2009-01-22 02:12:52 +08:00
|
|
|
|
2018-05-21 18:35:04 +08:00
|
|
|
# These are used when we want to do substitutions without confusing Make
|
|
|
|
NULL :=
|
|
|
|
SPACE := $(NULL) #
|
2016-06-01 12:25:15 +08:00
|
|
|
COMMA := ,
|
|
|
|
|
2009-10-07 03:11:14 +08:00
|
|
|
# Don't use implicit rules or variables
|
|
|
|
# we have explicit rules for everything
|
|
|
|
MAKEFLAGS += -rR
|
|
|
|
|
|
|
|
# Files with this suffixes are final, don't try to generate them
|
|
|
|
# using implicit rules
|
2016-11-03 03:46:13 +08:00
|
|
|
%/trace-events:
|
|
|
|
%.hx:
|
|
|
|
%.py:
|
|
|
|
%.objs:
|
2009-10-07 03:11:14 +08:00
|
|
|
%.d:
|
|
|
|
%.h:
|
|
|
|
%.c:
|
2014-02-06 01:27:27 +08:00
|
|
|
%.cc:
|
2013-08-07 23:39:36 +08:00
|
|
|
%.cpp:
|
2009-10-07 03:11:14 +08:00
|
|
|
%.m:
|
|
|
|
%.mak:
|
2016-09-05 16:52:19 +08:00
|
|
|
clean-target:
|
2009-10-07 03:11:14 +08:00
|
|
|
|
2009-11-20 03:07:52 +08:00
|
|
|
# Flags for dependency generation
|
2015-08-09 17:39:53 +08:00
|
|
|
QEMU_DGFLAGS += -MMD -MP -MT $@ -MF $(@D)/$(*F).d
|
2009-10-11 21:08:57 +08:00
|
|
|
|
make: move top level dir to end of include search path
Currently the search path is
1. source dir corresponding to input file (implicit by compiler)
2. top level build dir
3. top level source dir
4. top level source include/ dir
5. source dir corresponding to input file
6. build dir corresponding to output file
Search item 5 is an effective no-op, since it duplicates item 1.
When srcdir == builddir, item 6 also duplicates item 1, which
causes a semantic difference between VPATH and non-VPATH builds.
Thus to ensure consistent semantics we need item 6 to be present
immediately after item 1. e.g.
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file
3. top level build dir
4. top level source dir
5. top level source include/ dir
When srcdir == builddir, items 1 & 2 collapse into one, and items
3 & 4 collapse into one, but the overall search order is still
consistent with srcdir != builddir
A further complication is that while most of the source files
are built with a current directory of $BUILD_DIR, target dependant
files are built with a current directory of $BUILD_DIR/$TARGET.
As a result, search item 2 resolves to a different location for
target independant vs target dependant files. For example when
building 'migration/ram.o', the use of '-I$(@D)' (which expands
to '-Imigration') would not find '$BUILD_DIR/migration', but
rather '$BUILD_DIR/$TARGET/migration'.
If there are generated headers files to be used by the migration
code in '$BUILD_DIR/migration', these will not be found by the
relative include, an absolute include is needed instead. This
has not been a problem so far, since nothing has been generating
headers in sub-dirs, but the trace code will shortly be doing
that. So it is needed to list '-I$(BUILD_DIR)/$(@D)' as well as
'-I$(@D)' to ensure both directories are searched when building
target dependant code. So the search order ends up being:
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file (absolute)
3. build dir corresponding to output file (relative to cwd)
4. top level build dir
5. top level source dir
6. top level source include/ dir
One final complication is that the absolute '-I$(BUILD_DIR)/$(@D)'
will sometimes end up pointing to a non-existant directory if
that sub-dir does not have any target-independant files to be
built. Rather than try to dynamically filter this, a simple
'mkdir' ensures $(BUILD_DIR)/$(@D) is guaranteed to exist at
all times.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20170125161417.31949-2-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2017-01-26 00:14:10 +08:00
|
|
|
# Compiler searches the source file dir first, but in vpath builds
|
|
|
|
# we need to make it search the build dir too, before any other
|
|
|
|
# explicit search paths. There are two search locations in the build
|
|
|
|
# dir, one absolute and the other relative to the compiler working
|
|
|
|
# directory. These are the same for target-independent files, but
|
|
|
|
# different for target-dependent ones.
|
2018-03-21 23:22:07 +08:00
|
|
|
QEMU_LOCAL_INCLUDES = -iquote $(BUILD_DIR)/$(@D) -iquote $(@D)
|
2012-09-17 16:21:52 +08:00
|
|
|
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
WL_U := -Wl,-u,
|
2014-09-19 03:55:08 +08:00
|
|
|
find-symbols = $(if $1, $(sort $(shell $(NM) -P -g $1 | $2)))
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
defined-symbols = $(call find-symbols,$1,awk '$$2!="U"{print $$1}')
|
|
|
|
undefined-symbols = $(call find-symbols,$1,awk '$$2=="U"{print $$1}')
|
|
|
|
|
|
|
|
# All the .mo objects in -m variables are also added into corresponding -y
|
|
|
|
# variable in unnest-vars, but filtered out here, when LINK is called.
|
|
|
|
#
|
|
|
|
# The .mo objects are supposed to be linked as a DSO, for module build. So here
|
|
|
|
# they are only used as a placeholders to generate those "archive undefined"
|
|
|
|
# symbol options (-Wl,-u,$symbol_name), which are the archive functions
|
|
|
|
# referenced by the code in the DSO.
|
|
|
|
#
|
|
|
|
# Also the presence in -y variables will also guarantee they are built before
|
|
|
|
# linking executables that will load them. So we can look up symbol reference
|
|
|
|
# in LINK.
|
|
|
|
#
|
|
|
|
# This is necessary because the exectuable itself may not use the function, in
|
|
|
|
# which case the function would not be linked in. Then the DSO loading will
|
|
|
|
# fail because of the missing symbol.
|
|
|
|
process-archive-undefs = $(filter-out %.a %.mo,$1) \
|
|
|
|
$(addprefix $(WL_U), \
|
|
|
|
$(filter $(call defined-symbols,$(filter %.a, $1)), \
|
|
|
|
$(call undefined-symbols,$(filter %.mo,$1)))) \
|
|
|
|
$(filter %.a,$1)
|
|
|
|
|
2016-07-27 14:26:15 +08:00
|
|
|
extract-libs = $(strip $(foreach o,$(filter-out %.mo,$1),$($o-libs)))
|
2014-02-10 14:48:56 +08:00
|
|
|
expand-objs = $(strip $(sort $(filter %.o,$1)) \
|
|
|
|
$(foreach o,$(filter %.mo,$1),$($o-objs)) \
|
|
|
|
$(filter-out %.o %.mo,$1))
|
2014-02-10 14:48:53 +08:00
|
|
|
|
2010-01-07 03:24:05 +08:00
|
|
|
%.o: %.c
|
make: move top level dir to end of include search path
Currently the search path is
1. source dir corresponding to input file (implicit by compiler)
2. top level build dir
3. top level source dir
4. top level source include/ dir
5. source dir corresponding to input file
6. build dir corresponding to output file
Search item 5 is an effective no-op, since it duplicates item 1.
When srcdir == builddir, item 6 also duplicates item 1, which
causes a semantic difference between VPATH and non-VPATH builds.
Thus to ensure consistent semantics we need item 6 to be present
immediately after item 1. e.g.
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file
3. top level build dir
4. top level source dir
5. top level source include/ dir
When srcdir == builddir, items 1 & 2 collapse into one, and items
3 & 4 collapse into one, but the overall search order is still
consistent with srcdir != builddir
A further complication is that while most of the source files
are built with a current directory of $BUILD_DIR, target dependant
files are built with a current directory of $BUILD_DIR/$TARGET.
As a result, search item 2 resolves to a different location for
target independant vs target dependant files. For example when
building 'migration/ram.o', the use of '-I$(@D)' (which expands
to '-Imigration') would not find '$BUILD_DIR/migration', but
rather '$BUILD_DIR/$TARGET/migration'.
If there are generated headers files to be used by the migration
code in '$BUILD_DIR/migration', these will not be found by the
relative include, an absolute include is needed instead. This
has not been a problem so far, since nothing has been generating
headers in sub-dirs, but the trace code will shortly be doing
that. So it is needed to list '-I$(BUILD_DIR)/$(@D)' as well as
'-I$(@D)' to ensure both directories are searched when building
target dependant code. So the search order ends up being:
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file (absolute)
3. build dir corresponding to output file (relative to cwd)
4. top level build dir
5. top level source dir
6. top level source include/ dir
One final complication is that the absolute '-I$(BUILD_DIR)/$(@D)'
will sometimes end up pointing to a non-existant directory if
that sub-dir does not have any target-independant files to be
built. Rather than try to dynamically filter this, a simple
'mkdir' ensures $(BUILD_DIR)/$(@D) is guaranteed to exist at
all times.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20170125161417.31949-2-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2017-01-26 00:14:10 +08:00
|
|
|
$(call quiet-command,$(CC) $(QEMU_LOCAL_INCLUDES) $(QEMU_INCLUDES) \
|
|
|
|
$(QEMU_CFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) $($@-cflags) \
|
|
|
|
-c -o $@ $<,"CC","$(TARGET_DIR)$@")
|
2013-04-27 06:25:31 +08:00
|
|
|
%.o: %.rc
|
2016-10-05 00:27:21 +08:00
|
|
|
$(call quiet-command,$(WINDRES) -I. -o $@ $<,"RC","$(TARGET_DIR)$@")
|
2009-01-22 02:12:52 +08:00
|
|
|
|
2014-02-06 01:27:27 +08:00
|
|
|
# If we have a CXX we might have some C++ objects, in which case we
|
|
|
|
# must link with the C++ compiler, not the plain C compiler.
|
|
|
|
LINKPROG = $(or $(CXX),$(CC))
|
|
|
|
|
build-sys: clean up flags included in the linker command line
Some of the CFLAGS that are discovered during configure, for example
compiler warnings, are being included on the linker command line because
QEMU_CFLAGS is added to it. Other flags, such as the -m32, appear twice
because they are included in both QEMU_CFLAGS and LDFLAGS. All this
leads to confusion with respect to what goes in which Makefile variables
(and we have plenty).
So, introduce QEMU_LDFLAGS for flags discovered by configure, following
the lead of QEMU_CFLAGS, and stop adding to it:
1) options that are already in CFLAGS, for example "-g"
2) duplicate options
At the same time, options that _are_ needed by both compiler and linker
must now be added to both QEMU_CFLAGS and QEMU_LDFLAGS, which is clearer.
This is mostly -fsanitize options. For now, --extra-cflags has this behavior
(but --extra-cxxflags does not).
Meson will not include CFLAGS on the linker command line, do the same in our
build system as well.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-12-11 22:34:27 +08:00
|
|
|
LINK = $(call quiet-command, $(LINKPROG) $(CFLAGS) $(QEMU_LDFLAGS) -o $@ \
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
$(call process-archive-undefs, $1) \
|
2016-10-05 00:27:21 +08:00
|
|
|
$(version-obj-y) $(call extract-libs,$1) $(LIBS),"LINK","$(TARGET_DIR)$@")
|
2011-05-15 16:51:28 +08:00
|
|
|
|
2016-06-24 01:39:18 +08:00
|
|
|
%.o: %.S
|
make: move top level dir to end of include search path
Currently the search path is
1. source dir corresponding to input file (implicit by compiler)
2. top level build dir
3. top level source dir
4. top level source include/ dir
5. source dir corresponding to input file
6. build dir corresponding to output file
Search item 5 is an effective no-op, since it duplicates item 1.
When srcdir == builddir, item 6 also duplicates item 1, which
causes a semantic difference between VPATH and non-VPATH builds.
Thus to ensure consistent semantics we need item 6 to be present
immediately after item 1. e.g.
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file
3. top level build dir
4. top level source dir
5. top level source include/ dir
When srcdir == builddir, items 1 & 2 collapse into one, and items
3 & 4 collapse into one, but the overall search order is still
consistent with srcdir != builddir
A further complication is that while most of the source files
are built with a current directory of $BUILD_DIR, target dependant
files are built with a current directory of $BUILD_DIR/$TARGET.
As a result, search item 2 resolves to a different location for
target independant vs target dependant files. For example when
building 'migration/ram.o', the use of '-I$(@D)' (which expands
to '-Imigration') would not find '$BUILD_DIR/migration', but
rather '$BUILD_DIR/$TARGET/migration'.
If there are generated headers files to be used by the migration
code in '$BUILD_DIR/migration', these will not be found by the
relative include, an absolute include is needed instead. This
has not been a problem so far, since nothing has been generating
headers in sub-dirs, but the trace code will shortly be doing
that. So it is needed to list '-I$(BUILD_DIR)/$(@D)' as well as
'-I$(@D)' to ensure both directories are searched when building
target dependant code. So the search order ends up being:
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file (absolute)
3. build dir corresponding to output file (relative to cwd)
4. top level build dir
5. top level source dir
6. top level source include/ dir
One final complication is that the absolute '-I$(BUILD_DIR)/$(@D)'
will sometimes end up pointing to a non-existant directory if
that sub-dir does not have any target-independant files to be
built. Rather than try to dynamically filter this, a simple
'mkdir' ensures $(BUILD_DIR)/$(@D) is guaranteed to exist at
all times.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20170125161417.31949-2-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2017-01-26 00:14:10 +08:00
|
|
|
$(call quiet-command,$(CCAS) $(QEMU_LOCAL_INCLUDES) $(QEMU_INCLUDES) \
|
|
|
|
$(QEMU_CFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) \
|
|
|
|
-c -o $@ $<,"CCAS","$(TARGET_DIR)$@")
|
2009-01-22 02:12:52 +08:00
|
|
|
|
2014-02-06 01:27:27 +08:00
|
|
|
%.o: %.cc
|
make: move top level dir to end of include search path
Currently the search path is
1. source dir corresponding to input file (implicit by compiler)
2. top level build dir
3. top level source dir
4. top level source include/ dir
5. source dir corresponding to input file
6. build dir corresponding to output file
Search item 5 is an effective no-op, since it duplicates item 1.
When srcdir == builddir, item 6 also duplicates item 1, which
causes a semantic difference between VPATH and non-VPATH builds.
Thus to ensure consistent semantics we need item 6 to be present
immediately after item 1. e.g.
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file
3. top level build dir
4. top level source dir
5. top level source include/ dir
When srcdir == builddir, items 1 & 2 collapse into one, and items
3 & 4 collapse into one, but the overall search order is still
consistent with srcdir != builddir
A further complication is that while most of the source files
are built with a current directory of $BUILD_DIR, target dependant
files are built with a current directory of $BUILD_DIR/$TARGET.
As a result, search item 2 resolves to a different location for
target independant vs target dependant files. For example when
building 'migration/ram.o', the use of '-I$(@D)' (which expands
to '-Imigration') would not find '$BUILD_DIR/migration', but
rather '$BUILD_DIR/$TARGET/migration'.
If there are generated headers files to be used by the migration
code in '$BUILD_DIR/migration', these will not be found by the
relative include, an absolute include is needed instead. This
has not been a problem so far, since nothing has been generating
headers in sub-dirs, but the trace code will shortly be doing
that. So it is needed to list '-I$(BUILD_DIR)/$(@D)' as well as
'-I$(@D)' to ensure both directories are searched when building
target dependant code. So the search order ends up being:
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file (absolute)
3. build dir corresponding to output file (relative to cwd)
4. top level build dir
5. top level source dir
6. top level source include/ dir
One final complication is that the absolute '-I$(BUILD_DIR)/$(@D)'
will sometimes end up pointing to a non-existant directory if
that sub-dir does not have any target-independant files to be
built. Rather than try to dynamically filter this, a simple
'mkdir' ensures $(BUILD_DIR)/$(@D) is guaranteed to exist at
all times.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20170125161417.31949-2-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2017-01-26 00:14:10 +08:00
|
|
|
$(call quiet-command,$(CXX) $(QEMU_LOCAL_INCLUDES) $(QEMU_INCLUDES) \
|
|
|
|
$(QEMU_CXXFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) $($@-cflags) \
|
|
|
|
-c -o $@ $<,"CXX","$(TARGET_DIR)$@")
|
2014-02-06 01:27:27 +08:00
|
|
|
|
2013-08-07 23:39:36 +08:00
|
|
|
%.o: %.cpp
|
make: move top level dir to end of include search path
Currently the search path is
1. source dir corresponding to input file (implicit by compiler)
2. top level build dir
3. top level source dir
4. top level source include/ dir
5. source dir corresponding to input file
6. build dir corresponding to output file
Search item 5 is an effective no-op, since it duplicates item 1.
When srcdir == builddir, item 6 also duplicates item 1, which
causes a semantic difference between VPATH and non-VPATH builds.
Thus to ensure consistent semantics we need item 6 to be present
immediately after item 1. e.g.
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file
3. top level build dir
4. top level source dir
5. top level source include/ dir
When srcdir == builddir, items 1 & 2 collapse into one, and items
3 & 4 collapse into one, but the overall search order is still
consistent with srcdir != builddir
A further complication is that while most of the source files
are built with a current directory of $BUILD_DIR, target dependant
files are built with a current directory of $BUILD_DIR/$TARGET.
As a result, search item 2 resolves to a different location for
target independant vs target dependant files. For example when
building 'migration/ram.o', the use of '-I$(@D)' (which expands
to '-Imigration') would not find '$BUILD_DIR/migration', but
rather '$BUILD_DIR/$TARGET/migration'.
If there are generated headers files to be used by the migration
code in '$BUILD_DIR/migration', these will not be found by the
relative include, an absolute include is needed instead. This
has not been a problem so far, since nothing has been generating
headers in sub-dirs, but the trace code will shortly be doing
that. So it is needed to list '-I$(BUILD_DIR)/$(@D)' as well as
'-I$(@D)' to ensure both directories are searched when building
target dependant code. So the search order ends up being:
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file (absolute)
3. build dir corresponding to output file (relative to cwd)
4. top level build dir
5. top level source dir
6. top level source include/ dir
One final complication is that the absolute '-I$(BUILD_DIR)/$(@D)'
will sometimes end up pointing to a non-existant directory if
that sub-dir does not have any target-independant files to be
built. Rather than try to dynamically filter this, a simple
'mkdir' ensures $(BUILD_DIR)/$(@D) is guaranteed to exist at
all times.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20170125161417.31949-2-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2017-01-26 00:14:10 +08:00
|
|
|
$(call quiet-command,$(CXX) $(QEMU_LOCAL_INCLUDES) $(QEMU_INCLUDES) \
|
|
|
|
$(QEMU_CXXFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) $($@-cflags) \
|
|
|
|
-c -o $@ $<,"CXX","$(TARGET_DIR)$@")
|
2013-08-07 23:39:36 +08:00
|
|
|
|
2009-01-22 02:12:52 +08:00
|
|
|
%.o: %.m
|
make: move top level dir to end of include search path
Currently the search path is
1. source dir corresponding to input file (implicit by compiler)
2. top level build dir
3. top level source dir
4. top level source include/ dir
5. source dir corresponding to input file
6. build dir corresponding to output file
Search item 5 is an effective no-op, since it duplicates item 1.
When srcdir == builddir, item 6 also duplicates item 1, which
causes a semantic difference between VPATH and non-VPATH builds.
Thus to ensure consistent semantics we need item 6 to be present
immediately after item 1. e.g.
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file
3. top level build dir
4. top level source dir
5. top level source include/ dir
When srcdir == builddir, items 1 & 2 collapse into one, and items
3 & 4 collapse into one, but the overall search order is still
consistent with srcdir != builddir
A further complication is that while most of the source files
are built with a current directory of $BUILD_DIR, target dependant
files are built with a current directory of $BUILD_DIR/$TARGET.
As a result, search item 2 resolves to a different location for
target independant vs target dependant files. For example when
building 'migration/ram.o', the use of '-I$(@D)' (which expands
to '-Imigration') would not find '$BUILD_DIR/migration', but
rather '$BUILD_DIR/$TARGET/migration'.
If there are generated headers files to be used by the migration
code in '$BUILD_DIR/migration', these will not be found by the
relative include, an absolute include is needed instead. This
has not been a problem so far, since nothing has been generating
headers in sub-dirs, but the trace code will shortly be doing
that. So it is needed to list '-I$(BUILD_DIR)/$(@D)' as well as
'-I$(@D)' to ensure both directories are searched when building
target dependant code. So the search order ends up being:
1. source dir corresponding to input file (implicit by compiler)
2. build dir corresponding to output file (absolute)
3. build dir corresponding to output file (relative to cwd)
4. top level build dir
5. top level source dir
6. top level source include/ dir
One final complication is that the absolute '-I$(BUILD_DIR)/$(@D)'
will sometimes end up pointing to a non-existant directory if
that sub-dir does not have any target-independant files to be
built. Rather than try to dynamically filter this, a simple
'mkdir' ensures $(BUILD_DIR)/$(@D) is guaranteed to exist at
all times.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20170125161417.31949-2-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
2017-01-26 00:14:10 +08:00
|
|
|
$(call quiet-command,$(OBJCC) $(QEMU_LOCAL_INCLUDES) $(QEMU_INCLUDES) \
|
|
|
|
$(QEMU_CFLAGS) $(QEMU_DGFLAGS) $(CFLAGS) $($@-cflags) \
|
|
|
|
-c -o $@ $<,"OBJC","$(TARGET_DIR)$@")
|
2009-01-22 02:12:52 +08:00
|
|
|
|
2012-12-21 16:23:18 +08:00
|
|
|
%.o: %.dtrace
|
2016-10-05 00:27:21 +08:00
|
|
|
$(call quiet-command,dtrace -o $@ -G -s $<,"GEN","$(TARGET_DIR)$@")
|
2012-12-21 16:23:18 +08:00
|
|
|
|
2015-05-07 14:55:15 +08:00
|
|
|
DSO_OBJ_CFLAGS := -fPIC -DBUILD_DSO
|
|
|
|
module-common.o: CFLAGS += $(DSO_OBJ_CFLAGS)
|
build-sys: clean up flags included in the linker command line
Some of the CFLAGS that are discovered during configure, for example
compiler warnings, are being included on the linker command line because
QEMU_CFLAGS is added to it. Other flags, such as the -m32, appear twice
because they are included in both QEMU_CFLAGS and LDFLAGS. All this
leads to confusion with respect to what goes in which Makefile variables
(and we have plenty).
So, introduce QEMU_LDFLAGS for flags discovered by configure, following
the lead of QEMU_CFLAGS, and stop adding to it:
1) options that are already in CFLAGS, for example "-g"
2) duplicate options
At the same time, options that _are_ needed by both compiler and linker
must now be added to both QEMU_CFLAGS and QEMU_LDFLAGS, which is clearer.
This is mostly -fsanitize options. For now, --extra-cflags has this behavior
(but --extra-cxxflags does not).
Meson will not include CFLAGS on the linker command line, do the same in our
build system as well.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2019-12-11 22:34:27 +08:00
|
|
|
%$(DSOSUF): QEMU_LDFLAGS += $(LDFLAGS_SHARED)
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
%$(DSOSUF): %.mo
|
2014-02-10 14:48:56 +08:00
|
|
|
$(call LINK,$^)
|
2014-02-10 14:48:57 +08:00
|
|
|
@# Copy to build root so modules can be loaded when program started without install
|
2016-10-05 00:27:21 +08:00
|
|
|
$(if $(findstring /,$@),$(call quiet-command,cp $@ $(subst /,-,$@),"CP","$(subst /,-,$@)"))
|
2014-02-10 14:48:56 +08:00
|
|
|
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
|
2016-11-29 23:37:20 +08:00
|
|
|
LD_REL := $(CC) -nostdlib $(LD_REL_FLAGS)
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
|
|
|
|
%.mo:
|
2016-10-05 00:27:21 +08:00
|
|
|
$(call quiet-command,$(LD_REL) -o $@ $^,"LD","$(TARGET_DIR)$@")
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
|
2014-02-10 14:48:56 +08:00
|
|
|
.PHONY: modules
|
|
|
|
modules:
|
|
|
|
|
2009-01-22 02:13:02 +08:00
|
|
|
%$(EXESUF): %.o
|
2016-02-17 22:59:36 +08:00
|
|
|
$(call LINK,$(filter %.o %.a %.mo, $^))
|
2009-01-22 02:13:09 +08:00
|
|
|
|
2009-01-22 02:13:16 +08:00
|
|
|
%.a:
|
2016-10-05 00:27:21 +08:00
|
|
|
$(call quiet-command,rm -f $@ && $(AR) rcs $@ $^,"AR","$(TARGET_DIR)$@")
|
2009-01-22 02:13:16 +08:00
|
|
|
|
2016-10-05 00:27:21 +08:00
|
|
|
# Usage: $(call quiet-command,command and args,"NAME","args to print")
|
|
|
|
# This will run "command and args", and either:
|
|
|
|
# if V=1 just print the whole command and args
|
|
|
|
# otherwise print the 'quiet' output in the format " NAME args to print"
|
|
|
|
# NAME should be a short name of the command, 7 letters or fewer.
|
|
|
|
# If called with only a single argument, will print nothing in quiet mode.
|
2018-11-30 01:45:31 +08:00
|
|
|
quiet-command-run = $(if $(V),,$(if $2,printf " %-7s %s\n" $2 $3 && ))$1
|
|
|
|
quiet-@ = $(if $(V),,@)
|
|
|
|
quiet-command = $(quiet-@)$(call quiet-command-run,$1,$2,$3)
|
2009-07-21 20:11:18 +08:00
|
|
|
|
|
|
|
# cc-option
|
2009-08-31 06:48:45 +08:00
|
|
|
# Usage: CFLAGS+=$(call cc-option, -falign-functions=0, -malign-functions=0)
|
2009-07-21 20:11:18 +08:00
|
|
|
|
2009-09-12 00:45:40 +08:00
|
|
|
cc-option = $(if $(shell $(CC) $1 $2 -S -o /dev/null -xc /dev/null \
|
|
|
|
>/dev/null 2>&1 && echo OK), $2, $3)
|
2016-07-21 03:36:49 +08:00
|
|
|
cc-c-option = $(if $(shell $(CC) $1 $2 -c -o /dev/null -xc /dev/null \
|
|
|
|
>/dev/null 2>&1 && echo OK), $2, $3)
|
2009-10-07 08:40:58 +08:00
|
|
|
|
2019-05-24 21:09:42 +08:00
|
|
|
VPATH_SUFFIXES = %.c %.h %.S %.cc %.cpp %.m %.mak %.texi %.sh %.rc Kconfig% %.json.in
|
2010-04-27 05:52:23 +08:00
|
|
|
set-vpath = $(if $1,$(foreach PATTERN,$(VPATH_SUFFIXES),$(eval vpath $(PATTERN) $1)))
|
2009-12-21 17:06:55 +08:00
|
|
|
|
2014-06-22 14:55:23 +08:00
|
|
|
# install-prog list, dir
|
|
|
|
define install-prog
|
|
|
|
$(INSTALL_DIR) "$2"
|
|
|
|
$(INSTALL_PROG) $1 "$2"
|
|
|
|
$(if $(STRIP),$(STRIP) $(foreach T,$1,"$2/$(notdir $T)"),)
|
|
|
|
endef
|
|
|
|
|
2010-10-21 16:18:40 +08:00
|
|
|
# find-in-path
|
|
|
|
# Usage: $(call find-in-path, prog)
|
|
|
|
# Looks in the PATH if the argument contains no slash, else only considers one
|
|
|
|
# specific directory. Returns an # empty string if the program doesn't exist
|
|
|
|
# there.
|
2016-09-23 20:35:08 +08:00
|
|
|
find-in-path = $(if $(findstring /, $1), \
|
2010-10-21 16:18:40 +08:00
|
|
|
$(wildcard $1), \
|
|
|
|
$(wildcard $(patsubst %, %/$1, $(subst :, ,$(PATH)))))
|
|
|
|
|
2013-09-14 01:25:51 +08:00
|
|
|
# Logical functions (for operating on y/n values like CONFIG_FOO vars)
|
|
|
|
# Inputs to these must be either "y" (true) or "n" or "" (both false)
|
|
|
|
# Output is always either "y" or "n".
|
|
|
|
# Usage: $(call land,$(CONFIG_FOO),$(CONFIG_BAR))
|
|
|
|
# Logical NOT
|
|
|
|
lnot = $(if $(subst n,,$1),n,y)
|
|
|
|
# Logical AND
|
|
|
|
land = $(if $(findstring yy,$1$2),y,n)
|
|
|
|
# Logical OR
|
|
|
|
lor = $(if $(findstring y,$1$2),y,n)
|
|
|
|
# Logical XOR (note that this is the inverse of leqv)
|
|
|
|
lxor = $(if $(filter $(call lnot,$1),$(call lnot,$2)),n,y)
|
|
|
|
# Logical equivalence (note that leqv "","n" is true)
|
|
|
|
leqv = $(if $(filter $(call lnot,$1),$(call lnot,$2)),y,n)
|
|
|
|
# Logical if: like make's $(if) but with an leqv-like test
|
|
|
|
lif = $(if $(subst n,,$1),$2,$3)
|
|
|
|
|
2013-09-14 01:25:52 +08:00
|
|
|
# String testing functions: inputs to these can be any string;
|
|
|
|
# the output is always either "y" or "n". Leading and trailing whitespace
|
|
|
|
# is ignored when comparing strings.
|
|
|
|
# String equality
|
|
|
|
eq = $(if $(subst $2,,$1)$(subst $1,,$2),n,y)
|
|
|
|
# String inequality
|
|
|
|
ne = $(if $(subst $2,,$1)$(subst $1,,$2),y,n)
|
|
|
|
# Emptiness/non-emptiness tests:
|
|
|
|
isempty = $(if $1,n,y)
|
|
|
|
notempty = $(if $1,y,n)
|
|
|
|
|
2012-04-19 02:15:45 +08:00
|
|
|
# Generate files with tracetool
|
|
|
|
TRACETOOL=$(PYTHON) $(SRC_PATH)/scripts/tracetool.py
|
|
|
|
|
2009-10-07 08:40:58 +08:00
|
|
|
# Generate timestamp files for .h include files
|
|
|
|
|
2013-01-15 19:12:35 +08:00
|
|
|
config-%.h: config-%.h-timestamp
|
|
|
|
@cmp $< $@ >/dev/null 2>&1 || cp $< $@
|
2009-10-07 08:40:58 +08:00
|
|
|
|
2016-06-07 19:25:24 +08:00
|
|
|
config-%.h-timestamp: config-%.mak $(SRC_PATH)/scripts/create_config
|
2016-10-05 00:27:21 +08:00
|
|
|
$(call quiet-command, sh $(SRC_PATH)/scripts/create_config < $< > $@,"GEN","$(TARGET_DIR)config-$*.h")
|
2009-12-08 03:04:52 +08:00
|
|
|
|
2013-01-15 19:27:54 +08:00
|
|
|
.PHONY: clean-timestamp
|
|
|
|
clean-timestamp:
|
|
|
|
rm -f *.timestamp
|
|
|
|
clean: clean-timestamp
|
|
|
|
|
2009-12-08 03:04:52 +08:00
|
|
|
# will delete the target of a rule if commands exit with a nonzero exit status
|
|
|
|
.DELETE_ON_ERROR:
|
2012-05-22 19:41:27 +08:00
|
|
|
|
2014-05-27 15:54:19 +08:00
|
|
|
# save-vars
|
|
|
|
# Usage: $(call save-vars, vars)
|
|
|
|
# Save each variable $v in $vars as save-vars-$v, save their object's
|
2016-11-02 23:10:23 +08:00
|
|
|
# variables, then clear $v. saved-vars-$v contains the variables that
|
|
|
|
# where saved for the objects, in order to speedup load-vars.
|
2014-05-27 15:54:19 +08:00
|
|
|
define save-vars
|
|
|
|
$(foreach v,$1,
|
|
|
|
$(eval save-vars-$v := $(value $v))
|
2016-11-02 23:10:23 +08:00
|
|
|
$(eval saved-vars-$v := $(foreach o,$($v), \
|
|
|
|
$(if $($o-cflags), $o-cflags $(eval save-vars-$o-cflags := $($o-cflags))$(eval $o-cflags := )) \
|
|
|
|
$(if $($o-libs), $o-libs $(eval save-vars-$o-libs := $($o-libs))$(eval $o-libs := )) \
|
|
|
|
$(if $($o-objs), $o-objs $(eval save-vars-$o-objs := $($o-objs))$(eval $o-objs := ))))
|
2014-05-27 15:54:19 +08:00
|
|
|
$(eval $v := ))
|
2014-02-10 14:48:53 +08:00
|
|
|
endef
|
|
|
|
|
2014-05-27 15:54:19 +08:00
|
|
|
# load-vars
|
|
|
|
# Usage: $(call load-vars, vars, add_var)
|
|
|
|
# Load the saved value for each variable in @vars, and the per object
|
|
|
|
# variables.
|
|
|
|
# Append @add_var's current value to the loaded value.
|
|
|
|
define load-vars
|
|
|
|
$(eval $2-new-value := $(value $2))
|
|
|
|
$(foreach v,$1,
|
|
|
|
$(eval $v := $(value save-vars-$v))
|
2016-11-02 23:10:23 +08:00
|
|
|
$(foreach o,$(saved-vars-$v),
|
|
|
|
$(eval $o := $(save-vars-$o)) $(eval save-vars-$o := ))
|
|
|
|
$(eval save-vars-$v := )
|
|
|
|
$(eval saved-vars-$v := ))
|
2014-05-27 15:54:19 +08:00
|
|
|
$(eval $2 := $(value $2) $($2-new-value))
|
2012-05-22 19:41:27 +08:00
|
|
|
endef
|
|
|
|
|
2014-05-27 15:54:19 +08:00
|
|
|
# fix-paths
|
|
|
|
# Usage: $(call fix-paths, obj_path, src_path, vars)
|
|
|
|
# Add prefix @obj_path to all objects in @vars, and add prefix @src_path to all
|
|
|
|
# directories in @vars.
|
|
|
|
define fix-paths
|
|
|
|
$(foreach v,$3,
|
|
|
|
$(foreach o,$($v),
|
|
|
|
$(if $($o-libs),
|
|
|
|
$(eval $1$o-libs := $($o-libs)))
|
|
|
|
$(if $($o-cflags),
|
|
|
|
$(eval $1$o-cflags := $($o-cflags)))
|
|
|
|
$(if $($o-objs),
|
|
|
|
$(eval $1$o-objs := $(addprefix $1,$($o-objs)))))
|
|
|
|
$(eval $v := $(addprefix $1,$(filter-out %/,$($v))) \
|
|
|
|
$(addprefix $2,$(filter %/,$($v)))))
|
2012-05-22 19:41:27 +08:00
|
|
|
endef
|
|
|
|
|
2014-05-27 15:54:19 +08:00
|
|
|
# unnest-var-recursive
|
|
|
|
# Usage: $(call unnest-var-recursive, obj_prefix, vars, var)
|
|
|
|
#
|
|
|
|
# Unnest @var by including subdir Makefile.objs, while protect others in @vars
|
|
|
|
# unchanged.
|
|
|
|
#
|
|
|
|
# @obj_prefix is the starting point of object path prefix.
|
|
|
|
#
|
|
|
|
define unnest-var-recursive
|
|
|
|
$(eval dirs := $(sort $(filter %/,$($3))))
|
|
|
|
$(eval $3 := $(filter-out %/,$($3)))
|
|
|
|
$(foreach d,$(dirs:%/=%),
|
|
|
|
$(call save-vars,$2)
|
|
|
|
$(eval obj := $(if $1,$1/)$d)
|
|
|
|
$(eval -include $(SRC_PATH)/$d/Makefile.objs)
|
|
|
|
$(call fix-paths,$(if $1,$1/)$d/,$d/,$2)
|
|
|
|
$(call load-vars,$2,$3)
|
|
|
|
$(call unnest-var-recursive,$1,$2,$3))
|
2014-02-10 14:48:56 +08:00
|
|
|
endef
|
|
|
|
|
2014-05-27 15:54:19 +08:00
|
|
|
# unnest-vars
|
|
|
|
# Usage: $(call unnest-vars, obj_prefix, vars)
|
|
|
|
#
|
|
|
|
# @obj_prefix: object path prefix, can be empty, or '..', etc. Don't include
|
|
|
|
# ending '/'.
|
|
|
|
#
|
|
|
|
# @vars: the list of variable names to unnest.
|
|
|
|
#
|
|
|
|
# This macro will scan subdirectories's Makefile.objs, include them, to build
|
|
|
|
# up each variable listed in @vars.
|
|
|
|
#
|
|
|
|
# Per object and per module cflags and libs are saved with relative path fixed
|
|
|
|
# as well, those variables include -libs, -cflags and -objs. Items in -objs are
|
|
|
|
# also fixed to relative path against SRC_PATH plus the prefix @obj_prefix.
|
|
|
|
#
|
|
|
|
# All nested variables postfixed by -m in names are treated as DSO variables,
|
|
|
|
# and will be built as modules, if enabled.
|
|
|
|
#
|
|
|
|
# A simple example of the unnest:
|
|
|
|
#
|
|
|
|
# obj_prefix = ..
|
|
|
|
# vars = hot cold
|
|
|
|
# hot = fire.o sun.o season/
|
|
|
|
# cold = snow.o water/ season/
|
|
|
|
#
|
|
|
|
# Unnest through a faked source directory structure:
|
|
|
|
#
|
|
|
|
# SRC_PATH
|
|
|
|
# ├── water
|
|
|
|
# │ └── Makefile.objs──────────────────┐
|
|
|
|
# │ │ hot += steam.o │
|
|
|
|
# │ │ cold += ice.mo │
|
|
|
|
# │ │ ice.mo-libs := -licemaker │
|
|
|
|
# │ │ ice.mo-objs := ice1.o ice2.o │
|
|
|
|
# │ └──────────────────────────────┘
|
|
|
|
# │
|
|
|
|
# └── season
|
|
|
|
# └── Makefile.objs──────┐
|
|
|
|
# │ hot += summer.o │
|
|
|
|
# │ cold += winter.o │
|
|
|
|
# └──────────────────┘
|
|
|
|
#
|
|
|
|
# In the end, the result will be:
|
|
|
|
#
|
|
|
|
# hot = ../fire.o ../sun.o ../season/summer.o
|
|
|
|
# cold = ../snow.o ../water/ice.mo ../season/winter.o
|
|
|
|
# ../water/ice.mo-libs = -licemaker
|
|
|
|
# ../water/ice.mo-objs = ../water/ice1.o ../water/ice2.o
|
|
|
|
#
|
2019-02-13 21:02:40 +08:00
|
|
|
# Note that 'hot' didn't include 'water/' in the input, so 'steam.o' is not
|
2014-05-27 15:54:19 +08:00
|
|
|
# included.
|
|
|
|
#
|
2012-05-22 19:41:27 +08:00
|
|
|
define unnest-vars
|
2014-05-27 15:54:19 +08:00
|
|
|
# In the case of target build (i.e. $1 == ..), fix path for top level
|
|
|
|
# Makefile.objs objects
|
|
|
|
$(if $1,$(call fix-paths,$1/,,$2))
|
|
|
|
|
|
|
|
# Descend and include every subdir Makefile.objs
|
2015-01-12 12:43:09 +08:00
|
|
|
$(foreach v, $2,
|
|
|
|
$(call unnest-var-recursive,$1,$2,$v)
|
|
|
|
# Pass the .mo-cflags and .mo-libs along to its member objects
|
|
|
|
$(foreach o, $(filter %.mo,$($v)),
|
|
|
|
$(foreach p,$($o-objs),
|
|
|
|
$(if $($o-cflags), $(eval $p-cflags += $($o-cflags)))
|
|
|
|
$(if $($o-libs), $(eval $p-libs += $($o-libs))))))
|
|
|
|
|
|
|
|
# For all %.mo objects that are directly added into -y, just expand them
|
|
|
|
$(foreach v,$(filter %-y,$2),
|
|
|
|
$(eval $v := $(foreach o,$($v),$(if $($o-objs),$($o-objs),$o))))
|
2014-05-27 15:54:19 +08:00
|
|
|
|
|
|
|
$(foreach v,$(filter %-m,$2),
|
|
|
|
# All .o found in *-m variables are single object modules, create .mo
|
|
|
|
# for them
|
|
|
|
$(foreach o,$(filter %.o,$($v)),
|
|
|
|
$(eval $(o:%.o=%.mo)-objs := $o))
|
|
|
|
# Now unify .o in -m variable to .mo
|
|
|
|
$(eval $v := $($v:%.o=%.mo))
|
|
|
|
$(eval modules-m += $($v))
|
|
|
|
|
|
|
|
# For module build, build shared libraries during "make modules"
|
|
|
|
# For non-module build, add -m to -y
|
|
|
|
$(if $(CONFIG_MODULES),
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
$(foreach o,$($v),
|
2015-05-07 14:55:15 +08:00
|
|
|
$(eval $($o-objs): CFLAGS += $(DSO_OBJ_CFLAGS))
|
rules.mak: Fix DSO build by pulling in archive symbols
This fixes an issue with module build system. block/iscsi.so is
currently broken:
$ ~/build/last/qemu-img
Failed to open module: /home/fam/build/master/block-iscsi.so:
undefined symbol: qmp_query_uuid
qemu-img: Not enough arguments
Try 'qemu-img --help' for more information
To fix this, we should (at least) let qemu-img link qmp_query_uuid from
libqemustub.a. (There are a few other symbols missing, as well.)
This patch changes the linking rules to:
1) Build ".mo" with "ld -r -o $@ $^" for each ".so", and later build .so
with it.
2) Always build all the .mo before linking the executables. This is
achieved by adding those .mo files to the executables' "-y"
variables.
3) When linking an executable, those .mo files in its "-y" variables are
filtered out, and replaced by one or more -Wl,-u,$symbol flags. This
is done in the added macro "process-archive-undefs".
These "-Wl,-u,$symbol" flags will force ld to pull in the function
definition from the archives when linking.
Note that the .mo objects, that are actually meant to be linked in
the executables, are already expanded in unnest-vars, before the
linking command. So we are safe to simply filter out .mo for the
purpose of pulling undefined symbols.
process-archive-undefs works as this: For each ".mo", find all the
undefined symbols in it, filter ones that are defined in the
archives. For each of these symbols, generate a "-Wl,-u,$symbol" in
the link command, and put them before archive names in the command
line.
Suggested-by: H.J. Lu <hjl.tools@gmail.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2014-09-01 18:35:10 +08:00
|
|
|
$(eval $o: $($o-objs)))
|
|
|
|
$(eval $(patsubst %-m,%-y,$v) += $($v))
|
2014-05-27 15:54:19 +08:00
|
|
|
$(eval modules: $($v:%.mo=%$(DSOSUF))),
|
|
|
|
$(eval $(patsubst %-m,%-y,$v) += $(call expand-objs, $($v)))))
|
|
|
|
|
|
|
|
# Post-process all the unnested vars
|
|
|
|
$(foreach v,$2,
|
|
|
|
$(foreach o, $(filter %.mo,$($v)),
|
|
|
|
# Find all the .mo objects in variables and add dependency rules
|
|
|
|
# according to .mo-objs. Report error if not set
|
|
|
|
$(if $($o-objs),
|
|
|
|
$(eval $(o:%.mo=%$(DSOSUF)): module-common.o $($o-objs)),
|
2015-01-12 12:43:09 +08:00
|
|
|
$(error $o added in $v but $o-objs is not set)))
|
2014-05-27 15:54:19 +08:00
|
|
|
$(shell mkdir -p ./ $(sort $(dir $($v))))
|
|
|
|
# Include all the .d files
|
2015-08-09 17:39:59 +08:00
|
|
|
$(eval -include $(patsubst %.o,%.d,$(patsubst %.mo,%.d,$($v))))
|
2014-05-27 15:54:19 +08:00
|
|
|
$(eval $v := $(filter-out %/,$($v))))
|
2012-05-22 19:41:27 +08:00
|
|
|
endef
|
2017-01-13 22:41:33 +08:00
|
|
|
|
|
|
|
TEXI2MAN = $(call quiet-command, \
|
2017-06-06 22:55:19 +08:00
|
|
|
perl -Ww -- $(SRC_PATH)/scripts/texi2pod.pl $(TEXI2PODFLAGS) $< $@.pod && \
|
2017-01-13 22:41:33 +08:00
|
|
|
$(POD2MAN) --section=$(subst .,,$(suffix $@)) --center=" " --release=" " $@.pod > $@, \
|
|
|
|
"GEN","$@")
|
|
|
|
|
|
|
|
%.1:
|
|
|
|
$(call TEXI2MAN)
|
2017-01-13 22:41:35 +08:00
|
|
|
%.7:
|
|
|
|
$(call TEXI2MAN)
|
2017-01-13 22:41:33 +08:00
|
|
|
%.8:
|
|
|
|
$(call TEXI2MAN)
|
2019-05-24 21:09:42 +08:00
|
|
|
|
|
|
|
GEN_SUBST = $(call quiet-command, \
|
|
|
|
sed -e "s!@libexecdir@!$(libexecdir)!g" < $< > $@, \
|
|
|
|
"GEN","$@")
|
|
|
|
|
|
|
|
%.json: %.json.in
|
|
|
|
$(call GEN_SUBST)
|
2020-01-25 00:25:59 +08:00
|
|
|
|
|
|
|
# Support for building multiple output files by atomically executing
|
|
|
|
# a single rule which depends on several input files (so the rule
|
|
|
|
# will be executed exactly once, not once per output file, and
|
|
|
|
# not multiple times in parallel.) For more explanation see:
|
|
|
|
# https://www.cmcrossroads.com/article/atomic-rules-gnu-make
|
|
|
|
|
|
|
|
# Given a space-separated list of filenames, create the name of
|
|
|
|
# a 'sentinel' file to use to indicate that they have been built.
|
|
|
|
# We use fixed text on the end to avoid accidentally triggering
|
|
|
|
# automatic pattern rules, and . on the start to make the file
|
|
|
|
# not show up in ls output.
|
|
|
|
sentinel = .$(subst $(SPACE),_,$(subst /,_,$1)).sentinel.
|
|
|
|
|
|
|
|
# Define an atomic rule that builds multiple outputs from multiple inputs.
|
|
|
|
# To use:
|
|
|
|
# $(call atomic,out1 out2 ...,in1 in2 ...)
|
|
|
|
# <TAB>rule to do the operation
|
|
|
|
#
|
|
|
|
# Make 4.3 will have native support for this, and you would be able
|
|
|
|
# to instead write:
|
|
|
|
# out1 out2 ... &: in1 in2 ...
|
|
|
|
# <TAB>rule to do the operation
|
|
|
|
#
|
|
|
|
# The way this works is that it creates a make rule
|
|
|
|
# "out1 out2 ... : sentinel-file ; @:" which says that the sentinel
|
|
|
|
# depends on the dependencies, and the rule to do that is "do nothing".
|
|
|
|
# Then we have a rule
|
|
|
|
# "sentinel-file : in1 in2 ..."
|
|
|
|
# whose commands start with "touch sentinel-file" and then continue
|
|
|
|
# with the rule text provided by the user of this 'atomic' function.
|
|
|
|
# The foreach... is there to delete the sentinel file if any of the
|
|
|
|
# output files don't exist, so that we correctly rebuild in that situation.
|
|
|
|
atomic = $(eval $1: $(call sentinel,$1) ; @:) \
|
|
|
|
$(call sentinel,$1) : $2 ; @touch $$@ \
|
|
|
|
$(foreach t,$1,$(if $(wildcard $t),,$(shell rm -f $(call sentinel,$1))))
|
2020-03-07 01:04:56 +08:00
|
|
|
|
|
|
|
print-%:
|
|
|
|
@echo '$*=$($*)'
|