Date:   Wed, 2 Nov 2016 23:14:52 -0600
From:   Dmitry Vyukov <dvyukov@...gle.com>
To:     Alexei Starovoitov <ast@...nel.org>,
        netdev <netdev@...r.kernel.org>,
        Daniel Borkmann <daniel@...earbox.net>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        syzkaller <syzkaller@...glegroups.com>
Subject: bpf: kernel BUG in htab_elem_free

Here we go.

The following program triggers kernel BUG in htab_elem_free.
On commit 0c183d92b20b5c84ca655b45ef57b3318b83eb9e (Oct 31).
Run as "while true; do ./a.out; done".

------------[ cut here ]------------
kernel BUG at mm/slub.c:3866!
invalid opcode: 0000 [#1] SMP KASAN
Modules linked in:
CPU: 1 PID: 1542 Comm: kworker/1:2 Not tainted 4.9.0-rc3+ #20
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
Workqueue: events bpf_map_free_deferred
task: ffff88003b9c0040 task.stack: ffff88003cb70000
RIP: 0010:[<ffffffff814c9f00>]  [<ffffffff814c9f00>] kfree+0x140/0x1a0
RSP: 0018:ffff88003cb77c50  EFLAGS: 00010246
RAX: ffffea0000fb0aa0 RBX: ffff88003ec2a1a8 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 1ffff10007b50401 RDI: ffff88003ec2a1a8
RBP: ffff88003cb77c70 R08: 0000000000021800 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffffea0000fb0a80
R13: ffffffff81392bcb R14: 0000000000000000 R15: ffff88003ec2a1a8
FS:  0000000000000000(0000) GS:ffff88003ed00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000205d7000 CR3: 0000000037d29000 CR4: 00000000000006e0
Stack:
 dffffc0000000000 ffff88003da82008 ffff88003b75bb88 0000000000000000
 ffff88003cb77ce0 ffffffff81392bcb ffffffff81acf4f8 ffff88003b75bc04
 ffff88003b75bbe0 ffffed00076eb772 ffff88003b75bb90 000000003cb77ce0
Call Trace:
 [<     inline     >] htab_elem_free kernel/bpf/hashtab.c:388
 [<     inline     >] delete_all_elements kernel/bpf/hashtab.c:690
 [<ffffffff81392bcb>] htab_map_free+0x30b/0x470 kernel/bpf/hashtab.c:711
 [<ffffffff8137e9dc>] bpf_map_free_deferred+0xac/0xd0 kernel/bpf/syscall.c:97
 [<ffffffff8114f937>] process_one_work+0x8a7/0x1300 kernel/workqueue.c:2096
 [<ffffffff8115047d>] worker_thread+0xed/0x14e0 kernel/workqueue.c:2230
 [<ffffffff811619dc>] kthread+0x1ec/0x260 kernel/kthread.c:209
 [<ffffffff82fa78c5>] ret_from_fork+0x25/0x30 arch/x86/entry/entry_64.S:433
Code: 83 c4 18 48 89 da 4c 89 ee ff d0 49 8b 04 24 48 85 c0 75 e6 e9
e9 fe ff ff 49 8b 04 24 f6 c4 40 75 0b 49 8b 44 24 20 a8 01 75 02 <0f>
0b 48 89 df e8 56 35 00 00 49 8b 04 24 31 f6 f6 c4 40 74 05
RIP  [<     inline     >] PageCompound ./include/linux/page-flags.h:157
RIP  [<ffffffff814c9f00>] kfree+0x140/0x1a0 mm/slub.c:3866
 RSP <ffff88003cb77c50>
---[ end trace 1dc58d6aeb2596aa ]---
==================================================================
BUG: KASAN: stack-out-of-bounds in complete+0x68/0x70 at addr ffff88003cb77ed8
Read of size 4 by task kworker/1:2/1542
page:ffffea0000f2ddc0 count:0 mapcount:0 mapping:          (null) index:0x0
flags: 0x100000000000000()
page dumped because: kasan: bad access detected
CPU: 1 PID: 1542 Comm: kworker/1:2 Tainted: G      D         4.9.0-rc3+ #20
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
 ffff88003cb77ce0 ffffffff81acf609 ffffed000796efdb ffffed000796efdb
 0000000000000004 0000000000000000 ffff88003cb77d60 ffffffff814cdbfb
 ffff88003c8d97c8 dffffc0000000000 ffffffff811dd038 0000000000000097
Call Trace:
 [<     inline     >] __dump_stack lib/dump_stack.c:15
 [<ffffffff81acf609>] dump_stack+0x83/0xba lib/dump_stack.c:51
 [<     inline     >] kasan_report_error mm/kasan/report.c:204
 [<ffffffff814cdbfb>] kasan_report+0x4cb/0x500 mm/kasan/report.c:303
 [<ffffffff814cdc84>] __asan_report_load4_noabort+0x14/0x20
mm/kasan/report.c:328
 [<ffffffff811dd038>] complete+0x68/0x70 kernel/sched/completion.c:34
 [<     inline     >] complete_vfork_done kernel/fork.c:1030
 [<ffffffff810fa8c2>] mm_release+0x222/0x3f0 kernel/fork.c:1114
 [<     inline     >] exit_mm kernel/exit.c:467
 [<ffffffff8110b501>] do_exit+0x3a1/0x2960 kernel/exit.c:815
 [<ffffffff82fa8b97>] rewind_stack_do_exit+0x17/0x20
arch/x86/entry/entry_64.S:1526
Memory state around the buggy address:
 ffff88003cb77d80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 ffff88003cb77e00: 00 00 00 00 00 00 00 00 f1 f1 f1 f1 00 f4 f4 f4
>ffff88003cb77e80: f2 f2 f2 f2 00 f4 f4 f4 f2 f2 f2 f2 00 00 f4 f4
                                                    ^
 ffff88003cb77f00: f3 f3 f3 f3 00 00 00 00 00 00 00 00 00 00 00 00
 ffff88003cb77f80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
==================================================================
BUG: unable to handle kernel
paging request at ffffffffffffffd8
IP: [<ffffffff81163c5d>] kthread_data+0x4d/0x70 kernel/kthread.c:137
PGD 360d067 [   48.581115] PUD 360f067
PMD 0 [   48.581840]
Oops: 0000 [#2] SMP KASAN
Modules linked in:
CPU: 1 PID: 1542 Comm: kworker/1:2 Tainted: G    B D         4.9.0-rc3+ #20
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
task: ffff88003b9c0040 task.stack: ffff88003cb70000
RIP: 0010:[<ffffffff81163c5d>]  [<ffffffff81163c5d>] kthread_data+0x4d/0x70
RSP: 0018:ffff88003cb77c78  EFLAGS: 00010046
RAX: dffffc0000000000 RBX: 0000000000000000 RCX: 0000000000000000
RDX: 1ffffffffffffffb RSI: ffff88003b9c00c0 RDI: ffffffffffffffd8
RBP: ffff88003cb77c80 R08: ffff88003ed20a48 R09: ffff88003ed20a40
R10: 0000000000000000 R11: 0000000000000000 R12: ffff88003ed20980
R13: ffff88003b9c0040 R14: ffff88003b9c0094 R15: 0000000000000040
FS:  0000000000000000(0000) GS:ffff88003ed00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000028 CR3: 000000000360c000 CR4: 00000000000006e0
Stack:
 ffff88003b9c0040 ffff88003cb77ca0 ffffffff81155e77 0000000000020980
 ffff88003ed20980 ffff88003cb77db8 ffffffff82f9b1a2 0000000000000000
 ffff88003ddd2670 00ff88003ddd2640 ffff88003ed211f8 1ffff1000796ef9e
Call Trace:
 [<ffffffff81155e77>] wq_worker_sleeping+0x17/0x210 kernel/workqueue.c:876
 [<ffffffff82f9b1a2>] __schedule+0xc62/0x1730 kernel/sched/core.c:3380
 [<ffffffff8118c781>] do_task_dead+0x81/0xa0 kernel/sched/core.c:3431
 [<ffffffff8110c7f8>] do_exit+0x1698/0x2960 kernel/exit.c:885
 [<ffffffff82fa8b97>] rewind_stack_do_exit+0x17/0x20
arch/x86/entry/entry_64.S:1526
Code: c1 ea 03 80 3c 02 00 75 29 48 8b 9b 30 05 00 00 48 b8 00 00 00
00 00 fc ff df 48 8d 7b d8 48 89 fa 48 c1 ea 03 80 3c 02 00 75 0e <48>
8b 43 d8 5b 5d c3 e8 27 a0 36 00 eb d0 e8 20 a0 36 00 eb eb
RIP  [<ffffffff81163c5d>] kthread_data+0x4d/0x70 kernel/kthread.c:137
 RSP <ffff88003cb77c78>
CR2: ffffffffffffffd8
---[ end trace 1dc58d6aeb2596ab ]---
Fixing recursive fault but reboot is needed!


// autogenerated by syzkaller (http://github.com/google/syzkaller)

#ifndef __NR_bpf
#define __NR_bpf 321
#endif

#include <fcntl.h>
#include <pthread.h>
#include <signal.h>
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/ioctl.h>
#include <sys/stat.h>
#include <sys/syscall.h>
#include <sys/types.h>
#include <unistd.h>

long r[14];
void* thr(void* arg)
{
  switch ((long)arg) {
  case 0:
    r[0] =
        syscall(__NR_mmap, 0x20000000ul, 0x16000ul, 0x3ul,
                        0x32ul, 0xfffffffffffffffful, 0x0ul, 0, 0, 0);
    break;
  case 1:
    (*(uint32_t*)0x20011000 = (uint32_t)0x1);
    (*(uint32_t*)0x20011004 = (uint32_t)0x8);
    (*(uint32_t*)0x20011008 = (uint32_t)0x3);
    (*(uint32_t*)0x2001100c = (uint32_t)0x1);
    (*(uint32_t*)0x20011010 = (uint32_t)0x1);
    r[6] = syscall(__NR_bpf, 0x0ul, 0x20011000ul, 0x14ul, 0, 0,
                           0, 0, 0, 0);
    break;
  case 2:
    (*(uint32_t*)0x20013000 = r[6]);
    (*(uint64_t*)0x20013008 = (uint64_t)0x20013fb3);
    (*(uint64_t*)0x20013010 = (uint64_t)0x20012ff1);
    (*(uint64_t*)0x20013018 = (uint64_t)0x0);
    (memcpy(
        (void*)0x20013fb3,
        "\x3e\x51\x32\xbe\xd5\x24\x20\xb2\x50\x7a\x4d\xb5\xec\xb3\x8f"
        "\x65\x7f\xac\x61\x9a\xf0\x29\x3f\x77\x07\x2f\x2f\x60\xe9\x78"
        "\xc5\x79\x45\x16\x67\xf6\x64\xb4\xd5\xb2\x11\x88\x5f\x4f\x32"
        "\xba\xa8\x80\x8f\x7a\xea\x01\x1d\xe4\x08\xa4\x65\x73\x07\x91"
        "\x48\xd5\xc3\xf2\xc4\x08\x29\x8f\x88\x95\xc3\xd5\xf6\x86\x0f"
        "\x42\xab\x05\xf7\xfa\x2b\x12\x78\xb3\x4d\x17\x8c\x27\x57\x8b"
        "\x79\xdc\x4f\x8a\x7a\xf5\x8c\x8a\xc2\x18\x03\xa0\xf9\x5f\x7a",
        105));
    (memcpy(
        (void*)0x20012ff1,
        "\xb1\xb0\x4b\x14\x9e\xfa\xbc\xb2\xaf\x4b\x4a\x02\xbc\x9b\xc5",
        15));
    r[13] = syscall(__NR_bpf, 0x2ul, 0x20013000ul, 0x20ul, 0, 0,
                            0, 0, 0, 0);
    break;
  }
  return 0;
}

int main()
{
  long i;
  pthread_t th[6];

  memset(r, -1, sizeof(r));
  srand(getpid());
  for (i = 0; i < 3; i++) {
    pthread_create(&th[i], 0, thr, (void*)i);
    usleep(10000);
  }
  for (i = 0; i < 3; i++) {
    pthread_create(&th[3 + i], 0, thr, (void*)i);
    if (rand() % 2)
      usleep(rand() % 10000);
  }
  usleep(100000);
  return 0;
}

Powered by blists - more mailing lists