bug#23327: 25.0.92; show-trailing-whitespace uses too much cpu

bug#23327: 25.0.92; show-trailing-whitespace uses too much cpu

[Mohammed Sadik]

Enabling show-trailing-whitespace and navigating through buffer uses too
much CPU.  An increase of about 4-6% of CPU on my 2.4 GHz Quad-core
system.

How to reproduce:

1. Open a large source code file (I opened cc-engine.el from Emacs source).
2. Enable show-trailing-whitespace (setq show-trailing-whitespace t)
3. Press and hold the arrow keys to navigate. See the processor usage
   (use top/htop or similar program).
4. Now disable show-trailing-whitespace (setq show-trailing-whitespace
   nil)
5. Repeat 3

This also happens when pressing any other key (eg:while typing).

Also this creates minor glitches in displaying text on buffer
(sudden disappearance and appearence of some text) when CPU is already
enough busy.



In GNU Emacs 25.0.92.1 (x86_64-unknown-linux-gnu, GTK+ Version 3.18.9)
 of 2016-04-20 built on fedora.localdomain
Windowing system distributor 'Fedora Project', version 11.0.11800000
Configured features:
XPM JPEG TIFF GIF PNG SOUND DBUS GSETTINGS NOTIFY ACL LIBSELINUX GNUTLS
LIBXML2 FREETYPE XFT ZLIB TOOLKIT_SCROLL_BARS GTK3 X11

Important settings:
  value of $LC_MONETARY: en_IN.UTF-8
  value of $LC_NUMERIC: en_IN.UTF-8
  value of $LC_TIME: en_IN.UTF-8
  value of $LANG: en_US.UTF-8
  value of $XMODIFIERS: @im=ibus
  locale-coding-system: utf-8-unix

Major mode: Emacs-Lisp

Minor modes in effect:
  tooltip-mode: t
  global-eldoc-mode: t
  electric-indent-mode: t
  mouse-wheel-mode: t
  tool-bar-mode: t
  menu-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: t

Recent messages:
For information about GNU Emacs and the GNU system, type C-h C-a.
t
Mark set [2 times]
nil
Making completion list...

Load-path shadows:
None found.

Features:
(shadow sort mail-extr emacsbug message dired format-spec rfc822 mml
mml-sec epg epg-config gnus-util mm-decode mm-bodies mm-encode
mail-parse rfc2231 mailabbrev gmm-utils mailheader sendmail rfc2047
rfc2045 ietf-drums mm-util help-fns help-mode easymenu cl-loaddefs pcase
cl-lib mail-prsvr mail-utils time-date mule-util tooltip eldoc electric
uniquify ediff-hook vc-hooks lisp-float-type mwheel x-win
term/common-win x-dnd tool-bar dnd fontset image regexp-opt fringe
tabulated-list newcomment elisp-mode lisp-mode prog-mode register page
menu-bar rfn-eshadow timer select scroll-bar mouse jit-lock font-lock
syntax facemenu font-core frame cl-generic cham georgian utf-8-lang
misc-lang vietnamese tibetan thai tai-viet lao korean japanese eucjp-ms
cp51932 hebrew greek romanian slovak czech european ethiopic indian
cyrillic chinese charscript case-table epa-hook jka-cmpr-hook help
simple abbrev minibuffer cl-preloaded nadvice loaddefs button faces
cus-face macroexp files text-properties overlay sha1 md5 base64 format
env code-pages mule custom widget hashtable-print-readable backquote
dbusbind inotify dynamic-setting system-font-setting font-render-setting
move-toolbar gtk x-toolkit x multi-tty make-network-process emacs)

Memory information:
((conses 16 89437 9618)
 (symbols 48 19612 0)
 (miscs 40 83 185)
 (strings 32 14290 4341)
 (string-bytes 1 410796)
 (vectors 16 11670)
 (vector-slots 8 421100 7008)
 (floats 8 166 119)
 (intervals 56 962 0)
 (buffers 976 13)
 (heap 1024 35720 1005))

bug#23327: 25.0.92; show-trailing-whitespace uses too much cpu

[Eli Zaretskii]

> Date: Thu, 21 Apr 2016 11:33:55 +0530
> From: Mohammed Sadik <sadiq@sadiqpk.org>
> 
> Enabling show-trailing-whitespace and navigating through buffer uses too
> much CPU.  An increase of about 4-6% of CPU on my 2.4 GHz Quad-core
> system.
> 
> How to reproduce:
> 
> 1. Open a large source code file (I opened cc-engine.el from Emacs source).
> 2. Enable show-trailing-whitespace (setq show-trailing-whitespace t)
> 3. Press and hold the arrow keys to navigate. See the processor usage
>    (use top/htop or similar program).
> 4. Now disable show-trailing-whitespace (setq show-trailing-whitespace
>    nil)
> 5. Repeat 3
> 
> This also happens when pressing any other key (eg:while typing).

The Emacs display engine employs several aggressive optimizations for
simple operations like cursor motion and inserting a single
character.  When you turn on show-trailing-whitespace, most of these
optimizations are disabled, so it's very small wonder that redisplay
becomes more expensive.

The reason that these optimizations are disable is this feature of
show-trailing-whitespace:

     This feature does not apply when point is at the end of the line
  containing the whitespace.  Strictly speaking, that is trailing
  whitespace nonetheless, but displaying it specially in that case looks
  ugly while you are typing in new text.  In this special case, the
  location of point is enough to show you that the spaces are present.

Because of this, when you move the cursor, the display engine cannot
assume that the display remains unchanged, because it needs to see if
the cursor happens to be after a stretch of trailing whitespace, in
which case, that whole stretch of whitespace needs to be redrawn as
well.  For the same reason, other optimizations, like those which
reuse portions of the previously displayed window, are abandoned.

I've looked at the optimizations we disable in this case, and I cannot
see anything wrong in those decisions, given the way this feature
works.

We could try adding an optional feature which doesn't turn off the
special display of trailing whitespace when the cursor is at the end
of a line that ends in whitespace, that might allow us to re-enable
these optimizations.  Would users like that?

Other than that, I don't see how this could be helped, except by some
thorough redesign of how this feature works.  I don't immediately see
how to do that, but maybe someone else does, and is motivated to work
on this.

> Also this creates minor glitches in displaying text on buffer
> (sudden disappearance and appearence of some text) when CPU is already
> enough busy.

If your CPU is very busy, then adding a few percents to it might
indeed produce such jerky redisplay, I think.