(toplevel patch) Deal with multilib.out. Really.

(toplevel patch) Deal with multilib.out. Really.

[Nathanael Nerode]

Final version.  This is currently being tested to see if it generates any of
the known problems generated by previous tries.  If it's OK, can I commit it?

I should also add multilib.out to a clean target, but I can't stand to 
even think about it tonight.

	* Makefile.tpl: There is only one multilib.out.  Rebuild it with
	move-if-change.
	* Makefile.in: Regenerate.
	* configure.in: Remove unneeded stuff.

Index: Makefile.tpl
===================================================================
RCS file: /cvs/gcc/gcc/Makefile.tpl,v
retrieving revision 1.21
diff -u -r1.21 Makefile.tpl
--- Makefile.tpl	19 Dec 2002 20:23:50 -0000	1.21
+++ Makefile.tpl	20 Dec 2002 02:20:12 -0000
@@ -784,7 +784,6 @@
 # built are.
 TAGS: do-TAGS
 
-
 # --------------------------------------
 # Modules which run on the build machine
 # --------------------------------------
@@ -968,13 +967,10 @@
 maybe-configure-target-[+module+]:
 configure-target-[+module+]: $(TARGET_SUBDIR)/[+module+]/Makefile
 
-# Don't manually override CC_FOR_TARGET at make time; get it set right
-# at configure time.  Otherwise multilibs may be wrong.
-$(TARGET_SUBDIR)/[+module+]/multilib.out: maybe-all-gcc
+# There's only one multilib.out.  Cleverer subdirs shouldn't need it copied.
+$(TARGET_SUBDIR)/[+module+]/multilib.out: multilib.out
 	@[ -d $(TARGET_SUBDIR)/[+module+] ] || mkdir $(TARGET_SUBDIR)/[+module+];\
-	r=`${PWD}`; export r; \
-	echo "Configuring multilibs for [+module+]"; \
-	$(CC_FOR_TARGET) --print-multi-lib > $(TARGET_SUBDIR)/[+module+]/multilib.out 2> /dev/null
+	cp multilib.out $(TARGET_SUBDIR)/[+module+]/multilib.out
 
 $(TARGET_SUBDIR)/[+module+]/Makefile: config.status $(TARGET_SUBDIR)/[+module+]/multilib.out
 	@[ -d $(TARGET_SUBDIR)/[+module+] ] || mkdir $(TARGET_SUBDIR)/[+module+];\
@@ -1394,6 +1390,21 @@
 # --------------------------------
 # Regenerating top level configury
 # --------------------------------
+
+# Multilib.out tells target dirs what multilibs they should build.
+# There is really only one copy.  We use the 'timestamp' method to
+# work around various timestamp bugs on some systems.
+# We use move-if-change so that it's only considered updated when it
+# actually changes, because it has to depend on a phony target.
+multilib.out: multilib.ts
+	@true
+
+multilib.ts: maybe-all-gcc
+	r=`${PWD}`; export r; \
+	echo "Checking multilib configuration..."; \
+	$(CC_FOR_TARGET) --print-multi-lib > multilib.tmp 2> /dev/null ; \
+	$(SHELL) $(srcdir)/move-if-change multilib.tmp  $(TARGET_SUBDIR)/[+module+]/multilib.out ; \
+	echo timestamp > multilib.ts
 
 # Rebuilding Makefile.in, using autogen.
 $(srcdir)/Makefile.in: # $(srcdir)/Makefile.tpl $(srcdir)/Makefile.def
Index: configure.in
===================================================================
RCS file: /cvs/gcc/gcc/configure.in,v
retrieving revision 1.196
diff -u -r1.196 configure.in
--- configure.in	19 Dec 2002 20:23:50 -0000	1.196
+++ configure.in	20 Dec 2002 02:20:21 -0000
@@ -1698,12 +1698,6 @@
 qqCXX_FOR_TARGET=`echo "$qCXX_FOR_TARGET" | sed -e 's,[$][$],$$$$,g'`
 qqRAW_CXX_FOR_TARGET=`echo "$qRAW_CXX_FOR_TARGET" | sed -e 's,[$][$],$$$$,g'`
 
-# BASE_CC_FOR_TARGET is the actual program file of CC_FOR_TARGET,
-# for use in dependencies.  Replace $$r with a relative
-# path if it's present.
-set dummy $CC_FOR_TARGET; cc_for_target_first_word=$2
-BASE_CC_FOR_TARGET=`echo "$cc_for_target_first_word" | sed -e 's,[$][$]r,.,'`
-
 sedtemp=sed.$$
 cat >$sedtemp <<EOF
 s%@GDB_TK@%${GDB_TK}%
@@ -1722,7 +1716,6 @@
 s%@target_configargs@%${targargs}%
 s%@FLAGS_FOR_TARGET@%${FLAGS_FOR_TARGET}%
 s%@CC_FOR_TARGET@%\$(STAGE_CC_WRAPPER) ${CC_FOR_TARGET}%
-s%@BASE_CC_FOR_TARGET@%${BASE_CC_FOR_TARGET}%
 s%@GCJ_FOR_TARGET@%\$(STAGE_CC_WRAPPER) ${GCJ_FOR_TARGET}%
 s%@CXX_FOR_TARGET@%\$(STAGE_CC_WRAPPER) ${qCXX_FOR_TARGET}%
 s%@RAW_CXX_FOR_TARGET@%\$(STAGE_CC_WRAPPER) ${qRAW_CXX_FOR_TARGET}%

Re: (toplevel patch) Deal with multilib.out. Really.

[Raja R Harinath]

Hi,

Nathanael Nerode <neroden@twcny.rr.com> writes:

> +# Multilib.out tells target dirs what multilibs they should build.
> +# There is really only one copy.  We use the 'timestamp' method to
> +# work around various timestamp bugs on some systems.
> +# We use move-if-change so that it's only considered updated when it
> +# actually changes, because it has to depend on a phony target.
> +multilib.out: multilib.ts
> +	@true

To round out this idiom, this should be

  multilib.out: multilib.ts
	@if test -f multilib.out; then :; \
        else rm -f multilib.ts; $(MAKE) multilib.ts; fi

So that

  rm multilib.out
  make

works.  

Also, IIRC discussions on the automake list about the use of 'then :;
else' form rather than 'test ! -f' -- some shells use the exit code of
the 'if' test when they fall through a missing 'else' clause.

- Hari
-- 
Raja R Harinath ------------------------------ harinath@cs.umn.edu