[PATCH v2 0/7] More n_tty fixes

From: Peter Hurley
Date: Sun Jan 10 2016 - 00:46:08 EST

Changes from v2:
added patch 7 from Johannes Stezenbach which fixes the
TTY_DO_WRITE_WAKEUP logic to prevent missed SIGIO and reduce
excessive SIGIO

original message follows:

Hi Greg,

This series collects up several fixes for N_TTY.

The first patch fixes read() when VMIN > 0 & VTIME = 0 (so called "record mode").
By ripping out a marginal optimization, the overall code is greatly simplified
and "record mode" read()s won't hang :)

Patch 2 removes the pointless fasync() notification to line disciplines.
Patches 3,4 & 5 fix hangup races with enabling/disabling signal-driven i/o.
Patch 6 is a minor cleanup for a condition test that can never be true.

Peter Hurley (7):
n_tty: Always wake up read()/poll() if new input
tty, n_tty: Remove fasync() ldisc notification
tty: Add fasync() hung up file operation
tty: Fix ioctl(FIOASYNC) on hungup file
n_tty: Fix stuck write wakeup
n_tty: Remove tty count checks from unthrottle
tty: n_tty: fix SIGIO for output

drivers/tty/n_tty.c | 49 +++++------------------------------------------
drivers/tty/tty_io.c | 19 +++++++++---------
include/linux/tty_ldisc.h | 6 ------
3 files changed, 14 insertions(+), 60 deletions(-)