ipv6/flowlabel: wait rcu grace period before put_pid()
authorEric Dumazet <edumazet@google.com>
Sat, 27 Apr 2019 23:49:06 +0000 (16:49 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 8 May 2019 05:19:04 +0000 (07:19 +0200)
[ Upstream commit 6c0afef5fb0c27758f4d52b2210c61b6bd8b4470 ]

syzbot was able to catch a use-after-free read in pid_nr_ns() [1]

ip6fl_seq_show() seems to use RCU protection, dereferencing fl->owner.pid
but fl_free() releases fl->owner.pid before rcu grace period is started.

[1]

BUG: KASAN: use-after-free in pid_nr_ns+0x128/0x140 kernel/pid.c:407
Read of size 4 at addr ffff888094012a04 by task syz-executor.0/18087

CPU: 0 PID: 18087 Comm: syz-executor.0 Not tainted 5.1.0-rc6+ #89
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0x172/0x1f0 lib/dump_stack.c:113
 print_address_description.cold+0x7c/0x20d mm/kasan/report.c:187
 kasan_report.cold+0x1b/0x40 mm/kasan/report.c:317
 __asan_report_load4_noabort+0x14/0x20 mm/kasan/generic_report.c:131
 pid_nr_ns+0x128/0x140 kernel/pid.c:407
 ip6fl_seq_show+0x2f8/0x4f0 net/ipv6/ip6_flowlabel.c:794
 seq_read+0xad3/0x1130 fs/seq_file.c:268
 proc_reg_read+0x1fe/0x2c0 fs/proc/inode.c:227
 do_loop_readv_writev fs/read_write.c:701 [inline]
 do_loop_readv_writev fs/read_write.c:688 [inline]
 do_iter_read+0x4a9/0x660 fs/read_write.c:922
 vfs_readv+0xf0/0x160 fs/read_write.c:984
 kernel_readv fs/splice.c:358 [inline]
 default_file_splice_read+0x475/0x890 fs/splice.c:413
 do_splice_to+0x12a/0x190 fs/splice.c:876
 splice_direct_to_actor+0x2d2/0x970 fs/splice.c:953
 do_splice_direct+0x1da/0x2a0 fs/splice.c:1062
 do_sendfile+0x597/0xd00 fs/read_write.c:1443
 __do_sys_sendfile64 fs/read_write.c:1498 [inline]
 __se_sys_sendfile64 fs/read_write.c:1490 [inline]
 __x64_sys_sendfile64+0x15a/0x220 fs/read_write.c:1490
 do_syscall_64+0x103/0x610 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x458da9
Code: ad b8 fb ff c3 66 2e 0f 1f 84 00 00 00 00 00 66 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 0f 83 7b b8 fb ff c3 66 2e 0f 1f 84 00 00 00 00
RSP: 002b:00007f300d24bc78 EFLAGS: 00000246 ORIG_RAX: 0000000000000028
RAX: ffffffffffffffda RBX: 0000000000000004 RCX: 0000000000458da9
RDX: 00000000200000c0 RSI: 0000000000000008 RDI: 0000000000000007
RBP: 000000000073bf00 R08: 0000000000000000 R09: 0000000000000000
R10: 000000000000005a R11: 0000000000000246 R12: 00007f300d24c6d4
R13: 00000000004c5fa3 R14: 00000000004da748 R15: 00000000ffffffff

Allocated by task 17543:
 save_stack+0x45/0xd0 mm/kasan/common.c:75
 set_track mm/kasan/common.c:87 [inline]
 __kasan_kmalloc mm/kasan/common.c:497 [inline]
 __kasan_kmalloc.constprop.0+0xcf/0xe0 mm/kasan/common.c:470
 kasan_slab_alloc+0xf/0x20 mm/kasan/common.c:505
 slab_post_alloc_hook mm/slab.h:437 [inline]
 slab_alloc mm/slab.c:3393 [inline]
 kmem_cache_alloc+0x11a/0x6f0 mm/slab.c:3555
 alloc_pid+0x55/0x8f0 kernel/pid.c:168
 copy_process.part.0+0x3b08/0x7980 kernel/fork.c:1932
 copy_process kernel/fork.c:1709 [inline]
 _do_fork+0x257/0xfd0 kernel/fork.c:2226
 __do_sys_clone kernel/fork.c:2333 [inline]
 __se_sys_clone kernel/fork.c:2327 [inline]
 __x64_sys_clone+0xbf/0x150 kernel/fork.c:2327
 do_syscall_64+0x103/0x610 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

Freed by task 7789:
 save_stack+0x45/0xd0 mm/kasan/common.c:75
 set_track mm/kasan/common.c:87 [inline]
 __kasan_slab_free+0x102/0x150 mm/kasan/common.c:459
 kasan_slab_free+0xe/0x10 mm/kasan/common.c:467
 __cache_free mm/slab.c:3499 [inline]
 kmem_cache_free+0x86/0x260 mm/slab.c:3765
 put_pid.part.0+0x111/0x150 kernel/pid.c:111
 put_pid+0x20/0x30 kernel/pid.c:105
 fl_free+0xbe/0xe0 net/ipv6/ip6_flowlabel.c:102
 ip6_fl_gc+0x295/0x3e0 net/ipv6/ip6_flowlabel.c:152
 call_timer_fn+0x190/0x720 kernel/time/timer.c:1325
 expire_timers kernel/time/timer.c:1362 [inline]
 __run_timers kernel/time/timer.c:1681 [inline]
 __run_timers kernel/time/timer.c:1649 [inline]
 run_timer_softirq+0x652/0x1700 kernel/time/timer.c:1694
 __do_softirq+0x266/0x95a kernel/softirq.c:293

The buggy address belongs to the object at ffff888094012a00
 which belongs to the cache pid_2 of size 88
The buggy address is located 4 bytes inside of
 88-byte region [ffff888094012a00ffff888094012a58)
The buggy address belongs to the page:
page:ffffea0002500480 count:1 mapcount:0 mapping:ffff88809a483080 index:0xffff888094012980
flags: 0x1fffc0000000200(slab)
raw: 01fffc0000000200 ffffea00018a3508 ffffea0002524a88 ffff88809a483080
raw: ffff888094012980 ffff888094012000 000000010000001b 0000000000000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
 ffff888094012900: fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc
 ffff888094012980: fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc
>ffff888094012a00: fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc
                   ^
 ffff888094012a80: fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc
 ffff888094012b00: fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc

Fixes: 4f82f45730c6 ("net ip6 flowlabel: Make owner a union of struct pid * and kuid_t")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/ipv6/ip6_flowlabel.c

index b82e439..9cf8661 100644 (file)
@@ -94,15 +94,21 @@ static struct ip6_flowlabel *fl_lookup(struct net *net, __be32 label)
        return fl;
 }
 
+static void fl_free_rcu(struct rcu_head *head)
+{
+       struct ip6_flowlabel *fl = container_of(head, struct ip6_flowlabel, rcu);
+
+       if (fl->share == IPV6_FL_S_PROCESS)
+               put_pid(fl->owner.pid);
+       kfree(fl->opt);
+       kfree(fl);
+}
+
 
 static void fl_free(struct ip6_flowlabel *fl)
 {
-       if (fl) {
-               if (fl->share == IPV6_FL_S_PROCESS)
-                       put_pid(fl->owner.pid);
-               kfree(fl->opt);
-               kfree_rcu(fl, rcu);
-       }
+       if (fl)
+               call_rcu(&fl->rcu, fl_free_rcu);
 }
 
 static void fl_release(struct ip6_flowlabel *fl)