linux-kernel - [Question] How to deal D state on request_wait

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <5ae3557a-0b87-e6c0-be41-8441026c400c@huawei.com>
Date:   Tue, 17 Nov 2020 17:01:25 +0800
From:   Haotian Li <lihaotian9@...wei.com>
To:     <miklos@...redi.hu>, <mszeredi@...hat.com>, <virtio-fs@...hat.com>
CC:     <linux-fsdevel@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        "liuzhiqiang (I)" <liuzhiqiang26@...wei.com>,
        linfeilong <linfeilong@...wei.com>, <lixiaokeng@...wei.com>
Subject: [Question] How to deal D state on request_wait_answer?

Hi
    We recently detected a bug in virtiofs.  Here, we created a
virtual machine as guest with Qemu and virtiofsd. We mounted
virtiofs on guest, for example /home/virtiofs. Then we killed
the virtiofsd in host and accessed /home/virtiofs in guest later.
This casued a process with D state which could not be killed.
The stack of the process as following:
      wait_event_interruptible
[<0>] request_wait_answer+0x9d/0x210 [fuse]
[<0>] __fuse_request_send+0x75/0x80 [fuse]
[<0>] fuse_simple_request+0x164/0x270 [fuse]
[<0>] fuse_do_getattr+0xd5/0x2a0 [fuse]
[<0>] vfs_statx+0x89/0xe0
[<0>] __do_sys_newstat+0x39/0x70
[<0>] do_syscall_64+0x55/0x1c0
[<0>] entry_SYSCALL_64_after_hwframe+0x44/0xa9
    We have no idea to deal with the bug. Here, we have some
question about that:
    1. Why not use timeout mechanism?
    2. If timeout mechanism is used, the process will enter
 wait_event after wait_event_interruptible_timeout?