* [PATCH 1/5] libgloss: convert top level to automake
2022-02-05 5:46 [PATCH 0/5] start converting libgloss to non-recursive automake Mike Frysinger
@ 2022-02-05 5:46 ` Mike Frysinger
2022-02-05 5:46 ` [PATCH 2/5] libgloss: merge doc into top-level Makefile Mike Frysinger
` (4 subsequent siblings)
5 siblings, 0 replies; 14+ messages in thread
From: Mike Frysinger @ 2022-02-05 5:46 UTC (permalink / raw)
To: newlib
The top level dir isn't doing anything interesting, just recursing into
subdirs. So this change isn't terribly exciting. But it sets us up for
doing more fun stuff in follow up commits.
[TODO] Check test targets
---
libgloss/Makefile.am | 38 ++
libgloss/Makefile.in | 842 +++++++++++++++++++++++++++++--------
libgloss/config/default.mh | 3 +
libgloss/configure.ac | 2 +-
4 files changed, 715 insertions(+), 170 deletions(-)
create mode 100644 libgloss/Makefile.am
diff --git a/libgloss/Makefile.am b/libgloss/Makefile.am
new file mode 100644
index 000000000000..0779da76dd76
--- /dev/null
+++ b/libgloss/Makefile.am
@@ -0,0 +1,38 @@
+## Top level makefile for libgloss
+
+ACLOCAL_AMFLAGS = -I . -I .. -I ../config
+
+# Variables that will accumulate in subdirs.
+PHONY =
+
+SUBDIRS = @subdirs@ .
+
+tooldir = $(exec_prefix)/$(target_alias)
+
+## These are roughly topologically sorted in order to make porting more
+## streamlined.
+FLAGS_TO_PASS = \
+ "CC=$(CC)" \
+ "CFLAGS=$(CFLAGS)" \
+ "CFLAGS_FOR_TARGET=$(CFLAGS_FOR_TARGET)" \
+ "CCASFLAGS=$(CCASFLAGS)" \
+ "AR=$(AR)" \
+ "RANLIB=$(RANLIB)" \
+ "AR_FLAGS=$(AR_FLAGS)" \
+ "MAKEINFO=$(MAKEINFO)" \
+ "AS=$(AS)" \
+ "LD=$(LD)" \
+ "CROSS_CFLAGS=$(CROSS_CFLAGS)" \
+ "TARGET_CFLAGS=$(TARGET_CFLAGS)" \
+ "exec_prefix=$(exec_prefix)" \
+ "prefix=$(prefix)" \
+ "tooldir=$(tooldir)" \
+ "infodir=$(infodir)" \
+ "libdir=$(libdir)" \
+ "top_toollibdir=$(toollibdir)" \
+ "INSTALL=$(INSTALL)" \
+ "INSTALL_PROGRAM=$(INSTALL_PROGRAM)" \
+ "INSTALL_DATA=$(INSTALL_DATA)" \
+ "DESTDIR=$(DESTDIR)"
+
+include $(top_srcdir)/../multilib.am
diff --git a/libgloss/config/default.mh b/libgloss/config/default.mh
index 987dc390dc50..4e7f106f9366 100644
--- a/libgloss/config/default.mh
+++ b/libgloss/config/default.mh
@@ -29,3 +29,6 @@ AR_FLAGS = rc
#
.c.s:
$(CC) $(CFLAGS_FOR_TARGET) -S $(INCLUDES) $(CFLAGS) $<
+
+# We don't build docs in subdirs, so stub out the rules.
+.PHONY: doc docs dvi html install-html info install-info clean-info pdf install-pdf
diff --git a/libgloss/configure.ac b/libgloss/configure.ac
index 1fa877729f9b..e4f0b50860b6 100644
--- a/libgloss/configure.ac
+++ b/libgloss/configure.ac
@@ -11,7 +11,7 @@ AC_CONFIG_AUX_DIR(..)
AC_PROG_INSTALL
AC_CANONICAL_SYSTEM
-AM_INIT_AUTOMAKE([foreign no-define 1.15])
+AM_INIT_AUTOMAKE([foreign no-dist no-define subdir-objects 1.15.1])
if test -z "${with_multisubdir}" ; then
AC_CONFIG_FILES([doc/Makefile])
--
2.34.1
^ permalink raw reply [flat|nested] 14+ messages in thread* [PATCH 2/5] libgloss: merge doc into top-level Makefile
2022-02-05 5:46 [PATCH 0/5] start converting libgloss to non-recursive automake Mike Frysinger
2022-02-05 5:46 ` [PATCH 1/5] libgloss: convert top level to automake Mike Frysinger
@ 2022-02-05 5:46 ` Mike Frysinger
2022-02-05 5:46 ` [PATCH 3/5] libgloss: merge libgloss " Mike Frysinger
` (3 subsequent siblings)
5 siblings, 0 replies; 14+ messages in thread
From: Mike Frysinger @ 2022-02-05 5:46 UTC (permalink / raw)
To: newlib
Avoid a recursive make with this tiny subdir to speed things up a bit.
---
libgloss/Makefile.am | 7 +
libgloss/Makefile.in | 403 +++++++++++++++++++++++++++++++++-----
libgloss/configure | 22 ++-
libgloss/configure.ac | 7 +-
libgloss/doc/Makefile.in | 175 -----------------
libgloss/doc/Makefile.inc | 1 +
6 files changed, 388 insertions(+), 227 deletions(-)
delete mode 100644 libgloss/doc/Makefile.in
create mode 100644 libgloss/doc/Makefile.inc
diff --git a/libgloss/Makefile.am b/libgloss/Makefile.am
index 0779da76dd76..81ba50a70f8c 100644
--- a/libgloss/Makefile.am
+++ b/libgloss/Makefile.am
@@ -3,6 +3,7 @@
ACLOCAL_AMFLAGS = -I . -I .. -I ../config
# Variables that will accumulate in subdirs.
+info_TEXINFOS =
PHONY =
SUBDIRS = @subdirs@ .
@@ -36,3 +37,9 @@ FLAGS_TO_PASS = \
"DESTDIR=$(DESTDIR)"
include $(top_srcdir)/../multilib.am
+
+TEXINFO_TEX = ../texinfo/texinfo.tex
+
+if HAVE_DOC
+include doc/Makefile.inc
+endif
diff --git a/libgloss/configure.ac b/libgloss/configure.ac
index e4f0b50860b6..5657f094a197 100644
--- a/libgloss/configure.ac
+++ b/libgloss/configure.ac
@@ -13,9 +13,12 @@ AC_PROG_INSTALL
AC_CANONICAL_SYSTEM
AM_INIT_AUTOMAKE([foreign no-dist no-define subdir-objects 1.15.1])
-if test -z "${with_multisubdir}" ; then
- AC_CONFIG_FILES([doc/Makefile])
+if test -z "${with_multisubdir}"; then
+ have_multisubdir=no
+else
+ have_multisubdir=yes
fi
+AM_CONDITIONAL(HAVE_DOC, test x$have_multisubdir = xno)
AC_NO_EXECUTABLES
diff --git a/libgloss/doc/Makefile.inc b/libgloss/doc/Makefile.inc
new file mode 100644
index 000000000000..fa5e30fbf2bc
--- /dev/null
+++ b/libgloss/doc/Makefile.inc
@@ -0,0 +1 @@
+info_TEXINFOS += %D%/porting.texi
--
2.34.1
^ permalink raw reply [flat|nested] 14+ messages in thread* [PATCH 3/5] libgloss: merge libgloss into top-level Makefile
2022-02-05 5:46 [PATCH 0/5] start converting libgloss to non-recursive automake Mike Frysinger
2022-02-05 5:46 ` [PATCH 1/5] libgloss: convert top level to automake Mike Frysinger
2022-02-05 5:46 ` [PATCH 2/5] libgloss: merge doc into top-level Makefile Mike Frysinger
@ 2022-02-05 5:46 ` Mike Frysinger
2022-02-07 11:14 ` Mike Frysinger
2022-02-05 5:46 ` [PATCH 4/5] libgloss: merge iq2000 " Mike Frysinger
` (2 subsequent siblings)
5 siblings, 1 reply; 14+ messages in thread
From: Mike Frysinger @ 2022-02-05 5:46 UTC (permalink / raw)
To: newlib
Avoid a recursive make to speed things up a bit.
---
libgloss/Makefile.am | 13 ++
libgloss/Makefile.in | 384 ++++++++++++++++++++++++++++-----
libgloss/configure | 18 +-
libgloss/configure.ac | 5 +-
libgloss/libnosys/Makefile.in | 115 ----------
libgloss/libnosys/Makefile.inc | 27 +++
6 files changed, 383 insertions(+), 179 deletions(-)
delete mode 100644 libgloss/libnosys/Makefile.in
create mode 100644 libgloss/libnosys/Makefile.inc
diff --git a/libgloss/Makefile.am b/libgloss/Makefile.am
index 81ba50a70f8c..1209ec9d8d85 100644
--- a/libgloss/Makefile.am
+++ b/libgloss/Makefile.am
@@ -3,6 +3,7 @@
ACLOCAL_AMFLAGS = -I . -I .. -I ../config
# Variables that will accumulate in subdirs.
+check_PROGRAMS =
info_TEXINFOS =
PHONY =
@@ -10,6 +11,15 @@ SUBDIRS = @subdirs@ .
tooldir = $(exec_prefix)/$(target_alias)
+multilibtooldir = $(tooldir)/lib$(MULTISUBDIR)
+multilibtool_DATA =
+multilibtool_LIBRARIES =
+
+# A fake library so automake will generate rules for plain objects that we want
+# to install (e.g. our crt0.o objects).
+noinst_LIBRARIES = libobjs.a
+libobjs_a_SOURCES =
+
## These are roughly topologically sorted in order to make porting more
## streamlined.
FLAGS_TO_PASS = \
@@ -43,3 +53,6 @@ TEXINFO_TEX = ../texinfo/texinfo.tex
if HAVE_DOC
include doc/Makefile.inc
endif
+if CONFIG_LIBNOSYS
+include libnosys/Makefile.inc
+endif
diff --git a/libgloss/configure.ac b/libgloss/configure.ac
index 5657f094a197..1c04d92a44aa 100644
--- a/libgloss/configure.ac
+++ b/libgloss/configure.ac
@@ -230,10 +230,7 @@ dnl if test "${config_testsuite}" = "true";
dnl then AC_CONFIG_SUBDIRS([testsuite])
dnl fi
-if test "${config_libnosys}" = "true"; then
- AC_CONFIG_FILES([libnosys/Makefile])
- subdirs="$subdirs libnosys"
-fi
+AM_CONDITIONAL(CONFIG_LIBNOSYS, test x$config_libnosys = xtrue)
LIB_AC_PROG_CC
AS=${AS-as}
diff --git a/libgloss/libnosys/Makefile.inc b/libgloss/libnosys/Makefile.inc
new file mode 100644
index 000000000000..5e69072e7872
--- /dev/null
+++ b/libgloss/libnosys/Makefile.inc
@@ -0,0 +1,27 @@
+multilibtool_DATA += %D%/nosys.specs
+multilibtool_LIBRARIES += %D%/libnosys.a
+%C%_libnosys_a_SOURCES = \
+ %D%/chown.c \
+ %D%/close.c \
+ %D%/environ.c \
+ %D%/errno.c \
+ %D%/execve.c \
+ %D%/fork.c \
+ %D%/fstat.c \
+ %D%/getpid.c \
+ %D%/gettod.c \
+ %D%/isatty.c \
+ %D%/kill.c \
+ %D%/link.c \
+ %D%/lseek.c \
+ %D%/open.c \
+ %D%/read.c \
+ %D%/readlink.c \
+ %D%/sbrk.c \
+ %D%/stat.c \
+ %D%/symlink.c \
+ %D%/times.c \
+ %D%/unlink.c \
+ %D%/wait.c \
+ %D%/write.c \
+ %D%/_exit.c
--
2.34.1
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 3/5] libgloss: merge libgloss into top-level Makefile
2022-02-05 5:46 ` [PATCH 3/5] libgloss: merge libgloss " Mike Frysinger
@ 2022-02-07 11:14 ` Mike Frysinger
2022-02-07 11:49 ` Corinna Vinschen
0 siblings, 1 reply; 14+ messages in thread
From: Mike Frysinger @ 2022-02-07 11:14 UTC (permalink / raw)
To: newlib
[-- Attachment #1: Type: text/plain, Size: 834 bytes --]
On 05 Feb 2022 00:46, Mike Frysinger wrote:
> +if CONFIG_LIBNOSYS
> +include libnosys/Makefile.inc
> +endif
> --- a/libgloss/configure.ac
> +++ b/libgloss/configure.ac
> @@ -230,10 +230,7 @@ dnl if test "${config_testsuite}" = "true";
> dnl then AC_CONFIG_SUBDIRS([testsuite])
> dnl fi
>
> -if test "${config_libnosys}" = "true"; then
> - AC_CONFIG_FILES([libnosys/Makefile])
> - subdirs="$subdirs libnosys"
> -fi
> +AM_CONDITIONAL(CONFIG_LIBNOSYS, test x$config_libnosys = xtrue)
>
> LIB_AC_PROG_CC
> AS=${AS-as}
i'm not super happy with the CONFIG_xxx naming. newlib doesn't seem to have
an existing convention for me to copy & paste. there is LIB*_MACHINE_DIR,
but those correspond to specific machine/xxx/ subdirs.
i guess i could use the newlib pattern and name these HAVE_xxx_DIR.
-mike
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 3/5] libgloss: merge libgloss into top-level Makefile
2022-02-07 11:14 ` Mike Frysinger
@ 2022-02-07 11:49 ` Corinna Vinschen
0 siblings, 0 replies; 14+ messages in thread
From: Corinna Vinschen @ 2022-02-07 11:49 UTC (permalink / raw)
To: newlib
On Feb 7 06:14, Mike Frysinger wrote:
> On 05 Feb 2022 00:46, Mike Frysinger wrote:
> > +if CONFIG_LIBNOSYS
> > +include libnosys/Makefile.inc
> > +endif
> > --- a/libgloss/configure.ac
> > +++ b/libgloss/configure.ac
> > @@ -230,10 +230,7 @@ dnl if test "${config_testsuite}" = "true";
> > dnl then AC_CONFIG_SUBDIRS([testsuite])
> > dnl fi
> >
> > -if test "${config_libnosys}" = "true"; then
> > - AC_CONFIG_FILES([libnosys/Makefile])
> > - subdirs="$subdirs libnosys"
> > -fi
> > +AM_CONDITIONAL(CONFIG_LIBNOSYS, test x$config_libnosys = xtrue)
> >
> > LIB_AC_PROG_CC
> > AS=${AS-as}
>
> i'm not super happy with the CONFIG_xxx naming. newlib doesn't seem to have
> an existing convention for me to copy & paste. there is LIB*_MACHINE_DIR,
> but those correspond to specific machine/xxx/ subdirs.
>
> i guess i could use the newlib pattern and name these HAVE_xxx_DIR.
> -mike
Just keep it as is for now, it's easier to follow. Name changes can go
into another patch(set).
Corinna
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH 4/5] libgloss: merge iq2000 into top-level Makefile
2022-02-05 5:46 [PATCH 0/5] start converting libgloss to non-recursive automake Mike Frysinger
` (2 preceding siblings ...)
2022-02-05 5:46 ` [PATCH 3/5] libgloss: merge libgloss " Mike Frysinger
@ 2022-02-05 5:46 ` Mike Frysinger
2022-02-05 5:46 ` [PATCH 5/5] libgloss: merge bfin " Mike Frysinger
2022-02-07 11:36 ` [PATCH 0/5] start converting libgloss to non-recursive automake Corinna Vinschen
5 siblings, 0 replies; 14+ messages in thread
From: Mike Frysinger @ 2022-02-05 5:46 UTC (permalink / raw)
To: newlib
Avoid a recursive make to speed things up a bit.
---
libgloss/Makefile.am | 3 +
libgloss/Makefile.in | 181 +++++++++++++++++++++++++++++----
libgloss/aclocal.m4 | 20 ++++
libgloss/configure | 189 +++++++++++++++++++++++++++++++----
libgloss/configure.ac | 7 +-
libgloss/iq2000/Makefile.in | 149 ---------------------------
libgloss/iq2000/Makefile.inc | 11 ++
7 files changed, 368 insertions(+), 192 deletions(-)
delete mode 100644 libgloss/iq2000/Makefile.in
create mode 100644 libgloss/iq2000/Makefile.inc
diff --git a/libgloss/Makefile.am b/libgloss/Makefile.am
index 1209ec9d8d85..42d7b7ffb4b8 100644
--- a/libgloss/Makefile.am
+++ b/libgloss/Makefile.am
@@ -53,6 +53,9 @@ TEXINFO_TEX = ../texinfo/texinfo.tex
if HAVE_DOC
include doc/Makefile.inc
endif
+if CONFIG_IQ2000
+include iq2000/Makefile.inc
+endif
if CONFIG_LIBNOSYS
include libnosys/Makefile.inc
endif
diff --git a/libgloss/configure.ac b/libgloss/configure.ac
index 1c04d92a44aa..21871f954444 100644
--- a/libgloss/configure.ac
+++ b/libgloss/configure.ac
@@ -207,8 +207,7 @@ case "${target}" in
config_testsuite=false
;;
iq2000-*-*)
- AC_CONFIG_FILES([iq2000/Makefile])
- subdirs="$subdirs iq2000"
+ config_iq2000=true
;;
or1k-*-* | or1knd-*-* )
AC_CONFIG_FILES([or1k/Makefile])
@@ -224,6 +223,8 @@ case "${target}" in
;;
esac
+AM_CONDITIONAL(CONFIG_IQ2000, test x$config_iq2000 = xtrue)
+
dnl For now, don't bother configuring testsuite
dnl
dnl if test "${config_testsuite}" = "true";
@@ -241,7 +242,7 @@ AC_SUBST(AR)
LD=${LD-ld}
AC_SUBST(LD)
AC_PROG_RANLIB
-LIB_AM_PROG_AS
+AM_PROG_AS
AC_DEFINE(HAVE_GNU_LD, 1, [Using GNU LD])
diff --git a/libgloss/iq2000/Makefile.inc b/libgloss/iq2000/Makefile.inc
new file mode 100644
index 000000000000..97c91421cbed
--- /dev/null
+++ b/libgloss/iq2000/Makefile.inc
@@ -0,0 +1,11 @@
+multilibtool_DATA += %D%/crt0.o %D%/sim.ld
+multilibtool_LIBRARIES += %D%/libeval.a
+libobjs_a_SOURCES += %D%/crt0.S %D%/test.o
+
+%C%_libeval_a_SOURCES = \
+ %D%/_exit.c %D%/access.c %D%/chmod.c %D%/close.c %D%/crt1.c \
+ %D%/fstat.c %D%/getpid.c %D%/gettime.c %D%/isatty.c \
+ %D%/kill.c %D%/lseek.c %D%/open.c %D%/read.c \
+ %D%/sbrk.c %D%/stat.c %D%/time.c %D%/trap.c %D%/unlink.c %D%/utime.c %D%/write.c
+
+check_PROGRAMS += %D%/test
--
2.34.1
^ permalink raw reply [flat|nested] 14+ messages in thread* [PATCH 5/5] libgloss: merge bfin into top-level Makefile
2022-02-05 5:46 [PATCH 0/5] start converting libgloss to non-recursive automake Mike Frysinger
` (3 preceding siblings ...)
2022-02-05 5:46 ` [PATCH 4/5] libgloss: merge iq2000 " Mike Frysinger
@ 2022-02-05 5:46 ` Mike Frysinger
2022-02-07 11:44 ` Corinna Vinschen
2022-02-07 11:36 ` [PATCH 0/5] start converting libgloss to non-recursive automake Corinna Vinschen
5 siblings, 1 reply; 14+ messages in thread
From: Mike Frysinger @ 2022-02-05 5:46 UTC (permalink / raw)
To: newlib
Avoid a recursive make to speed things up a bit.
---
libgloss/Makefile.am | 9 +
libgloss/Makefile.in | 342 +++++++++++++++++++++++++++++++------
libgloss/bfin/Makefile.in | 207 ----------------------
libgloss/bfin/Makefile.inc | 71 ++++++++
libgloss/configure | 33 +++-
libgloss/configure.ac | 5 +-
6 files changed, 406 insertions(+), 261 deletions(-)
delete mode 100644 libgloss/bfin/Makefile.in
create mode 100644 libgloss/bfin/Makefile.inc
diff --git a/libgloss/Makefile.am b/libgloss/Makefile.am
index 42d7b7ffb4b8..b180309ca238 100644
--- a/libgloss/Makefile.am
+++ b/libgloss/Makefile.am
@@ -15,6 +15,12 @@ multilibtooldir = $(tooldir)/lib$(MULTISUBDIR)
multilibtool_DATA =
multilibtool_LIBRARIES =
+includetooldir = $(tooldir)/include
+includetool_DATA =
+
+includesystooldir = $(tooldir)/include/sys
+includesystool_DATA =
+
# A fake library so automake will generate rules for plain objects that we want
# to install (e.g. our crt0.o objects).
noinst_LIBRARIES = libobjs.a
@@ -53,6 +59,9 @@ TEXINFO_TEX = ../texinfo/texinfo.tex
if HAVE_DOC
include doc/Makefile.inc
endif
+if CONFIG_BFIN
+include bfin/Makefile.inc
+endif
if CONFIG_IQ2000
include iq2000/Makefile.inc
endif
diff --git a/libgloss/bfin/Makefile.inc b/libgloss/bfin/Makefile.inc
new file mode 100644
index 000000000000..35e880809a0e
--- /dev/null
+++ b/libgloss/bfin/Makefile.inc
@@ -0,0 +1,71 @@
+# Here is all of the simulator stuff.
+multilibtool_LIBRARIES += %D%/libsim.a
+%C%_libsim_a_SOURCES = \
+ %D%/clear_cache_range.c \
+ %D%/syscalls.c
+%C%_libsim_a_CPPFLAGS = \
+ $(AM_CPPFLAGS) \
+ -I$(srcdir)/%D/include
+
+check_PROGRAMS += %D%/sim-test
+
+# Here is all of the development board stuff.
+multilibtool_DATA += \
+ %D%/crt0.o \
+ %D%/basiccrt.o \
+ %D%/basiccrts.o \
+ %D%/basiccrt561.o \
+ %D%/basiccrt561s.o \
+ %D%/basiccrt561b.o
+# %D%/basiccrt60x.o \
+# %D%/basiccrt60xs.o \
+# %D%/basiccrt60xc1.o
+libobjs_a_SOURCES += \
+ %D%/crt0.S \
+ %D%/basiccrt.S
+
+if !HAVE_MULTISUBDIR
+multilibtool_DATA += \
+ %D%/bfin-common-sc.ld %D%/bfin-common-mc.ld %D%/bfin-common-mc0.ld \
+ %D%/bf504.ld %D%/bf506.ld \
+ %D%/bf512.ld %D%/bf514.ld %D%/bf516.ld %D%/bf518.ld \
+ %D%/bf522.ld %D%/bf523.ld %D%/bf524.ld %D%/bf525.ld %D%/bf526.ld %D%/bf527.ld \
+ %D%/bf531.ld %D%/bf532.ld %D%/bf533.ld \
+ %D%/bf534.ld %D%/bf536.ld %D%/bf537.ld \
+ %D%/bf538.ld %D%/bf539.ld \
+ %D%/bf542.ld %D%/bf544.ld %D%/bf547.ld %D%/bf548.ld %D%/bf549.ld \
+ %D%/bf561.ld %D%/bf561a.ld %D%/bf561b.ld %D%/bf561m.ld \
+ %D%/bf606.ld %D%/bf606c0.ld %D%/bf606c1.ld %D%/bf606m.ld \
+ %D%/bf607.ld %D%/bf607c0.ld %D%/bf607c1.ld %D%/bf607m.ld \
+ %D%/bf608.ld %D%/bf608c0.ld %D%/bf608c1.ld %D%/bf608m.ld \
+ %D%/bf609.ld %D%/bf609c0.ld %D%/bf609c1.ld %D%/bf609m.ld \
+ %D%/bf592.ld
+
+includetool_DATA += $(wildcard $(srcdir)/%D%/include/*.h)
+includesystool_DATA += $(wildcard $(srcdir)/%D%/include/sys/*.h)
+endif
+
+multilibtool_LIBRARIES += %D%/libbfinbsp.a
+%C%_libbfinbsp_a_SOURCES = \
+ %D%/clear_cache_range.c \
+ %D%/_exit.c
+%C%_libbfinbsp_a_CPPFLAGS = \
+ $(AM_CPPFLAGS) \
+ -I$(srcdir)/%D/include
+
+%D%/basiccrt.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),,-mcpu=bf532-any)
+%D%/basiccrts.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),,-mcpu=bf532-any) -D__BFIN_SDRAM
+%D%/basiccrt561.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),-mcpu=bf561-any,-mcpu=bf561-none)
+%D%/basiccrt561s.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),-mcpu=bf561-any,-mcpu=bf561-none) -D__BFIN_SDRAM
+%D%/basiccrt561b.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),-mcpu=bf561-any,-mcpu=bf561-none) -D__ADSPBF561_COREB__
+%D%/basiccrt60x.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),-mcpu=bf608-any,-mcpu=bf608-none)
+%D%/basiccrt60xs.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),-mcpu=bf608-any,-mcpu=bf608-none) -D__BFIN_SDRAM
+%D%/basiccrt60xc1.$(OBJEXT): %D%/basiccrt.S
+ $(AM_V_CC)$(COMPILE) -o $@ -c $< $(if $(findstring mcpu=,$(CFLAGS)),-mcpu=bf608-any,-mcpu=bf608-none) -D__ADSPBF60x_CORE1__
diff --git a/libgloss/configure.ac b/libgloss/configure.ac
index 21871f954444..11b115e2a7a0 100644
--- a/libgloss/configure.ac
+++ b/libgloss/configure.ac
@@ -19,6 +19,7 @@ else
have_multisubdir=yes
fi
AM_CONDITIONAL(HAVE_DOC, test x$have_multisubdir = xno)
+AM_CONDITIONAL(HAVE_MULTISUBDIR, test x$have_multisubdir = xyes)
AC_NO_EXECUTABLES
@@ -103,8 +104,7 @@ case "${target}" in
AC_CONFIG_SUBDIRS([mn10300])
;;
bfin-*-*)
- AC_CONFIG_FILES([bfin/Makefile])
- subdirs="$subdirs bfin"
+ config_bfin=true
;;
cr16-*-*)
AC_CONFIG_FILES([cr16/Makefile])
@@ -223,6 +223,7 @@ case "${target}" in
;;
esac
+AM_CONDITIONAL(CONFIG_BFIN, test x$config_bfin = xtrue)
AM_CONDITIONAL(CONFIG_IQ2000, test x$config_iq2000 = xtrue)
dnl For now, don't bother configuring testsuite
--
2.34.1
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 5/5] libgloss: merge bfin into top-level Makefile
2022-02-05 5:46 ` [PATCH 5/5] libgloss: merge bfin " Mike Frysinger
@ 2022-02-07 11:44 ` Corinna Vinschen
2022-02-07 12:07 ` Corinna Vinschen
0 siblings, 1 reply; 14+ messages in thread
From: Corinna Vinschen @ 2022-02-07 11:44 UTC (permalink / raw)
To: newlib
On Feb 5 00:46, Mike Frysinger wrote:
> Avoid a recursive make to speed things up a bit.
> ---
> libgloss/Makefile.am | 9 +
> libgloss/Makefile.in | 342 +++++++++++++++++++++++++++++++------
> libgloss/bfin/Makefile.in | 207 ----------------------
> libgloss/bfin/Makefile.inc | 71 ++++++++
> libgloss/configure | 33 +++-
> libgloss/configure.ac | 5 +-
> 6 files changed, 406 insertions(+), 261 deletions(-)
> delete mode 100644 libgloss/bfin/Makefile.in
> create mode 100644 libgloss/bfin/Makefile.inc
>
> diff --git a/libgloss/Makefile.am b/libgloss/Makefile.am
> index 42d7b7ffb4b8..b180309ca238 100644
> --- a/libgloss/Makefile.am
> +++ b/libgloss/Makefile.am
> @@ -15,6 +15,12 @@ multilibtooldir = $(tooldir)/lib$(MULTISUBDIR)
> multilibtool_DATA =
> multilibtool_LIBRARIES =
>
> +includetooldir = $(tooldir)/include
> +includetool_DATA =
> +
> +includesystooldir = $(tooldir)/include/sys
> +includesystool_DATA =
> +
> # A fake library so automake will generate rules for plain objects that we want
> # to install (e.g. our crt0.o objects).
> noinst_LIBRARIES = libobjs.a
> @@ -53,6 +59,9 @@ TEXINFO_TEX = ../texinfo/texinfo.tex
> if HAVE_DOC
> include doc/Makefile.inc
> endif
> +if CONFIG_BFIN
> +include bfin/Makefile.inc
> +endif
> if CONFIG_IQ2000
> include iq2000/Makefile.inc
> endif
> diff --git a/libgloss/bfin/Makefile.inc b/libgloss/bfin/Makefile.inc
> new file mode 100644
> index 000000000000..35e880809a0e
> --- /dev/null
> +++ b/libgloss/bfin/Makefile.inc
> @@ -0,0 +1,71 @@
> +# Here is all of the simulator stuff.
> +multilibtool_LIBRARIES += %D%/libsim.a
> +%C%_libsim_a_SOURCES = \
> + %D%/clear_cache_range.c \
> + %D%/syscalls.c
> +%C%_libsim_a_CPPFLAGS = \
> + $(AM_CPPFLAGS) \
> + -I$(srcdir)/%D/include
> +
> +check_PROGRAMS += %D%/sim-test
> +
> +# Here is all of the development board stuff.
> +multilibtool_DATA += \
> + %D%/crt0.o \
> + %D%/basiccrt.o \
> + %D%/basiccrts.o \
> + %D%/basiccrt561.o \
> + %D%/basiccrt561s.o \
> + %D%/basiccrt561b.o
> +# %D%/basiccrt60x.o \
> +# %D%/basiccrt60xs.o \
> +# %D%/basiccrt60xc1.o
Not sure how you guys actually building and using this stuff think about
this, but I, for one, am a little put off by this change. The old code
in bfin/Makefile.in had a few nice definitions like BOARD_BSP,
BOARD_CRT0S, SIM_BSP, SIM_CRT0, etc. Why didn't you take them over?
That's not saying we can't push this patch, but somehow it's losing
readability. Now it's just a list of arbitrary files to create.
Corinna
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 5/5] libgloss: merge bfin into top-level Makefile
2022-02-07 11:44 ` Corinna Vinschen
@ 2022-02-07 12:07 ` Corinna Vinschen
2022-02-07 13:11 ` Mike Frysinger
0 siblings, 1 reply; 14+ messages in thread
From: Corinna Vinschen @ 2022-02-07 12:07 UTC (permalink / raw)
To: newlib
On Feb 7 12:44, Corinna Vinschen wrote:
> On Feb 5 00:46, Mike Frysinger wrote:
> > Avoid a recursive make to speed things up a bit.
> > ---
> > libgloss/Makefile.am | 9 +
> > libgloss/Makefile.in | 342 +++++++++++++++++++++++++++++++------
> > libgloss/bfin/Makefile.in | 207 ----------------------
> > libgloss/bfin/Makefile.inc | 71 ++++++++
> > libgloss/configure | 33 +++-
> > libgloss/configure.ac | 5 +-
> > 6 files changed, 406 insertions(+), 261 deletions(-)
> > delete mode 100644 libgloss/bfin/Makefile.in
> > create mode 100644 libgloss/bfin/Makefile.inc
> > + [...]
> > +# Here is all of the development board stuff.
> > +multilibtool_DATA += \
> > + %D%/crt0.o \
> > + %D%/basiccrt.o \
> > + %D%/basiccrts.o \
> > + %D%/basiccrt561.o \
> > + %D%/basiccrt561s.o \
> > + %D%/basiccrt561b.o
> > +# %D%/basiccrt60x.o \
> > +# %D%/basiccrt60xs.o \
> > +# %D%/basiccrt60xc1.o
>
> Not sure how you guys actually building and using this stuff think about
> this, but I, for one, am a little put off by this change. The old code
> in bfin/Makefile.in had a few nice definitions like BOARD_BSP,
> BOARD_CRT0S, SIM_BSP, SIM_CRT0, etc. Why didn't you take them over?
>
> That's not saying we can't push this patch, but somehow it's losing
> readability. Now it's just a list of arbitrary files to create.
Also, what about the install-sim / install-board split? Do we know
for sure that they are not used by somebody?
Corinna
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 5/5] libgloss: merge bfin into top-level Makefile
2022-02-07 12:07 ` Corinna Vinschen
@ 2022-02-07 13:11 ` Mike Frysinger
0 siblings, 0 replies; 14+ messages in thread
From: Mike Frysinger @ 2022-02-07 13:11 UTC (permalink / raw)
To: newlib
[-- Attachment #1: Type: text/plain, Size: 4303 bytes --]
On 07 Feb 2022 13:07, Corinna Vinschen wrote:
> On Feb 7 12:44, Corinna Vinschen wrote:
> > On Feb 5 00:46, Mike Frysinger wrote:
> > > Avoid a recursive make to speed things up a bit.
> > > ---
> > > libgloss/Makefile.am | 9 +
> > > libgloss/Makefile.in | 342 +++++++++++++++++++++++++++++++------
> > > libgloss/bfin/Makefile.in | 207 ----------------------
> > > libgloss/bfin/Makefile.inc | 71 ++++++++
> > > libgloss/configure | 33 +++-
> > > libgloss/configure.ac | 5 +-
> > > 6 files changed, 406 insertions(+), 261 deletions(-)
> > > delete mode 100644 libgloss/bfin/Makefile.in
> > > create mode 100644 libgloss/bfin/Makefile.inc
> > > + [...]
> > > +# Here is all of the development board stuff.
> > > +multilibtool_DATA += \
> > > + %D%/crt0.o \
> > > + %D%/basiccrt.o \
> > > + %D%/basiccrts.o \
> > > + %D%/basiccrt561.o \
> > > + %D%/basiccrt561s.o \
> > > + %D%/basiccrt561b.o
> > > +# %D%/basiccrt60x.o \
> > > +# %D%/basiccrt60xs.o \
> > > +# %D%/basiccrt60xc1.o
> >
> > Not sure how you guys actually building and using this stuff think about
> > this, but I, for one, am a little put off by this change. The old code
> > in bfin/Makefile.in had a few nice definitions like BOARD_BSP,
> > BOARD_CRT0S, SIM_BSP, SIM_CRT0, etc. Why didn't you take them over?
> >
> > That's not saying we can't push this patch, but somehow it's losing
> > readability. Now it's just a list of arbitrary files to create.
>
> Also, what about the install-sim / install-board split? Do we know
> for sure that they are not used by somebody?
tl;dr: i think it's fine.
the bsp/sim logic looks like blind copy & paste between some subdirs.
today, we have about 50 targets in libgloss. "install-sim" exists in
only 11 of them. "install-board" only exists in bfin. neither of the
rules are exposed in the topdir, so you'd have to be in libgloss/bfin/
and run `make install-xxx` for them to work. but that won't really work
because (1) the multilib multiplex logic only exists in the top-level
libgloss/ dir and (2) none of the other vars (e.g. compiler settings)
are exported there ... instead, we rely on the recursive make out of the
top-level libgloss/ to pass things down via $(FLAGS_TO_PASS).
for bfin specifically, tbh, the bsp side of things has always been a bit
underdeveloped. we never had a lot of customer requests, and anyone who
did use newlib would write their own glue logic. other than the basic
crt, there's only 2 functions in there ... clear_cache_range is internal
hook for gcc to workaround hardware anomalies, and exit which just hangs
the CPU while triggering JTAG ICE exceptions. even the basic crt was
mostly copied from the proprietary side of the company, and has had some
feedback from people that it kind of sucks. so i think at this point,
the bfin bsp is likely to not gain a whole lot more.
on the otherhand, the bfin sim has seen a lot of work which is why it has
a bit more logic to it.
but let's look at your "readability" feedback:
* we still have sep libsim.a & libbfinbsp.a targets to keep the diff env
separate since our linker scripts refer to them directly. that takes
care of the *_BSP & *_OBJS & *_LDFLAGS variables.
* the Makefile.in today has sep install targets, although we always install
both. the new Makefile.inc doesn't need them since the variables imply
whether things are installed. that takes care of install-* and *_INSTALL.
* that leaves *_SCRIPTS & *_CRT0 variables. i'm adding them to the same
multilibtool_DATA variable which is what you highlighted above. i could
add some variables for these if you really wanted, it's just that automake
constrains the namespace a little, as does recursive make. so it'd have
to be something like:
%C%_sim_crts = %D%/crt0.o
%C%_bsp_crts = %D%/basiccrt.o ......
multilibtool_DATA += $(%C%_sim_crts) $(%C%_bsp_crts)
and at this point, considering how simple the existing logic is, and how
unlikely it is to change, it didn't seem worth adding sep vars. i could
at least keep separate comment blocks if you wanted. so i guess move the
multilibtool_DATA += %D%/crt0.o
to before the "development board stuff" section.
-mike
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 0/5] start converting libgloss to non-recursive automake
2022-02-05 5:46 [PATCH 0/5] start converting libgloss to non-recursive automake Mike Frysinger
` (4 preceding siblings ...)
2022-02-05 5:46 ` [PATCH 5/5] libgloss: merge bfin " Mike Frysinger
@ 2022-02-07 11:36 ` Corinna Vinschen
2022-02-07 12:45 ` Mike Frysinger
5 siblings, 1 reply; 14+ messages in thread
From: Corinna Vinschen @ 2022-02-07 11:36 UTC (permalink / raw)
To: newlib
On Feb 5 00:46, Mike Frysinger wrote:
> This converts the top-level to automake, and then cuts a few subdirs over
> to non-recursive automake. I didn't cut all the subdirs over to automake,
> but should be easy to convert more over.
>
> Mike Frysinger (5):
> libgloss: convert top level to automake
> libgloss: merge doc into top-level Makefile
> libgloss: merge libgloss into top-level Makefile
> libgloss: merge iq2000 into top-level Makefile
> libgloss: merge bfin into top-level Makefile
$ autoreconf
configure.ac:46: warning: macro 'AM_ENABLE_MULTILIB' not found in library
configure.ac:31: warning: macro 'AM_ENABLE_MULTILIB' not found in library
[...]
Problem? A dependency to another not yet applied patch?
Corinna
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 0/5] start converting libgloss to non-recursive automake
2022-02-07 11:36 ` [PATCH 0/5] start converting libgloss to non-recursive automake Corinna Vinschen
@ 2022-02-07 12:45 ` Mike Frysinger
2022-02-08 9:15 ` Corinna Vinschen
0 siblings, 1 reply; 14+ messages in thread
From: Mike Frysinger @ 2022-02-07 12:45 UTC (permalink / raw)
To: newlib
[-- Attachment #1: Type: text/plain, Size: 2116 bytes --]
On 07 Feb 2022 12:36, Corinna Vinschen wrote:
> On Feb 5 00:46, Mike Frysinger wrote:
> > This converts the top-level to automake, and then cuts a few subdirs over
> > to non-recursive automake. I didn't cut all the subdirs over to automake,
> > but should be easy to convert more over.
> >
> > Mike Frysinger (5):
> > libgloss: convert top level to automake
> > libgloss: merge doc into top-level Makefile
> > libgloss: merge libgloss into top-level Makefile
> > libgloss: merge iq2000 into top-level Makefile
> > libgloss: merge bfin into top-level Makefile
>
> $ autoreconf
> configure.ac:46: warning: macro 'AM_ENABLE_MULTILIB' not found in library
> configure.ac:31: warning: macro 'AM_ENABLE_MULTILIB' not found in library
> [...]
>
> Problem? A dependency to another not yet applied patch?
this isn't new behavior. although i understand the mismatch in expectations.
if you run `autoreconf` in libgloss/ today, you will see these same warnings.
it's because the only way to have aclocal search extra dirs for m4 macros is
either pass the -I flags to autoreconf directly, or have a Makefile.am with
ACLOCAL_AMFLAGS set in it (which autoreconf extracts and passes to aclocal).
w/out automake, ACLOCAL_AMFLAGS doesn't work, so you have to add all the -I
flags yourself with absolute paths:
$ autoreconf -I $PWD -I $PWD/.. -I $PWD/../config
yes, i'm converting the top-level to automake, so the ACLOCAL_AMFLAGS are
picked up from there, but that's only the top-level. autoreconf will go
into each subdir and run autoreconf in there, and that's where these are
coming from. once all the subdirs convert over to automake, then you can
go back to running plain `autoreconf` and it'll all just work for you.
if you run `autoreconf -v`, it's a little more obvious that it's coming
from the subdirs as it'll show:
autoreconf-2.69: configure.ac: adding subdirectory aarch64 to autoreconf
autoreconf-2.69: Entering directory `aarch64'
autoreconf-2.69: running: aclocal
configure.ac:46: warning: macro 'AM_ENABLE_MULTILIB' not found in library
-mike
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 0/5] start converting libgloss to non-recursive automake
2022-02-07 12:45 ` Mike Frysinger
@ 2022-02-08 9:15 ` Corinna Vinschen
0 siblings, 0 replies; 14+ messages in thread
From: Corinna Vinschen @ 2022-02-08 9:15 UTC (permalink / raw)
To: newlib
On Feb 7 07:45, Mike Frysinger wrote:
> On 07 Feb 2022 12:36, Corinna Vinschen wrote:
> > On Feb 5 00:46, Mike Frysinger wrote:
> > > This converts the top-level to automake, and then cuts a few subdirs over
> > > to non-recursive automake. I didn't cut all the subdirs over to automake,
> > > but should be easy to convert more over.
> > >
> > > Mike Frysinger (5):
> > > libgloss: convert top level to automake
> > > libgloss: merge doc into top-level Makefile
> > > libgloss: merge libgloss into top-level Makefile
> > > libgloss: merge iq2000 into top-level Makefile
> > > libgloss: merge bfin into top-level Makefile
> >
> > $ autoreconf
> > configure.ac:46: warning: macro 'AM_ENABLE_MULTILIB' not found in library
> > configure.ac:31: warning: macro 'AM_ENABLE_MULTILIB' not found in library
> > [...]
> >
> > Problem? A dependency to another not yet applied patch?
>
> this isn't new behavior. although i understand the mismatch in expectations.
>
> if you run `autoreconf` in libgloss/ today, you will see these same warnings.
> it's because the only way to have aclocal search extra dirs for m4 macros is
> either pass the -I flags to autoreconf directly, or have a Makefile.am with
> ACLOCAL_AMFLAGS set in it (which autoreconf extracts and passes to aclocal).
>
> w/out automake, ACLOCAL_AMFLAGS doesn't work, so you have to add all the -I
> flags yourself with absolute paths:
> $ autoreconf -I $PWD -I $PWD/.. -I $PWD/../config
>
> yes, i'm converting the top-level to automake, so the ACLOCAL_AMFLAGS are
> picked up from there, but that's only the top-level. autoreconf will go
> into each subdir and run autoreconf in there, and that's where these are
> coming from. once all the subdirs convert over to automake, then you can
> go back to running plain `autoreconf` and it'll all just work for you.
>
> if you run `autoreconf -v`, it's a little more obvious that it's coming
> from the subdirs as it'll show:
> autoreconf-2.69: configure.ac: adding subdirectory aarch64 to autoreconf
> autoreconf-2.69: Entering directory `aarch64'
> autoreconf-2.69: running: aclocal
> configure.ac:46: warning: macro 'AM_ENABLE_MULTILIB' not found in library
> -mike
Ok, thanks for this and your other reply. GTG.
Corinna
^ permalink raw reply [flat|nested] 14+ messages in thread