$(if $(KBUILD_SRC),$(srctree)/) was a useful strategy
to omit a long absolute path for in-source-tree build
prior to commit 890676c6
(kbuild: Use relative path when building in the source tree).

Now $(srctree) is "." when building in the source tree.
It would not be annoying to add "$(srctree)/" all the time.
Signed-off-by: NMasahiro Yamada <yamada.m@jp.panasonic.com>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

Commit c353acba ("kbuild: make: fix if_changed when command contains
backslashes") attempted to handle backslashes in *.cmd files, but it
only handled double backslashes for some reason. Changing make-cmd to also
handle single backslashes fixes rebuilds with dash, but it breaks bash
again. The reason is that the two shells disagree about the
interpretation of backslash sequences in the echo builtin. The way out
of this is to print the command with printf '%s\n'. While at it,
document what the individual parts of make-cmd do and why.
Reported-and-tested-by: NKonstantin Khlebnikov <koct9i@gmail.com>
Reviewed-by: NSam Ravnborg <sam@ravnborg.org>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

Signed-off-by: NMasahiro Yamada <yamada.m@jp.panasonic.com>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

To check the linker version. Used by the LTO makefile.
Signed-off-by: NAndi Kleen <ak@linux.intel.com>
Link: http://lkml.kernel.org/r/1391846481-31491-9-git-send-email-ak@linux.intel.comSigned-off-by: NH. Peter Anvin <hpa@linux.intel.com>

The kbuild's ld-option function is broken because
the command
  $(CC) /dev/null -c -o "$$TMPO"
does not create object file!

I have used a relatively old mips gcc 3.4.6 cross-compiler
and a relatively new gcc 4.7.2 to check this fact
but the results are the same.

EXAMPLE:
  $ rm /tmp/1.o
  $ mips-linux-gcc /dev/null -c -o /tmp/1.o
  mips-linux-gcc: /dev/null: linker input file unused because linking not done
  $ ls -la /tmp/1.o
  ls: cannot access /tmp/1.o: No such file or directory

We can easily fix the problem by adding
the '-x c' compiler option.

EXAMPLE:
  $ rm /tmp/1.o
  $ mips-linux-gcc -x c /dev/null -c -o /tmp/1.o
  $ ls -la /tmp/1.o
  -rw-r--r-- 1 antony antony 778 Apr  2 20:40 /tmp/1.o

Also fix wrong ld-option example.
Signed-off-by: NAntony Pavlov <antonynpavlov@gmail.com>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

The call if_changed mechanism does not work when the command contains
backslashes.  This basically is an issue with lzo and bzip2 compressed
kernels.  The compressed binaries do not contain the uncompressed image
size, so these use size_append to append the size.  This results in
backslashes in the executed command.  With this if_changed always
detects a change in the command and rebuilds the compressed image even
if nothing has changed.

Fix this by escaping backslashes in make-cmd
Signed-off-by: NSascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: NJan Luebbe <jlu@pengutronix.de>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Bernhard Walle <bernhard@bwalle.de>
Cc: Michal Marek <mmarek@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The correct syntax for gcc -x is "gcc -x assembler", not
"gcc -xassembler". Even though the latter happens to work, the former
is what is documented in the manual page and thus what gcc wrappers
such as icecream do expect.

This isn't a cosmetic change. The missing space prevents icecream from
recognizing compilation tasks it can't handle, leading to silent kernel
miscompilations.

Besides me, credits go to Michael Matz and Dirk Mueller for
investigating the miscompilation issue and tracking it down to this
incorrect -x parameter syntax.
Signed-off-by: NJean Delvare <jdelvare@suse.de>
Acked-by: NIngo Molnar <mingo@kernel.org>
Cc: stable@vger.kernel.org
Cc: Bernhard Walle <bernhard@bwalle.de>
Cc: Michal Marek <mmarek@suse.cz>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

"echo -e" is a GNU extension. When cross-compiling the kernel on a
BSD-like operating system (Mac OS X in my case), this doesn't work.

One could install a GNU version of echo, put that in the $PATH before
the system echo and use "/usr/bin/env echo", but the solution with
printf is simpler.

Since it is no disadvantage on Linux, I hope that gets accepted even if
cross-compiling the Linux kernel on another Unix operating system is
quite a rare use case.
Signed-off-by: NBernhard Walle <bernhard@bwalle.de>
Andreas Bießmann <andreas@biessmann.de>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

On the same model as `basetarget', it represents the filename of first
prerequisite with directory and extension stripped.
Signed-off-by: NArnaud Lacombe <lacombar@gmail.com>

Based on a patch by Rabin Vincent.

Fix building with KBUILD_NOCMDDEP=1, which currently does not work
because it does not build built-in.o with no dependencies:

  LD      fs/notify/built-in.o
ld: cannot find fs/notify/dnotify/built-in.o: No such file or directory
ld: cannot find fs/notify/inotify/built-in.o: No such file or directory
ld: cannot find fs/notify/fanotify/built-in.o: No such file or directory
Reported-and-tested-by: NRabin Vincent <rabin@rab.in>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

Starting with 4.4, gcc will happily accept -Wno-<anything> in the
cc-option test and complain later when compiling a file that has some
other warning. This rather unexpected behavior is intentional as per
http://gcc.gnu.org/PR28322, so work around it by testing for support of
the opposite option (without the no-). Introduce a new Makefile function
cc-disable-warning that does this and update two uses of cc-option in
the toplevel Makefile.
Reported-and-tested-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NMichal Marek <mmarek@suse.cz>

The D option of ar is only available in newer versions.
Signed-off-by: NMichal Marek <mmarek@suse.cz>

To make it easier for module-init-tools and scripts like mkinitrd to
distinguish builtin and missing modules, install a modules.builtin file
listing all builtin modules. This is done by generating an additional
config file (tristate.conf) with tristate options set to uppercase 'Y'
or 'M'. If we source that config file, the builtin modules appear in
obj-Y.
Signed-off-by: NMichal Marek <mmarek@suse.cz>

Alek reported that on Ubuntu, where dash is used, 'echo -e'
can't work, so let's use non-builtin echo in this case.
Reported-by: NAlek Du <alek.du@intel.com>
Signed-off-by: NWANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

The arch/*/boot/Makefile use cc-options to check for GCC command options
and cc-options use the hardened specs when checking for GCC command
options.  When -fPIE is pass to cc1 it can't use -ffreestanding or
-fno-toplevel-reorder.  Then it fail to build stuff with -ffreestanding
and -fno-toplevel-reorder.

Thanks to Fredric Johansson for finding the main problem behind a failed
build using a hardened toolchain.
Signed-off-by: NMagnus Granberg <zorry@ume.nu>
Signed-off-by: NJory A. Pratt <anarchy@gentoo.org>
Cc: Fredric Johansson <johansson_fredric@hotmail.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Cc: <stable@kernel.org>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

ld-option is used to check if $(LD) supports a specific option.

Based on patch from Andi Kleen.

Cc: Andi Kleen <ak@linux.intel.com>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>
First use is to check if option -X is supported (upcoming patch).
Theis is ne

ld-option is misnamed as it test options to gcc, not to ld.
Renamed it to reflect this.

Cc: Andi Kleen <andi@firstfloor.org>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

Convert a few echos in the build system to new $(kecho) so we get correct
output according to build verbosity.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
[sam: added kecho in a few more places for O=... builds]
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

There is a bunch of places in the build system where we do 'echo' to show
some nice status lines.  This means we still get output when running in
silent mode.  So declare a new KECHO variable that only does 'echo' when we
are in a suitable verbose build mode.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
[sam: added Documentation]
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

When adding extra -I options with O=... we could
end up in a situation where there were no parameters to -I.
So we had a commandline that looked like this:

    ... -I -Wall ...

This had the undesired side effect that gcc assumed "-Wall"
was a path to look for include files so this options was
effectively ignored.

This happens only when we build the generated module.mod.c files
as part of the final modules builds and is as such harmless
with current kbuild.
This bug was exposed when we rearranged the options to gcc.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

David Sanders wrote:

> I'm getting this error:
> as: unrecognized option `-mtune=generic32'
> I have binutils 2.17.

Use -c instead of -S in cc-option and cc-option-yn, so we can probe
options related to the assembler.
Signed-off-by: NH. Peter Anvin <hpa@zytor.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: kbuild devel <kbuild-devel@lists.sourceforge.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

When make -s support were added to filechk to
combination created with make V=1 were not
covered.
Fix it by explicitly cover this case too.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>
Cc: Mike Frysinger <vapier@gentoo.org>

Signed-off-by: NMike Frysinger <vapier@gentoo.org>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

cc-cross-prefix is useful for the architecture that like
to provide a default CROSS_COMPILE value,
but may have several to select between.

Sample usage:

ifneq ($(SUBARCH),$(ARCH))
        ifeq ($(CROSS_COMPILE),)
               CROSS_COMPILE := $(call cc-cross-prefix, m68k-linux-gnu- m68k-linux-)
        endif
endif

Actual usage by the different archs will taken care of later.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

The variable AFLAGS is a wellknown variable and the usage by
kbuild may result in unexpected behaviour.
On top of that several people over time has asked for a way to
pass in additional flags to gcc.

This patch replace use of AFLAGS with KBUILD_AFLAGS all over
the tree.

Patch was tested on following architectures:
alpha, arm, i386, x86_64, mips, sparc, sparc64, ia64, m68k, s390
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

The variable CFLAGS is a wellknown variable and the usage by
kbuild may result in unexpected behaviour.
On top of that several people over time has asked for a way to
pass in additional flags to gcc.

This patch replace use of CFLAGS with KBUILD_CFLAGS all over the
tree and enabling one to use:
make CFLAGS=...
to specify additional gcc commandline options.

One usecase is when trying to find gcc bugs but other
use cases has been requested too.

Patch was tested on following architectures:
alpha, arm, i386, x86_64, mips, sparc, sparc64, ia64, m68k

Test was simple to do a defconfig build, apply the patch and check
that nothing got rebuild.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

Prints a six-digit string including the GCC patchlevel.  Also fix
the 'usage' comment for cc-version.
Signed-off-by: NSegher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

This adds the remaining changes which should have been part of the
review process.

 - the define command is inappropriate (it's primarily for rule
   definitions)
 - execute commands in the current dir as all other commands
 - .*.tmp (but not .*.null) files are also removed up by "make clean"
 - printf has other side effects, just use "echo -e"
 - proper quoting
 - proper indentation
Signed-off-by: NRoman Zippel <zippel@linux-m68k.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Michal Ostrowski points out what the real problem was: the spaces at the
start of the definition of the 'checker-shell' make function.

Cc: Michal Ostrowski <mostrows@watson.ibm.com>
Acked-by: NDavid Miller <davem@davemloft.net>
Acked-by: NGeert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Acked-by: NOleg Verych <olecom@flower.upol.cz>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

It looks like GNU make version 3.80 (but apparently not 3.81) adds
leading whitespace to the result of the checker-shell execution.  This
strips them off explicitly.

Also, don't bother symlinking the output file to /dev/null.  It's likely
as expensive as just writing the temp-file, and we need to remove it
anyway afterwards.
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Do not use whitespace in arguments of functions in makefiles, as they
propagate further without notice.  Thus we get

  + echo ' y'

  instead of

  + echo y

Fix misleading comments.
Signed-off-by: NOleg Verych <olecom@flower.upol.cz>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

 GNU binutils, root users, tmpfiles, external modules ro builds must
 be fixed to do the right thing now.

Cc: Roman Zippel <zippel@linux-m68k.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Horst Schirmeier <horst@schirmeier.com>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Daniel Drake <dsd@gentoo.org>
Cc: Andi Kleen <ak@suse.de>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NOleg Verych <olecom@flower.upol.cz>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The as-instr/ld-option need to create temporary files, but create them in the
output directory, when compiling external modules.  Reformat them a bit and
use $(CC) instead of $(AS) as the former is used by kbuild to assemble files.
Signed-off-by: NRoman Zippel <zippel@linux-m68k.org>
Cc: Andi Kleen <ak@suse.de>
Cc: Jan Beulich <jbeulich@novell.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: <jpdenheijer@gmail.com>
Cc: Horst Schirmeier <horst@schirmeier.com>
Cc: Daniel Drake <dsd@gentoo.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Current gcc generates calls not jumps to noreturn functions. When that happens the
return address can point to the next function, which confuses the unwinder.

This patch works around it by marking asynchronous exception
frames in contrast normal call frames in the unwind information.  Then teach
the unwinder to decode this.

For normal call frames the unwinder now subtracts one from the address which avoids
this problem.  The standard libgcc unwinder uses the same trick.

It doesn't include adjustment of the printed address (i.e. for the original
example, it'd still be kernel_math_error+0 that gets displayed, but the
unwinder wouldn't get confused anymore.

This only works with binutils 2.6.17+ and some versions of H.J.Lu's 2.6.16
unfortunately because earlier binutils don't support .cfi_signal_frame

[AK: added automatic detection of the new binutils and wrote description]
Signed-off-by: NJan Beulich <jbeulich@novell.com>
Signed-off-by: NAndi Kleen <ak@suse.de>

... instead of using a CONFIG option. The config option still controls
if the resulting executable actually has unwind information.

This is useful to prevent compilation errors when users select
CONFIG_STACK_UNWIND on old binutils and also allows to use
CFI in the future for non kernel debugging applications.

Cc: jbeulich@novell.com
Cc: sam@ravnborg.org
Signed-off-by: NAndi Kleen <ak@suse.de>

tell why a a target got build
   enabled by make V=2
      Output (listed in the order they are checked):
         (1) - due to target is PHONY
         (2) - due to target missing
         (3) - due to: file1.h file2.h
         (4) - due to command line change
         (5) - due to missing .cmd file
         (6) - due to target not in $(targets)
(1) We always build PHONY targets
(2) No target, so we better build it
(3) Prerequisite is newer than target
(4) The command line stored in the file named dir/.target.cmd
    differed from actual command line. This happens when compiler
    options changes
(5) No dir/.target.cmd file (used to store command line)
(6) No dir/.target.cmd file and target not listed in $(targets)
    This is a good hint that there is a bug in the kbuild file

This patch is inspired by a patch from: Milton Miller <miltonm@bga.com>

Cc: Milton Miller <miltonm@bga.com>
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

Consistently decide when to rebuild a target across all of
if_changed, if_changed_dep, if_changed_rule.
PHONY targets are now treated alike (ignored) for all targets

While add it make Kbuild.include almost readable by factoring out a few
bits to some common variables and reuse this in Makefile.build.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

The possibility to specify an optional parameter did not work out as
expected and it was not used - so remove the possibility.
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>

The latest toolchains can produce a new ELF section in DSOs and
dynamically-linked executables.  The new section ".gnu.hash" replaces
".hash", and allows for more efficient runtime symbol lookups by the
dynamic linker.  The new ld option --hash-style={sysv|gnu|both} controls
whether to produce the old ".hash", the new ".gnu.hash", or both.  In some
new systems such as Fedora Core 6, gcc by default passes --hash-style=gnu
to the linker, so that a standard invocation of "gcc -shared" results in
producing a DSO with only ".gnu.hash".  The new ".gnu.hash" sections need
to be dealt with the same way as ".hash" sections in all respects; only the
dynamic linker cares about their contents.  To work with older dynamic
linkers (i.e.  preexisting releases of glibc), a binary must have the old
".hash" section.  The --hash-style=both option produces binaries that a new
dynamic linker can use more efficiently, but an old dynamic linker can
still handle.

The new section runs afoul of the custom linker scripts used to build vDSO
images for the kernel.  On ia64, the failure mode for this is a boot-time
panic because the vDSO's PT_IA_64_UNWIND segment winds up ill-formed.

This patch addresses the problem in two ways.

First, it mentions ".gnu.hash" in all the linker scripts alongside ".hash".
 This produces correct vDSO images with --hash-style=sysv (or old tools),
with --hash-style=gnu, or with --hash-style=both.

Second, it passes the --hash-style=sysv option when building the vDSO
images, so that ".gnu.hash" is not actually produced.  This is the most
conservative choice for compatibility with any old userland.  There is some
concern that some ancient glibc builds (though not any known old production
system) might choke on --hash-style=both binaries.  The optimizations
provided by the new style of hash section do not really matter for a DSO
with a tiny number of symbols, as the vDSO has.  If someone wants to use
=gnu or =both for their vDSO builds and worry less about that
compatibility, just change the option and the linker script changes will
make any choice work fine.
Signed-off-by: NRoland McGrath <roland@redhat.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

kbuild used $¤(*F to get filename of target without extension.
This was used in several places all over kbuild, but introducing
make -rR broke his for all cases where we specified full path to
target/prerequsite. It is assumed that make -rR disables old style
suffix-rules which is why is suddenly failed.

ia64 was impacted by this change because several div* routines in
arch/ia64/lib are build using explicit paths and then kbuild failed.

Thanks to David Mosberger-Tang <David.Mosberger@acm.org> for an explanation
what was the root-cause and for testing on ia64.

This patch also fixes two uses of $(*F) in arch/um
Signed-off-by: NSam Ravnborg <sam@ravnborg.org>