linux-kernel - Query Regarding init block up due to tty_wait_until

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: <9863fc6d-bae1-4847-161a-40df4e914081@codeaurora.org>
Date:   Mon, 5 Mar 2018 17:06:38 +0530
From:   "Kohli, Gaurav" <gkohli@...eaurora.org>
To:     jslaby@...e.com, mikey@...ling.org,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc:     Alan Cox <gnomes@...rguk.ukuu.org.uk>,
        linux-kernel@...r.kernel.org, linux-arm-msm@...r.kernel.org
Subject: Query Regarding init block up due to tty_wait_until_sent

Hi,

We have seen few instances, where init is getting blocked due to wait in below call:

-002|schedule()
-003|schedule_timeout()  -> timeout for 30 seconds
-004|tty_wait_until_sent()
-005|tty_port_close_start.part.3()
-006|tty_port_close()
-007|uart_close()
-008|tty_name(inline)
-008|tty_release()
-009|__fput()
-010|____fput()
-011|task_work_run()
-012|do_notify_resume()
-013|work_pending(asm)
  ---|end of frame

When i have checked the data structures, it seems there is pending data available in buffer, but uart_port got seems to free.

   (struct uart_state *)0xFFFFFFC3EDCE4000 = 0xFFFFFFC3EDCE4000 -> (
     port = (
       buf = (head = 0xFFFFFFC3E772BC80, work = (data = (counter = 1024), entry = (next = 0xFFFFFFC3EDCE4010, prev = 0xFFFFFFC3E
       tty = 0xFFFFFFC3E772C280,
       itty = 0xFFFFFFC3E772C280,
       ops = 0xFFFFFF89D3E7EBB0,
       lock = (rlock = (raw_lock = (owner = 3, next = 3), magic = 3735899821, owner_cpu = 4294967295, owner = 0xFFFFFFFFFFFFFFFF
       blocked_open = 0,
       count = 0,
       open_wait = (lock = (rlock = (raw_lock = (owner = 0, next = 0), magic = 3735899821, owner_cpu = 4294967295, owner = 0xFFF
       delta_msr_wait = (lock = (rlock = (raw_lock = (owner = 0, next = 0), magic = 3735899821, owner_cpu = 4294967295, owner =
       flags = 0,
       iflags = 5,
       console = 1,
       low_latency = 0,
       mutex = (count = (counter = 1), wait_lock = (rlock = (raw_lock = (owner = 4, next = 4), magic = 3735899821, owner_cpu = 4
       buf_mutex = (count = (counter = 1), wait_lock = (rlock = (raw_lock = (owner = 0, next = 0), magic = 3735899821, owner_cpu
       xmit_buf = 0x0,
       close_delay = 50,
       closing_wait = 3000,
       drain_delay = 0,
       kref = (refcount = (counter = 1))),
     pm_state = UART_PM_STATE_ON = 0x0,
     xmit = (buf = 0xFFFFFFC3E7507000, head = 1, tail = 0),
     refcount = (counter = 1),
     remove_wait = (lock = (rlock = (raw_lock = (owner = 0, next = 0), magic = 3735899821, owner_cpu = 4294967295, owner = 0xFFF
     uart_port = 0xFFFFFF89D58582A0 -> (
       lock = (rlock = (raw_lock = (owner = 604, next = 604), magic = 3735899821, owner_cpu = 4294967295, owner = 0xFFFFFFFFFFFF
       iobase = 0,
       membase = 0xFFFFFF800C7C4000,
       serial_in = 0x0,
       serial_out = 0x0,
       set_termios = 0x0,
  
So in that case write of data won't be possible, So to avoid that do we have to reduce timeout which seems high in current case.
And how can we avoid this if data is available in tty, but uart_port seems to be free.

Please let me know, In case further info is required.

Regards
Gaurav

-- 
Qualcomm India Private Limited, on behalf of Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project.