Merge branch 'maint'

* maint:
  config headers: remove stale comment in makefile fragment
  NEWS: wording and quoting fixlets in few older entries
  config headers: don't emit rules for headers not generated by autoheader
  docs: fix minor typo: s/expending/expanding/
  sync: update files from upstream with "make fetch"
  maint: post-release minor version bump
  maint: typo fixes s/lies into/lies in/
  release: stable release 1.12.4
  NEWS: minor fix

Signed-off-by: Stefano Lattarini <stefano.lattarini@gmail.com>

8 files changed, 170 insertions, 84 deletions

diff --git a/NEWS b/NEWS
index 46aa8cd3d..15c93326d 100644
--- a/NEWS
+++ b/NEWS
@@ -149,7 +149,7 @@ New in 1.13:
 
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-New in 1.12.4:
+New in 1.12.5:
 
 * WARNING: Future backward-incompatibilities!
 
@@ -205,13 +205,25 @@ New in 1.12.4:
     use the options '--automake-acdir' and '--system-acdir' instead
     (which have been introduced in Automake 1.11.2).
 
-  - The 'missing' script will not try anymore to update the timestamp
+  - The 'missing' script will no longer try to update the timestamp
     of out-of-date files that require a maintainer-specific tool to be
     remade, in case the user lacks such a tool (or has a too-old version
     of it).  In fact, starting from Automake 1.13, all it'll do will be
     giving more useful warnings than a bare "command not found" from a
     make recipe would.
 
+Bugs fixed in 1.12.5:
+
+* Long-standing bugs:
+
+  - Automake no longer generates spurious remake rules invoking autoheader
+    to regenerate the template corresponding to header files specified after
+    the first one in AC_CONFIG_HEADERS (automake bug#12495).
+
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+New in 1.12.4:
+
 * Warnings and deprecations:
 
   - Warnings in the 'obsolete' category are enabled by default both in
@@ -227,9 +239,8 @@ New in 1.12.3:
 
 * Miscellaneous changes:
 
-  - The '.m4' files provided by Automake does not define serial numbers
-    anymore.  This should cause no difference in the behaviour of aclocal
-    though.
+  - The '.m4' files provided by Automake no longer define serial numbers.
+    This should cause no difference in the behaviour of aclocal though.
 
   - Some testsuite weaknesses and spurious failures have been fixed.
 
@@ -322,7 +333,7 @@ New in 1.12.1:
   - Use of the long-deprecated two- and three-arguments invocation forms
     of the AM_INIT_AUTOMAKE macro now elicits a warning in the 'obsolete'
     category.  Starting from some future major Automake release (likely
-    post-1.13), such usages won't be allowed anymore.
+    post-1.13), such usages will be no longer allowed.
 
   - Support for the "Cygnus-style" trees (enabled by the 'cygnus' option) is
     now deprecated (its use triggers a warning in the 'obsolete' category).
@@ -336,8 +347,8 @@ New in 1.12.1:
 * Miscellaneous changes:
 
   - The Automake test cases now require a proper POSIX-conforming shell.
-    Older non-POSIX Bourne shells (like Solaris 10 /bin/sh) won't be
-    accepted anymore.  In most cases, the user shouldn't have to specify
+    Older non-POSIX Bourne shells (like Solaris 10 /bin/sh) will be no
+    longer accepted.  In most cases, the user shouldn't have to specify
     such POSIX shell explicitly, since it will be looked up at configure
     time.  Still, when this lookup fails, or when the user wants to
     override its conclusion, the variable 'AM_TEST_RUNNER_SHELL' can be
@@ -448,7 +459,7 @@ New in 1.12:
     the '--add-missing' option, or manually copy the 'test-driver' script
     into their tree.  The second, and more important, implication is that
     now, when the 'parallel-tests' option is in use, TESTS_ENVIRONMENT can
-    not be used anymore to define a test runner, and the command specified
+    no longer be used to define a test runner, and the command specified
     in LOG_COMPILER (and <ext>_LOG_COMPILER) must be a *real* executable
     program or script.  For example, this is still a valid usage (albeit
     a little contorted):
@@ -461,7 +472,7 @@ New in 1.12:
         fi;
       LOG_COMPILER = $(SHELL) $$maybe_errexit
 
-    while this is not anymore:
+    OTOH, this is no longer a valid usage:
 
       TESTS_ENVIRONMENT = \
         $(SHELL) `test -n '$(STRICT_TESTS_CHECKING)' && echo ' -e'`
@@ -562,9 +573,9 @@ Bugs fixed in 1.12:
   - The AM_COND_IF macro also works if the shell expression for the
     conditional is no longer valid for the condition.
 
-  - The automake-provided parallel testsuite harness does not fail anymore
-    with BSD make used in parallel mode when there are test scripts in a
-    subdirectory, like in:
+  - The automake-provided parallel testsuite harness no longer fails
+    with BSD make used in parallel mode when there are test scripts in
+    a subdirectory, like in:
 
       TESTS = sub/foo.test sub/bar.test
 
@@ -639,7 +650,7 @@ Bugs fixed in 1.11.4:
 * Bugs introduced by 1.11.2:
 
   - A definition of 'noinst_PYTHON' before 'python_PYTHON' (or similar)
-    don't cause spurious failures upon "make install" anymore.
+    no longer cause spurious failures upon "make install".
 
   - The user can now instruct the 'uninstall-info' rule not to update
     the '${infodir}/dir' file by exporting the environment variable
@@ -657,9 +668,9 @@ Bugs fixed in 1.11.4:
     '-I' is non-existent, aclocal will now create it before trying to copy
     files in it.
 
-  - An empty declaration of a "foo_PRIMARY" don't cause anymore the
-    generated install rules to create an empty $(foodir) directory;
-    for example, if Makefile.am contains something like:
+  - An empty declaration of a "foo_PRIMARY" no longer cause the generated
+    install rules to create an empty $(foodir) directory; for example, if
+    Makefile.am contains something like:
 
       pkglibexec_SCRIPTS =
       if FALSE
@@ -677,15 +688,15 @@ New in 1.11.3:
   - Automake's own build system is more silent by default, making use of
     the 'silent-rules' option.
 
-  - The master copy of the `gnupload' script is now maintained in gnulib,
+  - The master copy of the 'gnupload' script is now maintained in gnulib,
     not in automake.
 
-  - The `missing' script doesn't try to wrap calls to `tar' anymore.
+  - The 'missing' script no longer tries to wrap calls to 'tar'.
 
-  - "make dist" doesn't wrap `tar' invocations with the `missing' script
-    anymore.  Similarly, the obsolescent variable `$(AMTAR)' (which you
-    shouldn't be using BTW ;-) does not invoke the missing script anymore
-    to wrap tar, but simply invokes the `tar' program itself.
+  - "make dist" no longer wraps 'tar' invocations with the 'missing'
+    script.  Similarly, the obsolescent variable '$(AMTAR)' (which you
+    shouldn't be using BTW ;-) no longer invokes the 'missing' script
+    to wrap tar, but simply invokes the 'tar' program itself.
 
   - "make dist" can now create lzip-compressed tarballs.
 
@@ -711,24 +722,24 @@ Bugs fixed in 1.11.3:
 * Bugs introduced by 1.11.2:
 
   - Automake now correctly recognizes the prefix/primary combination
-   `pkglibexec_SCRIPTS' as valid.
+   'pkglibexec_SCRIPTS' as valid.
 
-  - The parallel-tests harness doesn't trip anymore on sed implementations
+  - The parallel-tests harness now doesn't trip on sed implementations
     with stricter limits on the length of input lines (problem seen at
     least on Solaris 8).
 
 * Long-standing bugs:
 
   - The "deleted header file problem" for *.am files is avoided by stub
-    rules.  This allows `make' to trigger a rerun of `automake' also if
-    some previously needed `.am' file has been removed.
+    rules.  This allows 'make' to trigger a rerun of 'automake' also if
+    some previously needed '.am' file has been removed.
 
-  - The `silent-rules' option now generates working makefiles even
-    for the uncommon `make' implementations that do not support the
-    nested-variables extension to POSIX 2008.  For such `make'
+  - The 'silent-rules' option now generates working makefiles even
+    for the uncommon 'make' implementations that do not support the
+    nested-variables extension to POSIX 2008.  For such 'make'
     implementations, whether a build is silent is determined at
     configure time, and cannot be overridden at make time with
-    `make V=0' or `make V=1'.
+    "make V=0" or "make V=1".
 
   - Vala support now works better in VPATH setups.
 
diff --git a/THANKS b/THANKS
index ca95db802..88f539c8f 100644
--- a/THANKS
+++ b/THANKS
@@ -143,6 +143,7 @@ Harald Dunkel                   harald@CoWare.com
 Harlan Stenn                    Harlan.Stenn@pfcs.com
 He Li                           tippa000@yahoo.com
 Henrik Frystyk Nielsen          frystyk@w3.org
+Hib Eris                        hib@hiberis.nl
 Ian Lance Taylor                ian@cygnus.com
 Ignacy Gawedzki                 i@lri.fr
 Илья Н. Голубев                 gin@mo.msk.ru
diff --git a/automake.in b/automake.in
index b16585508..9c0c8b9cc 100644
--- a/automake.in
+++ b/automake.in
@@ -2535,8 +2535,8 @@ sub handle_programs
       my($xlink, $vlink) = &define_per_target_linker_variable ($linker, $xname);
       $vlink = verbose_flag ($vlink || 'GEN');
 
-      # If the resulting program lies into a subdirectory,
-      # make sure this directory will exist.
+      # If the resulting program lies in a subdirectory,
+      # ensure that the directory exists before we need it.
       my $dirstamp = require_build_directory_maybe ($one_file);
 
       $libtool_clean_directories{dirname ($one_file)} = 1;
@@ -2646,7 +2646,7 @@ sub handle_libraries
       &handle_source_transform ($xlib, $onelib, $obj, $where,
 				NONLIBTOOL => 1, LIBTOOL => 0);
 
-      # If the resulting library lies into a subdirectory,
+      # If the resulting library lies in a subdirectory,
       # make sure this directory will exist.
       my $dirstamp = require_build_directory_maybe ($onelib);
       my $verbose = verbose_flag ('AR');
@@ -2890,7 +2890,7 @@ sub handle_ltlibraries
 	    }
 	}
 
-      # If the resulting library lies into a subdirectory,
+      # If the resulting library lies in a subdirectory,
       # make sure this directory will exist.
       my $dirstamp = require_build_directory_maybe ($onelib);
 
@@ -3082,7 +3082,7 @@ sub output_texinfo_build_rules ($$$@)
   # extension).
   my $generic_info = ($generic && $dsfx) ? 1 : 0;
 
-  # If the resulting file lie into a subdirectory,
+  # If the resulting file lies in a subdirectory,
   # make sure this directory will exist.
   my $dirstamp = require_build_directory_maybe ($dest);
 
@@ -4114,6 +4114,7 @@ sub handle_configure ($$$@)
 	    file_contents ('remake-hdr',
 			   new Automake::Location,
 			   FILES            => "@files",
+			   'FIRST-HDR'      => ($hdr_index == 1),
 			   CONFIG_H         => $cn_sans_dir,
 			   CONFIG_HIN       => $ins[0],
 			   CONFIG_H_DEPS    => "@ins",
diff --git a/doc/automake.texi b/doc/automake.texi
index 13db2c1c5..7058545ec 100644
--- a/doc/automake.texi
+++ b/doc/automake.texi
@@ -2956,9 +2956,11 @@ Automake will require the sources file declared with
 macro.
 
 @item AC_CONFIG_HEADERS
-Automake will generate rules to rebuild these headers.  Older versions
-of Automake required the use of @code{AM_CONFIG_HEADER}; this is no
-longer the case, and that macro has indeed been removed.
+Automake will generate rules to rebuild these headers from the
+corresponding templates (usually, the template for a @file{foo.h}
+header being @file{foo.h.in}).  Older versions of Automake
+required the use of @code{AM_CONFIG_HEADER}; this is no longer
+the case, and that macro has indeed been removed.
 
 As with @code{AC_CONFIG_FILES} (@pxref{Requirements}), parts of the
 specification using shell variables will be ignored as far as
@@ -11817,7 +11819,7 @@ Although @samp{$(wildcard ...)} works with GNU @command{make}, it is
 not portable to other @command{make} implementations.
 
 The only way Automake could support @command{$(wildcard ...)} is by
-expending @command{$(wildcard ...)} when @command{automake} is run.
+expanding @command{$(wildcard ...)} when @command{automake} is run.
 The resulting @file{Makefile.in}s would be portable since they would
 list all files and not use @samp{$(wildcard ...)}.  However that
 means developers would need to remember to run @command{automake} each
diff --git a/lib/am/remake-hdr.am b/lib/am/remake-hdr.am
index f61400a79..155e7b4cb 100644
--- a/lib/am/remake-hdr.am
+++ b/lib/am/remake-hdr.am
@@ -14,9 +14,6 @@
 ## You should have received a copy of the GNU General Public License
 ## along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
-## Explicitly look in srcdir for benefit of non-GNU makes.
-
-
 %CONFIG_H%: %STAMP%
 ## Recover from removal of CONFIG_HEADER
 	@if test ! -f $@; then rm -f %STAMP%; else :; fi
@@ -30,7 +27,7 @@
 
 ## Only the first file of AC_CONFIG_HEADERS is assumed to be generated
 ## by autoheader.
-if %?FIRST%
+if %?FIRST-HDR%
 %CONFIG_HIN%: %MAINTAINER-MODE% $(am__configure_deps) %FILES%
 ## Cater to parallel BSD make.
 	($(am__cd) $(top_srcdir) && $(AUTOHEADER))
@@ -71,4 +68,4 @@ if %?FIRST%
 ## by config.status, there is no reason to make things complex for
 ## config.hin.
 	touch $@
-endif %?FIRST%
+endif %?FIRST-HDR%
diff --git a/lib/texinfo.tex b/lib/texinfo.tex
index f458ba726..f3093d085 100644
--- a/lib/texinfo.tex
+++ b/lib/texinfo.tex
@@ -3,7 +3,7 @@
 % Load plain if necessary, i.e., if running under initex.
 \expandafter\ifx\csname fmtname\endcsname\relax\input plain\fi
 %
-\def\texinfoversion{2012-09-04.17}
+\def\texinfoversion{2012-09-12.16}
 %
 % Copyright 1985, 1986, 1988, 1990, 1991, 1992, 1993, 1994, 1995,
 % 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
@@ -2272,8 +2272,6 @@ end
 
 \gdef\markupsetcodequoteleft{\let`\codequoteleft}
 \gdef\markupsetcodequoteright{\let'\codequoteright}
-
-\gdef\markupsetnoligaturesquoteleft{\let`\noligaturesquoteleft}
 }
 
 \let\markupsetuplqcode \markupsetcodequoteleft
@@ -2282,6 +2280,9 @@ end
 \let\markupsetuplqexample \markupsetcodequoteleft
 \let\markupsetuprqexample \markupsetcodequoteright
 %
+\let\markupsetuplqkbd     \markupsetcodequoteleft
+\let\markupsetuprqkbd     \markupsetcodequoteright
+%
 \let\markupsetuplqsamp \markupsetcodequoteleft
 \let\markupsetuprqsamp \markupsetcodequoteright
 %
@@ -2291,8 +2292,6 @@ end
 \let\markupsetuplqverbatim \markupsetcodequoteleft
 \let\markupsetuprqverbatim \markupsetcodequoteright
 
-\let\markupsetuplqkbd \markupsetnoligaturesquoteleft
-
 % Allow an option to not use regular directed right quote/apostrophe
 % (char 0x27), but instead the undirected quote from cmtt (char 0x0d).
 % The undirected quote is ugly, so don't make it the default, but it
@@ -2382,8 +2381,7 @@ end
   \aftersmartic
 }
 
-% like \smartslanted except unconditionally uses \ttsl, and no ic.
-% @var is set to this for defun arguments.
+% Unconditional use \ttsl, and no ic.  @var is set to this for defuns.
 \def\ttslanted#1{{\ttsl #1}}
 
 % @cite is like \smartslanted except unconditionally use \sl.  We never want
@@ -2695,10 +2693,6 @@ end
   \let\email=\uref
 \fi
 
-% @kbd is like @code, except that if the argument is just one @key command,
-% then @kbd has no effect.
-\def\kbd#1{{\setupmarkupstyle{kbd}\def\look{#1}\expandafter\kbdfoo\look??\par}}
-
 % @kbdinputstyle -- arg is `distinct' (@kbd uses slanted tty font always),
 %   `example' (@kbd uses ttsl only inside of @example and friends),
 %   or `code' (@kbd uses normal tty font always).
@@ -2722,11 +2716,17 @@ end
 % Default is `distinct'.
 \kbdinputstyle distinct
 
+% @kbd is like @code, except that if the argument is just one @key command,
+% then @kbd has no effect.
+\def\kbd#1{{\def\look{#1}\expandafter\kbdsub\look??\par}}
+
 \def\xkey{\key}
-\def\kbdfoo#1#2#3\par{\def\one{#1}\def\three{#3}\def\threex{??}%
-\ifx\one\xkey\ifx\threex\three \key{#2}%
-\else{\tclose{\kbdfont\setupmarkupstyle{kbd}\look}}\fi
-\else{\tclose{\kbdfont\setupmarkupstyle{kbd}\look}}\fi}
+\def\kbdsub#1#2#3\par{%
+  \def\one{#1}\def\three{#3}\def\threex{??}%
+  \ifx\one\xkey\ifx\threex\three \key{#2}%
+  \else{\tclose{\kbdfont\setupmarkupstyle{kbd}\look}}\fi
+  \else{\tclose{\kbdfont\setupmarkupstyle{kbd}\look}}\fi
+}
 
 % definition of @key that produces a lozenge.  Doesn't adjust to text size.
 %\setfont\keyrm\rmshape{8}{1000}{OT1}
@@ -3272,7 +3272,8 @@ end
 % Settings used for typesetting titles: no hyphenation, no indentation,
 % don't worry much about spacing, ragged right.  This should be used
 % inside a \vbox, and fonts need to be set appropriately first.  Because
-% it is always used for titles, nothing else, we call \rmisbold.
+% it is always used for titles, nothing else, we call \rmisbold.  \par
+% should be specified before the end of the \vbox, since a vbox is a group.
 % 
 \def\raggedtitlesettings{%
   \rmisbold
@@ -3289,7 +3290,7 @@ end
 
 \parseargdef\title{%
   \checkenv\titlepage
-  \vbox{\titlefonts \raggedtitlesettings #1}
+  \vbox{\titlefonts \raggedtitlesettings #1\par}%
   % print a rule at the page bottom also.
   \finishedtitlepagefalse
   \vskip4pt \hrule height 4pt width \hsize \vskip4pt
@@ -4256,7 +4257,7 @@ end
 }
 \def\ifcmddefinedfail{\doignore{ifcommanddefined}}
 
-% @ifcommandnotdefined CMD ... handlded similar to @ifclear above.
+% @ifcommandnotdefined CMD ... handled similar to @ifclear above.
 \makecond{ifcommandnotdefined}
 \def\ifcommandnotdefined{%
   \parsearg{\doifcmddefined{\else \let\next=\ifcmdnotdefinedfail}}}
@@ -5592,14 +5593,6 @@ end
 
 % Define @majorheading, @heading and @subheading
 
-% NOTE on use of \vbox for chapter headings, section headings, and such:
-%       1) We use \vbox rather than the earlier \line to permit
-%          overlong headings to fold.
-%       2) \hyphenpenalty is set to 10000 because hyphenation in a
-%          heading is obnoxious; this forbids it.
-%       3) Likewise, headings look best if no \parindent is used, and
-%          if justification is not attempted.  Hence \raggedright.
-
 \def\majorheading{%
   {\advance\chapheadingskip by 10pt \chapbreak }%
   \parsearg\chapheadingzzz
@@ -5607,10 +5600,8 @@ end
 
 \def\chapheading{\chapbreak \parsearg\chapheadingzzz}
 \def\chapheadingzzz#1{%
-  {\chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000
-                    \parindent=0pt\ptexraggedright
-                    \rmisbold #1\hfill}}%
-  \bigskip \par\penalty 200\relax
+  \vbox{\chapfonts \raggedtitlesettings #1\par}%
+  \nobreak\bigskip \nobreak
   \suppressfirstparagraphindent
 }
 
@@ -5769,8 +5760,7 @@ end
     %
     % Typeset the actual heading.
     \nobreak % Avoid page breaks at the interline glue.
-    \vbox{\hyphenpenalty=10000 \tolerance=5000 \parindent=0pt \ptexraggedright
-          \hangindent=\wd0 \centerparametersmaybe
+    \vbox{\raggedtitlesettings \hangindent=\wd0 \centerparametersmaybe
           \unhbox0 #1\par}%
   }%
   \nobreak\bigskip % no page break after a chapter title
@@ -5792,18 +5782,18 @@ end
 \def\setchapterstyle #1 {\csname CHAPF#1\endcsname}
 %
 \def\unnchfopen #1{%
-\chapoddpage {\chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000
-                       \parindent=0pt\ptexraggedright
-                       \rmisbold #1\hfill}}\bigskip \par\nobreak
+  \chapoddpage
+  \vbox{\chapfonts \raggedtitlesettings #1\par}%
+  \nobreak\bigskip\nobreak
 }
 \def\chfopen #1#2{\chapoddpage {\chapfonts
 \vbox to 3in{\vfil \hbox to\hsize{\hfil #2} \hbox to\hsize{\hfil #1} \vfil}}%
 \par\penalty 5000 %
 }
 \def\centerchfopen #1{%
-\chapoddpage {\chapfonts \vbox{\hyphenpenalty=10000\tolerance=5000
-                       \parindent=0pt
-                       \hfill {\rmisbold #1}\hfill}}\bigskip \par\nobreak
+  \chapoddpage
+  \vbox{\chapfonts \raggedtitlesettings \hfill #1\hfill}%
+  \nobreak\bigskip \nobreak
 }
 \def\CHAPFopen{%
   \global\let\chapmacro=\chfopen
@@ -7082,7 +7072,10 @@ end
   \df \sl \hyphenchar\font=0
   %
   % On the other hand, if an argument has two dashes (for instance), we
-  % want a way to get ttsl.  Let's try @var for that.
+  % want a way to get ttsl.  We used to recommend @var for that, so
+  % leave the code in, but it's strange for @var to lead to typewriter.
+  % Nowadays we recommend @code, since the difference between a ttsl hyphen
+  % and a tt hyphen is pretty tiny.  @code also disables ?` !`.
   \def\var##1{{\setupmarkupstyle{var}\ttslanted{##1}}}%
   #1%
   \sl\hyphenchar\font=45
diff --git a/t/autohdr-subdir-pr12495.sh b/t/autohdr-subdir-pr12495.sh
new file mode 100755
index 000000000..77d2522bc
--- /dev/null
+++ b/t/autohdr-subdir-pr12495.sh
@@ -0,0 +1,80 @@
+#! /bin/sh
+# Copyright (C) 2012 Free Software Foundation, Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2, or (at your option)
+# any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# Related to automake bug#12495: Automake shouldn't generate useless
+# remake rules for AC_CONFIG_HEADERS arguments after the first one,
+# not even when subdirs are involved.
+
+. ./defs || exit 1
+
+cat >> configure.ac << 'END'
+AC_CONFIG_HEADERS([a.h b.h sub/c.h])
+AC_CONFIG_FILES([sub/Makefile])
+AC_OUTPUT
+END
+
+mkdir sub
+echo SUBDIRS = sub > Makefile.am
+: > sub/Makefile.am
+
+$ACLOCAL
+$AUTOCONF
+$AUTOHEADER
+# Even if an AC_CONFIG_HEADERS invocation is passed several files in
+# the first argument, only the first one is considered by autoheader
+# for automatic generation.  Otherwise, the present test case would
+test -f a.h.in && test ! -f c.h.in && test ! -f sub/c.h.in \
+  || fatal_ "unexpected autoheader behavior with multiple" \
+            "AC_CONFIG_HEADERS arguments"
+# Automake should require the missing headers though.
+AUTOMAKE_fails -Wno-error -Wnone
+grep "^configure\.ac:4:.* required file 'b.h.in' not found" stderr
+grep "^configure\.ac:4:.* required file 'sub/c.h.in' not found" stderr
+: > b.h.in
+: > sub/c.h.in
+$AUTOMAKE
+
+./configure
+
+# Automake should regenerate this.
+grep '^$(srcdir)/a\.h\.in:' Makefile.in
+# But not these.
+grep '[bc]\.h\.in.*:' Makefile.in sub/Makefile.in && exit 1
+
+test -f a.h && test -f b.h && test -f sub/c.h \
+  || fatal_ "unexpected ./configure behavior with multiple" \
+            "AC_CONFIG_HEADERS arguments"
+
+rm -f a.h.in a.h
+$MAKE
+test -f a.h.in
+test -f a.h
+
+ocwd=$(pwd)
+for x in b c; do
+  test $x = b || cd sub
+  rm -f $x.h.in
+  $MAKE $x.h.in 2>stderr && { cat stderr >&2; exit 1; }
+  cat stderr >&2
+  test ! -f $x.h.in
+  if using_gmake; then
+    grep "No rule to make target [\`\"']$x\.h\.in[\`\"']" stderr
+  fi
+  : > $x.h.in
+  cd "$ocwd" || fatal_ "cannot chdir back"
+done
+
+:
diff --git a/t/list-of-tests.mk b/t/list-of-tests.mk
index 3708ef7db..c3b8cf45b 100644
--- a/t/list-of-tests.mk
+++ b/t/list-of-tests.mk
@@ -161,6 +161,7 @@ t/autodist-stamp-vti.sh \
 t/autohdr.sh \
 t/autohdr3.sh \
 t/autohdr4.sh \
+t/autohdr-subdir-pr12495.sh \
 t/autohdrdry.sh \
 t/automake-cmdline.tap \
 t/auxdir.sh \