Per-package host/target directory support

The pkg-config wrapper script is currently generated with absolute paths to $(STAGING_DIR). However, this will not work properly with per-package SDK, and each package will be built with a different STAGING_DIR value.

In order to fix this, we adjust how the pkg-config wrapper script is generated, so that it uses a relative path to itself: the sysroot (i.e STAGING_DIR) is always located in $(path of pkg-config)/../$(STAGING_SUBDIR).

This change is independent from the per-package SDK work, and could be applied independently from it.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- None
Changes since v1:
- As suggested by Arnout, simplify the generation of the pkg-config script by doing only a replacement on @STAGING_SUBDIR@, the rest being encoded inside the pkg-config script.

---
 package/pkgconf/pkg-config.in | 5 ++++-
 package/pkgconf/pkgconf.mk    | 3 +--
 2 files changed, 5 insertions(+), 3 deletions(-)

package/pkgconf/pkg-config.in +4 -1

index 4dec48789a..b9ce0935cc 100644
@@ -1,2 +1,5 @@
 #!/bin/sh
-PKG_CONFIG_LIBDIR=${PKG_CONFIG_LIBDIR:- at PKG_CONFIG_LIBDIR@} PKG_CONFIG_SYSROOT_DIR=${PKG_CONFIG_SYSROOT_DIR:- at STAGING_DIR@} $(dirname $0)/pkgconf @STATIC@ $@
+PKGCONFDIR=$(dirname $0)
+DEFAULT_PKG_CONFIG_LIBDIR=${PKGCONFDIR}/../@STAGING_SUBDIR@/usr/lib/pkgconfig:${PKGCONFDIR}/../@STAGING_SUBDIR@/usr/share/pkgconfig
+DEFAULT_PKG_CONFIG_SYSROOT_DIR=${PKGCONFDIR}/../@STAGING_SUBDIR@
+PKG_CONFIG_LIBDIR=${PKG_CONFIG_LIBDIR:-${DEFAULT_PKG_CONFIG_LIBDIR}} PKG_CONFIG_SYSROOT_DIR=${PKG_CONFIG_SYSROOT_DIR:-${DEFAULT_PKG_CONFIG_SYSROOT_DIR}} ${PKGCONFDIR}/pkgconf @STATIC@ $@

package/pkgconf/pkgconf.mk +1 -2

index cc190d26da..00b2d017ee 100644
@@ -19,8 +19,7 @@ endef
 define HOST_PKGCONF_INSTALL_WRAPPER
 	$(INSTALL) -m 0755 -D package/pkgconf/pkg-config.in \
 		$(HOST_DIR)/bin/pkg-config
-	$(SED) 's, at PKG_CONFIG_LIBDIR@,$(STAGING_DIR)/usr/lib/pkgconfig:$(STAGING_DIR)/usr/share/pkgconfig,' \
-		-e 's, at STAGING_DIR@,$(STAGING_DIR),' \
+	$(SED) 's, at STAGING_SUBDIR@,$(STAGING_SUBDIR),g' \
 		$(HOST_DIR)/bin/pkg-config
 endef
 

Signature · 1 line

2.13.6

The pkg-config wrapper script is currently generated with absolute paths to $(STAGING_DIR). However, this will not work properly with per-package SDK, and each package will be built with a different STAGING_DIR value.

In order to fix this, we adjust how the pkg-config wrapper script is generated, so that it uses a relative path to itself: the sysroot (i.e STAGING_DIR) is always located in $(path of pkg-config)/../$(STAGING_SUBDIR).

This change is independent from the per-package SDK work, and could be applied independently from it.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- None
Changes since v1:
- As suggested by Arnout, simplify the generation of the pkg-config script by doing only a replacement on @STAGING_SUBDIR@, the rest being encoded inside the pkg-config script.

---
 package/pkgconf/pkg-config.in | 5 ++++-
 package/pkgconf/pkgconf.mk    | 3 +--
 2 files changed, 5 insertions(+), 3 deletions(-)

The upcoming per-package SDK functionality is heavily based on the fact that HOST_DIR, STAGING_DIR and TARGET_DIR are evaluated during the configure/build/install steps of the packages. Therefore, any evaluation-during-assignment using := is going to cause problems, and need to be turned into evaluation-during-use using =.

This patch fix up one such instance in the external toolchain code.

This change is independent from the per-package SDK functionality, and could be applied separately.

Signed-off-by Thomas Petazzoni <[email protected]>

Reviewed-by Arnout Vandecappelle (Essensium/Mind) <[email protected]>

---
Changes since v2:
- None
Changes since v1:
- Added Arnout Reviewed-by.

---
 toolchain/toolchain-external/pkg-toolchain-external.mk | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

toolchain/toolchain-external/pkg-toolchain-external.mk +1 -1

index dc0588c536..b9ad1720a1 100644
@@ -77,7 +77,7 @@ ifneq ($(TOOLCHAIN_EXTERNAL_PREFIX),)
 TOOLCHAIN_EXTERNAL_BIN := $(dir $(shell which $(TOOLCHAIN_EXTERNAL_PREFIX)-gcc))
 endif
 else
-TOOLCHAIN_EXTERNAL_BIN := $(TOOLCHAIN_EXTERNAL_INSTALL_DIR)/bin
+TOOLCHAIN_EXTERNAL_BIN = $(TOOLCHAIN_EXTERNAL_INSTALL_DIR)/bin
 endif
 
 # If this is a buildroot toolchain, it already has a wrapper which we want to

Signature · 1 line

2.13.6

The upcoming per-package SDK functionality is heavily based on the fact that HOST_DIR, STAGING_DIR and TARGET_DIR are evaluated during the configure/build/install steps of the packages. Therefore, any evaluation-during-assignment using := is going to cause problems, and need to be turned into evaluation-during-use using =.

This patch fix up one such instance in the external toolchain code.

This change is independent from the per-package SDK functionality, and could be applied separately.

Signed-off-by Thomas Petazzoni <[email protected]>

Reviewed-by Arnout Vandecappelle (Essensium/Mind) <[email protected]>

---
Changes since v2:
- None
Changes since v1:
- Added Arnout Reviewed-by.

---
 toolchain/toolchain-external/pkg-toolchain-external.mk | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

As part of the per-package SDK work, we want to avoid having logic that installs files to the global HOST_DIR, and instead do it inside packages. One thing that gets installed to the global HOST_DIR is the minimal "skeleton" that we create in host:

- the "usr" symbolic link for backward compatibility

- the "lib" directory, and its lib64 or lib32 symbolic links

This commit moves this logic to a new host-skeleton package, and makes all packages (except itself) depend on it. We also make sure that this host-skeleton package doesn't depend on host-patchelf, because host-patchelf depends on host-skeleton.

Signed-off-by Thomas Petazzoni <[email protected]>

package/skeleton/skeleton.mk +0 -0

index d380f41649..a32bacd0b3 100644
@@ -11,4 +11,16 @@
 SKELETON_ADD_TOOLCHAIN_DEPENDENCY = NO
 SKELETON_ADD_SKELETON_DEPENDENCY = NO

Currently, the toolchainfile.cmake and Buildroot.cmake files are installed outside of any package, just triggered by the toolchain target.

As part of the per-package SDK effort, we are trying to avoid anything that installs to the global $(HOST_DIR), and this is one of the remaining files installed in $(HOST_DIR) outside of any package. We fix this by installing such files as part of the toolchain package post-install staging hooks.

Yes, a post-install staging hook to install things to $(HOST_DIR) is a bit weird, but the toolchain infrastructure is made of target packages only, and they all install a lot of stuff to $(HOST_DIR) already.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- None
Changes since v1:
- New patch

---
 package/pkg-cmake.mk             | 12 +++++++-----
 toolchain/toolchain/toolchain.mk |  3 ---
 2 files changed, 7 insertions(+), 8 deletions(-)

package/pkg-cmake.mk +7 -5

index 6739704e3c..6d4f948d07 100644
@@ -244,8 +244,8 @@ endif
 # based on the toolchainfile.cmake file's location: $(HOST_DIR)/share/buildroot
 # In all the other variables, HOST_DIR will be replaced by RELOCATED_HOST_DIR,
 # so we have to strip "$(HOST_DIR)/" from the paths that contain it.
-$(HOST_DIR)/share/buildroot/toolchainfile.cmake:
-	@mkdir -p $(@D)
+define TOOLCHAIN_CMAKE_INSTALL_FILES
+	@mkdir -p $(HOST_DIR)/share/buildroot
 	sed \
 		-e 's#@@STAGING_SUBDIR@@#$(call qstrip,$(STAGING_SUBDIR))#' \
 		-e 's#@@TARGET_CFLAGS@@#$(call qstrip,$(TARGET_CFLAGS))#' \

@@ -259,7 +259,9 @@ $(HOST_DIR)/share/buildroot/toolchainfile.cmake:
 		-e 's#@@TOOLCHAIN_HAS_FORTRAN@@#$(if $(BR2_TOOLCHAIN_HAS_FORTRAN),1,0)#' \
 		-e 's#@@CMAKE_BUILD_TYPE@@#$(if $(BR2_ENABLE_DEBUG),Debug,Release)#' \
 		$(TOPDIR)/support/misc/toolchainfile.cmake.in \
-		> $@
+		> $(HOST_DIR)/share/buildroot/toolchainfile.cmake
+	$(Q)$(INSTALL) -D -m 0644 support/misc/Buildroot.cmake \
+		$(HOST_DIR)/share/buildroot/Platform/Buildroot.cmake
+endef
 
-$(HOST_DIR)/share/buildroot/Platform/Buildroot.cmake:
-	$(Q)$(INSTALL) -D -m 0644 support/misc/Buildroot.cmake $(@)
+TOOLCHAIN_POST_INSTALL_STAGING_HOOKS += TOOLCHAIN_CMAKE_INSTALL_FILES

toolchain/toolchain/toolchain.mk +0 -3

index b55b0c712c..283e0a74ee 100644
@@ -38,6 +38,3 @@ TOOLCHAIN_INSTALL_STAGING = YES
 endif
 
 $(eval $(virtual-package))
-
-toolchain: $(HOST_DIR)/share/buildroot/toolchainfile.cmake
-toolchain: $(HOST_DIR)/share/buildroot/Platform/Buildroot.cmake

Signature · 1 line

2.13.6

On 2017-12-01 21:53 +0100, Thomas Petazzoni spake thusly:

Quoted 10 lines

Currently, the toolchainfile.cmake and Buildroot.cmake files are installed outside of any package, just triggered by the toolchain target.

As part of the per-package SDK effort, we are trying to avoid anything that installs to the global $(HOST_DIR), and this is one of the remaining files installed in $(HOST_DIR) outside of any package. We fix this by installing such files as part of the toolchain package post-install staging hooks.

Quoted 3 lines

Yes, a post-install staging hook to install things to $(HOST_DIR) is a bit weird, but the toolchain infrastructure is made of target packages only, and they all install a lot of stuff to $(HOST_DIR) already.

What I am more concerned about, is that the cmake package is registering a hook for another package, toolchain.

Have you tried to run check-package on that?

Quoted 11 lines

On 2017-12-01 21:53 +0100, Thomas Petazzoni spake thusly:

Quoted 10 lines

Currently, the toolchainfile.cmake and Buildroot.cmake files are installed outside of any package, just triggered by the toolchain target.

As part of the per-package SDK effort, we are trying to avoid anything that installs to the global $(HOST_DIR), and this is one of the remaining files installed in $(HOST_DIR) outside of any package. We fix this by installing such files as part of the toolchain package post-install staging hooks.

Quoted 6 lines

Quoted 3 lines

Yes, a post-install staging hook to install things to $(HOST_DIR) is a bit weird, but the toolchain infrastructure is made of target packages only, and they all install a lot of stuff to $(HOST_DIR) already.

What I am more concerned about, is that the cmake package is registering a hook for another package, toolchain.

Do we want to introduce a toolchain-cmake package, that all CMake packages would depend on, just for the sake of installing those files ?

If we would do this, then a normal toolchain build (i.e without building any CMake package) would no longer generate the toolchainfile.cmake file, which is quite useful for SDKs. People could indeed enable that package explicitly when they are building a SDK.

But all in all, I'm not sure it's really worth the hassle. But I'm open to comments on that.

Or perhaps you had other suggestions in mind?

Quoted 1 line

Have you tried to run check-package on that?

Nope, I did not, but it does not complain:

$ ./utils/check-package package/pkg-cmake.mk 
0 lines processed
0 warnings generated

Thanks for the review!

Thomas

Signature · 3 lines

Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

The installation to target, staging and images as well as the host installation are all done in 4 different make targets. We don't have any place currently to run something at the end of a package installation, i.e once all of target/staging/images (for target) or host (for host) installation have completed.

We will need such a step to fix the RPATH of binaries in the host directory. Indeed, while normally only host packages should install things in $(HOST_DIR), in practice a number of target packages (ex:
qt4, qt5base) do install things in $(HOST_DIR).

Therefore, we want to run our RPATH fixing logic at the end of all installation steps. To achieve this, this commit introduces a new make target, .stamp_installed, which depends on .stamp_{target,staging,images}_installed for target packages and .stamp_host_installed for host packages.

This new step currently doesn't do anything, but the actual logic is added in a follow-up commit.

This change in fact makes the overall step sequencing more logical:
for all steps the dependencies were done on the stamp file targets, except for the install step where we were using phony targets to make $(1)-install depends on $(1)-install-{target,staging,images,host}. So now, the phony targets $(1)-install-{target,staging,images,host} are only human-readable targets that are shortcuts to the corresponding stamp file targets, and not usded anymore for the step sequencing. This makes them more consistent with the other phony $(1)-{build,configure,patch,...} phony targets.

It also makes $(1)-install-host more consistent with other phony
targets: $(1)-install-target only existed if the package had <pkg>_INSTALL_TARGET = YES, but $(1)-install-host existed unconditionally. Now it only exists if it's a host package.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- None
Changes since v1:
- New patch.

---
 package/pkg-generic.mk | 44 +++++++++++++++++++++++++++-----------------
 1 file changed, 27 insertions(+), 17 deletions(-)

package/pkg-generic.mk +27 -17

index 0019814276..2b821e9bdf 100644
@@ -340,6 +340,12 @@ $(BUILD_DIR)/%/.stamp_target_installed:
 	@$(call step_end,install-target)
 	$(Q)touch $@
 
+# Finalize installation
+$(BUILD_DIR)/%/.stamp_installed:
+	@$(call step_start,install)
+	@$(call step_end,install)
+	$(Q)touch $@
+
 # Remove package sources
 $(BUILD_DIR)/%/.stamp_dircleaned:
 	rm -Rf $(@D)

@@ -595,6 +601,7 @@ $(2)_TARGET_INSTALL_TARGET =	$$($(2)_DIR)/.stamp_target_installed
 $(2)_TARGET_INSTALL_STAGING =	$$($(2)_DIR)/.stamp_staging_installed
 $(2)_TARGET_INSTALL_IMAGES =	$$($(2)_DIR)/.stamp_images_installed
 $(2)_TARGET_INSTALL_HOST =      $$($(2)_DIR)/.stamp_host_installed
+$(2)_TARGET_INSTALL =      	$$($(2)_DIR)/.stamp_installed
 $(2)_TARGET_BUILD =		$$($(2)_DIR)/.stamp_built
 $(2)_TARGET_CONFIGURE =		$$($(2)_DIR)/.stamp_configured
 $(2)_TARGET_RSYNC =	        $$($(2)_DIR)/.stamp_rsynced

@@ -642,38 +649,40 @@ $(2)_ROOTFS_POST_CMD_HOOKS      ?=
 # human-friendly targets and target sequencing
 $(1):			$(1)-install
 
+$(1)-install:			$$($(2)_TARGET_INSTALL)
+
 ifeq ($$($(2)_TYPE),host)
-$(1)-install:	        $(1)-install-host
-else
-$(1)-install:		$(1)-install-staging $(1)-install-target $(1)-install-images
-endif
+
+$(1)-install-host:		$$($(2)_TARGET_INSTALL_HOST)
+$$($(2)_TARGET_INSTALL):	$$($(2)_TARGET_INSTALL_HOST)
+
+else # host vs. target
 
 ifeq ($$($(2)_INSTALL_TARGET),YES)
 $(1)-install-target:		$$($(2)_TARGET_INSTALL_TARGET)
-$$($(2)_TARGET_INSTALL_TARGET):	$$($(2)_TARGET_BUILD)
-else
-$(1)-install-target:
+$$($(2)_TARGET_INSTALL):	$$($(2)_TARGET_INSTALL_TARGET)
 endif
 
 ifeq ($$($(2)_INSTALL_STAGING),YES)
-$(1)-install-staging:			$$($(2)_TARGET_INSTALL_STAGING)
-$$($(2)_TARGET_INSTALL_STAGING):	$$($(2)_TARGET_BUILD)
-# Some packages use install-staging stuff for install-target
-$$($(2)_TARGET_INSTALL_TARGET):		$$($(2)_TARGET_INSTALL_STAGING)
-else
-$(1)-install-staging:
+$(1)-install-staging:		$$($(2)_TARGET_INSTALL_STAGING)
+$$($(2)_TARGET_INSTALL):	$$($(2)_TARGET_INSTALL_STAGING)
 endif
 
 ifeq ($$($(2)_INSTALL_IMAGES),YES)
 $(1)-install-images:		$$($(2)_TARGET_INSTALL_IMAGES)
-$$($(2)_TARGET_INSTALL_IMAGES):	$$($(2)_TARGET_BUILD)
-else
-$(1)-install-images:
+$$($(2)_TARGET_INSTALL):	$$($(2)_TARGET_INSTALL_IMAGES)
 endif
 
-$(1)-install-host:		$$($(2)_TARGET_INSTALL_HOST)
+endif # host vs.target
+
+$$($(2)_TARGET_INSTALL_TARGET):	$$($(2)_TARGET_BUILD)
+$$($(2)_TARGET_INSTALL_STAGING):$$($(2)_TARGET_BUILD)
+$$($(2)_TARGET_INSTALL_IMAGES):	$$($(2)_TARGET_BUILD)
 $$($(2)_TARGET_INSTALL_HOST):	$$($(2)_TARGET_BUILD)
 
+# Some packages use install-staging stuff for install-target
+$$($(2)_TARGET_INSTALL_TARGET):		$$($(2)_TARGET_INSTALL_STAGING)
+
 $(1)-build:		$$($(2)_TARGET_BUILD)
 $$($(2)_TARGET_BUILD):	$$($(2)_TARGET_CONFIGURE)
 

@@ -813,6 +822,7 @@ $(1)-reconfigure:	$(1)-clean-for-reconfigure $(1)
 
 # define the PKG variable for all targets, containing the
 # uppercase package variable prefix
+$$($(2)_TARGET_INSTALL):		PKG=$(2)
 $$($(2)_TARGET_INSTALL_TARGET):		PKG=$(2)
 $$($(2)_TARGET_INSTALL_STAGING):	PKG=$(2)
 $$($(2)_TARGET_INSTALL_IMAGES):		PKG=$(2)

Signature · 1 line

2.13.6

Quoted 6 lines · Thomas Petazzoni

package/pkg-generic.mk +3 -0

+# Some packages use install-staging stuff for install-target
+$$($(2)_TARGET_INSTALL_TARGET):                $$($(2)_TARGET_INSTALL_STAGING)
+
 $(1)-build:            $$($(2)_TARGET_BUILD)
 $$($(2)_TARGET_BUILD): $$($(2)_TARGET_CONFIGURE)

Review $(1)-reconfigure: $(1)-clean-for-reconfigure $(1) · line 822

Quoted 2 lines

(quoted diff) +0 -0

@@ -813,6 +822,7 @@ $(1)-reconfigure:   $(1)-clean-for-reconfigure $(1)
 

Should the pkgname-dirclean target also remove its per-package folder ?

Extract dependencies are dependencies that must be ready before the extract step of a package, i.e for tools that are needed to extract packages themselves. Current examples of such tools are host-tar, host-lzip and host-xz.

They are currently handled through DEPENDENCIES_HOST_PREREQ. However, this mechanism has a number of drawbacks:

- First and foremost, because host-tar/host-lzip/host-xz are not listed in the dependencies of packages, the package infrastructure does not know it should rsync them in the context of per-package SDK.

- Second, there is no dependency handling *between* them. I.e, we have no mechanism that says host-tar should be built before host-lzip, while it is in fact the case: if you need to build host-lzip, you need to extract a tarball, so you may need host-tar if your system tarball is not capable enough.

For those reasons, it makes sense to add explicit support for "extract dependencies" in the package infrastructure, through the <pkg>_EXTRACT_DEPENDENCIES variable. It is unlikely this variable will ever be used by a package .mk file, but it will be used internally by the package infrastructure.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 docs/manual/adding-packages-generic.txt | 7 +++++++
 package/pkg-generic.mk                  | 8 +++++++-
 2 files changed, 14 insertions(+), 1 deletion(-)

docs/manual/adding-packages-generic.txt +0 -0

index 63ea51bf89..7f42cb09e9 100644

docs/manual/adding-packages-generic.txt +2 -0

+  and should typically not be used directly by packages.
+
 * +LIBFOO_PATCH_DEPENDENCIES+ lists the dependencies (in terms of
   package name) that are required for the current package to be
   patched. These dependencies are guaranteed to be extracted and

package/pkg-generic.mk +7 -1

index 2b821e9bdf..824c34bec9 100644
@@ -589,8 +589,13 @@ endif
 
 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
+$(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))
 $(2)_FINAL_PATCH_DEPENDENCIES = $$(sort $$($(2)_PATCH_DEPENDENCIES))
-$(2)_FINAL_ALL_DEPENDENCIES = $$(sort $$($(2)_FINAL_DEPENDENCIES) $$($(2)_FINAL_PATCH_DEPENDENCIES))
+$(2)_FINAL_ALL_DEPENDENCIES = \
+	$$(sort \
+		$$($(2)_FINAL_DEPENDENCIES) \
+		$$($(2)_FINAL_EXTRACT_DEPENDENCIES) \
+		$$($(2)_FINAL_PATCH_DEPENDENCIES))
 
 $(2)_INSTALL_STAGING		?= NO
 $(2)_INSTALL_IMAGES		?= NO

@@ -716,6 +721,7 @@ $$($(2)_TARGET_PATCH):  | $$(patsubst %,%-patch,$$($(2)_FINAL_PATCH_DEPENDENCIES
 
 $(1)-extract:			$$($(2)_TARGET_EXTRACT)
 $$($(2)_TARGET_EXTRACT):	$$($(2)_TARGET_SOURCE)
+$$($(2)_TARGET_EXTRACT): | $$($(2)_FINAL_EXTRACT_DEPENDENCIES)
 
 $(1)-depends:		$$($(2)_FINAL_DEPENDENCIES)
 

Signature · 1 line

2.13.6

This moves the host-tar dependency handling from
DEPENDENCY_HOST_PREREQ to an extract dependency.

To achieve that, check-host-tar.mk fills in the

BR2_TAR_HOST_DEPENDENCY with host-tar if building a host-tar is needed. The name BR2_TAR_HOST_DEPENDENCY has been chosen because it matches the name BR2_CMAKE_HOST_DEPENDENCY already used in check-host-cmake.mk.

The BR2_TAR_HOST_DEPENDENCY is added to all packages, except host-tar itself (obviously) and host-skeleton, because we depend on host-skeleton to install host-tar properly in HOST_DIR.

In addition, we modify tar.mk to explicitly build host-tar without
ccache: since ccache source code is available as a tarball, ccache will obviously depend on host-tar if the system tar is insufficient.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 package/pkg-generic.mk                 | 4 ++++
 package/tar/tar.mk                     | 6 ++++++
 support/dependencies/check-host-tar.mk | 2 +-
 3 files changed, 11 insertions(+), 1 deletion(-)

package/pkg-generic.mk +0 -0

index 824c34bec9..86dd2561d4 100644

package/pkg-generic.mk +0 -0

 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
 $(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))

package/tar/tar.mk +6 -0

index cb2ddc0eca..813eabed14 100644
@@ -44,4 +44,10 @@ define HOST_TAR_EXTRACT_CMDS
 	mv $(@D)/tar-$(TAR_VERSION)/* $(@D)
 	rmdir $(@D)/tar-$(TAR_VERSION)
 endef
+
+# we are built before ccache
+HOST_TAR_CONF_ENV = \
+	CC="$(HOSTCC_NOCCACHE)" \
+	CXX="$(HOSTCXX_NOCCACHE)"
+
 $(eval $(host-autotools-package))

support/dependencies/check-host-tar.mk +0 -0

index ad0b32e277..d07f727c4c 100644

support/dependencies/check-host-tar.mk +0 -0

 endif

Signature · 1 line

2.13.6

Quoted 1 line

To achieve that, check-host-tar.mk fills in the

s/$/ variable/

Quoted 3 lines

The BR2_TAR_HOST_DEPENDENCY is added to all packages, except host-tar itself (obviously) and host-skeleton, because we depend on host-skeleton to install host-tar properly in HOST_DIR.

As we discussed on IRC, there is also the case for the tar filesystem, which may require host-tar.

Granted, it is very very unlikely that we end up with no package that require host-tar, but still...

package/pkg-generic.mk ifneq ($(1),host-skeleton) · line 592

Quoted 3 lines

package/pkg-generic.mk +3 -0

+ifeq ($(filter host-tar host-skeleton,$(1)),)
+$(2)_EXTRACT_DEPENDENCIES += $(BR2_TAR_HOST_DEPENDENCY)
+endif  

So, even packages that are available as a zip file will get the dependency on host-tar?

support/dependencies/check-host-tar.mk line 5

Quoted 4 lines

ifeq (,$(call suitable-host-package,tar,$(TAR)))

support/dependencies/check-host-tar.mk +1 -1

-DEPENDENCIES_HOST_PREREQ += host-tar
 TAR = $(HOST_DIR)/bin/tar
+BR2_TAR_HOST_DEPENDENCY += host-tar  

Why assign-append here, instead of simply assign?

Quoted 2 lines

Quoted 1 line

To achieve that, check-host-tar.mk fills in the

s/$/ variable/

ACK, will fix.

Quoted 9 lines

Quoted 3 lines

The BR2_TAR_HOST_DEPENDENCY is added to all packages, except host-tar itself (obviously) and host-skeleton, because we depend on host-skeleton to install host-tar properly in HOST_DIR.

As we discussed on IRC, there is also the case for the tar filesystem, which may require host-tar.

Granted, it is very very unlikely that we end up with no package that require host-tar, but still...

Yes, I agree, the tar filesystem should depend on $(BR2_TAR_HOST_DEPENDENCY) as well. I'll fix that.

Quoted 6 lines

package/pkg-generic.mk ifneq ($(1),host-skeleton) · line 592

Quoted 3 lines

package/pkg-generic.mk +3 -0

+ifeq ($(filter host-tar host-skeleton,$(1)),)
+$(2)_EXTRACT_DEPENDENCIES += $(BR2_TAR_HOST_DEPENDENCY)
+endif  

So, even packages that are available as a zip file will get the dependency on host-tar?

Yes. Like host-xz/host-lzip, my series does not attempt to switch to more fine-grained dependencies, it only intends to move to proper package dependencies. Using more fine-grained dependencies is left for future work.

Quoted 5 lines

support/dependencies/check-host-tar.mk line 5

Quoted 4 lines

ifeq (,$(call suitable-host-package,tar,$(TAR)))

support/dependencies/check-host-tar.mk +1 -1

-DEPENDENCIES_HOST_PREREQ += host-tar
 TAR = $(HOST_DIR)/bin/tar
+BR2_TAR_HOST_DEPENDENCY += host-tar  

Why assign-append here, instead of simply assign?

Good point, will fix.

This moves the host-xz dependency handling from
DEPENDENCY_HOST_PREREQ to an extract dependency.

To achieve that, check-host-xz.mk fills in the BR2_XZ_HOST_DEPENDENCY with host-tar if building a host-tar is needed. The name BR2_XZ_HOST_DEPENDENCY has been chosen because it matches the name BR2_CMAKE_HOST_DEPENDENCY already used in check-host-cmake.mk.

The BR2_XZ_HOST_DEPENDENCY is added to all packages, except:

- host-xz, because we would otherwise depend on ourself.

- host-tar, because xz itself is delivered as a tarball, so we need to have host-xz depend on host-tar, and not host-tar depend on host-xz

- host-skeleton, because we need to have host-xz depend on host-skeleton, and not the opposite.

In addition, we modify xz.mk to explicitly build host-xz without ccache. We generally took the approach of building host-ccache *after* all the extractors have been built.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 package/pkg-generic.mk                   | 4 ++++
 package/xz/xz.mk                         | 5 +++++
 support/dependencies/check-host-xzcat.mk | 2 +-
 3 files changed, 10 insertions(+), 1 deletion(-)

package/pkg-generic.mk +0 -0

index 86dd2561d4..dfd4719d29 100644

package/pkg-generic.mk +0 -0

 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
 $(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))

package/xz/xz.mk +5 -0

index e8116f24ae..bcdac13ee3 100644
@@ -18,5 +18,10 @@ else
 XZ_CONF_OPTS = --disable-threads
 endif
 
+# we are built before ccache
+HOST_XZ_CONF_ENV = \
+	CC="$(HOSTCC_NOCCACHE)" \
+	CXX="$(HOSTCXX_NOCCACHE)"
+
 $(eval $(autotools-package))
 $(eval $(host-autotools-package))

support/dependencies/check-host-xzcat.mk +1 -1

index 9be75c7311..e5d72f95dd 100644
@@ -2,7 +2,7 @@
 # If it is not present, build our own host-xzcat
 
 ifeq (,$(call suitable-host-package,xzcat,$(XZCAT)))
-DEPENDENCIES_HOST_PREREQ += host-xz
+BR2_XZCAT_HOST_DEPENDENCY = host-xz
 EXTRACTOR_DEPENDENCY_PRECHECKED_EXTENSIONS += .xz .lzma
 XZCAT = $(HOST_DIR)/bin/xzcat
 endif

Signature · 1 line

2.13.6

package/pkg-generic.mk ifeq ($(filter host-tar host-skeleton,$(1)),) · line 596

Quoted 3 lines

package/pkg-generic.mk +3 -0

+ifeq ($(filter host-tar host-skeleton host-xz,$(1)),)
+$(2)_EXTRACT_DEPENDENCIES += $(BR2_XZCAT_HOST_DEPENDENCY)
+endif  

So, all packages will now depend on this, even those that are not compressed with xz?

Currently, filesystems that want to compress with xz all depend on host-xz unconditioanlly. Do we want to change that as well, so they only depend on it if needed?

Hello,

Quoted 6 lines · Yann E. MORIN

package/pkg-generic.mk ifeq ($(filter host-tar host-skeleton,$(1)),) · line 596

Quoted 3 lines

package/pkg-generic.mk +3 -0

+ifeq ($(filter host-tar host-skeleton host-xz,$(1)),)
+$(2)_EXTRACT_DEPENDENCIES += $(BR2_XZCAT_HOST_DEPENDENCY)
+endif  

So, all packages will now depend on this, even those that are not compressed with xz?

Yes, if there is no xzcat available on the system. This is exactly what the current behavior is with DEPENDENCIES_HOST_PREREQ. My series does not intend to solve this problem, which is a separate one.

I do believe however that by moving those dependencies (host-xz and host-lzip) as proper package dependencies, it will be easier to have follow-up patches that add the dependency only if lzip or xz are needed. But again, my series does not intend to fix that, it's a separate problem.

Quoted 3 lines

Currently, filesystems that want to compress with xz all depend on host-xz unconditioanlly. Do we want to change that as well, so they only depend on it if needed?

I don't understand this part, because filesystems do not depend on host-xz unconditionally. They only depend on it if xz compression of the filesystem image is requested. From fs/common.mk:

ifeq ($$(BR2_TARGET_ROOTFS_$(2)_XZ),y)
ROOTFS_$(2)_DEPENDENCIES += host-xz
ROOTFS_$(2)_COMPRESS_EXT = .xz
ROOTFS_$(2)_COMPRESS_CMD = xz -9 -C crc32 -c
endif

Best regards,

Thomas Petazzoni

Signature · 3 lines

Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

Quoted 3 lines

Yes, if there is no xzcat available on the system. This is exactly what the current behavior is with DEPENDENCIES_HOST_PREREQ. My series does not intend to solve this problem, which is a separate one.

I somehow interpreted your patches in my head as "only build them when needed". ;-)

Quoted 8 lines

Quoted 3 lines

Currently, filesystems that want to compress with xz all depend on host-xz unconditioanlly. Do we want to change that as well, so they only depend on it if needed?

I don't understand this part, because filesystems do not depend on host-xz unconditionally. They only depend on it if xz compression of the filesystem image is requested. From fs/common.mk:

Quoted 5 lines

ifeq ($$(BR2_TARGET_ROOTFS_$(2)_XZ),y)
ROOTFS_$(2)_DEPENDENCIES += host-xz
ROOTFS_$(2)_COMPRESS_EXT = .xz
ROOTFS_$(2)_COMPRESS_CMD = xz -9 -C crc32 -c
endif

Well, first, let's assume that when the host has a suitable xzcat, it also has the xz command as well; I think it's not too far-fetched an assumption.

Then, we have BR2_XZCAT_HOST_DEPENDENCY empty, because we do have a suitable xzcat, and thus a suitable xz. Yet, when the user requires their filesystem to be xz-compressed, we still forcefully depend on host-xz, even though we do have a suitable xz command.

What I mean is that, probably, we will want to make it:

    ROOTFS_$(2)_DEPENDENCIES += $$(BR2_XZCAT_HOST_DEPENDENCY)

Quoted 8 lines

Quoted 8 lines

Quoted 3 lines

Currently, filesystems that want to compress with xz all depend on host-xz unconditioanlly. Do we want to change that as well, so they only depend on it if needed?

I don't understand this part, because filesystems do not depend on host-xz unconditionally. They only depend on it if xz compression of the filesystem image is requested. From fs/common.mk:

Quoted 5 lines

Quoted 5 lines

ifeq ($$(BR2_TARGET_ROOTFS_$(2)_XZ),y)
ROOTFS_$(2)_DEPENDENCIES += host-xz
ROOTFS_$(2)_COMPRESS_EXT = .xz
ROOTFS_$(2)_COMPRESS_CMD = xz -9 -C crc32 -c
endif

Quoted 3 lines

What I mean is that, probably, we will want to make it:

    ROOTFS_$(2)_DEPENDENCIES += $$(BR2_XZCAT_HOST_DEPENDENCY)

Yes, I thought about this too. But again, it's a separate issue: the filesystem code already depends on host-xz, regardless of whether xz is already available on the system.

This moves the host-xz dependency handling from
DEPENDENCY_HOST_PREREQ to an extract dependency.

To achieve that, check-host-xz.mk fills in the BR2_XZ_HOST_DEPENDENCY with host-tar if building a host-tar is needed. The name BR2_XZ_HOST_DEPENDENCY has been chosen because it matches the name BR2_CMAKE_HOST_DEPENDENCY already used in check-host-cmake.mk.

The BR2_XZ_HOST_DEPENDENCY is added to all packages, except:

- host-xz, because we would otherwise depend on ourself.

- host-tar, because xz itself is delivered as a tarball, so we need to have host-xz depend on host-tar, and not host-tar depend on host-xz

- host-skeleton, because we need to have host-xz depend on host-skeleton, and not the opposite.

In addition, we modify xz.mk to explicitly build host-xz without ccache. We generally took the approach of building host-ccache *after* all the extractors have been built.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 package/pkg-generic.mk                   | 4 ++++
 package/xz/xz.mk                         | 5 +++++
 support/dependencies/check-host-xzcat.mk | 2 +-
 3 files changed, 10 insertions(+), 1 deletion(-)

package/pkg-generic.mk +4 -0

index 86dd2561d4..dfd4719d29 100644
@@ -591,6 +591,10 @@ ifeq ($(filter host-tar host-skeleton,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_TAR_HOST_DEPENDENCY)
 endif
 
+ifeq ($(filter host-tar host-skeleton host-xz,$(1)),)
+$(2)_EXTRACT_DEPENDENCIES += $(BR2_XZCAT_HOST_DEPENDENCY)
+endif
+
 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
 $(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))

package/xz/xz.mk +5 -0

index e8116f24ae..bcdac13ee3 100644
@@ -18,5 +18,10 @@ else
 XZ_CONF_OPTS = --disable-threads
 endif
 
+# we are built before ccache
+HOST_XZ_CONF_ENV = \
+	CC="$(HOSTCC_NOCCACHE)" \
+	CXX="$(HOSTCXX_NOCCACHE)"
+
 $(eval $(autotools-package))
 $(eval $(host-autotools-package))

support/dependencies/check-host-xzcat.mk +1 -1

index 9be75c7311..e5d72f95dd 100644
@@ -2,7 +2,7 @@
 # If it is not present, build our own host-xzcat
 
 ifeq (,$(call suitable-host-package,xzcat,$(XZCAT)))
-DEPENDENCIES_HOST_PREREQ += host-xz
+BR2_XZCAT_HOST_DEPENDENCY = host-xz
 EXTRACTOR_DEPENDENCY_PRECHECKED_EXTENSIONS += .xz .lzma
 XZCAT = $(HOST_DIR)/bin/xzcat
 endif

Signature · 1 line

2.13.6

This moves the host-xz dependency handling from
DEPENDENCY_HOST_PREREQ to an extract dependency.

To achieve that, check-host-xz.mk fills in the BR2_LZIP_HOST_DEPENDENCY with host-tar if building a host-tar is needed. The name BR2_LZIP_HOST_DEPENDENCY has been chosen because it matches the name BR2_CMAKE_HOST_DEPENDENCY already used in check-host-cmake.mk.

The BR2_LZIP_HOST_DEPENDENCY is added to all packages, except:

- host-lzip, because we would otherwise depend on ourself.

- host-tar, because lzip itself is delivered as a tarball, so we need to have host-lzip depend on host-tar, and not host-tar depend on host-lzip

- host-skeleton, because we need to have host-lzip depend on host-skeleton, and not the opposite.

We also mutually exclude host-lzip and host-xz from dependending on each other, to avoid a circular dependency.

In addition, we modify lzip.mk to explicitly build host-lzip without ccache. We generally took the approach of building host-ccache *after* all the extractors have been built.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 package/lzip/lzip.mk                    | 2 +-
 package/pkg-generic.mk                  | 6 +++++-
 support/dependencies/check-host-lzip.mk | 2 +-
 3 files changed, 7 insertions(+), 3 deletions(-)

package/lzip/lzip.mk +1 -1

index b7ba5dd21d..a16a22214c 100644
@@ -16,7 +16,7 @@ endef
 
 define HOST_LZIP_CONFIGURE_CMDS
 	(cd $(@D); $(HOST_MAKE_ENV) ./configure --prefix=$(HOST_DIR) \
-		$(HOST_CONFIGURE_OPTS) )
+		$(HOST_CONFIGURE_OPTS) CC="$(HOSTCC_NOCCACHE)" CXX="$(HOSTCXX_NOCCACHE)")
 endef
 
 define LZIP_BUILD_CMDS

package/pkg-generic.mk +5 -1

index dfd4719d29..e623afe7bc 100644
@@ -591,10 +591,14 @@ ifeq ($(filter host-tar host-skeleton,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_TAR_HOST_DEPENDENCY)
 endif
 
-ifeq ($(filter host-tar host-skeleton host-xz,$(1)),)
+ifeq ($(filter host-tar host-skeleton host-xz host-lzip,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_XZCAT_HOST_DEPENDENCY)
 endif
 
+ifeq ($(filter host-tar host-skeleton host-xz host-lzip,$(1)),)
+$(2)_EXTRACT_DEPENDENCIES += $(BR2_LZIP_HOST_DEPENDENCY)
+endif
+
 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
 $(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))

support/dependencies/check-host-lzip.mk +1 -1

index 00cdd0a236..cdd784058c 100644
@@ -1,5 +1,5 @@
 ifeq (,$(call suitable-host-package,lzip,$(LZCAT)))
-DEPENDENCIES_HOST_PREREQ += host-lzip
+BR2_LZIP_HOST_DEPENDENCY = host-lzip
 EXTRACTOR_DEPENDENCY_PRECHECKED_EXTENSIONS += .lz
 LZCAT = $(HOST_DIR)/bin/lzip -d -c
 endif

Signature · 1 line

2.13.6

Quoted 6 lines

To achieve that, check-host-xz.mk fills in the BR2_LZIP_HOST_DEPENDENCY with host-tar if building a host-tar is needed. The name BR2_LZIP_HOST_DEPENDENCY has been chosen because it matches the name BR2_CMAKE_HOST_DEPENDENCY already used in check-host-cmake.mk.

Quoted 1 line

The BR2_LZIP_HOST_DEPENDENCY is added to all packages, except:

So even when no package compresed with lzip is enabled?

That was a gripe of yours, that host-lzip was always built even when not needed...

This moves the host-xz dependency handling from
DEPENDENCY_HOST_PREREQ to an extract dependency.

To achieve that, check-host-xz.mk fills in the BR2_LZIP_HOST_DEPENDENCY with host-tar if building a host-tar is needed. The name BR2_LZIP_HOST_DEPENDENCY has been chosen because it matches the name BR2_CMAKE_HOST_DEPENDENCY already used in check-host-cmake.mk.

The BR2_LZIP_HOST_DEPENDENCY is added to all packages, except:

- host-lzip, because we would otherwise depend on ourself.

- host-tar, because lzip itself is delivered as a tarball, so we need to have host-lzip depend on host-tar, and not host-tar depend on host-lzip

- host-skeleton, because we need to have host-lzip depend on host-skeleton, and not the opposite.

We also mutually exclude host-lzip and host-xz from dependending on each other, to avoid a circular dependency.

In addition, we modify lzip.mk to explicitly build host-lzip without ccache. We generally took the approach of building host-ccache *after* all the extractors have been built.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 package/lzip/lzip.mk                    | 2 +-
 package/pkg-generic.mk                  | 6 +++++-
 support/dependencies/check-host-lzip.mk | 2 +-
 3 files changed, 7 insertions(+), 3 deletions(-)

package/lzip/lzip.mk +1 -1

index b7ba5dd21d..a16a22214c 100644
@@ -16,7 +16,7 @@ endef
 
 define HOST_LZIP_CONFIGURE_CMDS
 	(cd $(@D); $(HOST_MAKE_ENV) ./configure --prefix=$(HOST_DIR) \
-		$(HOST_CONFIGURE_OPTS) )
+		$(HOST_CONFIGURE_OPTS) CC="$(HOSTCC_NOCCACHE)" CXX="$(HOSTCXX_NOCCACHE)")
 endef
 
 define LZIP_BUILD_CMDS

package/pkg-generic.mk +5 -1

index dfd4719d29..e623afe7bc 100644
@@ -591,10 +591,14 @@ ifeq ($(filter host-tar host-skeleton,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_TAR_HOST_DEPENDENCY)
 endif
 
-ifeq ($(filter host-tar host-skeleton host-xz,$(1)),)
+ifeq ($(filter host-tar host-skeleton host-xz host-lzip,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_XZCAT_HOST_DEPENDENCY)
 endif
 
+ifeq ($(filter host-tar host-skeleton host-xz host-lzip,$(1)),)
+$(2)_EXTRACT_DEPENDENCIES += $(BR2_LZIP_HOST_DEPENDENCY)
+endif
+
 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
 $(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))

support/dependencies/check-host-lzip.mk +1 -1

index 00cdd0a236..cdd784058c 100644
@@ -1,5 +1,5 @@
 ifeq (,$(call suitable-host-package,lzip,$(LZCAT)))
-DEPENDENCIES_HOST_PREREQ += host-lzip
+BR2_LZIP_HOST_DEPENDENCY = host-lzip
 EXTRACTOR_DEPENDENCY_PRECHECKED_EXTENSIONS += .lz
 LZCAT = $(HOST_DIR)/bin/lzip -d -c
 endif

Signature · 1 line

2.13.6

This moves the host-ccache dependency handling from DEPENDENCIES_HOST_PREREQ to a proper package dependency. When BR2_CCACHE=y, we add host-ccache as a regular dependency of all packages except:

- The extractor packages host-tar, host-xz and host-lzip

- host-ccache itself

- host-skeleton, because all packages depend on it

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 package/ccache/ccache.mk             | 5 +++++
 package/pkg-generic.mk               | 6 ++++++
 support/dependencies/dependencies.mk | 4 ----
 3 files changed, 11 insertions(+), 4 deletions(-)

package/ccache/ccache.mk +5 -0

index afbec44fac..d6ef455e8b 100644
@@ -21,6 +21,11 @@ CCACHE_LICENSE_FILES = LICENSE.txt GPL-3.0.txt
 # has zero dependency besides the C library.
 HOST_CCACHE_CONF_OPTS += --with-bundled-zlib
 
+# We are ccache, so we can't use ccache
+HOST_CCACHE_CONF_ENV = \
+	CC="$(HOSTCC_NOCCACHE)" \
+	CXX="$(HOSTCXX_NOCCACHE)"
+
 # Patch host-ccache as follows:
 #  - Use BR_CACHE_DIR instead of CCACHE_DIR, because CCACHE_DIR
 #    is already used by autotargets for the ccache package.

package/pkg-generic.mk +0 -0

index e623afe7bc..cb5889c9ef 100644

package/pkg-generic.mk +3 -0

+endif
+endif
+
 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
 $(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))

support/dependencies/dependencies.mk +0 -4

index ef2ae9b7e1..1a4b5df9f2 100644
@@ -14,10 +14,6 @@ $(shell support/dependencies/check-host-$(1).sh $(2))
 endef
 -include $(sort $(wildcard support/dependencies/check-host-*.mk))
 
-ifeq ($(BR2_CCACHE),y)
-DEPENDENCIES_HOST_PREREQ += host-ccache
-endif
-
 core-dependencies:
 	@HOSTCC="$(firstword $(HOSTCC))" MAKE="$(MAKE)" \
 		DL_TOOLS="$(sort $(DL_TOOLS_DEPENDENCIES))" \

Signature · 1 line

2.13.6

package/pkg-generic.mk ifeq ($(filter host-tar host-skeleton host-xz host-lzip,$(1)),) · line 604

Quoted 3 lines

package/pkg-generic.mk +3 -0

+ifeq ($(BR2_CCACHE),y)
+ifeq ($(filter host-tar host-skeleton host-xz host-lzip,$(1)),)
+$(2)_DEPENDENCIES += host-ccache  

Sorry, I must be dumb, but I don't see where you exclude host-ccache from being its own dependency...

This commit moves the host-fakedate dependency handling from DEPENDENCIES_HOST_PREREQ to a proper regular dependency handled by the package infrastructure.

host-fakedate is added as dependency to all packages, except host-skeleton, because we depend on it.

In addition, we make sure that host-fakedate does not grow a dependency on host-{tar,xz,lzip,ccache} to avoid circular dependencies. host-fakedate does not need any extraction tool and does not need to build C/C++ code (the source code is just a shell script available in Buildroot).

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 Makefile               |  1 -
 package/pkg-generic.mk | 14 ++++++++++----
 2 files changed, 10 insertions(+), 5 deletions(-)

Makefile +0 -1

index 56bf083098..f31834682c 100644
@@ -255,7 +255,6 @@ export LC_ALL = C
 export GZIP = -n
 BR2_VERSION_GIT_EPOCH = $(shell GIT_DIR=$(TOPDIR)/.git $(GIT) log -1 --format=%at)
 export SOURCE_DATE_EPOCH ?= $(if $(wildcard $(TOPDIR)/.git),$(BR2_VERSION_GIT_EPOCH),$(BR2_VERSION_EPOCH))
-DEPENDENCIES_HOST_PREREQ += host-fakedate
 endif
 
 # To put more focus on warnings, be less verbose as default

package/pkg-generic.mk +2 -2

index cb5889c9ef..7c5d951af9 100644
@@ -587,24 +587,30 @@ ifneq ($(1),host-skeleton)
 $(2)_DEPENDENCIES += host-skeleton
 endif
 
-ifeq ($(filter host-tar host-skeleton,$(1)),)
+ifeq ($(filter host-tar host-skeleton host-fakedate,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_TAR_HOST_DEPENDENCY)
 endif
 
-ifeq ($(filter host-tar host-skeleton host-xz host-lzip,$(1)),)
+ifeq ($(filter host-tar host-skeleton host-xz host-lzip host-fakedate,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_XZCAT_HOST_DEPENDENCY)

package/pkg-generic.mk +6 -0

 $(2)_DEPENDENCIES += host-ccache
 endif
 endif
 
+ifeq ($(BR2_REPRODUCIBLE),y)
+ifeq ($(filter host-skeleton,$(1)),)
+$(2)_DEPENDENCIES += host-fakedate
+endif
+endif
+
 # Eliminate duplicates in dependencies
 $(2)_FINAL_DEPENDENCIES = $$(sort $$($(2)_DEPENDENCIES))
 $(2)_FINAL_EXTRACT_DEPENDENCIES = $$(sort $$($(2)_EXTRACT_DEPENDENCIES))

Signature · 1 line

2.13.6

Now that DEPENDENCIES_HOST_PREREQ is no longer used anywhere, we can kill it.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 Makefile                             |  2 --
 package/pkg-generic.mk               |  2 --
 support/dependencies/dependencies.mk | 10 +++-------
 3 files changed, 3 insertions(+), 11 deletions(-)

Makefile +0 -2

index f31834682c..090b3ba191 100644
@@ -485,8 +485,6 @@ include package/Makefile.in
 -include $(wildcard arch/arch.mk.*)
 include support/dependencies/dependencies.mk
 
-PACKAGES += $(DEPENDENCIES_HOST_PREREQ)
-
 include toolchain/*.mk
 include toolchain/*/*.mk
 

package/pkg-generic.mk +0 -2

index 7c5d951af9..8c6e34cc9f 100644
@@ -725,9 +725,7 @@ $(1)-configure:			$$($(2)_TARGET_CONFIGURE)
 $$($(2)_TARGET_CONFIGURE):	| $$($(2)_FINAL_DEPENDENCIES)
 
 $$($(2)_TARGET_SOURCE) $$($(2)_TARGET_RSYNC): | dirs prepare
-ifeq ($$(filter $(1),$$(DEPENDENCIES_HOST_PREREQ)),)
 $$($(2)_TARGET_SOURCE) $$($(2)_TARGET_RSYNC): | dependencies
-endif
 
 ifeq ($$($(2)_OVERRIDE_SRCDIR),)
 # In the normal case (no package override), the sequence of steps is

support/dependencies/dependencies.mk +3 -7

index 1a4b5df9f2..d8ec137b12 100644
@@ -14,18 +14,14 @@ $(shell support/dependencies/check-host-$(1).sh $(2))
 endef
 -include $(sort $(wildcard support/dependencies/check-host-*.mk))
 
-core-dependencies:
-	@HOSTCC="$(firstword $(HOSTCC))" MAKE="$(MAKE)" \
+dependencies:
+	@HOSTCC="$(firstword $(HOSTCC_NOCCACHE))" MAKE="$(MAKE)" \
 		DL_TOOLS="$(sort $(DL_TOOLS_DEPENDENCIES))" \
 		$(TOPDIR)/support/dependencies/dependencies.sh
 
-core-dependencies $(DEPENDENCIES_HOST_PREREQ): HOSTCC=$(HOSTCC_NOCCACHE)
-core-dependencies $(DEPENDENCIES_HOST_PREREQ): HOSTCXX=$(HOSTCXX_NOCCACHE)
-dependencies: core-dependencies $(DEPENDENCIES_HOST_PREREQ)
-
 ################################################################################
 #
 # Toplevel Makefile options
 #
 ################################################################################
-.PHONY: dependencies core-dependencies
+.PHONY: dependencies

Signature · 1 line

2.13.6

Currently, our strategy for the RPATH of host binaries is as follows:

- During the build, we force an absolute RPATH to be encoded into binaries, by passing -Wl,-rpath,$(HOST_DIR)/lib. This ensures that host binaries will find their libraries.

- In the "make sdk" target, when preparing the SDK to be relocatable, we use patchelf to replace those absolute RPATHs by relative RPATHs that use $ORIGIN.

Unfortunately, the use of absolute RPATH during the build is not compatible with the move to per-package SDK, where a different host directory is used for the build of each package. Therefore, we need to move to a different strategy for RPATH handling.

The new strategy is as follows:

- We no longer pass -Wl,-rpath,$(HOST_DIR)/lib when building packages, so the host binaries no longer have any rpath encoded.

- At the end of the installation of every package, we just the fix-rpath logic on the host binaries, so that all host binaries that don't already have $ORIGIN/../lib as their RPATH are updated to have such a RPATH.

In order to achieve this, the following changes are made:

- HOST_LDFLAGS no longer contains -Wl,-rpath,$(HOST_DIR)

- The hook that ran at the end of every package installation (check_host_rpath) to verify that our hardcoded RPATH is indeed present in all binaries is removed, as it is no longer needed: we will force $ORIGIN/../LIB as an RPATH in all binaries.

- host-patchelf is added as a dependency of all packages, except itself. This is necessary as all packages now need to run patchelf at the end of their installation process.

TODO: things like host-tar, host-lzip and host-ccache will have to be handled properly.

- "fix-rpath host" is called at the end of the package installation, in the recently introduced .stamp_installed, which guarantees to be executed after all of target, staging, images and host installations have completed. Indeed, while in theory only host packages install files into $(HOST_DIR), in practice a number of target packages also install host binaries. So to be reliable, we do the "fix-rpath host" logic at the end of the installation of every package.

- The fix-rpath script is modified:

* The "host" target now simply hardcodes setting $ORIGIN/../lib as an RPATH, instead of using --relative-to-file

* We simply exclude the patchelf program instead of making a copy of it: since patchelf depends only on the C/C++ libraries, it doesn't need to have a RPATH set to $ORIGIN/../lib

* Also, in order to avoid re-writing host binaries over and over again, we only set the RPATH if it is not already set.

This change is independent from the per-package SDK functionality, and could be applied separately.

Signed-off-by Thomas Petazzoni <[email protected]>

---
NOTE: an alternative to running "fix-rpath host" at the end of every package installation would be to run it only for host packages (unconditionally) and have a special flag that target packages can use to indicate that they have installed things to $(HOST_DIR). Not sure if this is worth the complexity, though.

NOTE2: is it OK to force $ORIGIN/../lib as the RPATH of all host binaries? This is OK for binaries in host/{bin,sbin} of course, but do we have host binaries in other places, for which this $ORIGIN/../lib RPATH wouldn't be appropriate?

Changes since v2:
- No changes

Changes since v1:
- Following the feedback from Yann and Arnout, the approach in v2 was completely changed: instead of using LD_LIBRARY_PATH, we now run patchelf at the end of each package installation to fix the RPATH.

---
 Makefile                  |  1 -
 package/Makefile.in       |  2 +-
 package/pkg-generic.mk    | 14 ++++++--------
 support/scripts/fix-rpath | 31 +++++++++++++++++++------------
 4 files changed, 26 insertions(+), 22 deletions(-)

Makefile +0 -1

index 090b3ba191..d8fa91120b 100644
@@ -555,7 +555,6 @@ world: target-post-image
 .PHONY: sdk
 sdk: world
 	@$(call MESSAGE,"Rendering the SDK relocatable")
-	$(TOPDIR)/support/scripts/fix-rpath host
 	$(TOPDIR)/support/scripts/fix-rpath staging
 	$(INSTALL) -m 755 $(TOPDIR)/support/misc/relocate-sdk.sh $(HOST_DIR)/relocate-sdk.sh
 	echo $(HOST_DIR) > $(HOST_DIR)/share/buildroot/sdk-location

package/Makefile.in +1 -1

index a1a5316051..e94a75c230 100644
@@ -220,7 +220,7 @@ HOST_CPPFLAGS  = -I$(HOST_DIR)/include
 HOST_CFLAGS   ?= -O2
 HOST_CFLAGS   += $(HOST_CPPFLAGS)
 HOST_CXXFLAGS += $(HOST_CFLAGS)
-HOST_LDFLAGS  += -L$(HOST_DIR)/lib -Wl,-rpath,$(HOST_DIR)/lib
+HOST_LDFLAGS  += -L$(HOST_DIR)/lib
 
 # The macros below are taken from linux 4.11 and adapted slightly.
 # Copy more when needed.

package/pkg-generic.mk +6 -8

index 8c6e34cc9f..f4a943e3ce 100644
@@ -118,14 +118,6 @@ endef
 
 GLOBAL_INSTRUMENTATION_HOOKS += check_bin_arch
 
-# This hook checks that host packages that need libraries that we build
-# have a proper DT_RPATH or DT_RUNPATH tag
-define check_host_rpath
-	$(if $(filter install-host,$(2)),\
-		$(if $(filter end,$(1)),support/scripts/check-host-rpath $(3) $(HOST_DIR)))
-endef
-GLOBAL_INSTRUMENTATION_HOOKS += check_host_rpath
-
 define step_check_build_dir_one
 	if [ -d $(2) ]; then \
 		printf "%s: installs files in %s\n" $(1) $(2) >&2; \

@@ -343,6 +335,8 @@ $(BUILD_DIR)/%/.stamp_target_installed:
 # Finalize installation
 $(BUILD_DIR)/%/.stamp_installed:
 	@$(call step_start,install)
+	$(if $(filter host-patchelf,$($(PKG)_FINAL_DEPENDENCIES)), \
+		$(TOPDIR)/support/scripts/fix-rpath host)
 	@$(call step_end,install)
 	$(Q)touch $@
 

@@ -587,6 +581,10 @@ ifneq ($(1),host-skeleton)
 $(2)_DEPENDENCIES += host-skeleton
 endif
 
+ifeq ($(filter host-patchelf host-skeleton host-tar host-xz host-lzip host-ccache host-fakedate,$(1)),)
+$(2)_DEPENDENCIES += host-patchelf
+endif
+
 ifeq ($(filter host-tar host-skeleton host-fakedate,$(1)),)
 $(2)_EXTRACT_DEPENDENCIES += $(BR2_TAR_HOST_DEPENDENCY)
 endif

support/scripts/fix-rpath +19 -12

index 15705a3b0d..5b2f24ed32 100755
@@ -61,7 +61,7 @@ main() {
     local rootdir
     local tree="${1}"
     local find_args=( )
-    local sanitize_extra_args=( )
+    local sanitize_args=( )
 
     if ! "${PATCHELF}" --version > /dev/null 2>&1; then
 	echo "Error: can't execute patchelf utility '${PATCHELF}'"

@@ -84,12 +84,14 @@ main() {
                 find_args+=( "-path" "${HOST_DIR}""${excludepath}" "-prune" "-o" )
             done
 
-            # do not process the patchelf binary but a copy to work-around "file in use"
+            # do not process the patchelf binary, as it is ourselves
+            # (and it doesn't need a rpath as it doesn't use libraries
+            # from HOST_DIR)
             find_args+=( "-path" "${PATCHELF}" "-prune" "-o" )
-            cp "${PATCHELF}" "${PATCHELF}.__to_be_patched"
 
             # we always want $ORIGIN-based rpaths to make it relocatable.
-            sanitize_extra_args+=( "--relative-to-file" )
+            sanitize_args+=( "--set-rpath" )
+            sanitize_args+=( "\$ORIGIN/../lib" )
             ;;
 
         staging)

@@ -101,14 +103,18 @@ main() {
             done
 
             # should be like for the target tree below
-            sanitize_extra_args+=( "--no-standard-lib-dirs" )
+            sanitize_args+=( "--no-standard-lib-dirs" )
+            sanitize_args+=( "--make-rpath-relative" )
+            sanitize_args+=( "${rootdir}" )
             ;;
 
         target)
             rootdir="${TARGET_DIR}"
             # we don't want $ORIGIN-based rpaths but absolute paths without rootdir.
             # we also want to remove rpaths pointing to /lib or /usr/lib.
-            sanitize_extra_args+=( "--no-standard-lib-dirs" )
+            sanitize_args+=( "--no-standard-lib-dirs" )
+            sanitize_args+=( "--make-rpath-relative" )
+            sanitize_args+=( "${rootdir}" )
             ;;
 
         *)

@@ -120,20 +126,21 @@ main() {
     find_args+=( "-type" "f" "-print" )
 
     while read file ; do
-        # check if it's an ELF file
-        if ${PATCHELF} --print-rpath "${file}" > /dev/null 2>&1; then
+        rpath=$(${PATCHELF} --print-rpath "${file}" 2>/dev/null)
+        if test $? -eq 0; then
+            # For host binaries, if the rpath is already correct, skip
+            if test "${tree}" = "host" -a "${rpath}" = "\$ORIGIN/../lib" ; then
+                continue
+            fi
             # make files writable if necessary
             changed=$(chmod -c u+w "${file}")
             # call patchelf to sanitize the rpath
-            ${PATCHELF} --make-rpath-relative "${rootdir}" ${sanitize_extra_args[@]} "${file}"
+            ${PATCHELF} ${sanitize_args[@]} "${file}"
             # restore the original permission
             test "${changed}" != "" && chmod u-w "${file}"
         fi
     done < <(find "${rootdir}" ${find_args[@]})
 
-    # Restore patched patchelf utility
-    test "${tree}" = "host" && mv "${PATCHELF}.__to_be_patched" "${PATCHELF}"
-
     # ignore errors
     return 0
 }

Signature · 1 line

2.13.6

Quoted 8 lines

In order to achieve this, the following changes are made:

- HOST_LDFLAGS no longer contains -Wl,-rpath,$(HOST_DIR)

- The hook that ran at the end of every package installation (check_host_rpath) to verify that our hardcoded RPATH is indeed present in all binaries is removed, as it is no longer needed: we will force $ORIGIN/../LIB as an RPATH in all binaries.

Then you can just get rid of support/scripts/check-host-rpath that is now no longer used.

Quoted 7 lines

- host-patchelf is added as a dependency of all packages, except itself. This is necessary as all packages now need to run patchelf at the end of their installation process.

TODO: things like host-tar, host-lzip and host-ccache will have to be handled properly.

Indeed, because the very moment that patchelf is available, the next package will trigger patchelf on all executables, which means that package will be responsible for touching files from previous packages, and thus will trigger the no-two-packages-should-touch-the-same-file check.

Quoted 2 lines

This change is independent from the per-package SDK functionality, and could be applied separately.

Except it requires the host-tar et al. stuff to be handled first.

Hello,

Quoted 11 lines · Yann E. MORIN

Quoted 8 lines

In order to achieve this, the following changes are made:

- HOST_LDFLAGS no longer contains -Wl,-rpath,$(HOST_DIR)

- The hook that ran at the end of every package installation (check_host_rpath) to verify that our hardcoded RPATH is indeed present in all binaries is removed, as it is no longer needed: we will force $ORIGIN/../LIB as an RPATH in all binaries.

Then you can just get rid of support/scripts/check-host-rpath that is now no longer used.

True.

Quoted 5 lines

Quoted 5 lines

- host-patchelf is added as a dependency of all packages, except itself. This is necessary as all packages now need to run patchelf at the end of their installation process.

Quoted 2 lines

Quoted 2 lines

TODO: things like host-tar, host-lzip and host-ccache will have to be handled properly.

In fact, this TODO was no longer accurate: I did take care of host-tar, host-lzip and host-ccache as part of this RFCv3. However...

Quoted 5 lines

Indeed, because the very moment that patchelf is available, the next package will trigger patchelf on all executables, which means that package will be responsible for touching files from previous packages, and thus will trigger the no-two-packages-should-touch-the-same-file check.

... I indeed didn't think of this problem. I see two possibilities to address this:

1. Have host-patchelf built earlier. I.e have only host-tar be a dependency of host-patchelf and nothing else (in my current patch series, host-patchelf is built after host-tar, host-xz, host-lzip, host-ccache). This requires using HOSTCC_NOCCACHE to build host-tar and host-patchelf, but I don't think it's a big deal. And then do a special case in fix-rpath to not fix the rpath of the tar program, since it should not have dependencies on libraries in HOST_DIR/lib

2. Make the RPATH fixing logic smarter, and don't set a RPATH to $ORIGIN/../lib if the program doesn't use a library that is in HOST_DIR/lib. This is perhaps the most pretty solution.

Quoted 4 lines

Quoted 2 lines

This change is independent from the per-package SDK functionality, and could be applied separately.

Except it requires the host-tar et al. stuff to be handled first.

Which is why the host-tar and al. stuff is earlier. What I meant with this sentence is that reworking the host RPATH handling can be applied *before* we switch to per-package SDK.

Best regards,

Thomas

Signature · 3 lines

Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

Hello,

On Fri, 29 Dec 2017 18:31:18 +0100, Yann E. MORIN wrote:

2. Make the RPATH fixing logic smarter, and don't set a RPATH to $ORIGIN/../lib if the program doesn't use a library that is in HOST_DIR/lib. This is perhaps the most pretty solution.

Thomas,

On Fri, Dec 1, 2017 at 9:53 PM, Thomas Petazzoni
<[email protected]> wrote:
[snip]

package/Makefile.in HOST_CPPFLAGS = -I$(HOST_DIR)/include · line 223

Quoted 8 lines

package/Makefile.in +1 -1

@@ -220,7 +220,7 @@ HOST_CPPFLAGS  = -I$(HOST_DIR)/include
 HOST_CFLAGS   ?= -O2
 HOST_CFLAGS   += $(HOST_CPPFLAGS)
 HOST_CXXFLAGS += $(HOST_CFLAGS)
-HOST_LDFLAGS  += -L$(HOST_DIR)/lib -Wl,-rpath,$(HOST_DIR)/lib
+HOST_LDFLAGS  += -L$(HOST_DIR)/lib

Reproduced with the following config
BR2_aarch64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
BR2_SYSTEM_DHCP="eth0"
BR2_LINUX_KERNEL=y
BR2_LINUX_KERNEL_CUSTOM_VERSION=y BR2_LINUX_KERNEL_CUSTOM_VERSION_VALUE="4.13.6" BR2_LINUX_KERNEL_USE_CUSTOM_CONFIG=y BR2_LINUX_KERNEL_CUSTOM_CONFIG_FILE="board/qemu/aarch64-virt/linux-4.13.config" BR2_PACKAGE_CHECKPOLICY=y
BR2_PACKAGE_POLICYCOREUTILS=y
BR2_PACKAGE_SETOOLS=y
BR2_PACKAGE_HOST_CHECKPOLICY=y

That build will first fail at host-libglib2 (mentioned in previous email on this thread) because the autotools test of pcre unicode will fail as it can't find the libraries. See patch
https://patchwork.ozlabs.org/patch/870130/

Next it will fail at setools because of a python header include now only pointing at the python host/include vs the full host/include. It seems setuptools by default inherits an include path to python headers install location. See patch
https://patchwork.ozlabs.org/patch/870131/.

Last it looks like the rpath still needs to be set so that the linker can evaluate the shared library's shared library dependency. I tested the changes below and it builds, plus doesn't leave an rpath in the executable (Thanks Roman).
<<<<<<<<< CUT

package/Makefile.in +1 -1

@@ -238,7 +238,7 @@ HOST_CPPFLAGS  = -I$(HOST_DIR)/include
 HOST_CFLAGS   ?= -O2
 HOST_CFLAGS   += $(HOST_CPPFLAGS)
 HOST_CXXFLAGS += $(HOST_CFLAGS)
-HOST_LDFLAGS  += -L$(HOST_DIR)/lib
+HOST_LDFLAGS  += -L$(HOST_DIR)/lib  -Wl,-rpath-link,$(HOST_DIR)/lib
 
 # The macros below are taken from linux 4.11 and adapted slightly.
 # Copy more when needed.

<<<<<<<<< CUT
The use of LD_LIBRARY_PATH set to the same value also works. Here's my understanding as pulled from the ld man page.

When using ELF or SunOS, one shared library may require another. This happens when an ld -shared link includes a shared library as one of the input files. When the linker encounters such a dependency when doing a non-shared, non-relocateable link, it will automatically try to locate the required shared library and include it in the link, if it is not included explicitly. In such a case, the -rpath-link option specifies the first set of directories to search. The -rpath-link option may specify a sequence of directory names either by specifying a list of names separated by colons, or by appearing multiple times. The linker uses the following search paths to locate required shared libraries.
1) Any directories specified by -rpath-link options.
2) Any directories specified by -rpath options. The difference between -rpath and -rpath-link is that directories specified by -rpath options are included in the executable and used at runtime, whereas the

-rpath-link option is only effective at link time.
3) On an ELF system, if the -rpath and rpath-link options were not used, search the contents of the environment variable LD_RUN_PATH.
4) On SunOS, if the -rpath option was not used, search any directories specified using -L options.
5) For a native linker, the contents of the environment variable LD_LIBRARY_PATH.
6) The default directories, normally /lib' and /usr/lib'.

Quoted 29 lines · Matthew Weber

Thomas,

On Fri, Dec 1, 2017 at 9:53 PM, Thomas Petazzoni
<[email protected]> wrote:
[snip]

package/Makefile.in HOST_CPPFLAGS = -I$(HOST_DIR)/include · line 223

Quoted 8 lines

package/Makefile.in +1 -1

@@ -220,7 +220,7 @@ HOST_CPPFLAGS  = -I$(HOST_DIR)/include
 HOST_CFLAGS   ?= -O2
 HOST_CFLAGS   += $(HOST_CPPFLAGS)
 HOST_CXXFLAGS += $(HOST_CFLAGS)
-HOST_LDFLAGS  += -L$(HOST_DIR)/lib -Wl,-rpath,$(HOST_DIR)/lib
+HOST_LDFLAGS  += -L$(HOST_DIR)/lib

Reproduced with the following config
BR2_aarch64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
BR2_SYSTEM_DHCP="eth0"
BR2_LINUX_KERNEL=y
BR2_LINUX_KERNEL_CUSTOM_VERSION=y BR2_LINUX_KERNEL_CUSTOM_VERSION_VALUE="4.13.6" BR2_LINUX_KERNEL_USE_CUSTOM_CONFIG=y BR2_LINUX_KERNEL_CUSTOM_CONFIG_FILE="board/qemu/aarch64-virt/linux-4.13.config" BR2_PACKAGE_CHECKPOLICY=y
BR2_PACKAGE_POLICYCOREUTILS=y
BR2_PACKAGE_SETOOLS=y
BR2_PACKAGE_HOST_CHECKPOLICY=y

Quoted 4 lines

That build will first fail at host-libglib2 (mentioned in previous email on this thread) because the autotools test of pcre unicode will fail as it can't find the libraries. See patch
https://patchwork.ozlabs.org/patch/870130/

IIUC this patch should no longer be needed if we do -Wl,-rpath-link instead, right?

Quoted 5 lines

Next it will fail at setools because of a python header include now only pointing at the python host/include vs the full host/include. It seems setuptools by default inherits an include path to python headers install location. See patch
https://patchwork.ozlabs.org/patch/870131/.

If setuptools completely ignores the CFLAGS we pass to it, then it should be
fixed in setuptools, not in a per-package patch...

Quoted 35 lines

Last it looks like the rpath still needs to be set so that the linker can evaluate the shared library's shared library dependency. I tested the changes below and it builds, plus doesn't leave an rpath in the executable (Thanks Roman).
<<<<<<<<< CUT

package/Makefile.in +1 -1

@@ -238,7 +238,7 @@ HOST_CPPFLAGS  = -I$(HOST_DIR)/include
 HOST_CFLAGS   ?= -O2
 HOST_CFLAGS   += $(HOST_CPPFLAGS)
 HOST_CXXFLAGS += $(HOST_CFLAGS)
-HOST_LDFLAGS  += -L$(HOST_DIR)/lib
+HOST_LDFLAGS  += -L$(HOST_DIR)/lib  -Wl,-rpath-link,$(HOST_DIR)/lib
 
 # The macros below are taken from linux 4.11 and adapted slightly.
 # Copy more when needed.

<<<<<<<<< CUT
The use of LD_LIBRARY_PATH set to the same value also works. Here's my understanding as pulled from the ld man page.

When using ELF or SunOS, one shared library may require another. This happens when an ld -shared link includes a shared library as one of the input files. When the linker encounters such a dependency when doing a non-shared, non-relocateable link, it will automatically try to locate the required shared library and include it in the link, if it is not included explicitly. In such a case, the -rpath-link option specifies the first set of directories to search. The -rpath-link option may specify a sequence of directory names either by specifying a list of names separated by colons, or by appearing multiple times. The linker uses the following search paths to locate required shared libraries.
1) Any directories specified by -rpath-link options.
2) Any directories specified by -rpath options. The difference between -rpath and -rpath-link is that directories specified by -rpath options are included in the executable and used at runtime, whereas the

Quoted 8 lines

-rpath-link option is only effective at link time.
3) On an ELF system, if the -rpath and rpath-link options were not used, search the contents of the environment variable LD_RUN_PATH.
4) On SunOS, if the -rpath option was not used, search any directories specified using -L options.
5) For a native linker, the contents of the environment variable LD_LIBRARY_PATH.
6) The default directories, normally /lib' and /usr/lib'.

I have read this section of the manpage several times over the years, and now
is the first time it makes sense to me :-)

So indeed, for NEEDED libraries within libraries that you link with, ld will
not look at -L but it will only look at -rpath-link and -rpath (the rest is not relevant in our context). So if we remove -rpath, it does make sense that we pass -rpath-link instead (we already pass -L as well).

That said, I still don't understand why it didn't work without -rpath-link.
Indeed, there *is* a DT_RUNPATH set in the library (at least according to the commit log of this patch, since fix-rpath is run at the end of installation), but it's set to $ORIGIN/../lib instead of to an absolute path. So, does ld not look at DT_RUNPATH? Or does it evaluate $ORIGIN in an unexpected way?

Finally, note that we said in the Buildroot meeting that this patch is in fact
probably NOT needed for per-package SDK, because the NEEDed libraries are in fact installed in the location specified by the absolute RPATH. It's just weird to have an RPATH referring outside of the per-package staging, but it's not an actual problem.

As we are going to move to per-package SDK, the location of CCACHE and therefore the definitions of HOSTCC and HOSTCXX need to be evaluated at the time of use and not at the time of assignment. Indeed, the value of HOST_DIR changes from one package to the other.

Therefore, we need to change from := to =.

In addition, while doing A := $(something) $(A) is possible, doing A = $(something) $(A) is not legal. So, instead of defining HOSTCC in terms of the current HOSTCC variable, we re-use HOSTCC_NOCCACHE instead.

Signed-off-by Thomas Petazzoni <[email protected]>

---
Changes since v2:
- New patch

---
 Makefile | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Makefile +3 -3

index d8fa91120b..e05a1ec09a 100644
@@ -446,11 +446,11 @@ BR_PATH = "$(HOST_DIR)/bin:$(HOST_DIR)/sbin:$(PATH)"
 TARGET_DIR_WARNING_FILE = $(TARGET_DIR)/THIS_IS_NOT_YOUR_ROOT_FILESYSTEM
 
 ifeq ($(BR2_CCACHE),y)
-CCACHE := $(HOST_DIR)/bin/ccache
+CCACHE = $(HOST_DIR)/bin/ccache
 BR_CACHE_DIR ?= $(call qstrip,$(BR2_CCACHE_DIR))
 export BR_CACHE_DIR
-HOSTCC := $(CCACHE) $(HOSTCC)
-HOSTCXX := $(CCACHE) $(HOSTCXX)
+HOSTCC = $(CCACHE) $(HOSTCC_NOCCACHE)
+HOSTCXX = $(CCACHE) $(HOSTCXX_NOCCACHE)
 else
 export BR_NO_CCACHE
 endif

Signature · 1 line

2.13.6

This commit implemnts the core of the move to per-package SDK and target directories. The main idea is that instead of having a global output/host and output/target in which all packages install files, we switch to per-package host and target folders, that only contain their explicit dependencies.

There are two main benefits:

- Packages will no longer discover dependencies that they do not explicitly indicate in their <pkg>_DEPENDENCIES variable.

- We can support top-level parallel build properly, because a package only "sees" its own host directory and target directory, isolated from the build of other packages that can happen in parallel.

It works as follows:

- A new output/per-package/ folder is created, which will contain one sub-folder per package, and inside it, a "host" folder and a "target" folder:

output/per-package/busybox/target
output/per-package/busybox/host
output/per-package/host-fakeroot/target
output/per-package/host-fakeroot/host

This output/per-package/ folder is PER_PACKAGE_DIR.

- The global TARGET_DIR and HOST_DIR variable now automatically point to the per-package directory when PKG is defined. So whenever a package references $(HOST_DIR) or $(TARGET_DIR) in its build process, it effectively references the per-package host/target directories. Note that STAGING_DIR is a sub-dir of HOST_DIR, so it is handled as well.

Using this patchset without any -j. I found a case with host-setools where it uses the python-package framework to build and using setuptools puts a default -I<abs path>/per-package/host-python/host/include. Previously this would have pointed at the full sysroot so now there is a missing include.

I've added it for setools and continued on testing at this point. I'm unsure if we would consider this a pkg-python.mk update or each package.....

Matt

On 06-02-18 15:45, Matthew Weber wrote:
[snip]

Quoted 1 line

Quoted 5 lines

Using this patchset without any -j. I found a case with host-setools where it uses the python-package framework to build and using setuptools puts a default -I<abs path>/per-package/host-python/host/include. Previously this would have pointed at the full sysroot so now there is a missing include.

There should also be a -I<abs path>/per-package/host-setools/host/include
because that's in HOST_CFLAGS. Or does this get hardcoded by setuptools?

BTW, I guess the problem is that it fails to find libselinux or libsepol includes?

Thomas Petazzoni (15):
pkgconf: use relative path to STAGING_DIR instead of absolute path
toolchain: post-pone evaluation of TOOLCHAIN_EXTERNAL_BIN

Makefile, skeleton: move the host skeleton logic to host-skeleton
package
pkg-cmake: install CMake files as part of a package

pkg-generic: add .stamp_installed step

package/pkg-generic: add the concept of extract dependency package/pkg-generic: handle host-tar as an extract dependency package/pkg-generic: handle host-xz as an extract dependency package/pkg-generic: handle host-lzip as an extract dependency package/pkg-generic: handle host-ccache as a regular dependency package/pkg-generic: handle host-fakedate as a regular dependency
core: kill DEPENDENCIES_HOST_PREREQ