RDI: 0000000000000008
> > RBP: 00005652ff487f60 R08: 0000000000000006 R09: b595b5b875e4bbae
> > R10: 000000000000010f R11: 0000000000000246 R12: 0000000000000002
> > R13: 00005652ff478840 R14: 0000000000000008 R15: 00005652ff466910
> > </TASK>
> >
> > Allocated by task 5112:
> > kasan_save_stack mm/kasan/common.c:47 [inline]
> > kasan_save_track+0x3f/0x80 mm/kasan/common.c:68
> > unpoison_slab_object mm/kasan/common.c:312 [inline]
> > __kasan_slab_alloc+0x66/0x80 mm/kasan/common.c:338
> > kasan_slab_alloc include/linux/kasan.h:201 [inline]
> > slab_post_alloc_hook mm/slub.c:3992 [inline]
> > slab_alloc_node mm/slub.c:4041 [inline]
> > kmem_cache_alloc_noprof+0x135/0x2a0 mm/slub.c:4048
> > alloc_buffer_head+0x2a/0x290 fs/buffer.c:3025
> > folio_alloc_buffers+0x241/0x5b0 fs/buffer.c:929
> > grow_dev_folio fs/buffer.c:1072 [inline]
> > grow_buffers fs/buffer.c:1113 [inline]
> > __getblk_slow fs/buffer.c:1139 [inline]
> > bdev_getblk+0x2a6/0x550 fs/buffer.c:1441
> > __bread_gfp+0x86/0x400 fs/buffer.c:1495
> > sb_bread include/linux/buffer_head.h:347 [inline]
> > sysv_fill_super+0x231/0x710 fs/sysv/super.c:379
> > mount_bdev+0x20a/0x2d0 fs/super.c:1679
> > legacy_get_tree+0xee/0x190 fs/fs_context.c:662
> > vfs_get_tree+0x90/0x2b0 fs/super.c:1800
> > do_new_mount+0x2be/0xb40 fs/namespace.c:3472
> > do_mount fs/namespace.c:3812 [inline]
> > __do_sys_mount fs/namespace.c:4020 [inline]
> > __se_sys_mount+0x2d6/0x3c0 fs/namespace.c:3997
> > do_syscall_x64 arch/x86/entry/common.c:52 [inline]
> > do_syscall_64+0xf3/0x230 arch/x86/entry/common.c:83
> > entry_SYSCALL_64_after_hwframe+0x77/0x7f
> >
> > Freed by task 79:
> > kasan_save_stack mm/kasan/common.c:47 [inline]
> > kasan_save_track+0x3f/0x80 mm/kasan/common.c:68
> > kasan_save_free_info+0x40/0x50 mm/kasan/generic.c:579
> > poison_slab_object+0xe0/0x150 mm/kasan/common.c:240
> > __kasan_slab_free+0x37/0x60 mm/kasan/common.c:256
> > kasan_slab_free include/linux/kasan.h:184 [inline]
> > slab_free_hook mm/slub.c:2256 [inline]
> > slab_free mm/slub.c:4477 [inline]
> > kmem_cache_free+0x145/0x350 mm/slub.c:4552
> > free_buffer_head+0x54/0x240 fs/buffer.c:3041
> > try_to_free_buffers+0x311/0x5f0 fs/buffer.c:2982
> > shrink_folio_list+0x26c2/0x8c90 mm/vmscan.c:1413
> > evict_folios+0x50f7/0x7780 mm/vmscan.c:4560
> > try_to_shrink_lruvec+0x9ab/0xbb0 mm/vmscan.c:4755
> > shrink_one+0x3b9/0x850 mm/vmscan.c:4793
> > shrink_many mm/vmscan.c:4856 [inline]
> > lru_gen_shrink_node mm/vmscan.c:4934 [inline]
> > shrink_node+0x3799/0x3de0 mm/vmscan.c:5914
> > kswapd_shrink_node mm/vmscan.c:6742 [inline]
> > balance_pgdat mm/vmscan.c:6934 [inline]
> > kswapd+0x1cbc/0x3720 mm/vmscan.c:7203
> > kthread+0x2f0/0x390 kernel/kthread.c:389
> > ret_from_fork+0x4b/0x80 arch/x86/kernel/process.c:147
> > ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:244
> >
> > The buggy address belongs to the object at ffff88801c9899f8
> > which belongs to the cache buffer_head of size 168
> > The buggy address is located 96 bytes inside of
> > freed 168-byte region [ffff88801c9899f8, ffff88801c989aa0)
> >
> > The buggy address belongs to the physical page:
> > page: refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1c989
> > flags: 0xfff00000000000(node=0|zone=1|lastcpupid=0x7ff)
> > page_type: 0xfdffffff(slab)
> > raw: 00fff00000000000 ffff88801b763c80 ffffea0000725d40 0000000000000006
> > raw: 0000000000000000 0000000080110011 00000001fdffffff 0000000000000000
> > page dumped because: kasan: bad access detected
> > page_owner tracks the page as allocated
> > page last allocated via order 0, migratetype Reclaimable, gfp_mask 0x152c50(GFP_NOFS|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_HARDWALL|__GFP_RECLAIMABLE), pid 1, tgid 1 (init), ts 28576033639, free_ts 0
> > set_page_owner include/linux/page_owner.h:32 [inline]
> > post_alloc_hook+0x1f3/0x230 mm/page_alloc.c:1500
> > prep_new_page mm/page_alloc.c:1508 [inline]
> > get_page_from_freelist+0x2e4c/0x2f10 mm/page_alloc.c:3446
> > __alloc_pages_noprof+0x256/0x6c0 mm/page_alloc.c:4702
> > __alloc_pages_node_noprof include/linux/gfp.h:269 [inline]
> > alloc_pages_node_noprof include/linux/gfp.h:296 [inline]
> > alloc_slab_page+0x5f/0x120 mm/slub.c:2325
> > allocate_slab+0x5a/0x2f0 mm/slub.c:2488
> > new_slab mm/slub.c:2541 [inline]
> > ___slab_alloc+0xcd1/0x14b0 mm/slub.c:3727
> > __slab_alloc+0x58/0xa0 mm/slub.c:3817
> > __slab_alloc_node mm/slub.c:3870 [inline]
> > slab_alloc_node mm/slub.c:4029 [inline]
> > kmem_cache_alloc_noprof+0x1c1/0x2a0 mm/slub.c:4048
> > alloc_buffer_head+0x2a/0x290 fs/buffer.c:3025
> > folio_alloc_buffers+0x241/0x5b0 fs/buffer.c:929
> > grow_dev_folio fs/buffer.c:1072 [inline]
> > grow_buffers fs/buffer.c:1113 [inline]
> > __getblk_slow fs/buffer.c:1139 [inline]
> > bdev_getblk+0x2a6/0x550 fs/buffer.c:1441
> > __getblk include/linux/buffer_head.h:381 [inline]
> > sb_getblk include/linux/buffer_head.h:387 [inline]
> > ext4_read_inode_bitmap+0x24c/0x12f0 fs/ext4/ialloc.c:145
> > __ext4_new_inode+0x106f/0x4260 fs/ext4/ialloc.c:1054
> > ext4_create+0x279/0x550 fs/ext4/namei.c:2832
> > lookup_open fs/namei.c:3578 [inline]
> > open_last_lookups fs/namei.c:3647 [inline]
> > path_openat+0x1a9a/0x3470 fs/namei.c:3883
> > do_filp_open+0x235/0x490 fs/namei.c:3913
> > page_owner free stack trace missing
> >
> > Memory state around the buggy address:
> > ffff88801c989900: fc fc fa fb fb fb fb fb fb fb fb fb fb fb fb fb
> > ffff88801c989980: fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc fa
> > >ffff88801c989a00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> > ^
> > ffff88801c989a80: fb fb fb fb fc fc fc fc fc fc fc fc 00 00 00 00
> > ffff88801c989b00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> > ==================================================================

Next message: Christophe Leroy: "Re: [PATCH] crypto: Removing CRYPTO_AES_GCM_P10."
Previous message: Jeff Layton: "[PATCH v7 11/11] tmpfs: add support for multigrain timestamps"
In reply to: Hillf Danton: "Re: [syzbot] [fs?] KASAN: slab-use-after-free Read in invalidate_bh_lru"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]