Re: [Buildroot] [RFC PATCH v3 2/2] package/sysdig: bump to 0.29.1

[Francis Laniel <flaniel@linux.microsoft.com>]

Le vendredi 22 avril 2022, 23:48:43 BST Arnout Vandecappelle a écrit :
>   Hi Francis,

Hi!

>   Thanks for continuing to work on this!

You are welcome!

> On 22/04/2022 16:31, Francis Laniel wrote:
> > sysdig 0.27.1 cannot be cross-compiled to, e.g., aarch64 because it uses
> > open() syscall [1].
> > This patch bumps its version to enable cross-compilation.
> > 
> > By doing so, we can remove the previous patches.
> > But this commit introduces new patches.
> > Indeed, libsinsp (lib system inspection), libscap (lib system capture) and
> > the kernel module were moved to falcosecurity/libs.
> > Sadly, it is not possible to compile sysdig with pre-compiled libs (like
> > we do in buildroot).
> > So, this contribution introduces a new patch to build sysdig with already
> > built falcosecurity/libs.
> > 
> > [1] https://marc.info/?l=buildroot&m=164951521629400
> > 
> > Signed-off-by: Francis Laniel <flaniel@linux.microsoft.com>
> > ---
> > 
> >   ...BUNDLED_DEPS-before-getting-nlohmann.patch | 51 ++++++++++++
> >   ...ATCH_COMMAND-to-fix-lua-types-and-fu.patch | 82 -------------------
> >   ...gainst-libabseil-cpp-which-grpc-uses.patch | 47 -----------
> >   ...e-Add-libsinsp.a-as-IMPORTED-library.patch | 59 +++++++++++++
> >   package/sysdig/Config.in                      | 31 ++-----
> >   package/sysdig/sysdig.hash                    |  4 +-
> >   package/sysdig/sysdig.mk                      | 55 ++++++-------
> >   7 files changed, 141 insertions(+), 188 deletions(-)
> >   create mode 100644
> >   package/sysdig/0001-cmake-Check-USE_BUNDLED_DEPS-before-getting-nlohman
> >   n.patch delete mode 100644
> >   package/sysdig/0001-libsinsp-Apply-PATCH_COMMAND-to-fix-lua-types-and-f
> >   u.patch delete mode 100644
> >   package/sysdig/0002-Link-against-libabseil-cpp-which-grpc-uses.patch
> >   create mode 100644
> >   package/sysdig/0002-cmake-Add-libsinsp.a-as-IMPORTED-library.patch> 
> > diff --git
> > a/package/sysdig/0001-cmake-Check-USE_BUNDLED_DEPS-before-getting-nlohman
> > n.patch
> > b/package/sysdig/0001-cmake-Check-USE_BUNDLED_DEPS-before-getting-nlohman
> > n.patch new file mode 100644
> > index 0000000000..29f03b1b61
> > --- /dev/null
> > +++
> > b/package/sysdig/0001-cmake-Check-USE_BUNDLED_DEPS-before-getting-nlohman
> > n.patch @@ -0,0 +1,51 @@
> > +From 0dbebd008c04d266dc41c4bec8280a0744fd5130 Mon Sep 17 00:00:00 2001
> > +From: Francis Laniel <flaniel@linux.microsoft.com>
> > +Date: Wed, 13 Apr 2022 18:01:11 +0100
> > +Subject: [PATCH 1/2] cmake: Check USE_BUNDLED_DEPS before getting
> 
>   There should be no N/M in patches. check-package reports this.
> 
> > + nlohmann-json.
> > +
> > +Signed-off-by: Francis Laniel <flaniel@linux.microsoft.com>
> 
>   Please send your patches upstream (i.e. create a merge request on the
> sysdig project) before even sending it to Buildroot, and add
> 
> Upstream: https://github.com/...
> 
> to the patch you submit to Buildroot.
> 
>   This way, it's easier to find if the patch can be removed when the package
> version is bumped.
> 
>   Also, for reviewing, it allows us to check if upstream has any comments on
> the patch. Upstream generally has a better idea of how things are supposed
> to work than we do. And if they end up committing an alternative approach,
> we can simply pick that one (while committing) instead of the one you
> submitted.
> > +---
> > + cmake/modules/nlohmann-json.cmake | 29 +++++++++++++++++++----------
> > + 1 file changed, 19 insertions(+), 10 deletions(-)
> > +
> > +diff --git a/cmake/modules/nlohmann-json.cmake
> > b/cmake/modules/nlohmann-json.cmake +index bb1279d7..feb0f071 100644
> > +--- a/cmake/modules/nlohmann-json.cmake
> > ++++ b/cmake/modules/nlohmann-json.cmake
> > +@@ -16,13 +16,22 @@
> > + # limitations under the License.
> > + #
> > +
> > +-set(NJSON_SRC "${PROJECT_BINARY_DIR}/njson-prefix/src/njson")
> > +-message(STATUS "Using bundled nlohmann-json in '${NJSON_SRC}'")
> > +-set(NJSON_INCLUDE_DIR "${NJSON_SRC}/single_include")
> > +-ExternalProject_Add(
> > +-  njson
> > +-  URL "https://github.com/nlohmann/json/archive/v3.3.0.tar.gz"
> > +-  URL_HASH
> > "SHA256=2fd1d207b4669a7843296c41d3b6ac5b23d00dec48dba507ba051d14564aa801"
> > +-  CONFIGURE_COMMAND ""
> > +-  BUILD_COMMAND ""
> > +-  INSTALL_COMMAND "")
> > ++if(NOT USE_BUNDLED_DEPS)
> > ++  find_path(NJSON_INCLUDE_DIR NAMES nlohmann/json.hpp)
> > ++  if(NJSON_INCLUDE_DIR)
> > ++    message(STATUS "Found njson: include: ${NJSON_INCLUDE_DIR}")
> > ++  else()
> > ++    message(FATAL_ERROR "Couldn't find system njson")
> > ++  endif()
> > ++else()
> > ++  set(NJSON_SRC "${PROJECT_BINARY_DIR}/njson-prefix/src/njson")
> > ++  message(STATUS "Using bundled nlohmann-json in '${NJSON_SRC}'")
> > ++  set(NJSON_INCLUDE_DIR "${NJSON_SRC}/single_include")
> > ++  ExternalProject_Add(
> > ++    njson
> > ++    URL "https://github.com/nlohmann/json/archive/v3.3.0.tar.gz"
> > ++    URL_HASH
> > "SHA256=2fd1d207b4669a7843296c41d3b6ac5b23d00dec48dba507ba051d14564aa801"
> > ++    CONFIGURE_COMMAND ""
> > ++    BUILD_COMMAND ""
> > ++    INSTALL_COMMAND "")
> > ++endif()
> > +--
> > +2.25.1
> > +
> > diff --git
> > a/package/sysdig/0001-libsinsp-Apply-PATCH_COMMAND-to-fix-lua-types-and-f
> > u.patch
> > b/package/sysdig/0001-libsinsp-Apply-PATCH_COMMAND-to-fix-lua-types-and-f
> > u.patch deleted file mode 100644
> > index 7873210281..0000000000
> > ---
> > a/package/sysdig/0001-libsinsp-Apply-PATCH_COMMAND-to-fix-lua-types-and-f
> > u.patch +++ /dev/null
> > @@ -1,82 +0,0 @@
> > -From cc8bccc3ebb90103900a7f0f2b085ddb723b8792 Mon Sep 17 00:00:00 2001
> > -From: Francis Laniel <flaniel@linux.microsoft.com>
> > -Date: Wed, 6 Apr 2022 16:54:37 +0100
> > -Subject: [PATCH] libsinsp: Apply PATCH_COMMAND to fix lua types and
> > function. -
> > -Buildroot luajit 5.1 seems to not have compatibility between luaL_reg and
> > -luaL_Reg.
> > -So, we apply sysdig CMakeLists.txt PATCH_COMMAND to fix this and lua
> > function -call as well.
> > -Note that, this PATCH_COMMAND was added in sysdig in:
> > -a064440394c9 ("Adding power support to Travis builds (#1566)")
> > -
> > -This patch is also present in kubernetes/minikube in:
> > -f036c279bc59 ("Add patch for compiling sysdig with system luajit")
> 
>   You didn't apply this patch to falcosecurity-libs. Is it no longer needed
> there?

I normally took into account all your comments in v4 I just sent.
I just wanted to precise here that the patch regarding luajit is indeed no 
more needed in falcosecurity-libs because they use luaL_Reg.

> [snip]
> 
> > diff --git
> > a/package/sysdig/0002-cmake-Add-libsinsp.a-as-IMPORTED-library.patch
> > b/package/sysdig/0002-cmake-Add-libsinsp.a-as-IMPORTED-library.patch new
> > file mode 100644
> > index 0000000000..30601c2430
> > --- /dev/null
> > +++ b/package/sysdig/0002-cmake-Add-libsinsp.a-as-IMPORTED-library.patch
> > @@ -0,0 +1,59 @@
> > +From e681028c5567582f9c66cf2b11234ab3a1cacc73 Mon Sep 17 00:00:00 2001
> > +From: Francis Laniel <flaniel@linux.microsoft.com>
> > +Date: Fri, 15 Apr 2022 14:17:48 +0100
> > +Subject: [PATCH 2/2] cmake: Add libsinsp.a as IMPORTED library.
> > +
> > +libsinsp.a is created by falcosecurity/libs which is downloaded while
> > building +sysdig.
> > +
> > +In buildroot, we do not want to depend of external download.
> > +This patch adds libsinsp.a as IMPORTER library and links all its
> > dependencies +against it.
> > +
> > +Signed-off-by: Francis Laniel <flaniel@linux.microsoft.com>
> > +---
> > + userspace/sysdig/CMakeLists.txt | 28 ++++++++++++++++++++++++++++
> > + 1 file changed, 28 insertions(+)
> > +
> > +diff --git a/userspace/sysdig/CMakeLists.txt
> > b/userspace/sysdig/CMakeLists.txt +index 401d65e3..ef339b71 100644
> > +--- a/userspace/sysdig/CMakeLists.txt
> > ++++ b/userspace/sysdig/CMakeLists.txt
> > +@@ -86,6 +86,34 @@ target_include_directories(
> > + if(NOT WIN32)
> > + 	include_directories(${PROJECT_BINARY_DIR}/driver/src)
> > +
> > ++	if(SCAP_STATIC_LIBRARY_PATH AND SINSP_STATIC_LIBRARY_PATH)
> 
>   If you add new CMake variables, use the SET command to define them and
> give them a help text.
> 
>   However, is it necessary to set them? Can't you just use find_library to
> find them in the sysroot? It's basically the same approach as for finding
> the JSON library in the other patch.
> 
>   Except... Apparently falcosecurity-libs doesn't actually install its
> stuff. It really is meant to be used as a submodule.
> 
>   OK, it looks like this is really trying to fight upstream too much. So
> then it's probably better to revert to your earlier proposal: turn
> falcosecurity-libs into a generic package that just downloads stuff and
> doesn't install, and do the build from the sysdig build. I think, since
> it's not using in-tree build, that it doesn't even write anything in the
> falcosecurity-libs source directory, right? So the ugliness is rather
> limited...
> 
> > ++		find_library(ABSL_SYNC_LIB NAMES absl_synchronization)
> > ++		find_library(B64_LIB NAMES b64)
> > ++		find_library(CARES_LIB NAMES cares)
> > ++		find_library(ELF_LIB NAMES elf)
> > ++		find_library(GTEST_LIB NAMES gtest)
> > ++		find_library(JQ_LIB NAMES jq)
> > ++		find_library(JSONCPP_LIB NAMES jsoncpp)
> > ++		find_library(LUAJIT_LIB NAMES luajit luajit-5.1)
> > ++		find_library(PROTOBUF_LIB NAMES protobuf)
> > ++		find_library(TBB_LIB NAMES tbb)
> > ++		find_library(ZLIB_LIB NAMES z)
> > ++
> > ++		find_package(CURL REQUIRED)
> > ++		find_package(gRPC REQUIRED)
> > ++		find_package(OpenSSL REQUIRED)
> > ++
> > ++		set(GPR_LIB gRPC::gpr)
> > ++		set(GRPC_LIB gRPC::grpc)
> > ++		set(GRPCPP_LIB gRPC::grpc++)
> > ++
> > ++		add_library(sinsp STATIC IMPORTED)
> > ++		set_target_properties(sinsp PROPERTIES
> > ++			IMPORTED_LOCATION "${SINSP_STATIC_LIBRARY_PATH}"
> > ++			INTERFACE_LINK_LIBRARIES
> > "${ABSL_SYNC_LIB};${B64_LIB};${CARES_LIB};${CURL_LIBRARIES};${ELF_LIB};${
> > GPR_LIB};${GRPC_LIB};${GRPCPP_LIB};${GTEST_LIB};${JQ_LIB};${JSONCPP_LIB};$
> > {LUAJIT_LIB};${OPENSSL_LIBRARIES};${PROTOBUF_LIB};${TBB_LIB};${SCAP_STATIC
> > _LIBRARY_PATH};${ZLIB_LIB}" ++		)
> > ++	endif()
> > ++
> > + 	target_link_libraries(sysdig
> > + 		sinsp
> > + 		"${YAMLCPP_LIB}")
> > +--
> > +2.25.1
> > +
> > diff --git a/package/sysdig/Config.in b/package/sysdig/Config.in
> > index c124054105..e00516a2b5 100644
> > --- a/package/sysdig/Config.in
> > +++ b/package/sysdig/Config.in
> > @@ -1,26 +1,11 @@
> > 
> >   config BR2_PACKAGE_SYSDIG
> >   
> >   	bool "sysdig"
> > 
> > -	depends on BR2_PACKAGE_PROTOBUF_ARCH_SUPPORTS # protobuf
> > 
> >   	depends on BR2_LINUX_KERNEL
> > 
> > -	depends on BR2_INSTALL_LIBSTDCPP # jsoncpp, protobuf, tbb
> > -	depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_8 # protobuf
> > -	depends on BR2_TOOLCHAIN_HAS_THREADS # elfutils, jq, protobuf, tbb
> > -	depends on !BR2_STATIC_LIBS # elfutils, protobuf, tbb
> > -	depends on BR2_USE_WCHAR # elfutils
> > -	depends on BR2_TOOLCHAIN_USES_GLIBC # elfutils
> > -	depends on BR2_PACKAGE_LUAINTERPRETER_ABI_VERSION_5_1
> 
>   All these dependencies should still stay, they just become indirect
> dependencies.
> 
>   Except, if you don't do any actual build of falcosecurity-libs and only do
> it in sysdig, then all these dependencies and selects should also stay in
> sysdig.
> 
>   Also, you should not make falcosecurity-libs a user-selectable option in
> that case. It still would need a Config.in entry, but without prompt.
> 
> > -	select BR2_PACKAGE_C_ARES > -	select BR2_PACKAGE_ELFUTILS
> > -	select BR2_PACKAGE_GRPC
> > -	select BR2_PACKAGE_JQ
> > -	select BR2_PACKAGE_JSONCPP
> > -	select BR2_PACKAGE_LIBB64
> > -	select BR2_PACKAGE_LIBCURL
> > +	depends on BR2_INSTALL_LIBSTDCPP # yaml-cpp
> > +	select BR2_PACKAGE_FALCOSECURITY_LIBS
> > 
> >   	select BR2_PACKAGE_NCURSES
> > 
> > -	select BR2_PACKAGE_OPENSSL
> > -	select BR2_PACKAGE_PROTOBUF
> > -	select BR2_PACKAGE_TBB
> > -	select BR2_PACKAGE_ZLIB
> > +	select BR2_PACKAGE_JSON_FOR_MODERN_CPP
> > +	select BR2_PACKAGE_YAML_CPP
> > 
> >   	help
> >   	
> >   	  Sysdig is open source, system-level exploration:
> >   	  capture system state and activity from a running Linux
> > 
> > @@ -30,10 +15,6 @@ config BR2_PACKAGE_SYSDIG
> > 
> >   	  https://github.com/draios/sysdig/wiki
> > 
> > -comment "sysdig needs a glibc toolchain w/ C++, threads, gcc >= 4.8,
> > dynamic library, a Linux kernel, and luajit or lua 5.1 to be built"
> > -	depends on BR2_PACKAGE_PROTOBUF_ARCH_SUPPORTS
> > +comment "sysdig needs a toolchain w/ C++, a Linux kernel and luajit or
> > lua 5.1 to be built"> 
> >   	depends on !BR2_LINUX_KERNEL || !BR2_INSTALL_LIBSTDCPP \
> > 
> > -		|| !BR2_TOOLCHAIN_HAS_THREADS \
> > -		|| !BR2_TOOLCHAIN_GCC_AT_LEAST_4_8 || BR2_STATIC_LIBS \
> > -		|| !BR2_TOOLCHAIN_USES_GLIBC \
> > -	    || !BR2_PACKAGE_LUAINTERPRETER_ABI_VERSION_5_1
> > +		|| !BR2_PACKAGE_LUAINTERPRETER_ABI_VERSION_5_1
> > diff --git a/package/sysdig/sysdig.hash b/package/sysdig/sysdig.hash
> > index 4ec46abfc3..380c9dce1e 100644
> > --- a/package/sysdig/sysdig.hash
> > +++ b/package/sysdig/sysdig.hash
> > @@ -1,5 +1,3 @@
> > 
> >   # sha256 locally computed
> > 
> > -sha256  b9d05854493d245a7a7e75f77fc654508f720aab5e5e8a3a932bd8eb54e49bda 
> > sysdig-0.27.1.tar.gz +sha256 
> > 3721c97f0da43c1a68d2962d30363525d8532ab5f2534dfefc86d175e17e423e 
> > sysdig-0.29.1.tar.gz> 
> >   sha256  a88fbf820b38b1c7fabc6efe291b8259e02ae21326f56fe31c6c9adf374b2702
> >    COPYING> 
> > -sha256  8177f97513213526df2cf6184d8ff986c675afb514d4e68a404010521b880643 
> > driver/GPL2.txt -sha256 
> > e3f6a43d097ed68176e9738de925d98b938e1bccf6d6bd1bcd79395eca8f76ed 
> > driver/MIT.txt diff --git a/package/sysdig/sysdig.mk
> > b/package/sysdig/sysdig.mk
> > index 9a9aaa35c3..c6487c2337 100644
> > --- a/package/sysdig/sysdig.mk
> > +++ b/package/sysdig/sysdig.mk
> > @@ -4,10 +4,10 @@
> > 
> >   #
> >   ########################################################################
> >   ########> 
> > -SYSDIG_VERSION = 0.27.1
> > +SYSDIG_VERSION = 0.29.1
> > 
> >   SYSDIG_SITE = $(call github,draios,sysdig,$(SYSDIG_VERSION))
> > 
> > -SYSDIG_LICENSE = Apache-2.0 (userspace), MIT or GPL-2.0 (driver)
> > -SYSDIG_LICENSE_FILES = COPYING driver/MIT.txt driver/GPL2.txt
> > +SYSDIG_LICENSE = Apache-2.0
> > +SYSDIG_LICENSE_FILE = COPYING
> > 
> >   SYSDIG_CPE_ID_VENDOR = sysdig
> >   SYSDIG_CONF_OPTS = \
> >   
> >   	-DENABLE_DKMS=OFF \
> > 
> > @@ -16,39 +16,32 @@ SYSDIG_CONF_OPTS = \
> > 
> >   SYSDIG_SUPPORTS_IN_SOURCE_BUILD = NO
> >   
> >   SYSDIG_DEPENDENCIES = \
> > 
> > -	c-ares \
> > -	elfutils \
> > -	grpc \
> > -	jq \
> > -	jsoncpp \
> > -	libb64 \
> > -	libcurl \
> > -	luainterpreter \
> > +	falcosecurity-libs \
> > 
> >   	ncurses \
> > 
> > -	openssl \
> > -	protobuf \
> > -	tbb \
> > -	zlib
> > +	json-for-modern-cpp \
> > +	yaml-cpp
> > 
> > -# sysdig creates the module Makefile from a template, which contains a
> > -# single place-holder, KBUILD_FLAGS, wich is only replaced with two
> > -# things:
> > -#   - debug flags, which we don't care about here,
> > -#   - 'sysdig-feature' flags, which are never set, so always empty
> > -# So, just replace the place-holder with the only meaningful value:
> > nothing. -define SYSDIG_MODULE_GEN_MAKEFILE
> > -	$(INSTALL) -m 0644 $(@D)/driver/Makefile.in $(@D)/driver/Makefile
> > -	$(SED) 's/@KBUILD_FLAGS@//;' $(@D)/driver/Makefile
> > -	$(SED) 's/@PROBE_NAME@/sysdig-probe/;' $(@D)/driver/Makefile
> > -endef
> > -SYSDIG_POST_PATCH_HOOKS += SYSDIG_MODULE_GEN_MAKEFILE
> > +# We need to set all these variables to avoid compiling
> > falcosecurity-libs
> > +# twice.
> > +SYSDIG_FALCOSECURITY_LIBS_SOURCE_DIR=$(BUILD_DIR)/falcosecurity-libs-e5c5
> > 3d648f3c4694385bbe488e7d47eaa36c229a
>   Put spaces around assignments. Doesn't check-package warn about this?
> 
>   Also, use $(FALCOSECURITY_LIBS_SRCDIR) instead of reconstructing it. And
> with that, there's actually no need to introduce a variable for it. In
> fact, I don't see the point for any of the variables defined below, they
> only obfuscate what is going on.
> 
> > +SYSDIG_FALCOSECURITY_LIBS_DRIVER_CONFIG_DIR=$(SYSDIG_FALCOSECURITY_LIBS_S
> > OURCE_DIR)/buildroot-build/driver/src
> > +SYSDIG_FALCOSECURITY_LIBS_LIBSCAP_SOURCE_DIR=$(SYSDIG_FALCOSECURITY_LIBS
> > _SOURCE_DIR)/userspace/libscap
> > +SYSDIG_FALCOSECURITY_LIBS_LIBSINSP_SOURCE_DIR=$(SYSDIG_FALCOSECURITY_LIB
> > S_SOURCE_DIR)/userspace/libsinsp
> > +SYSDIG_FALCOSECURITY_LIBS_CHISEL_SOURCE_DIR=$(SYSDIG_FALCOSECURITY_LIBS_
> > SOURCE_DIR)/userspace/chisel
> > +SYSDIG_FALCOSECURITY_LIBS_LIBSCAP_STATIC_LIB_PATH=$(SYSDIG_FALCOSECURITY
> > _LIBS_SOURCE_DIR)/buildroot-build/libscap/libscap.a
> > +SYSDIG_FALCOSECURITY_LIBS_LIBSINSP_STATIC_LIB_PATH=$(SYSDIG_FALCOSECURIT
> > Y_LIBS_SOURCE_DIR)/buildroot-build/libsinsp/libsinsp.a
> > 
> > -# Don't build the driver as part of the 'standard' procedure, we'll
> > -# build it on our own with the kernel-module infra.
> > -SYSDIG_CONF_OPTS += -DBUILD_DRIVER=OFF
> > +# SCAP_STATIC_LIBRARY_PATH and SINSP_STATIC_LIBRARY_PATH were added by
> > patch. +SYSDIG_CONF_OPTS +=
> > -DFALCOSECURITY_LIBS_SOURCE_DIR=$(SYSDIG_FALCOSECURITY_LIBS_SOURCE_DIR) \
>   Small nit: if you have a multiline assignment, split it like this:
> 
> SYSDIG_CONF_OPTS += \
> 	-D...
> 
>   Also, there's already a SYSDIG_CONF_OPTS assignment above, and both are
> unconditional. Merge them. Yes, I know this was already wrong before.
> 
> > +	-DHAVE_LIBSCAP=On \
> 
>   We use ON and OFF for CMake. Doesn't make a difference, but it's nice if
> things are consistent.
> 
>   Regards,
>   Arnout
> 
> > +	-DLIBSCAP_INCLUDE_DIRS="$
(SYSDIG_FALCOSECURITY_LIBS_LIBSCAP_SOURCE_DIR);
> > $(SYSDIG_FALCOSECURITY_LIBS_DRIVER_CONFIG_DIR)" \ +	-DHAVE_LIBSINSP=On \
> > +	-DLIBSINSP_INCLUDE_DIRS="$
(SYSDIG_FALCOSECURITY_LIBS_LIBSINSP_SOURCE_DIR
> > );$(SYSDIG_FALCOSECURITY_LIBS_CHISEL_SOURCE_DIR)" \
> > +	-DVALIJSON_INCLUDE=$(BUILD_DIR)/valijson-0.6/include/valijson \
> > +	-DSCAP_STATIC_LIBRARY_PATH=$
(SYSDIG_FALCOSECURITY_LIBS_LIBSCAP_STATIC_LI
> > B_PATH) \
> > +	-DSINSP_STATIC_LIBRARY_PATH=$
(SYSDIG_FALCOSECURITY_LIBS_LIBSINSP_STATIC
> > _LIB_PATH)> 
> >   SYSDIG_MODULE_SUBDIRS = driver
> >   SYSDIG_MODULE_MAKE_OPTS = KERNELDIR=$(LINUX_DIR)
> > 
> > -$(eval $(kernel-module))
> > 
> >   $(eval $(cmake-package))

Best regards.



_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot