So this is it. The big step why we did all the work over the past
kernel releases. Now everything is prepared, so nothing protects us
from doing that big step.

           |  |            \  \ nnnn/^l      |  |
           |  |             \  /     /       |  |
           |  '-,.__   =>    \/   ,-`    =>  |  '-,.__
           | O __.´´)        (  .`           | O __.´´)
            ~~~   ~~          ``              ~~~   ~~
The buffers are now in the tty_port structure and we can start
teaching the buffer helpers (insert char/string, flip etc.) to use
tty_port instead of tty_struct all around.
Signed-off-by: NJiri Slaby <jslaby@suse.cz>
Acked-by: NAlan Cox <alan@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

During the move of tty buffers from tty_struct to tty_port, we will
need to switch all users of buf to tty->port->buf. There are many
functions where this is accessed directly in their code many times.
Cache the tty->buf pointer in such functions now and change only
single lines in each function in the next patch.

Not that it is convenient for the next patch, but the code is now also
more readable.
Signed-off-by: NJiri Slaby <jslaby@suse.cz>
Acked-by: NAlan Cox <alan@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

They are only TTY buffers specific. And the buffers will go to
tty_port in the next patches. So to remove the need to have both
tty_port and tty_struct at some places, let us move the flags to
tty_port.
Signed-off-by: NJiri Slaby <jslaby@suse.cz>
Acked-by: NAlan Cox <alan@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

When low_latency flag is set the TTY receive flip buffer is copied to the
line discipline directly instead of using a work queue in the background.
Therefor only in case a workqueue is actually used for copying data to the
line discipline we'll have to flush the workqueue.

This prevents unnecessary spin lock/unlock on the workqueue spin lock that
can cause additional scheduling overhead on a PREEMPT_RT system. On a 200
MHz AT91SAM9261 processor setup this fixes about 100us of scheduling
overhead on the TTY read call.
Signed-off-by: NIvo Sieben <meltedpianoman@gmail.com>
Acked-by: NAlan Cox <alan@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

tty_buffer_request_room is well protected, but while after it returns,
 it releases the port->lock. tty->buf.tail might be modified
by either irq handler or other threads. The patch adds more protection
by holding the lock across tty buffer finding and buffer filling.
Signed-off-by: NAlek Du <alek.du@intel.com>
Signed-off-by: NXiaobing Tu <xiaobing.tu@intel.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The flush_to_ldisc() work entry has special logic to notice when it has
seen the original tail of the data queue, and it avoids continuing the
flush if it sees that _original_ tail rather than the current tail.

This logic can trigger in case somebody is constantly adding new data to
the tty while the flushing is active - and the intent is to avoid
excessive CPU usage while flushing the tty, especially as we used to do
this from a softirq context which made it non-preemptible.

However, since we no longer re-arm the work-queue from within itself
(because that causes other trouble: see commit a5660b41 "tty: fix
endless work loop when the buffer fills up"), this just leads to
possible hung tty's (most easily seen in SMP and with a test-program
that floods a pty with data - nobody seems to have reported this for any
real-life situation yet).

And since the workqueue isn't done from timers and softirq's any more,
it's doubtful whether the CPU useage issue is really relevant any more.
So just remove the logic entirely, and see if anybody ever notices.

Alternatively, we might want to re-introduce the "re-arm the work" for
just this case, but then we'd have to re-introduce the delayed work
model or some explicit timer, which really doesn't seem worth it for
this.
Reported-and-tested-by: NGuillaume Chazarain <guichaz@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This reverts commit b1c43f82.

It was broken in so many ways, and results in random odd pty issues.

It re-introduced the buggy schedule_work() in flush_to_ldisc() that can
cause endless work-loops (see commit a5660b41: "tty: fix endless
work loop when the buffer fills up").

It also used an "unsigned int" return value fo the ->receive_buf()
function, but then made multiple functions return a negative error code,
and didn't actually check for the error in the caller.

And it didn't actually work at all.  BenH bisected down odd tty behavior
to it:
  "It looks like the patch is causing some major malfunctions of the X
   server for me, possibly related to PTYs.  For example, cat'ing a
   large file in a gnome terminal hangs the kernel for -minutes- in a
   loop of what looks like flush_to_ldisc/workqueue code, (some ftrace
   data in the quoted bits further down).

   ...

   Some more data: It -looks- like what happens is that the
   flush_to_ldisc work queue entry constantly re-queues itself (because
   the PTY is full ?) and the workqueue thread will basically loop
   forver calling it without ever scheduling, thus starving the consumer
   process that could have emptied the PTY."

which is pretty much exactly the problem we fixed in a5660b41.

Milton Miller pointed out the 'unsigned int' issue.
Reported-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Reported-by: NMilton Miller <miltonm@bga.com>
Cc: Stefan Bigler <stefan.bigler@keymile.com>
Cc: Toby Gray <toby.gray@realvnc.com>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

it makes it simpler to keep track of the amount of
bytes received and simplifies how flush_to_ldisc counts
the remaining bytes. It also fixes a bug of lost bytes
on n_tty when flushing too many bytes via the USB
serial gadget driver.
Tested-by: NStefan Bigler <stefan.bigler@keymile.com>
Tested-by: NToby Gray <toby.gray@realvnc.com>
Signed-off-by: NFelipe Balbi <balbi@ti.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Commit f23eb2b2 ('tty: stop using "delayed_work" in the tty layer')
ended up causing hung machines on UP with no preemption, because the
work routine to flip the buffer data to the ldisc would endlessly re-arm
itself if the destination buffer had filled up.

With the delayed work, that only caused a timer-driving polling of the
tty state every timer tick, but without the delay we just ended up with
basically a busy loop instead.

Stop the insane polling, and instead make the code that opens up the
receive room re-schedule the buffer flip work.  That's what we should
have been doing anyway.

This same "poll for tty room" issue is almost certainly also the cause
of excessive kworker activity when idle reported by Dave Jones, who also
reported "flush_to_ldisc executing 2500 times a second" back in Nov 2010:

  http://lkml.org/lkml/2010/11/30/592

which is that silly flushing done every timer tick.  Wasting both power
and CPU for no good reason.
Reported-and-tested-by: NAlexander Beregalov <a.beregalov@gmail.com>
Reported-and-tested-by: NSitsofe Wheeler <sitsofe@yahoo.com>
Cc: Greg KH <gregkh@suse.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Dave Jones <davej@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Using delayed-work for tty flip buffers ends up causing us to wait for
the next tick to complete some actions.  That's usually not all that
noticeable, but for certain latency-critical workloads it ends up being
totally unacceptable.

As an extreme case of this, passing a token back-and-forth over a pty
will take two ticks per iteration, so even just a thousand iterations
will take 8 seconds assuming a common 250Hz configuration.

Avoiding the whole delayed work issue brings that ping-pong test-case
down to 0.009s on my machine.

In more practical terms, this latency has been a performance problem for
things like dive computer simulators (simulating the serial interface
using the ptys) and for other environments (Alan mentions a CP/M emulator).
Reported-by: NJef Driesen <jefdriesen@telenet.be>
Acked-by: NGreg KH <gregkh@suse.de>
Acked-by: NAlan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

There's a small window inside the flush_to_ldisc function,
where the tty is unlocked and calling ldisc's receive_buf
function. If in this window new buffer is added to the tty,
the processing might never leave the flush_to_ldisc function.

This scenario will hog the cpu, causing other tty processing
starving, and making it impossible to interface the computer
via tty.

I was able to exploit this via pty interface by sending only
control characters to the master input, causing the flush_to_ldisc
to be scheduled, but never actually generate any output.

To reproduce, please run multiple instances of following code.

- SNIP
#define _XOPEN_SOURCE
#include <stdlib.h>
#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>

int main(int argc, char **argv)
{
        int i, slave, master = getpt();
        char buf[8192];

        sprintf(buf, "%s", ptsname(master));
        grantpt(master);
        unlockpt(master);

        slave = open(buf, O_RDWR);
        if (slave < 0) {
                perror("open slave failed");
                return 1;
        }

        for(i = 0; i < sizeof(buf); i++)
                buf[i] = rand() % 32;

        while(1) {
                write(master, buf, sizeof(buf));
        }

        return 0;
}
- SNIP

The attached patch (based on -next tree) fixes this by checking on the
tty buffer tail. Once it's reached, the current work is rescheduled
and another could run.
Signed-off-by: NJiri Olsa <jolsa@redhat.com>
Cc: stable <stable@kernel.org>
Acked-by: NAlan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The tty code should be in its own subdirectory and not in the char
driver with all of the cruft that is currently there.

Based on work done by Arnd Bergmann <arnd@arndb.de>
Acked-by: NArnd Bergmann <arnd@arndb.de>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Comment was not updated when tty_insert_flip_string was generalised.
Signed-off-by: NJohan Hovold <jhovold@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

CC      drivers/char/tty_buffer.o
drivers/char/tty_buffer.c: In function ‘tty_insert_flip_string_fixed_flag’:
drivers/char/tty_buffer.c:251: warning: comparison of distinct pointer types lacks a cast
drivers/char/tty_buffer.c: In function ‘tty_insert_flip_string_flags’:
drivers/char/tty_buffer.c:288: warning: comparison of distinct pointer types lacks a cast

Fix it by replacing min() with min_t() in tty_insert_flip_string_flags and
					  tty_insert_flip_string_fixed_flag().
Signed-off-by: NFang Wenqi <antonf@turbolinux.com.cn>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The USB drivers often want to insert a series of bytes all with the same
flag set - provide a helper for this case.
Signed-off-by: NAlan Cox <alan@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

We allocate during interrupts so while our buffering is normally diced up
small anyway on some hardware at speed we can pressure the VM excessively
for page pairs. We don't really need big buffers to be linear so don't try
so hard.

In order to make this work well we will tidy up excess callers to request_room,
which cannot itself enforce this break up.
Signed-off-by: NAlan Cox <alan@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This way all flush_to_ldisc work is always done through the workqueues,
and we thus have a single point of serialization.  It also means that we
can avoid calling flush_to_ldisc() entirely if there was no delayed work
pending.

[ Side note: using workqueues and keventd as the single way to enter
  flush_to_ldisc() still doesn't absolutely guarantee that we can't have
  concurrency: keventd is multithreaded and has a thread per CPU, and
  while the WORK_STRUCT_PENDING bit guarantees a single work only being
  on the pending list once, the work might be both pending and _running_
  at the same time. Workqueues are not simple. ]

This was also confirmed to fix bugzilla #14388, even without the earlier
locking fix and cleanup (commit c8e33141: "tty: Make flush_to_ldisc()
locking more robust").  So both commits fix the same bug differently,
and either would have worked on its own.  But I'm committing them both
since they are cleanups independent of each other.
Reported-and-tested-by: NBoyan <btanastasov@yahoo.co.uk>
Acked-by: NAlan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The locking logic in this function is extremely subtle, and it broke
when we started doing potentially concurrent 'flush_to_ldisc()' calls in
commit e043e42b ("pty: avoid forcing
'low_latency' tty flag").

The code in flush_to_ldisc() used to set 'tty->buf.head' to NULL, with
the intention that this would then cause any other concurrent calls to
not do anything (locking note: we have to drop the buf.lock over the
call to ->receive_buf that can block, which is why we can have
concurrency here at all in the first place).

It also used to set the TTY_FLUSHING bit, which would then cause any
concurrent 'tty_buffer_flush()' to not free all the tty buffers and
clear 'tty->buf.tail'.  And with 'buf.head' being NULL, and 'buf.tail'
being non-NULL, new data would never touch 'buf.head'.

Does that sound a bit too subtle? It was.  If another concurrent call to
'flush_to_ldisc()' were to come in, the NULL buf.head would indeed cause
it to not process the buffer list, but it would still clear TTY_FLUSHING
afterwards, making the buffer protection against 'tty_buffer_flush()' no
longer work.

So this clears it all up.  We depend purely on TTY_FLUSHING for handling
re-entrancy, and stop playing games with the buffer list entirely.  In
fact, the buffer list handling is now robust enough that we could
probably stop doing the whole "protect against 'tty_buffer_flush()'"
thing entirely.

However, Alan also points out that we would probably be better off
simplifying the locking even further, and just take the tty ldisc_mutex
around all the buffer flushing calls.  That seems like a good idea, but
in the meantime this is a conceptually minimal fix (with the patch
itself being bigger than required just to clean the code up and make it
readable).

This fixes keyboard trouble under X:

	http://bugzilla.kernel.org/show_bug.cgi?id=14388Reported-and-tested-by: NFrédéric Meunier <fredlwm@gmail.com>
Reported-and-tested-by: NBoyan <btanastasov@yahoo.co.uk>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Paul Fulghum <paulkf@microgate.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This way all flush_to_ldisc work is always done through the workqueues,
and we thus have a single point of serialization.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

We really don't want to mark the pty as a low-latency device, because as
Alan points out, the ->write method can be called from an IRQ (ppp?),
and that means we can't use ->low_latency=1 as we take mutexes in the
low_latency case.

So rather than using low_latency to force the written data to be pushed
to the ldisc handling at 'write()' time, just make the reader side (or
the poll function) do the flush when it checks whether there is data to
be had.

This also fixes the problem with lost data in an emacs compile buffer
(bugzilla 13815), and we can thus revert the low_latency pty hack
(commit 3a542974: "pty: quickfix for the
pty ENXIO timing problems").
Signed-off-by: NOGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Tested-by: NAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
[ Modified to do the tty_flush_to_ldisc() inside input_available_p() so
  that it triggers for both read and poll()  - Linus]
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The two are basically independent chunks of code so lets split them up for
readability and sanity. It also makes the API boundaries much clearer.
Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>