This commit implements 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 directories, that only contain
their explicit dependencies.
There are two main benefits:
- Packages will now see only the dependencies they explicitly list in
their <pkg>_DEPENDENCIES variable, and the recursive dependencies
thereof.
- 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/ directory is created, which will contain
one sub-directory per package, and inside it, a "host" directory
and a "target" directory:
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/ directory 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.
- Of course, packages have dependencies, so those dependencies must
be installed in the per-package host and target directories. To do
so, we simply rsync (using hard links to save space and time) the
host and target directories of the direct dependencies of the
package to the current package host and target directories.
We only need to take care of direct dependencies (and not
recursively all dependencies), because we accumulate into those
per-package host and target directories the files installed by the
dependencies. Note that this only works because we make the
assumption that one package does *not* overwrite files installed by
another package.
This is done for "extract dependencies" at the beginning of the
extract step, and for "normal dependencies" at the beginning of the
configure step.
This is basically enough to make per-package SDK and target work. The
only gotcha is that at the end of the build, output/target and
output/host are empty, which means that:
- The filesystem image creation code cannot work.
- We don't have a SDK to build code outside of Buildroot.
In order to fix this, this commit extends the target-finalize step so
that it starts by populating output/target and output/host by
rsync-ing into them the target and host directories of all packages
listed in the $(PACKAGES) variable. It is necessary to do this
sequentially in the target-finalize step and not in each
package. Doing it in package installation means that it can be done in
parallel. In that case, there is a chance that two rsyncs are creating
the same hardlink or directory at the same time, which makes one of
them fail.
This change to per-package directories has an impact on the RPATH
built into the host binaries, as those RPATH now point to various
per-package host directories, and no longer to the global host
directory. We do not try to rewrite such RPATHs during the build as
having such RPATHs is perfectly fine, but we still need to handle two
fallouts from this change:
- The check-host-rpath script, which verifies at the end of each
package installation that it has the appropriate RPATH, is modified
to understand that a RPATH to $(PER_PACKAGE_DIR)/<pkg>/host/lib is
a correct RPAT.
- The fix-rpath script, which mungles the RPATH mainly for the SDK
preparation, is modified to rewrite the RPATH to not point to
per-package directories. Indeed the patchelf --make-rpath-relative
call only works if the RPATH points to the ROOTDIR passed as
argument, and this ROOTDIR is the global host directory. Rewriting
the RPATH to not point to per-package host directories prior to
this is an easy solution to this issue.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
As suggested by Arnout Vandecappelle, let's document the
elf_needs_rpath() and check_elf_has_rpath() functions, before we make
them a bit more complicated with per-package directory support.
Suggested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Inside the check_elf_has_rpath(), we check if the host binary has a
correct RPATH, which should be either an absolute path to
$(HOST_DIR)/lib, or a relative path using $ORIGIN. Those two
conditions are checked in a single statements, but as we are going to
add a third condition, let's split this up a bit:
- If we have a RPATH to $(HOST_DIR)/lib -> we're good, return 0
- If we have a RPATH to $ORIGIN/../lib -> we're good, return 0
- Otherwise, we will exit the loop, and return 1
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
We sanity-check the host executables that they have a correct RPATH
pointing to the host libraries.
This is currently done by looking for all files in $(HOST_DIR) that
match the 'ELF executable' pattern (a bit more complex, but that's
idea).
However, when an executable is built with -fPIE of -fpie, it no longer
appears to be an 'ELF executable', but it rather looks like an 'ELF
sheard object' (like if it were an library.
So, we miss those files.
It turns out that the problem is a real one, because quite a few
mainline distros, expecially those based on Debian for example, have
already switched to generating PIE code by default, and thus we miss on
a whole class of systems..
We fix that by simply looking if we can find an ELF interpreter in each
file. If we there is one, this is an ELF executable; if not, it may be
anything else: we don't care (not even about ELF libraries).
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Now all packages have been updated to install things in $(HOST_DIR)/lib
instead of $(HOST_DIR)/usr/lib, there should no longer be any reason
to have $(HOST_DIR)/usr/lib in the RPATH, so we don't allow it any more.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Since $(HOST_DIR)/usr/{bin,sbin} are now symlinks to
$(HOST_DIR)/{bin,sbin}, it makes no sense to check them - they are
already covered.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This is a step towards eliminating $(HOST_DIR)/usr. It allows us to
convert all packages installing things into $(HOST_DIR)/usr/lib without
affecting the rest.
To allow compatibility with packages that still use $(HOST_DIR)/usr as
the prefix, create a symlink from usr/lib to ../lib.
Note that the symlink creation will break when $(HOST_DIR)/usr/lib
already exists as a directory, i.e. when rebuilding in an existing
output directory. This is necessary: if we don't break it now, the
following commits (which remove the usr part from various variables)
_will_ break it.
At the same time as creating this symlink, we also have to update the
check-host-rpath script to accept both $(HOST_DIR)/usr/lib and
$(HOST_DIR)/lib, because depending on how the package derives the
path, it may be different.
Since there are some dependency chains that involve $(STAGING_DIR),
$(STAGING_DIR) may in fact be created before $(HOST_DIR). Since
$(STAGING_DIR) is a subdirectory of $(HOST_DIR), it is possible that the
newly added rule for $(HOST_DIR) never triggers. To make sure that the
rule does trigger, add an order-only dependency from $(STAGING_DIR) to
$(HOST_DIR).
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
"$ORIGIN/../../usr/lib" is also a valid RPATH for binaries in
"$hostdir/usr/bin". After RPATH sanitation, all RPATH
directories start with "$ORIGIN".
Signed-off-by: Wolfgang Grandegger <wg@grandegger.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
At least syslinux is installing stuff in HOST_DIR/sbin.
Cc: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Despite the comment saying so, the trailing '/' in the host directory is
not removed. Note however that it is properly removed from extracted
RPATH tags.
This is not visible when the host directory is our default $(O)/host
location, but breaks for user-supplied external host directory, when
the user leaves a trailing slash in the path.
Fix that.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When we build our host programs, and they depend on a host library we
also build, we want to ensure that program actually uses that library at
runtime, and not the one from the system.
We currently ensure that in two ways:
- we add a RPATH tag that points to our host library directory,
- we export LD_LIBRARY_PATH to point to that same directory.
With these two in place, we're pretty much confident that our host
libraries will be used by our host programs.
However, it turns our that not all the host programs we build end up
with an RPATH tag:
- some packages do not use our $(HOST_LDFLAGS)
- some packages' build system are oblivious to those LDFLAGS
In this case, there are two situations:
- the program is not linked to one of our host libraries: it in fact
does not need an RPATH tag [0]
- the program actually uses one of our host libraries: in that case it
should have had an RPATH tag pointing to the host directory.
For libraries, they only need an RPATH if they depend on another library
that is not installed in the standard library path. However, any system
library will already be in the standard library path, and any library we
install ourselves is in $(HOST_DIR)/usr/lib so already in RPATH.
We add a new support script that checks that all ELF executables have
a proper DT_RPATH (or DT_RUNPATH) tag when they link to our host
libraries, and reports those file that are missing an RPATH. If a file
missing an RPATH is an executable, the script aborts; if only libraries
are are missing an RPATH, the script does not abort.
[0] Except if it were to dlopen() it, of course, but the only program
I'm aware of that does that is openssl, and it has a correct RPATH tag.
[Peter: reworded as suggested by Arnout, fix HOT_DIR typo in comment]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <jacmet@uclibc.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
