The current logic deletes old project files e.g. sources,
but *after* updating the project hash.
This means that if a deletion fails, and the directory
is still there (e.g. src/coreboot/default/) afterward, it's
now a tainted archive, yet the hash has been updated, so
subsequent runs of the build system will cause unknown
errors.
This patch fixes that, by first copying the new hash to
a temporary file. *Then*, deletions are handled, and the
final hash file is updated afterward.
The code is now a bit more bloated as a result, but this will
reduce the risk of tainted sources being handled under fault
conditions.
Signed-off-by: Leah Rowe <leah@libreboot.org>
in that case, the previous tree-wide check will cover
it, so the current logic wastes computational time.
this patch therefore somewhat optimises the code.
Signed-off-by: Leah Rowe <leah@libreboot.org>
This fixes the following error on ./mk release:
cp: cannot stat 'elf/coreboot/default/d510mo/libgfxinit_txtmode/coreboot.rom': No such file or directory
I recently re-wrote the handling of coreboot images, and
I overlooked this entirely. When a given target specifies
release=n, it has to be skipped, so builds are not done.
The "release" variable is already checked, in mkcoreboottar.
Let's also put it in the other mkhelper functions, to make sure
there are no errors.
Signed-off-by: Leah Rowe <leah@libreboot.org>
This was a problem when I did it before, because individual
target builds weren't automatically re-compiled when needed.
The recent design improvements in lbmk enable this to be
done again.
Cached images in elf/ have no payloads, so they are a liability,
therefore they are padded by one byte to prevent flashing. This
solves the problem that the previous caching had.
With this change, modifying payloads can be tested without
needing to re-build coreboot each time.
The cached coreboot builds are also automatically re-built when
needed, which is another improvement this time, compared to
the last time coreboot builds were cached in this manner.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Detect when a config changes. This is done even if the
entire tree doesn't change.
This is already done per-tree if files change, but
individual project files don't change.
For example, if a grub.cfg changes, the given cached
build for that GRUB tree isn't deleted. Same thing if
a given U-Boot config doesn't change.
This patch fixes a longstanding design flaw of lbmk,
making auto-re-builds more reliable. This complements
another recent change, that deletes all target builds
of a given tree when the tree changes.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Target builds go inside a common directory for
the given tree now, which gets deleted, thus
deleting all target builds of that given tree.
Therefore, the deletion being removed is redundant.
Signed-off-by: Leah Rowe <leah@libreboot.org>
as opposed to target/
for example:
image the command:
./mk -b u-boot amd64coreboot
This would put the U-Boot binaries here:
elf/u-boot/amd64coreboot/default/
With this change, they now go here:
elf/u-boot/x86_64/amd64coreboot/default/
This solves a problem that existed previously, where
you could modify a given tree in a multi-tree project,
but cached builds for targets branching separately off
of each tree would not be deleted, and thus not re-built.
This accomplishes such a result, without needing to
further check hashes of individual targets.
The latter will still be done, in a future change, because
this change doesn't fix another problem:
If you change a given config, e.g. targetname "foo" which
uses tree "bar", elf/foo/ would not be removed automatically
for re-build.
So this change only deletes individual target builds when
their master tree changes.
Where the target and tree are the same, this also means
elf/tree/target/
for example: seabios/default would create binaries in:
elf/seabios/default/default/
not:
elf/seabios/default/
Signed-off-by: Leah Rowe <leah@libreboot.org>
it's 2019-2025 now, not 2019-2024, because i recently imported
new pcsx-redux upstream changes that go up to June 2025.
Signed-off-by: Leah Rowe <leah@libreboot.org>
it means nothing here. in context, if a non-zero return
is observed, we should not do anything here, which is
already the behaviour anyway, except that "break" means
nothing since we're not in a loop here.
where an error exit should be observed, x_ is used inside
the command given for fx_
Signed-off-by: Leah Rowe <leah@libreboot.org>
handle errors on sha512sum - also handle awk errors inside
the mini subshell, and provide overall error handling.
we know that the project.hash file should always exist, and
always be read no matter what; technically, the find command
that proceeds it might not yield any results, but an empty
file would then be produced.
the edge case of an empty file would have lead to an error
beforehand, when configuring the project in function,
configure_project(), so we've already got that covered.
Signed-off-by: Leah Rowe <leah@libreboot.org>
when reading old_pjhash, we need to error out where a read
error occurs. such an error is unlikely, but could occur under
certain edge cases.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Don't do no-op if it fails; fall back to "clean" instead,
and fail if that fails.
The no-op was there was not all projects have distclean,
but we do intend for them all to be cleaned.
We mitigate further error by only running make-clean if
a makefile exists.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Don't copy the files directly, because we might be doing
this from a work directory that has no files; in this case,
generic "unknown" variables are used, without generating
any files, so the current logic would produce an error.
However, we do need to create those dot files, because
we then rely on them for building release binaries.
The new logic maintains current behaviour, while fixing
this technical edge-case scenario via mitigation.
Signed-off-by: Leah Rowe <leah@libreboot.org>
stick it in git_prep, which both single- and multi-tree
projects will use, when downloading git repositories.
Signed-off-by: Leah Rowe <leah@libreboot.org>
The following execution will result in another printf
that says exactly what is being downloaded.
There is no need to inform the user twice about
what is being downloaded.
Signed-off-by: Leah Rowe <leah@libreboot.org>
A git-pull is performed immediately after git-fetch.
Git-pull already performs git-fetch as a prerequisite.
Signed-off-by: Leah Rowe <leah@libreboot.org>
the checks at the end of the function are mostly
superfluous, because bad_checksum() is immediately
called just beforehand, and performs the same checks.
Signed-off-by: Leah Rowe <leah@libreboot.org>
the current behaviour is a relic from the older lbmk
design, before recent auditing.
the current logic would cause xbmk to continue execution,
going into a child process with .git/ being a symlink.
The .git/ directory should never be a symlink, because
it is extremely error-prone.
Signed-off-by: Leah Rowe <leah@libreboot.org>
We rely on a non-zero exit on other try_ commands, which
works fine there because we then check the file afterward
and error out accordingly.
For git repositories, we assume that both mirrors are
identical and therefore once we get to the first clone
attempt, we assume that it must succeed.
Therefore, if it does not succeed, we must fail. This fixes
a regression I found in testing, where sometimes a failed
patching attempt would not result in an error exit, and
would therefore result in broken sources being present.
In practise, I always very closely watch the terminal when
testing xbmk, especially when updating project patches, so
we probably didn't introduce any broken sources in practice.
Signed-off-by: Leah Rowe <leah@libreboot.org>
This code was introduced to provide fault tolerance,
so that if I forgot to manually update the configs
myself, builds would still succeed, e.g. coreboot
builds.
However, there have been cases in the past where this
introduces settings we don't want, and in general we
do want to know when there is an error in the configs.
The policy should always be: fail early, fail hard.
This also mitigates bugs in U-Boot's build system; for
example, when I last attempted to update the U-Boot
tree for x86, make-oldconfig introduced a lot of junk
settings unrelated, which then introduced code that
would brick the board if you tried it on one, e.g.
it broke booting most Linux kernels via bootflow.
With this change, U-Boot will be easier to handle,
which normally requires manual configuration; the
automated make-oldconfig reconfiguration feature
breaks U-Boot. This will no longer occur, since we
no longer run it manually.
On the other hand, this feature has also prevented
other disastrous bugs in the past, such as when I
forgot to properly set the SPD size on T480; it was
set to 256 bytes, not 512 as is correct. Therefore,
this new design change means I must also be more
vigilant about config changes in project trees.
Signed-off-by: Leah Rowe <leah@libreboot.org>
it mainly does general tasks, like handling utils
and enabling ccache. the vfiles are a small part.
rename the function accordingly. it is called by
premake, so let's call it corebootpremake.
this change will also make sense when cherry-picked
into cbmk, which does not handle vfiles at all.
Signed-off-by: Leah Rowe <leah@libreboot.org>
we simply do not need to run the make-oldconfig command
at all, and after removing it, the "cook" function seemed
quite redundant so i merged it with mkvendorfiles()
Signed-off-by: Leah Rowe <leah@libreboot.org>
define it with a single variable, rather than several.
this allows several checks to be greatly simplified.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Right now, if cache/clone/PROJECT/ already exists,
the logic for pulling new changes doesn't execute,
and neither does the logic for updating remotes.
This is bad when updating revisions, because then
manual updating is required, defeating the purpose
of xbmk's own automation in this regard.
Fix it by only checking the cached download on files,
not Git repositories; the try_git function itself will
already perform this check, before updating remotes
and pulling in new commits from upstream.
The updating only happens when a given target directory
doesn't exist, e.g. src/flashprog/ or src/grub/default/,
so this won't slow down release builds for example.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Otherwise, an "unknown" version number is created.
This regression was caused by the recent optimisation
that reduces the amount of extra work done by init.sh
on child instances of xbmk.
As a result of those changes, now release.sh has to
do some minor initialisation of its own, such as this.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Otherwise, ./mk -d (without arguments) fails for GRUB,
which first requires running autoconf to get a Makefile.
Signed-off-by: Leah Rowe <leah@libreboot.org>
This reverts commit fb7aaa78bb.
it caused a few issues. will re-do later
the old code isn't really broken, just inefficient, because
several files are scanned twice, but in practise the overhead
isn't that great
The error occurs sometimes, when bruteforcing me.bin:
ERROR ./mk: Unhandled error for: mv /home/user/lbmk/tmp/me.bin /home/user/lbmk/cache/tmpdl/check
This revert should fix the issue, for now.
Leah Rowe