perf by itself is not a standalone package; instead, it is part of a
bigger package, linux-tools.
Even though perf is the only one to need kernel .config fixups, we still
do it in a generic way, as it blends nicely in the existing variables,
which all use a loop over all the tools.
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
perf auto-detects and uses the libbfd (from binutils) and openssl
libraries if they are detected and happen to be built before perf is,
but if they're not, or if per-package directories are enabled, it won't
detect these libraries. Explicitly add dependencies on these packages if
they are enabled, and disable the feature if not, so that the behavior
is deterministic.
Signed-off-by: Robert Hancock <hancock@sedsystems.ca>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Workaround for:
ld: hv_vss_daemon.o: unable to initialize decompress status for section .debug_info
ld: hv_vss_daemon.o: unable to initialize decompress status for section .debug_info
hv_vss_daemon.o: file not recognized: File format not recognized
Signed-off-by: Pascal de Bruijn <p.debruijn@unilogic.nl>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The hyperv integration services offer convenience features for guest
operating systems running on the microsoft hyperv virtualization
platform. They roughly are for HyperV what openvmtools are for VMWare.
The installed binary names are derived from what seems common in large
distros like RedHat:
linux kernel source name -> installed binary name
hv_vss_daemon -> hypervvssd
hv_kvp_daemon -> hypervkvpd
hv_fcopy_daemon -> hypervfcopyd
Each tool was introduced at different points in the kernel history, so
we need to check each of them.
We provide a single init script that is responsible for starting all
enabled programs. The global status will be the status of the last
program to fail to start, or empty (i.e. success) if they all started
successfuly.
However, we provide one systemd unit per program, because it is not easy
to use a single unit to start (and monitor) more than one executable.
Additionally, we do not provide a template that is filled at tinstall
time either, because it does not gain much (three simple units vs. a
template and some replacement code in the .mk).
Finally, the key-value daemon uses a few helper scripts to get/set the
network config. All are optional (their presence is checked before
running them), but one, hv_set_ifconfig. However, it is not strictly
speaking required either, so we just symlink it to /bin/true to avoid
any warning at runtime. Providing actual helpers is left to the end
user, to adapt to their own environment.
Signed-off-by: Pascal de Bruijn <p.debruijn@unilogic.nl>
[yann.morin.1998@free.fr:
- aggregate all three tools in a single sub-package
- introduce the main HV option, use a sub-option for each tool
- aggregate the three init scripts into one
- don't install the helpers; symlink the mandatory one
- don't create symlinks for systemd units (systemctl preset-all does
it for us now)
- expand commit log
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Some linux tools (e.g. the Microsoft HyperV convenience utilities) will
install programs tostart at boot time, so they need to be able to
install init files (systemd units, sysv init script, or openrc units).
Unlike the other commands, we are redefining the real _INSTALL_INIT_*
macros, rather than use hooks, to let the infra call those at the right
moment.
We must be careful about the openrc support, though: if two tools are
enabled, one which provides sysv scripts but no openrc config, and the
other which provides openrc config, and we are using openrc as init
system, then we want to use the sysv scripts from the former as well as
the openrc config of the latter. Thus we need to duplicate a bit the
openrc logic here.
Signed-off-by: Pascal de Bruijn <p.debruijn@unilogic.nl>
[yann.morin.1998@free.fr:
- define macros, not hooks
- introduce support for openrc too
- expand commit log
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
The indirect dependency through kmod was not tracked.
Detected with randconfig.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Sync builtins ops are strictly required by perf utility as it uses
atomic_xxx() functions.
Otherwise building fails like that:
|.../output/host/opt/ext-toolchain/bin/../lib/gcc/arc-snps-linux-uclibc/8.2.1/../../../../arc-snps-linux-uclibc/bin/ld:
|.../output/build/linux-5.0.7/tools/perf/libperf.a(libperf-in.o): in function `atomic_cmpxchg':
|.../output/build/linux-5.0.7/tools/include/asm-generic/atomic-gcc.h:69: undefined reference to `__sync_val_compare_and_swap_4'
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
[Thomas: drop Config.in comment]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Some kernel-selftests are using modprobe options (-n) that are not available
from busybox's modprobe, so make sure that BR2_PACKAGE_KMOD_TOOLS is selected.
[linux-4.19 selftests]$ git grep tput
drivers/gpu/drm_mm.sh:if ! /sbin/modprobe -n -q test-drm_mm; then
drivers/usb/usbip/usbip_test.sh:if ! /sbin/modprobe -q -n usbip_host; then
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Some kernel-selftests are using tput program, so make sure that
BR2_PACKAGE_NCURSES_TARGET_PROGS is selected.
[linux-4.19 selftests]$ git grep tput
[...]
futex/run.sh:tput setf 7 || tput setaf 7
futex/run.sh: tput sgr0
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Add support for selection and compiling the pcitest tool on buildroot.
This tool is available to be compiled since kernel 4.20.
Signed-off-by: Gustavo Pimentel <gustavo.pimentel@synopsys.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Perf profiling cannot be used if CONFIG_PERF_EVENTS is not enabled in the
kernel configuration. Similar to other tools, like ktap, we can enable the
right options automatically.
Signed-off-by: Jan Heylen <jan.heylen@nokia.com>
Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Since Linux 3.10, perf's NO_NEWT configuration option simply sets
NO_SLANG=1 and there is no dependency on libnewt.
We already handle NO_SLANG correctly based on whether or not
BR2_PACKAGE_SLANG is selected, so all we accomplish by setting NO_NEWT=1
is disabling perf's TUI when all of the dependencies are available.
To simplify all of this, introduce a new config option to enable the
perf TUI which depends on slang and add a check to prevent building the
TUI on versions which are too old. The check for NO_SLANG is equivalent
to checking if NO_NEWT is required as NO_SLANG was added in the same
commit that removed the libnewt dependency and deprecated NO_NEWT
(6692c262df4f, "perf tools: Remove dependency on libnewt", 2013-03-28).
Signed-off-by: John Keeping <john@metanate.com>
[Thomas: add missing BR2_USE_MMU dependency, coming from the slang
package. Noticed by Yann E. Morin.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Since Linux kernel commit [1], the build of the iio tool has been
changed to use the common Linux tools build system. The installation
directory is now given by DESTDIR, like for all other Linux tools.
We keep the INSTALL_DIR environment in the 'install' target to be
compatible with kernels older than 4.14.
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=18956cf2d78a8d4a5959e20240f04ce8d5a6c121
Signed-off-by: Julien BOIBESSOT <julien.boibessot@armadeus.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Passing just the endianness flag to LD is not enough. We need to pass
the right emulation flag which will set everything for us, not only the
endianness.
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Markus Mayer <mmayer@broadcom.com>
[Thomas:
- tweak Config.in help text as provided by Markus
- adjust commit log
- add missing select BR2_PACKAGE_NCURSES
- add missing dependency on host-pkgconf, used by the tmon Makefile
to find ncurses
- add logic to disable -fstack-protector when the toolchain doesn't
have SSP support.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Make inclusion ordering of all linux-tool-*.mk sub-makefiles explicit
instead of relying on alphabetical sort order. This is done by
renaming the Linux tools sub-makefiles to the format linux-tool-*.mk.in.
This causes the top-level Makefile to ignore the Linux tools
sub-makefiles.
Until now, the main Makefile included all linux-tool-*.mk files, as
well as linux-tools.mk, and it relied on alphabetical sorting to
include them in the proper order (linux-tool-*.mk before
linux-tools.mk).
Signed-off-by: Markus Mayer <mmayer@broadcom.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[Thomas: improve comment in the code as suggested by Yann.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This commit switches to use the new gettext logic, which involves:
- using TARGET_NLS_DEPENDENCIES instead of hand-encoded dependencies
on gettext/host-gettext
- using TARGET_NLS_LIBS to force linking against libintl
- dropping BR2_PACKAGE_GETTEXT selection. The BR2_USE_WCHAR
dependency was only needed for gettext, so it is no longer needed.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Marcin Niestroj <m.niestroj@grinn-global.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Marcin Niestroj <m.niestroj@grinn-global.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Because each tool's individual .mk files is included from the top-level
Makefile and we also include them from linux-tools.mk, they get
registered twice, and thus built twice, and thus installed twice.
We did include them from linux-tools.mk to guarantee they would be
included early and each tool had a chance to register itself before we
were to construct the build and install hooks.
However, the ordering is _currently_ guaranteed, in the C locale by the
files names, which we anyway sort using make's $(sort) function, which
always sorts in the C locale.
Beware if we are to ever rename those files in the future...
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The kernel source tree also contains the sources for various userland
tools, of which cpupower, perf or selftests.
Currently, we have support for building those tools as part of the
kernel build procedure. This looked the correct thing to do so far,
because, well, they *are* part of the kernel source tree and some
really have to be the same version as the kernel that will run.
However, this is causing quite a non-trivial-to-break circular
dependency in some configurations. For example, this defconfig fails to
build (similar to the one reported by Paul):
BR2_arm=y
BR2_cortex_a7=y
BR2_ARM_FPU_NEON_VFPV4=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_INIT_SYSTEMD=y
BR2_LINUX_KERNEL=y
BR2_LINUX_KERNEL_CUSTOM_GIT=y
BR2_LINUX_KERNEL_CUSTOM_REPO_URL="https://github.com/raspberrypi/linux.git"
BR2_LINUX_KERNEL_CUSTOM_REPO_VERSION="26f3b72a9c049be10e6af196252283e1f6ab9d1f"
BR2_LINUX_KERNEL_DEFCONFIG="bcm2709"
BR2_PACKAGE_LINUX_TOOLS_CPUPOWER=y
BR2_PACKAGE_CRYPTODEV=y
BR2_PACKAGE_OPENSSL=y
BR2_PACKAGE_LIBCURL=y
This causes a circular dependency, as explained by Thomas:
- When libcurl is enabled, systemd depends on it
- When OpenSSL is enabled, obviously, will use it for SSL support
- When cryptodev-linux is enabled, OpenSSL will depend on it to use
crypto accelerators supported in the kernel via cryptodev-linux.
- cryptodev-linux being a kernel module, it depends on linux
- linux by itself (the kernel) does not depend on pciutils, but the
linux tool "cpupower" (managed in linux-tool-cpupower) depends on
pciutils
- pciutils depends on udev when available
- udev is provided by systemd.
And indeed, during the build, we can see that make warns (it's only
reported as a *warning*, not as an actual error):
[...]
make[1]: Circular /home/ymorin/dev/buildroot/O/build/openssl-1.0.2h/.stamp_configured
<- cryptodev-linux dependency dropped.
>>> openssl 1.0.2h Downloading
[...]
So the build fails later on, when openssl is actually built:
eng_cryptodev.c:57:31: fatal error: crypto/cryptodev.h: No such file or directory
compilation terminated.
<builtin>: recipe for target 'eng_cryptodev.o' failed
Furthermore, graph-depends also detects the circular dependency, but
treats it as a hard-error:
Recursion detected for : cryptodev-linux
which is a dependency of: openssl
which is a dependency of: libcurl
which is a dependency of: systemd
which is a dependency of: udev
which is a dependency of: pciutils
which is a dependency of: linux
which is a dependency of: cryptodev-linux
Makefile:738: recipe for target 'graph-depends' failed
Of course, there is no way to break the loop without losing
functionality in either one of the involved packages *and* keep
our infrastructure and packages as-is.
The only solution is to break the loop at the linux-tools level, by
moving them away into their own package, so that the linux package will
no longer have the opportunity to depend on another package via a
dependency of one the tools.
All three linux tools are thus moved away to their own package.
The package infrastructure only knows of three types of packages: those
in package/ , in boot/ , in toolchain/ and the one in linux/ . So we
create that new linux-tools package in package/ so that we don't have to
fiddle with yet another special case in the infra. Still, we want its
configure options to appear in the kernel's sub-menu.
So, we make it a prompt-less package, with only the tools visible as
options of that package, but without the usual dependency on their
master symbol; they only depend on the Linux kernel.
Furthermore, because the kernel is such a huge pile of code, we would
not be very happy to extract it a second time just for the sake of a few
tools. We can't extract only the tools/ sub-directory from the kernel
source either, because some tools have hard-coded path to includes from
the kernel (arch and stuff).
Instead, we just use the linux source tree as our own build tree, and
ensure the linux tree is extracted and patched before linux-tools is
configured and built.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Paul Ashford <paul.ashford@zurria.co.uk>
[Thomas:
- fix typo #(@D) -> $(@D)
- fix the inclusion of the per-tool .mk files.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>