Re: [syzbot] memory leak in setup_mq_sysctls
From: Eric W. Biederman
Date: Tue Jun 21 2022 - 10:31:13 EST
Alexey Gladkov <gladkov.alexey@xxxxxxxxx> writes:
> On Sun, Jun 19, 2022 at 11:52:25PM -0700, syzbot wrote:
>> Hello,
>>
>> syzbot found the following issue on:
>>
>> HEAD commit: 979086f5e006 Merge tag 'fs.fixes.v5.19-rc3' of git://git.k..
>> git tree: upstream
>> console output: https://syzkaller.appspot.com/x/log.txt?x=1284331bf00000
>> kernel config: https://syzkaller.appspot.com/x/.config?x=c696a83383a77f81
>> dashboard link: https://syzkaller.appspot.com/bug?extid=b4b0d1b35442afbf6fd2
>> compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
>> syz repro: https://syzkaller.appspot.com/x/repro.syz?x=163e740ff00000
>> C reproducer: https://syzkaller.appspot.com/x/repro.c?x=132b758bf00000
>>
>> IMPORTANT: if you fix the issue, please add the following tag to the commit:
>> Reported-by: syzbot+b4b0d1b35442afbf6fd2@xxxxxxxxxxxxxxxxxxxxxxxxx
>
> I'm working on a fix that will remove this memory allocation entirely.
Hmm. The memory should be freed when the corresponding namespace exits.
I see retire_mq_sysctls is being called to free this memory. Alex do
you see any leaks when you read that code?
So it looks like either someone broke this in linux-next or there
is a bug in the memory leak detector, or something truly strange
like a memory stop is going on.
I don't see any changes to the ipc subdirectory since v5.19-rc1 in
commit 979086f5e006 ("Merge tag 'fs.fixes.v5.19-rc3' of git://git.k.." )
so the idea that the code is broken in linux-next is out.
Which leaves the memory leak detector having trouble with this,
or something like a memory stomp is causing problems.
Catalin is it possible that the clever use of ctl_table_arg to hold the
reference to the table before it is freed is confusing the memory leak
detector? The idiom is old enough I don't expect so, but I have seen
bugs lurk for a long time.
Which leaves just a memory stomp or something even stranger in the code.
syzkaller can you reproduce this on Linus's branch?
Eric
>> executing program
>> BUG: memory leak
>> unreferenced object 0xffff888112fc9200 (size 512):
>> comm "syz-executor237", pid 3648, jiffies 4294970469 (age 12.270s)
>> hex dump (first 32 bytes):
>> ef d3 60 85 ff ff ff ff 0c 9b d2 12 81 88 ff ff ..`.............
>> 04 00 00 00 a4 01 00 00 00 00 00 00 00 00 00 00 ................
>> backtrace:
>> [<ffffffff814b6eb3>] kmemdup+0x23/0x50 mm/util.c:129
>> [<ffffffff82219a9b>] kmemdup include/linux/fortify-string.h:456 [inline]
>> [<ffffffff82219a9b>] setup_mq_sysctls+0x4b/0x1c0 ipc/mq_sysctl.c:89
>> [<ffffffff822197f2>] create_ipc_ns ipc/namespace.c:63 [inline]
>> [<ffffffff822197f2>] copy_ipcs+0x292/0x390 ipc/namespace.c:91
>> [<ffffffff8127de7c>] create_new_namespaces+0xdc/0x4f0 kernel/nsproxy.c:90
>> [<ffffffff8127e89b>] unshare_nsproxy_namespaces+0x9b/0x120 kernel/nsproxy.c:226
>> [<ffffffff8123f92e>] ksys_unshare+0x2fe/0x600 kernel/fork.c:3165
>> [<ffffffff8123fc42>] __do_sys_unshare kernel/fork.c:3236 [inline]
>> [<ffffffff8123fc42>] __se_sys_unshare kernel/fork.c:3234 [inline]
>> [<ffffffff8123fc42>] __x64_sys_unshare+0x12/0x20 kernel/fork.c:3234
>> [<ffffffff845aab45>] do_syscall_x64 arch/x86/entry/common.c:50 [inline]
>> [<ffffffff845aab45>] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
>> [<ffffffff8460006a>] entry_SYSCALL_64_after_hwframe+0x46/0xb0
>>
>> BUG: memory leak
>> unreferenced object 0xffff888112fd5f00 (size 256):
>> comm "syz-executor237", pid 3648, jiffies 4294970469 (age 12.270s)
>> hex dump (first 32 bytes):
>> 00 92 fc 12 81 88 ff ff 00 00 00 00 01 00 00 00 ................
>> 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
>> backtrace:
>> [<ffffffff816fea1b>] kmalloc include/linux/slab.h:605 [inline]
>> [<ffffffff816fea1b>] kzalloc include/linux/slab.h:733 [inline]
>> [<ffffffff816fea1b>] __register_sysctl_table+0x7b/0x7f0 fs/proc/proc_sysctl.c:1344
>> [<ffffffff82219b7a>] setup_mq_sysctls+0x12a/0x1c0 ipc/mq_sysctl.c:112
>> [<ffffffff822197f2>] create_ipc_ns ipc/namespace.c:63 [inline]
>> [<ffffffff822197f2>] copy_ipcs+0x292/0x390 ipc/namespace.c:91
>> [<ffffffff8127de7c>] create_new_namespaces+0xdc/0x4f0 kernel/nsproxy.c:90
>> [<ffffffff8127e89b>] unshare_nsproxy_namespaces+0x9b/0x120 kernel/nsproxy.c:226
>> [<ffffffff8123f92e>] ksys_unshare+0x2fe/0x600 kernel/fork.c:3165
>> [<ffffffff8123fc42>] __do_sys_unshare kernel/fork.c:3236 [inline]
>> [<ffffffff8123fc42>] __se_sys_unshare kernel/fork.c:3234 [inline]
>> [<ffffffff8123fc42>] __x64_sys_unshare+0x12/0x20 kernel/fork.c:3234
>> [<ffffffff845aab45>] do_syscall_x64 arch/x86/entry/common.c:50 [inline]
>> [<ffffffff845aab45>] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
>> [<ffffffff8460006a>] entry_SYSCALL_64_after_hwframe+0x46/0xb0
>>
>> BUG: memory leak
>> unreferenced object 0xffff888112fbba00 (size 256):
>> comm "syz-executor237", pid 3648, jiffies 4294970469 (age 12.270s)
>> hex dump (first 32 bytes):
>> 78 ba fb 12 81 88 ff ff 00 00 00 00 01 00 00 00 x...............
>> 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
>> backtrace:
>> [<ffffffff816fef49>] kmalloc include/linux/slab.h:605 [inline]
>> [<ffffffff816fef49>] kzalloc include/linux/slab.h:733 [inline]
>> [<ffffffff816fef49>] new_dir fs/proc/proc_sysctl.c:978 [inline]
>> [<ffffffff816fef49>] get_subdir fs/proc/proc_sysctl.c:1022 [inline]
>> [<ffffffff816fef49>] __register_sysctl_table+0x5a9/0x7f0 fs/proc/proc_sysctl.c:1373
>> [<ffffffff82219b7a>] setup_mq_sysctls+0x12a/0x1c0 ipc/mq_sysctl.c:112
>> [<ffffffff822197f2>] create_ipc_ns ipc/namespace.c:63 [inline]
>> [<ffffffff822197f2>] copy_ipcs+0x292/0x390 ipc/namespace.c:91
>> [<ffffffff8127de7c>] create_new_namespaces+0xdc/0x4f0 kernel/nsproxy.c:90
>> [<ffffffff8127e89b>] unshare_nsproxy_namespaces+0x9b/0x120 kernel/nsproxy.c:226
>> [<ffffffff8123f92e>] ksys_unshare+0x2fe/0x600 kernel/fork.c:3165
>> [<ffffffff8123fc42>] __do_sys_unshare kernel/fork.c:3236 [inline]
>> [<ffffffff8123fc42>] __se_sys_unshare kernel/fork.c:3234 [inline]
>> [<ffffffff8123fc42>] __x64_sys_unshare+0x12/0x20 kernel/fork.c:3234
>> [<ffffffff845aab45>] do_syscall_x64 arch/x86/entry/common.c:50 [inline]
>> [<ffffffff845aab45>] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
>> [<ffffffff8460006a>] entry_SYSCALL_64_after_hwframe+0x46/0xb0
>>
>> BUG: memory leak
>> unreferenced object 0xffff888112fbb900 (size 256):
>> comm "syz-executor237", pid 3648, jiffies 4294970469 (age 12.270s)
>> hex dump (first 32 bytes):
>> 78 b9 fb 12 81 88 ff ff 00 00 00 00 01 00 00 00 x...............
>> 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
>> backtrace:
>> [<ffffffff816fef49>] kmalloc include/linux/slab.h:605 [inline]
>> [<ffffffff816fef49>] kzalloc include/linux/slab.h:733 [inline]
>> [<ffffffff816fef49>] new_dir fs/proc/proc_sysctl.c:978 [inline]
>> [<ffffffff816fef49>] get_subdir fs/proc/proc_sysctl.c:1022 [inline]
>> [<ffffffff816fef49>] __register_sysctl_table+0x5a9/0x7f0 fs/proc/proc_sysctl.c:1373
>> [<ffffffff82219b7a>] setup_mq_sysctls+0x12a/0x1c0 ipc/mq_sysctl.c:112
>> [<ffffffff822197f2>] create_ipc_ns ipc/namespace.c:63 [inline]
>> [<ffffffff822197f2>] copy_ipcs+0x292/0x390 ipc/namespace.c:91
>> [<ffffffff8127de7c>] create_new_namespaces+0xdc/0x4f0 kernel/nsproxy.c:90
>> [<ffffffff8127e89b>] unshare_nsproxy_namespaces+0x9b/0x120 kernel/nsproxy.c:226
>> [<ffffffff8123f92e>] ksys_unshare+0x2fe/0x600 kernel/fork.c:3165
>> [<ffffffff8123fc42>] __do_sys_unshare kernel/fork.c:3236 [inline]
>> [<ffffffff8123fc42>] __se_sys_unshare kernel/fork.c:3234 [inline]
>> [<ffffffff8123fc42>] __x64_sys_unshare+0x12/0x20 kernel/fork.c:3234
>> [<ffffffff845aab45>] do_syscall_x64 arch/x86/entry/common.c:50 [inline]
>> [<ffffffff845aab45>] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
>> [<ffffffff8460006a>] entry_SYSCALL_64_after_hwframe+0x46/0xb0
>>
>>
>>
>> ---
>> This report is generated by a bot. It may contain errors.
>> See https://goo.gl/tpsmEJ for more information about syzbot.
>> syzbot engineers can be reached at syzkaller@xxxxxxxxxxxxxxxx.
>>
>> syzbot will keep track of this issue. See:
>> https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
>> syzbot can test patches for this issue, for details see:
>> https://goo.gl/tpsmEJ#testing-patches
>>