linux-stable.git/drivers/tty/tty_buffer.c, branch v4.4.293

tty: tty_buffer: Fix the softlockup issue in flush_to_ldisc

2021-11-26T10:58:42+00:00

[ Upstream commit 3968ddcf05fb4b9409cd1859feb06a5b0550a1c1 ]

When running ltp testcase(ltp/testcases/kernel/pty/pty04.c) with arm64, there is a soft lockup,
which look like this one:

  Workqueue: events_unbound flush_to_ldisc
  Call trace:
   dump_backtrace+0x0/0x1ec
   show_stack+0x24/0x30
   dump_stack+0xd0/0x128
   panic+0x15c/0x374
   watchdog_timer_fn+0x2b8/0x304
   __run_hrtimer+0x88/0x2c0
   __hrtimer_run_queues+0xa4/0x120
   hrtimer_interrupt+0xfc/0x270
   arch_timer_handler_phys+0x40/0x50
   handle_percpu_devid_irq+0x94/0x220
   __handle_domain_irq+0x88/0xf0
   gic_handle_irq+0x84/0xfc
   el1_irq+0xc8/0x180
   slip_unesc+0x80/0x214 [slip]
   tty_ldisc_receive_buf+0x64/0x80
   tty_port_default_receive_buf+0x50/0x90
   flush_to_ldisc+0xbc/0x110
   process_one_work+0x1d4/0x4b0
   worker_thread+0x180/0x430
   kthread+0x11c/0x120

In the testcase pty04, The first process call the write syscall to send
data to the pty master. At the same time, the workqueue will do the
flush_to_ldisc to pop data in a loop until there is no more data left.
When the sender and workqueue running in different core, the sender sends
data fastly in full time which will result in workqueue doing work in loop
for a long time and occuring softlockup in flush_to_ldisc with kernel
configured without preempt. So I add need_resched check and cond_resched
in the flush_to_ldisc loop to avoid it.

Signed-off-by: Guanghui Feng 
Link: https://lore.kernel.org/r/1633961304-24759-1-git-send-email-guanghuifeng@linux.alibaba.com
Signed-off-by: Greg Kroah-Hartman 
Signed-off-by: Sasha Levin

tty: increase the default flip buffer limit to 2*640K

2019-04-27T07:33:52+00:00

[ Upstream commit 7ab57b76ebf632bf2231ccabe26bea33868118c6 ]

We increase the default limit for buffer memory allocation by a factor of
10 to 640K to prevent data loss when using fast serial interfaces.

For example when using RS485 without flow-control at speeds of 1Mbit/s
an upwards we've run into problems such as applications being too slow
to read out this buffer (on embedded devices based on imx53 or imx6).

If you want to write transmitted data to a slow SD card and thus have
realtime requirements, this limit can become a problem.

That shouldn't be the case and 640K buffers fix such problems for us.

This value is a maximum limit for allocation only. It has no effect
on systems that currently run fine. When transmission is slow enough
applications and hardware can keep up and increasing this limit
doesn't change anything.

It only _allows_ to allocate more than 2*64K in cases we currently fail to
allocate memory despite having some.

Signed-off-by: Manfred Schlaegl 
Signed-off-by: Martin Kepplinger 
Signed-off-by: Greg Kroah-Hartman 
Signed-off-by: Sasha Levin

tty: wipe buffer.

2018-12-01T08:46:40+00:00

commit c9a8e5fce009e3c601a43c49ea9dbcb25d1ffac5 upstream.

After we are done with the tty buffer, zero it out.

Reported-by: aszlig 
Tested-by: Milan Broz 
Tested-by: Daniel Zatovic 
Tested-by: aszlig 
Cc: Willy Tarreau 
Signed-off-by: Greg Kroah-Hartman

tty: fix __tty_insert_flip_char regression

2017-09-27T09:00:13+00:00

commit 8a5a90a2a477b86a3dc2eaa5a706db9bfdd647ca upstream.

Sergey noticed a small but fatal mistake in __tty_insert_flip_char,
leading to an oops in an interrupt handler when using any serial
port.

The problem is that I accidentally took the tty_buffer pointer
before calling __tty_buffer_request_room(), which replaces the
buffer. This moves the pointer lookup to the right place after
allocating the new buffer space.

Fixes: 979990c62848 ("tty: improve tty_insert_flip_char() fast path")
Reported-by: Sergey Senozhatsky 
Tested-by: Sergey Senozhatsky 
Signed-off-by: Arnd Bergmann 
Signed-off-by: Greg Kroah-Hartman

tty: improve tty_insert_flip_char() slow path

2017-09-27T09:00:13+00:00

commit 065ea0a7afd64d6cf3464bdd1d8cd227527e2045 upstream.

While working on improving the fast path of tty_insert_flip_char(),
I noticed that by calling tty_buffer_request_room(), we needlessly
move to the separate flag buffer mode for the tty, even when all
characters use TTY_NORMAL as the flag.

This changes the code to call __tty_buffer_request_room() with the
correct flag, which will then allocate a regular buffer when it rounds
out of space but no special flags have been used. I'm guessing that
this is the behavior that Peter Hurley intended when he introduced
the compacted flip buffers.

Fixes: acc0f67f307f ("tty: Halve flip buffer GFP_ATOMIC memory consumption")
Cc: Peter Hurley 
Signed-off-by: Arnd Bergmann 
Signed-off-by: Greg Kroah-Hartman

tty: improve tty_insert_flip_char() fast path

2017-09-27T09:00:13+00:00

commit 979990c6284814617d8f2179d197f72ff62b5d85 upstream.

kernelci.org reports a crazy stack usage for the VT code when CONFIG_KASAN
is enabled:

drivers/tty/vt/keyboard.c: In function 'kbd_keycode':
drivers/tty/vt/keyboard.c:1452:1: error: the frame size of 2240 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]

The problem is that tty_insert_flip_char() gets inlined many times into
kbd_keycode(), and also into other functions, and each copy requires 128
bytes for stack redzone to check for a possible out-of-bounds access on
the 'ch' and 'flags' arguments that are passed into
tty_insert_flip_string_flags as a variable-length string.

This introduces a new __tty_insert_flip_char() function for the slow
path, which receives the two arguments by value. This completely avoids
the problem and the stack usage goes back down to around 100 bytes.

Without KASAN, this is also slightly better, as we don't have to
spill the arguments to the stack but can simply pass 'ch' and 'flag'
in registers, saving a few bytes in .text for each call site.

This should be backported to linux-4.0 or later, which first introduced
the stack sanitizer in the kernel.

Fixes: c420f167db8c ("kasan: enable stack instrumentation")
Signed-off-by: Arnd Bergmann 
Signed-off-by: Greg Kroah-Hartman

Fix OpenSSH pty regression on close

2016-06-01T19:15:52+00:00

commit 0f40fbbcc34e093255a2b2d70b6b0fb48c3f39aa upstream.

OpenSSH expects the (non-blocking) read() of pty master to return
EAGAIN only if it has received all of the slave-side output after
it has received SIGCHLD. This used to work on pre-3.12 kernels.

This fix effectively forces non-blocking read() and poll() to
block for parallel i/o to complete for all ttys. It also unwinds
these changes:

1) f8747d4a466ab2cafe56112c51b3379f9fdb7a12
   tty: Fix pty master read() after slave closes

2) 52bce7f8d4fc633c9a9d0646eef58ba6ae9a3b73
   pty, n_tty: Simplify input processing on final close

3) 1a48632ffed61352a7810ce089dc5a8bcd505a60
   pty: Fix input race when closing

Inspired by analysis and patch from Marc Aurele La France 

Reported-by: Volth 
Reported-by: Marc Aurele La France 
BugLink: https://bugzilla.mindrot.org/show_bug.cgi?id=52
BugLink: https://bugzilla.mindrot.org/show_bug.cgi?id=2492
Signed-off-by: Brian Bloniarz 
Reviewed-by: Peter Hurley 
Signed-off-by: Greg Kroah-Hartman

tty: Fix GPF in flush_to_ldisc()

2015-12-13T07:05:28+00:00

A line discipline which does not define a receive_buf() method can
can cause a GPF if data is ever received [1]. Oddly, this was known
to the author of n_tracesink in 2011, but never fixed.

[1] GPF report
    BUG: unable to handle kernel NULL pointer dereference at           (null)
    IP: [<          (null)>]           (null)
    PGD 3752d067 PUD 37a7b067 PMD 0
    Oops: 0010 [#1] SMP KASAN
    Modules linked in:
    CPU: 2 PID: 148 Comm: kworker/u10:2 Not tainted 4.4.0-rc2+ #51
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
    Workqueue: events_unbound flush_to_ldisc
    task: ffff88006da94440 ti: ffff88006db60000 task.ti: ffff88006db60000
    RIP: 0010:[<0000000000000000>]  [<          (null)>]           (null)
    RSP: 0018:ffff88006db67b50  EFLAGS: 00010246
    RAX: 0000000000000102 RBX: ffff88003ab32f88 RCX: 0000000000000102
    RDX: 0000000000000000 RSI: ffff88003ab330a6 RDI: ffff88003aabd388
    RBP: ffff88006db67c48 R08: ffff88003ab32f9c R09: ffff88003ab31fb0
    R10: ffff88003ab32fa8 R11: 0000000000000000 R12: dffffc0000000000
    R13: ffff88006db67c20 R14: ffffffff863df820 R15: ffff88003ab31fb8
    FS:  0000000000000000(0000) GS:ffff88006dc00000(0000) knlGS:0000000000000000
    CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
    CR2: 0000000000000000 CR3: 0000000037938000 CR4: 00000000000006e0
    Stack:
     ffffffff829f46f1 ffff88006da94bf8 ffff88006da94bf8 0000000000000000
     ffff88003ab31fb0 ffff88003aabd438 ffff88003ab31ff8 ffff88006430fd90
     ffff88003ab32f9c ffffed0007557a87 1ffff1000db6cf78 ffff88003ab32078
    Call Trace:
     [] process_one_work+0x8f1/0x17a0 kernel/workqueue.c:2030
     [] worker_thread+0xd4/0x1180 kernel/workqueue.c:2162
     [] kthread+0x1cf/0x270 drivers/block/aoe/aoecmd.c:1302
     [] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:468
    Code:  Bad RIP value.
    RIP  [<          (null)>]           (null)
     RSP 
    CR2: 0000000000000000
    ---[ end trace a587f8947e54d6ea ]---

Reported-by: Dmitry Vyukov 
Cc: 
Signed-off-by: Peter Hurley 
Signed-off-by: Greg Kroah-Hartman

tty: Use unbound workqueue for all input workers

2015-10-18T04:32:21+00:00

The commonly accepted wisdom that scheduling work on the same cpu
that handled interrupt i/o benefits from cache-locality is only
true if the cpu is idle (since bound kworkers are often the highest
vruntime and thus the lowest priority).

Measurements of scheduling via the unbound queue show lowered
worst-case latency responses of up to 5x over bound workqueue, without
increase in average latency or throughput.

pty i/o test measurements show >3x (!) reduced total running time; tests
previously taking ~8s now complete in <2.5s.

Signed-off-by: Peter Hurley 
Signed-off-by: Greg Kroah-Hartman

tty: Abstract tty buffer work

2015-10-18T04:32:21+00:00

Introduce API functions to restart and cancel tty buffer work, rather
than manipulate buffer work directly.

Signed-off-by: Peter Hurley 
Signed-off-by: Greg Kroah-Hartman