Race condition in ext4 (was Re: 4.11-rc1 acpi stomping ext4 slabs)

From: Nikolay Borisov
Date: Tue Mar 07 2017 - 10:05:44 EST



On 7.03.2017 11:38, Nikolay Borisov wrote:
>
>
> On 7.03.2017 00:35, Rafael J. Wysocki wrote:
>> On Mon, Mar 6, 2017 at 9:31 PM, Nikolay Borisov
>> <n.borisov.lkml@xxxxxxxxx> wrote:
>>> Hello,
>>>
>>> Booting 4.11-rc1 with kasan enabled and "slub_debug=F" produces the following errors:
>>>
>>> [ 7.070797] ==================================================================
>>> [ 7.071724] BUG: KASAN: slab-out-of-bounds in filldir+0xc3/0x160 at addr ffff88006bc2b0ae
>>> [ 7.071724] Read of size 20 by task systemd/1
>>> [ 7.071724] CPU: 1 PID: 1 Comm: systemd Not tainted 4.11.0-rc1-nbor #150
>>> [ 7.071724] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
>>> [ 7.071724] Call Trace:
>>> [ 7.071724] dump_stack+0x85/0xc9
>>> [ 7.071724] kasan_object_err+0x2c/0x90
>>> [ 7.071724] kasan_report+0x285/0x510
>>> [ 7.071724] check_memory_region+0x137/0x160
>>> [ 7.071724] kasan_check_read+0x11/0x20
>>> [ 7.071724] filldir+0xc3/0x160
>>> [ 7.071724] call_filldir+0x88/0x140
>>> [ 7.071724] ext4_readdir+0x757/0x920
>>> [ 7.071724] ? iterate_dir+0x49/0x190
>>> [ 7.071724] iterate_dir+0x7d/0x190
>>> [ 7.071724] ? entry_SYSCALL_64_fastpath+0x5/0xc6
>>> [ 7.071724] SyS_getdents+0xac/0x170
>>> [ 7.071724] ? filldir64+0x170/0x170
>>> [ 7.071724] entry_SYSCALL_64_fastpath+0x23/0xc6
>>> [ 7.071724] RIP: 0033:0x7fa37ca2dd3b
>>> [ 7.071724] RSP: 002b:00007ffc63daf400 EFLAGS: 00000206 ORIG_RAX: 000000000000004e
>>> [ 7.071724] RAX: ffffffffffffffda RBX: 0000000000000046 RCX: 00007fa37ca2dd3b
>>> [ 7.071724] RDX: 0000000000008000 RSI: 0000560b369e4a10 RDI: 0000000000000004
>>> [ 7.071724] RBP: 00007fa37cd29b20 R08: 00007fa37cd29bd8 R09: 0000000000000000
>>> [ 7.071724] R10: 000000000000008f R11: 0000000000000206 R12: 0000000000008041
>>> [ 7.071724] R13: 00007fa37cd29b78 R14: 000000000000270f R15: 00007fa37cd29b78
>>> [ 7.071724] Object at ffff88006bc2b080, in cache kmalloc-96 size: 96
>>> [ 7.071724] Allocated:
>>> [ 7.071724] PID = 1
>>> [ 7.071724] save_stack_trace+0x1b/0x20
>>> [ 7.071724] kasan_kmalloc.part.4+0x64/0xf0
>>> [ 7.071724] kasan_kmalloc+0x85/0xb0
>>> [ 7.071724] __kmalloc+0x12b/0x320
>>> [ 7.071724] ext4_htree_store_dirent+0x3e/0x120
>>> [ 7.071724] htree_dirblock_to_tree+0xb9/0x1a0
>>> [ 7.071724] ext4_htree_fill_tree+0xa3/0x310
>>> [ 7.071724] ext4_readdir+0x6a9/0x920
>>> [ 7.071724] iterate_dir+0x7d/0x190
>>> [ 7.071724] SyS_getdents+0xac/0x170
>>> [ 7.071724] entry_SYSCALL_64_fastpath+0x23/0xc6
>>> [ 7.071724] Freed:
>>> [ 7.071724] PID = 1
>>> [ 7.071724] save_stack_trace+0x1b/0x20
>>> [ 7.071724] kasan_slab_free+0xbe/0x190
>>> [ 7.071724] kfree+0xff/0x2f0
>>> [ 7.071724] acpi_ut_evaluate_object+0x18e/0x19d
>>> [ 7.071724] acpi_ut_execute_STA+0x26/0x53
>>> [ 7.071724] acpi_ns_get_device_callback+0x73/0x163
>>> [ 7.071724] acpi_ns_walk_namespace+0xc0/0x17a
>>> [ 7.071724] acpi_get_devices+0x66/0x7d
>>> [ 7.071724] pnpacpi_init+0x52/0x74
>>> [ 7.071724] do_one_initcall+0x51/0x1b0
>>> [ 7.071724] kernel_init_freeable+0x20a/0x2a1
>>> [ 7.071724] kernel_init+0xe/0x100
>>> [ 7.071724] ret_from_fork+0x31/0x40
>>> [ 7.071724] Memory state around the buggy address:
>>> [ 7.071724] ffff88006bc2af80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>>> [ 7.071724] ffff88006bc2b000: fb fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
>>> [ 7.071724] >ffff88006bc2b080: 00 00 00 00 00 00 00 00 05 fc fc fc fc fc fc fc
>>> [ 7.071724] ^
>>> [ 7.071724] ffff88006bc2b100: 00 00 00 00 00 00 00 00 00 04 fc fc fc fc fc fc
>>> [ 7.071724] ffff88006bc2b180: 00 00 00 00 00 00 00 00 00 00 fc fc fc fc fc fc
>>>
>>> Not killing the VM instantly produces a continuous stream of kasan errors. Most of them
>>> are identical to the one above, however there was one which was different:
>>>
>>> [ 5.846193] ==================================================================
>>> [ 5.846787] BUG: KASAN: slab-out-of-bounds in filldir+0xc3/0x160 at addr ffff88006c783eae
>>> [ 5.847177] Read of size 22 by task systemd/1
>>> [ 5.847177] CPU: 3 PID: 1 Comm: systemd Tainted: G B 4.11.0-rc1-nbor #150
>>> [ 5.847177] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
>>> [ 5.847177] Call Trace:
>>> [ 5.847177] dump_stack+0x85/0xc9
>>> [ 5.847177] kasan_object_err+0x2c/0x90
>>> [ 5.847177] kasan_report+0x285/0x510
>>> [ 5.847177] check_memory_region+0x137/0x160
>>> [ 5.847177] kasan_check_read+0x11/0x20
>>> [ 5.847177] filldir+0xc3/0x160
>>> [ 5.847177] call_filldir+0x88/0x140
>>> [ 5.847177] ext4_readdir+0x757/0x920
>>> [ 5.847177] ? iterate_dir+0x49/0x190
>>> [ 5.847177] iterate_dir+0x7d/0x190
>>> [ 5.847177] ? entry_SYSCALL_64_fastpath+0x5/0xc6
>>> [ 5.847177] SyS_getdents+0xac/0x170
>>> [ 5.847177] ? filldir64+0x170/0x170
>>> [ 5.847177] entry_SYSCALL_64_fastpath+0x23/0xc6
>>> [ 5.847177] RIP: 0033:0x7f9dbd4e1d3b
>>> [ 5.847177] RSP: 002b:00007ffee6b51a60 EFLAGS: 00000206 ORIG_RAX: 000000000000004e
>>> [ 5.847177] RAX: ffffffffffffffda RBX: 0000000000000046 RCX: 00007f9dbd4e1d3b
>>> [ 5.847177] RDX: 0000000000008000 RSI: 000055c046802a10 RDI: 0000000000000004
>>> [ 5.847177] RBP: 00007f9dbd7ddb20 R08: 00007f9dbd7ddbd8 R09: 0000000000000000
>>> [ 5.847177] R10: 000000000000008f R11: 0000000000000206 R12: 0000000000008041
>>> [ 5.847177] R13: 00007f9dbd7ddb78 R14: 000000000000270f R15: 00007f9dbd7ddb78
>>> [ 5.847177] Object at ffff88006c783e80, in cache kmalloc-96 size: 96
>>> [ 5.847177] Allocated:
>>> [ 5.847177] PID = 1
>>> [ 5.847177] save_stack_trace+0x1b/0x20
>>> [ 5.847177] kasan_kmalloc.part.4+0x64/0xf0
>>> [ 5.847177] kasan_kmalloc+0x85/0xb0
>>> [ 5.847177] __kmalloc+0x12b/0x320
>>> [ 5.847177] ext4_htree_store_dirent+0x3e/0x120
>>> [ 5.847177] htree_dirblock_to_tree+0xb9/0x1a0
>>> [ 5.847177] ext4_htree_fill_tree+0xa3/0x310
>>> [ 5.847177] ext4_readdir+0x6a9/0x920
>>> [ 5.847177] iterate_dir+0x7d/0x190
>>> [ 5.847177] SyS_getdents+0xac/0x170
>>> [ 5.847177] entry_SYSCALL_64_fastpath+0x23/0xc6
>>> [ 5.847177] Freed:
>>> [ 5.847177] PID = 1
>>> [ 5.847177] save_stack_trace+0x1b/0x20
>>> [ 5.847177] kasan_slab_free+0xbe/0x190
>>> [ 5.847177] kfree+0xff/0x2f0
>>> [ 5.847177] krealloc+0xac/0xc0
>>> [ 5.847177] create_trace_option_files+0x127/0x270
>>> [ 5.847177] __update_tracer_options+0x2c/0x40
>>> [ 5.847177] tracer_init_tracefs+0x1a4/0x1b7
>>> [ 5.847177] do_one_initcall+0x51/0x1b0
>>> [ 5.847177] kernel_init_freeable+0x20a/0x2a1
>>> [ 5.847177] kernel_init+0xe/0x100
>>> [ 5.847177] ret_from_fork+0x31/0x40
>>> [ 5.847177] Memory state around the buggy address:
>>>
>>> So the free path is different.
>>>
>>> On a different boot with slab_debug options omitted e.g. no debugging enabled for SLUB
>>> I got:
>>>
>>> [ 5.586620] ==================================================================
>>> [ 5.587445] BUG: KASAN: slab-out-of-bounds in filldir+0xc3/0x160 at addr ffff880000141aae
>>> [ 5.587584] Read of size 20 by task systemd/1
>>> [ 5.587584] CPU: 0 PID: 1 Comm: systemd Not tainted 4.11.0-rc1-nbor #148
>>> [ 5.587584] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
>>> [ 5.587584] Call Trace:
>>> [ 5.587584] dump_stack+0x85/0xc9
>>> [ 5.587584] kasan_object_err+0x2c/0x90
>>> [ 5.587584] kasan_report+0x285/0x510
>>> [ 5.587584] check_memory_region+0x137/0x160
>>> [ 5.587584] kasan_check_read+0x11/0x20
>>> [ 5.587584] filldir+0xc3/0x160
>>> [ 5.587584] call_filldir+0x88/0x140
>>> [ 5.587584] ext4_readdir+0x757/0x920
>>> [ 5.587584] ? iterate_dir+0x49/0x190
>>> [ 5.587584] iterate_dir+0x7d/0x190
>>> [ 5.587584] ? entry_SYSCALL_64_fastpath+0x5/0xc6
>>> [ 5.587584] SyS_getdents+0xac/0x170
>>> [ 5.587584] ? filldir64+0x170/0x170
>>> [ 5.587584] entry_SYSCALL_64_fastpath+0x23/0xc6
>>> [ 5.587584] RIP: 0033:0x7f71af785d3b
>>> [ 5.587584] RSP: 002b:00007ffeeda83390 EFLAGS: 00000206 ORIG_RAX: 000000000000004e
>>> [ 5.587584] RAX: ffffffffffffffda RBX: 0000000000000046 RCX: 00007f71af785d3b
>>> [ 5.587584] RDX: 0000000000008000 RSI: 0000561e6483ba10 RDI: 0000000000000004
>>> [ 5.587584] RBP: 00007f71afa81b20 R08: 00007f71afa81bd8 R09: 0000000000000000
>>> [ 5.587584] R10: 000000000000008f R11: 0000000000000206 R12: 0000000000008041
>>> [ 5.587584] R13: 00007f71afa81b78 R14: 000000000000270f R15: 00007f71afa81b78
>>> [ 5.587584] Object at ffff880000141a80, in cache kmalloc-96 size: 96
>>> [ 5.587584] Allocated:
>>> [ 5.587584] PID = 1
>>> [ 5.587584] save_stack_trace+0x1b/0x20
>>> [ 5.587584] kasan_kmalloc.part.4+0x64/0xf0
>>> [ 5.587584] kasan_kmalloc+0x85/0xb0
>>> [ 5.587584] __kmalloc+0x12b/0x320
>>> [ 5.587584] ext4_htree_store_dirent+0x3e/0x120
>>> [ 5.587584] htree_dirblock_to_tree+0xb9/0x1a0
>>> [ 5.587584] ext4_htree_fill_tree+0xa3/0x310
>>> [ 5.587584] ext4_readdir+0x6a9/0x920
>>> [ 5.587584] iterate_dir+0x7d/0x190
>>> [ 5.587584] SyS_getdents+0xac/0x170
>>> [ 5.587584] entry_SYSCALL_64_fastpath+0x23/0xc6
>>> [ 5.587584] Freed:
>>> [ 5.587584] PID = 1
>>> [ 5.587584] save_stack_trace+0x1b/0x20
>>> [ 5.587584] kasan_slab_free+0xbe/0x190
>>> [ 5.587584] kfree+0xff/0x2f0
>>> [ 5.587584] acpi_evaluate_object+0x26c/0x27e
>>> [ 5.587584] acpi_evaluate_integer+0x34/0x53
>>> [ 5.587584] acpi_get_node+0x2b/0x51
>>> [ 5.587584] pci_acpi_scan_root+0x2e/0x1d0
>>> [ 5.587584] acpi_pci_root_add+0x264/0x34b
>>> [ 5.587584] acpi_bus_attach+0xb6/0x15c
>>> [ 5.587584] acpi_bus_attach+0x123/0x15c
>>> [ 5.587584] acpi_bus_attach+0x123/0x15c
>>> [ 5.587584] acpi_bus_scan+0x5b/0x6b
>>> [ 5.587584] acpi_scan_init+0xcd/0x211
>>> [ 5.587584] acpi_init+0x2e0/0x309
>>> [ 5.587584] do_one_initcall+0x51/0x1b0
>>> [ 5.587584] kernel_init_freeable+0x20a/0x2a1
>>> [ 5.587584] kernel_init+0xe/0x100
>>> [ 5.587584] ret_from_fork+0x31/0x40
>>> [ 5.587584] Memory state around the buggy address:
>>> [ 5.587584] ffff880000141980: 00 00 00 00 00 00 00 00 00 fc fc fc fc fc fc fc
>>> [ 5.587584] ffff880000141a00: 00 00 00 00 00 00 00 00 00 03 fc fc fc fc fc fc
>>> [ 5.587584] >ffff880000141a80: 00 00 00 00 00 00 00 00 05 fc fc fc fc fc fc fc
>>> [ 5.587584] ^
>>> [ 5.587584] ffff880000141b00: fb fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc
>>> [ 5.587584] ffff880000141b80: 00 00 00 00 00 00 00 00 00 00 00 fc fc fc fc fc
>>>
>>> I'm not sure if this is an ext4 or ACPI problem.
>>
>> If this is a new bug, you can look for the first bad commit using git-bisect.
>
> So apparently this is an ext4 bug, since I managed to bisected it all the way to 1771c6e1a567ea0
> ("x86/kasan: instrument user memory access API"). So the first bad kernel is 4.7 and ACPI is not to blame .
>
> [EXT4 people might want to start looking from here]
>
> There the splat looks like :
>
> [ 17.003256] ==================================================================
> [ 17.004185] BUG: KASAN: slab-out-of-bounds in filldir+0xc8/0x170 at addr ffff88006a22560e
> [ 17.005177] Read of size 20 by task systemd/1
> [ 17.005708] =============================================================================
> [ 17.006688] BUG kmalloc-96 (Not tainted): kasan: bad access detected
> [ 17.007464] -----------------------------------------------------------------------------
> [ 17.007464]
> [ 17.008584] Disabling lock debugging due to kernel taint
> [ 17.009202] INFO: Allocated in ext4_htree_store_dirent+0x3e/0x120 age=0 cpu=2 pid=1
> [ 17.010080] ___slab_alloc+0x636/0x6a0
> [ 17.010514] __slab_alloc+0x4f/0x86
> [ 17.010927] __kmalloc+0x27a/0x340
> [ 17.011318] ext4_htree_store_dirent+0x3e/0x120
> [ 17.011796] htree_dirblock_to_tree+0x16a/0x190
> [ 17.012270] ext4_htree_fill_tree+0xaa/0x310
> [ 17.012735] ext4_readdir+0x698/0x950
> [ 17.013117] iterate_dir+0x7d/0x190
> [ 17.013485] SyS_getdents+0x91/0x120
> [ 17.013873] entry_SYSCALL_64_fastpath+0x23/0xc1
> [ 17.014360] INFO: Freed in ext4_ext_map_blocks+0x7f9/0x23e0 age=1 cpu=2 pid=1
> [ 17.015110] __slab_free+0x31b/0x440
> [ 17.015489] kfree+0x27f/0x2d0
> [ 17.015820] ext4_ext_map_blocks+0x7f9/0x23e0
> [ 17.016283] ext4_map_blocks+0x3b4/0x5b0
> [ 17.016699] ext4_getblk+0x54/0x1a0
> [ 17.017066] ext4_bread+0x13/0x90
> [ 17.017412] __ext4_read_dirblock+0x3f/0x380
> [ 17.017861] htree_dirblock_to_tree+0x48/0x190
> [ 17.018324] ext4_htree_fill_tree+0xaa/0x310
> [ 17.018776] ext4_readdir+0x698/0x950
> [ 17.019176] iterate_dir+0x7d/0x190
> [ 17.019551] SyS_getdents+0x91/0x120
> [ 17.019932] entry_SYSCALL_64_fastpath+0x23/0xc1
> [ 17.020431] INFO: Slab 0xffffea0001a88900 objects=20 used=17 fp=0xffff88006a224e10 flags=0x4080
> [ 17.021348] INFO: Object 0xffff88006a2255e0 @offset=5600 fp=0x45b282a2484c60d4
> [ 17.021348]
> [ 17.022264] Bytes b4 ffff88006a2255d0: 02 00 00 00 01 00 00 00 c9 ac fb ff 00 00 00 00 ................
> [ 17.023272] Object ffff88006a2255e0: d4 60 4c 48 a2 82 b2 45 18 8a 82 6a 00 88 ff ff .`LH...E...j....
> [ 17.024252] Object ffff88006a2255f0: 38 51 22 6a 00 88 ff ff 88 8b 82 6a 00 88 ff ff 8Q"j.......j....
> [ 17.025166] Object ffff88006a225600: 00 00 00 00 00 00 00 00 28 03 08 00 14 01 66 62 ........(.....fb
> [ 17.026129] Object ffff88006a225610: 64 65 76 2d 62 6c 61 63 6b 6c 69 73 74 2e 63 6f dev-blacklist.co
> [ 17.027156] Object ffff88006a225620: 6e 66 00 00 00 00 00 00 00 00 00 00 00 00 00 00 nf..............
> [ 17.028202] Object ffff88006a225630: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
> [ 17.029186] CPU: 2 PID: 1 Comm: systemd Tainted: G B 4.7.0-nbor #171
> [ 17.029972] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Ubuntu-1.8.2-1ubuntu1 04/01/2014
> [ 17.030959] 0000000000000000 ffff88006cd97c58 ffffffff8146bd4c ffff8800000946c0
> [ 17.031776] ffff88006a2255e0 ffff88006cd97c88 ffffffff81198d96 ffff8800000946c0
> [ 17.032597] ffffea0001a88900 ffff88006a2255e0 0000000000000000 ffff88006cd97cb0
> [ 17.033425] Call Trace:
> [ 17.033700] [<ffffffff8146bd4c>] dump_stack+0x85/0xc9
> [ 17.034246] [<ffffffff81198d96>] print_trailer+0x116/0x190
> [ 17.034840] [<ffffffff811992c1>] object_err+0x41/0x50
> [ 17.035390] [<ffffffff811a0a42>] kasan_report+0x282/0x530
> [ 17.035966] [<ffffffff8119ffa7>] check_memory_region+0x137/0x160
> [ 17.036594] [<ffffffff811a0041>] kasan_check_read+0x11/0x20
> [ 17.037181] [<ffffffff811ccc08>] filldir+0xc8/0x170 <--- if (copy_to_user(dirent->d_name, name, namlen))
> [ 17.037687] [<ffffffff8124af38>] call_filldir+0x88/0x140
> [ 17.038244] [<ffffffff8124b934>] ext4_readdir+0x714/0x950
> [ 17.038813] [<ffffffff811cccf9>] ? iterate_dir+0x49/0x190
> [ 17.039399] [<ffffffff811ccd2d>] iterate_dir+0x7d/0x190
> [ 17.039909] [<ffffffff811ccf71>] SyS_getdents+0x91/0x120
> [ 17.040428] [<ffffffff811ccb40>] ? filldir64+0x180/0x180
> [ 17.040977] [<ffffffff816d7d80>] entry_SYSCALL_64_fastpath+0x23/0xc1
> [ 17.041652] Memory state around the buggy address:
> [ 17.042136] ffff88006a225500: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
> [ 17.042924] ffff88006a225580: fc fc fc fc fc fc fc fc fc fc fc fc 00 00 00 00
> [ 17.043652] >ffff88006a225600: 00 00 00 00 05 fc fc fc fc fc fc fc fc fc fc fc
> [ 17.044382] ^
> [ 17.044814] ffff88006a225680: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
> [ 17.045559] ffff88006a225700: fc fc fc fc fc fc fc fc fc fc fc fc fc fc 00 00
> [ 17.046308] ==================================================================
>
>
> So the buffer containing the dentry name as received from ext4_htree_fill is actually
> freed.
>

So this is some sort of a race condition. The problem disappeared as
soon as I added the following line:
pr_info("%s:freeing %p\n", __func__, path);
right after the ext4_ext_drop_refs(path); call in out2: label.