bug#23368: 25.0.92; xftfont.c has got a bug

bug#23368: 25.0.92; xftfont.c has got a bug

[Masaru Nomiya]

Hello,

With Emacs 25.0.92, Emacs 25.0.93, and Emacs 25.1.50, Emacs start to
use 100% CPU.
That is, afetr I start Wanderlust(MUA) on Emacs, and try to open some
folders in it, Emacs starts to use 100% CPU, and unresposible.

I investigated the latest Emacs with git bisect.
As a result, I found that xftfont.c has got a bug as follows.


--------------------------------------------------------------------
masaru@linux-uw5l:/tmp/tes/emacs> git checkout emacs-25
Branch emacs-25 set up to track remote branch emacs-25 from origin.
Switched to a new branch 'emacs-25'

masaru@linux-uw5l:/tmp/tes/emacs> git bisect start

masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad

masaru@linux-uw5l:/tmp/tes/emacs> git bisect good emacs-25.0.91
Bisecting: 182 revisions left to test after this (roughly 8 steps)
[04f5525fbb961f6006280be340ca0f03df9e80a7] Make '.' associative, for easier sexp navigation

masaru@linux-uw5l:/tmp/tes/emacs> ./autogen.sh

masaru@linux-uw5l:/tmp/tes/emacs> ./configure

masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10

test --  NG

masaru@linux-uw5l:/tmp/tes/emacs> make clean

masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
Bisecting: 90 revisions left to test after this (roughly 7 steps)
[06a872b71d6580e0699600844bb9526602437990] Fix redisplay on a TTY after 'make-frame'
[...]
masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10

test  --  OK

masaru@linux-uw5l:/tmp/tes/emacs> make clean

masaru@linux-uw5l:/tmp/tes/emacs> git bisect good
Bisecting: 0 revisions left to test after this (roughly 1 step)
[ec10ef95cf2a5309ec7cb74f1383860a13daa546] * lisp/apropos.el (apropos-variable): Doc fix.  (Bug#22813).

masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10

test  --  NG

masaru@linux-uw5l:/tmp/tes/emacs> make clean

masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
Bisecting: 0 revisions left to test after this (roughly 0 steps)
[d2dd614716e34edb5891e58c029741cd6b32217d] Remove unneeded workaround in xftfont.c

---------------------------------------------------------------------

※ test = Start Emacs & Wanderlust, then try to open the folder in
Wanderlust. OK means "can open without any problem", NG means "Emacs
eat 100% CPU).

---
 Masaru Nomiya             mail-to: nomiya @ galaxy.dti.ne.jp

bug#23368: 25.0.92; xftfont.c has got a bug

[Masaru Nomiya]

Hello,

Sorry, I've fogotten to write a workaround.
My workaround is replacement of xftfont.c in Emacs 25.0.92, Emacs
25.0.93, and Emacs 25.1.50 with Emacs 25.0.91's.
Then Emacs (25.0.92, 25.0.93, and 25.1.50) works fine for me.

In the Message; 

  Subject    : bug#23368: 25.0.92; xftfont.c has got a bug
  Message-ID : <877ffmrzsv.wl-nomiya@galaxy.dti.ne.jp>
  Date & Time: Mon, 25 Apr 2016 15:02:40 +0900

Masaru Nomiya <nomiya@galaxy.dti.ne.jp> has written:

> Hello,

> With Emacs 25.0.92, Emacs 25.0.93, and Emacs 25.1.50, Emacs start to
> use 100% CPU.
> That is, afetr I start Wanderlust(MUA) on Emacs, and try to open some
> folders in it, Emacs starts to use 100% CPU, and unresposible.

> I investigated the latest Emacs with git bisect.
> As a result, I found that xftfont.c has got a bug as follows.


> --------------------------------------------------------------------
masaru@linux-uw5l:/tmp/tes/emacs> git checkout emacs-25
> Branch emacs-25 set up to track remote branch emacs-25 from origin.
> Switched to a new branch 'emacs-25'

masaru@linux-uw5l:/tmp/tes/emacs> git bisect start

masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad

masaru@linux-uw5l:/tmp/tes/emacs> git bisect good emacs-25.0.91
> Bisecting: 182 revisions left to test after this (roughly 8 steps)
> [04f5525fbb961f6006280be340ca0f03df9e80a7] Make '.' associative, for easier sexp navigation

masaru@linux-uw5l:/tmp/tes/emacs> ./autogen.sh

masaru@linux-uw5l:/tmp/tes/emacs> ./configure

masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10

> test --  NG

masaru@linux-uw5l:/tmp/tes/emacs> make clean

masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> Bisecting: 90 revisions left to test after this (roughly 7 steps)
> [06a872b71d6580e0699600844bb9526602437990] Fix redisplay on a TTY after 'make-frame'
> [...]
masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10

> test  --  OK

masaru@linux-uw5l:/tmp/tes/emacs> make clean

masaru@linux-uw5l:/tmp/tes/emacs> git bisect good
> Bisecting: 0 revisions left to test after this (roughly 1 step)
> [ec10ef95cf2a5309ec7cb74f1383860a13daa546] * lisp/apropos.el (apropos-variable): Doc fix.  (Bug#22813).

masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10

> test  --  NG

masaru@linux-uw5l:/tmp/tes/emacs> make clean

masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> Bisecting: 0 revisions left to test after this (roughly 0 steps)
> [d2dd614716e34edb5891e58c029741cd6b32217d] Remove unneeded workaround in xftfont.c

> ---------------------------------------------------------------------

> ※ test = Start Emacs & Wanderlust, then try to open the folder in
> Wanderlust. OK means "can open without any problem", NG means "Emacs
> eat 100% CPU).

> ---
>  Masaru Nomiya             mail-to: nomiya @ galaxy.dti.ne.jp

bug#23368: 25.0.92; xftfont.c has got a bug

[Eli Zaretskii]

> Date: Mon, 25 Apr 2016 15:02:40 +0900
> From: Masaru Nomiya <nomiya@galaxy.dti.ne.jp>
> 
> Hello,
> 
> With Emacs 25.0.92, Emacs 25.0.93, and Emacs 25.1.50, Emacs start to
> use 100% CPU.
> That is, afetr I start Wanderlust(MUA) on Emacs, and try to open some
> folders in it, Emacs starts to use 100% CPU, and unresposible.
> 
> I investigated the latest Emacs with git bisect.
> As a result, I found that xftfont.c has got a bug as follows.

CC'ing Handa-san.  This seems to be a duplicate of bug#23360.  Can you
tell which font(s) are used by Wanderlust when this happens?

> --------------------------------------------------------------------
> masaru@linux-uw5l:/tmp/tes/emacs> git checkout emacs-25
> Branch emacs-25 set up to track remote branch emacs-25 from origin.
> Switched to a new branch 'emacs-25'
> 
> masaru@linux-uw5l:/tmp/tes/emacs> git bisect start
> 
> masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> 
> masaru@linux-uw5l:/tmp/tes/emacs> git bisect good emacs-25.0.91
> Bisecting: 182 revisions left to test after this (roughly 8 steps)
> [04f5525fbb961f6006280be340ca0f03df9e80a7] Make '.' associative, for easier sexp navigation
> 
> masaru@linux-uw5l:/tmp/tes/emacs> ./autogen.sh
> 
> masaru@linux-uw5l:/tmp/tes/emacs> ./configure
> 
> masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10
> 
> test --  NG
> 
> masaru@linux-uw5l:/tmp/tes/emacs> make clean
> 
> masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> Bisecting: 90 revisions left to test after this (roughly 7 steps)
> [06a872b71d6580e0699600844bb9526602437990] Fix redisplay on a TTY after 'make-frame'
> [...]
> masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10
> 
> test  --  OK
> 
> masaru@linux-uw5l:/tmp/tes/emacs> make clean
> 
> masaru@linux-uw5l:/tmp/tes/emacs> git bisect good
> Bisecting: 0 revisions left to test after this (roughly 1 step)
> [ec10ef95cf2a5309ec7cb74f1383860a13daa546] * lisp/apropos.el (apropos-variable): Doc fix.  (Bug#22813).
> 
> masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10
> 
> test  --  NG
> 
> masaru@linux-uw5l:/tmp/tes/emacs> make clean
> 
> masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> Bisecting: 0 revisions left to test after this (roughly 0 steps)
> [d2dd614716e34edb5891e58c029741cd6b32217d] Remove unneeded workaround in xftfont.c
> 
> ---------------------------------------------------------------------
> 
> ※ test = Start Emacs & Wanderlust, then try to open the folder in
> Wanderlust. OK means "can open without any problem", NG means "Emacs
> eat 100% CPU).

Thanks, I hope Handa-san will be able to suggest a solution.  Failing
that, the only way out of this mess I can see is introduce a variable
that will control whether we employ that workaround in xftfont.c, and
let users whose fonts need this customize that variable.

bug#23368: 25.0.92; xftfont.c has got a bug

[fxiny]

[-- Attachment #1: Type: text/plain, Size: 886 bytes --]

hi 
that sounds familiar to me

are you using .emacs.desktop ?
cause i use it and i have two situations at times
yes i do use gnus wlust and mu4e

1) problems with monaco fonts overlapping as for bug 23360
temporarily solved by using xftfont.c from pretest 0.91
or a 2016-02-14 git clone i keep around

2) at times same 100 irresponsive cpu and swap loud grinding ;)
in a word : box completely freezes 
i can only kill it from remote if i am already logged
otherwise i can only hard reset
yes : terrible bug

you are giving me some clues

that is something i am chasing from some time

to get my box alive i have to delete .emacs.desktop
so i wonder if you can do the same
and if we can blame  xftfont.c or something else

in my case the freezes start if i
ctrl right mouse popping the menu up
options -> set default fonts 
to skip the the monaco fonts that as i said was overlapping



[-- Attachment #2: Type: text/html, Size: 1000 bytes --]

bug#23368: 25.0.92; xftfont.c has got a bug

[Masaru Nomiya]

Hello,

In the Message; 

  Subject    : bug#23368: 25.0.92; xftfont.c has got a bug
  Message-ID : <781740024.9864121461571328929.JavaMail.httpd@webmail-45.iol.local>
  Date & Time: Mon, 25 Apr 2016 10:02:08 +0200 (CEST)

fxiny <wtimew@libero.it> has written:

> hi
> that sounds familiar to me

> are you using .emacs.desktop ?

No, I don't.

> cause i use it and i have two situations at times
> yes i do use gnus wlust and mu4e

> 1) problems with monaco fonts overlapping as for bug 23360
> temporarily solved by using xftfont.c from pretest 0.91
> or a 2016-02-14 git clone i keep around

I didn't investigate which font(s?) causeed the problem.

> 2) at times same 100 irresponsive cpu and swap loud grinding ;)
> in a word : box completely freezes
> i can only kill it from remote if i am already logged
> otherwise i can only hard reset
> yes : terrible bug

Do you?

> you are giving me some clues

> that is something i am chasing from some time

> to get my box alive i have to delete .emacs.desktop
> so i wonder if you can do the same
> and if we can blame  xftfont.c or something else

> in my case the freezes start if i
> ctrl right mouse popping the menu up
> options -> set default fonts
> to skip the the monaco fonts that as i said was overlapping

I see.

Thanks,

---
 Masaru Nomiya             mail-to: nomiya @ galaxy.dti.ne.jp

bug#23368: 25.0.92; xftfont.c has got a bug

[Eli Zaretskii]

Ping!

> Date: Mon, 25 Apr 2016 11:00:36 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: 23368@debbugs.gnu.org
> 
> > Date: Mon, 25 Apr 2016 15:02:40 +0900
> > From: Masaru Nomiya <nomiya@galaxy.dti.ne.jp>
> > 
> > Hello,
> > 
> > With Emacs 25.0.92, Emacs 25.0.93, and Emacs 25.1.50, Emacs start to
> > use 100% CPU.
> > That is, afetr I start Wanderlust(MUA) on Emacs, and try to open some
> > folders in it, Emacs starts to use 100% CPU, and unresposible.
> > 
> > I investigated the latest Emacs with git bisect.
> > As a result, I found that xftfont.c has got a bug as follows.
> 
> CC'ing Handa-san.  This seems to be a duplicate of bug#23360.  Can you
> tell which font(s) are used by Wanderlust when this happens?
> 
> > --------------------------------------------------------------------
> > masaru@linux-uw5l:/tmp/tes/emacs> git checkout emacs-25
> > Branch emacs-25 set up to track remote branch emacs-25 from origin.
> > Switched to a new branch 'emacs-25'
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> git bisect start
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> git bisect good emacs-25.0.91
> > Bisecting: 182 revisions left to test after this (roughly 8 steps)
> > [04f5525fbb961f6006280be340ca0f03df9e80a7] Make '.' associative, for easier sexp navigation
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> ./autogen.sh
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> ./configure
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10
> > 
> > test --  NG
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> make clean
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> > Bisecting: 90 revisions left to test after this (roughly 7 steps)
> > [06a872b71d6580e0699600844bb9526602437990] Fix redisplay on a TTY after 'make-frame'
> > [...]
> > masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10
> > 
> > test  --  OK
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> make clean
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> git bisect good
> > Bisecting: 0 revisions left to test after this (roughly 1 step)
> > [ec10ef95cf2a5309ec7cb74f1383860a13daa546] * lisp/apropos.el (apropos-variable): Doc fix.  (Bug#22813).
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> make bootstrap -j10
> > 
> > test  --  NG
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> make clean
> > 
> > masaru@linux-uw5l:/tmp/tes/emacs> git bisect bad
> > Bisecting: 0 revisions left to test after this (roughly 0 steps)
> > [d2dd614716e34edb5891e58c029741cd6b32217d] Remove unneeded workaround in xftfont.c
> > 
> > ---------------------------------------------------------------------
> > 
> > ※ test = Start Emacs & Wanderlust, then try to open the folder in
> > Wanderlust. OK means "can open without any problem", NG means "Emacs
> > eat 100% CPU).
> 
> Thanks, I hope Handa-san will be able to suggest a solution.  Failing
> that, the only way out of this mess I can see is introduce a variable
> that will control whether we employ that workaround in xftfont.c, and
> let users whose fonts need this customize that variable.
> 
> 
> 
> 

bug#23368: 25.0.92; xftfont.c has got a bug

[handa]

In article <83zis2e0ob.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

> Ping!

Oops, sorry for the late response.

> > > Date: Mon, 25 Apr 2016 15:02:40 +0900
> > > From: Masaru Nomiya <nomiya@galaxy.dti.ne.jp>
> > > 
> > > Hello,
> > > 
> > > With Emacs 25.0.92, Emacs 25.0.93, and Emacs 25.1.50, Emacs start to
> > > use 100% CPU.
> > > That is, afetr I start Wanderlust(MUA) on Emacs, and try to open some
> > > folders in it, Emacs starts to use 100% CPU, and unresposible.
> > > 
> > > I investigated the latest Emacs with git bisect.
> > > As a result, I found that xftfont.c has got a bug as follows.
> > 
> > CC'ing Handa-san.  This seems to be a duplicate of bug#23360.  Can you
> > tell which font(s) are used by Wanderlust when this happens?

I re-read the latest xftfont_open of xftfont.c, but couldn't find a
problem.  And, as I'm not using Wanderlust, I couldn't reproduce the
problem.

Could someone please find a way to reproduce the problem on GNU/Linux
system by Emacs started with -Q.

---
K. Handa
handa@gnu.org

bug#23368: 25.0.92; xftfont.c has got a bug

[Masaru Nomiya]

Hello,

In the Message; 

  Subject    : Re: bug#23368: 25.0.92; xftfont.c has got a bug
  Message-ID : <87vb2n2sug.fsf@gnu.org>
  Date & Time: Mon, 09 May 2016 23:45:11 +0900

handa <handa@gnu.org> has written:

> In article <83zis2e0ob.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

> > Ping!

> Oops, sorry for the late response.

> > > > Date: Mon, 25 Apr 2016 15:02:40 +0900
> > > > From: Masaru Nomiya <nomiya@galaxy.dti.ne.jp>
> > > > 
> > > > Hello,
> > > > 
> > > > With Emacs 25.0.92, Emacs 25.0.93, and Emacs 25.1.50, Emacs start to
> > > > use 100% CPU.
> > > > That is, afetr I start Wanderlust(MUA) on Emacs, and try to open some
> > > > folders in it, Emacs starts to use 100% CPU, and unresposible.
> > > > 
> > > > I investigated the latest Emacs with git bisect.
> > > > As a result, I found that xftfont.c has got a bug as follows.
> > > 
> > > CC'ing Handa-san.  This seems to be a duplicate of bug#23360.  Can you
> > > tell which font(s) are used by Wanderlust when this happens?

> I re-read the latest xftfont_open of xftfont.c, but couldn't find a
> problem.  And, as I'm not using Wanderlust, I couldn't reproduce the
> problem.

Sorry for bother you.

> Could someone please find a way to reproduce the problem on GNU/Linux
> system by Emacs started with -Q.

I tested with -nw. I didn't get any problem. I.e. Wanderlust works
fine for me.

Is this give you any hint?

Thanks,

---
M. Nomiya

bug#23368: 25.0.92; xftfont.c has got a bug

[Eli Zaretskii]

> Date: Mon, 16 May 2016 16:52:54 +0900
> From: Masaru Nomiya <nomiya@galaxy.dti.ne.jp>
> Cc: eliz@gnu.org, 23368@debbugs.gnu.org
> 
> > > > CC'ing Handa-san.  This seems to be a duplicate of bug#23360.  Can you
> > > > tell which font(s) are used by Wanderlust when this happens?
> 
> > I re-read the latest xftfont_open of xftfont.c, but couldn't find a
> > problem.  And, as I'm not using Wanderlust, I couldn't reproduce the
> > problem.
> 
> Sorry for bother you.
> 
> > Could someone please find a way to reproduce the problem on GNU/Linux
> > system by Emacs started with -Q.
> 
> I tested with -nw. I didn't get any problem. I.e. Wanderlust works
> fine for me.
> 
> Is this give you any hint?

Thanks, but using -nw doesn't help with this problem.  The -nw session
doesn't use any font shaping engines, it uses text-mode stdio writes
to the screen, so using it doesn't add any useful information about
this problem.

I believe Handa-san asked for "emacs -Q", not "emacs -nw".  Can you
try that?

bug#23368: 25.0.92; xftfont.c has got a bug

[Eli Zaretskii]

> From: handa <handa@gnu.org>
> Cc: nomiya@galaxy.dti.ne.jp, 23368@debbugs.gnu.org
> Date: Mon, 09 May 2016 23:45:11 +0900
> 
> > > CC'ing Handa-san.  This seems to be a duplicate of bug#23360.  Can you
> > > tell which font(s) are used by Wanderlust when this happens?
> 
> I re-read the latest xftfont_open of xftfont.c, but couldn't find a
> problem.  And, as I'm not using Wanderlust, I couldn't reproduce the
> problem.
> 
> Could someone please find a way to reproduce the problem on GNU/Linux
> system by Emacs started with -Q.

Since we want to release Emacs 25.1 soon, I've added a variable,
xft-font-ascent-descent-override, that allows to work around the
problems reported in this and related bugs.  By default, the variable
is nil, which disables overriding the font ascent and descent values,
as this seems to be unnecessary on most systems, and produces annoying
results with some fonts.  On those systems where the override is still
needed, users can set the variable to a non-nil value to solve the
rendering problems.

With that, I consider the bug fixed for Emacs 25.1, and I'm removing
it from the list of bugs that block the release.

We would still want to find a proper fix for these problems, when
their cause becomes known.

Thanks.

bug#23368: Solved! (Was: Re: bug#23368: 25.0.92; xftfont.c has got a bug)

[野宮 賢 / NOMIYA Masaru]

Hello,

In the Message; 

  Subject    : Re: bug#23368: 25.0.92; xftfont.c has got a bug
  Message-ID : <87vb2n2sug.fsf@gnu.org>
  Date & Time: Mon, 09 May 2016 23:45:11 +0900

handa <handa@gnu.org> has written:

> In article <83zis2e0ob.fsf@gnu.org>, Eli Zaretskii <eliz@gnu.org> writes:

> > Ping!

> Oops, sorry for the late response.

> > > > Date: Mon, 25 Apr 2016 15:02:40 +0900
> > > > From: Masaru Nomiya <nomiya@galaxy.dti.ne.jp>
> > > > 
> > > > Hello,
> > > > 
> > > > With Emacs 25.0.92, Emacs 25.0.93, and Emacs 25.1.50, Emacs start to
> > > > use 100% CPU.
> > > > That is, afetr I start Wanderlust(MUA) on Emacs, and try to open some
> > > > folders in it, Emacs starts to use 100% CPU, and unresposible.
> > > > 
> > > > I investigated the latest Emacs with git bisect.
> > > > As a result, I found that xftfont.c has got a bug as follows.
> > > 
> > > CC'ing Handa-san.  This seems to be a duplicate of bug#23360.  Can you
> > > tell which font(s) are used by Wanderlust when this happens?

> I re-read the latest xftfont_open of xftfont.c, but couldn't find a
> problem.  And, as I'm not using Wanderlust, I couldn't reproduce the
> problem.

> Could someone please find a way to reproduce the problem on GNU/Linux
> system by Emacs started with -Q.

I checked all the related files one by one. :-)
I found thst the CPU 100% usage's cause is fontconfig.
In my system, the fontconfig is 2.11.1.
I reverted this to 2.10.2's fontconfig, then problem has disappeared!

I wonder if the 2.11.1's fontconfig has got a bug?
Maybe not.

But waht's the matter, I wonder.

Thanks,

---
 Masaru Nomiya             mail-to: nomiya @ galaxy.dti.ne.jp

bug#23368: Solved! (Was: Re: bug#23368: 25.0.92; xftfont.c has got a bug)

[Eli Zaretskii]

> Date: Sat, 28 May 2016 19:17:50 +0900
> From: 野宮　賢 / NOMIYA Masaru
>  <nomiya@galaxy.dti.ne.jp>
> Cc: eliz@gnu.org, 23368@debbugs.gnu.org
> 
> I checked all the related files one by one. :-)
> I found thst the CPU 100% usage's cause is fontconfig.
> In my system, the fontconfig is 2.11.1.
> I reverted this to 2.10.2's fontconfig, then problem has disappeared!
> 
> I wonder if the 2.11.1's fontconfig has got a bug?
> Maybe not.
> 
> But waht's the matter, I wonder.

Can you try the latest sources from the emacs-25 branch of the Emacs
repository, and see if setting xft-font-ascent-descent-override to a
non-nil value solves the problem even when fontconfig 2.11.1 is used?

Thanks.

bug#23368: Solved! (Was: Re: bug#23368: 25.0.92; xftfont.c has got a bug)

[Eli Zaretskii]

> Date: Sat, 28 May 2016 19:44:13 +0900
> From: 野宮　賢 / NOMIYA Masaru
>  <nomiya@galaxy.dti.ne.jp>
> 
> > Can you try the latest sources from the emacs-25 branch of the Emacs
> > repository, and see if setting xft-font-ascent-descent-override to a
> > non-nil value solves the problem even when fontconfig 2.11.1 is used?
> 
> Yes, solved!

Thanks.  So I'm marking the bug as done.