summaryrefslogtreecommitdiff
path: root/baserock-bootstrap
Commit message (Collapse)AuthorAgeFilesLines
* Fix baserock: aliased URLs to point at new locationLars Wirzenius2012-11-211-6/+6
|
* Update bootstrap script to use new namesJames Thomas2012-10-011-3/+3
|
* Use tarballs instead of bundlesDaniel Silverstone2012-09-251-5/+5
| | | | | | | This makes a non-backward-compatible change to morph which switches it to using tarballs instead of bundles when initialising cached git repositories. This is faster because it doesn't require index-pack --fix-thin operations on the machine running morph.
* Don't require .morph in triplet argumentsSam Thursfield2012-09-121-7/+7
| | | | | | | | | This gives us consistency with morphologies, where the triplets are repo|ref|morphology, not repo|ref|filename Anyone who runs 'morph build baserock:morphs master system.morph' will now see an error ending with 'was looking for system.morph.morph', which should make it clear where they have gone wrong.
* bootstrap: Fix to use morph build-morphologySam Thursfield2012-09-121-3/+3
|
* baserock-bootstrap: remove update-gits loopRichard Maw2012-07-021-23/+14
| | | | | | | | | This should no longer be needed, since gitano on our own server is more stable and faster than gitorious.org was for us. Ideally we would have update-gits have a set of repositories to update and if one fails, increment the fail count and move on to the next if it hasn't exceeded the retries limit.
* bootstrap: remove softfp from armv5 buildRichard Maw2012-06-251-1/+1
|
* bootstrap: don't assume arm is armv7Richard Maw2012-06-251-3/+11
| | | | | | We need to support ARMv5, so we need to set different flags. If uname -m has armv7 then use the configuration we had before, if it has armv5 then use one with -march=armv5 and without the fpu.
* bootstrap: turn stratum into tarball for fillerRichard Maw2012-06-201-10/+8
| | | | | | | | | | | | | | The change of strata into chunk lists broke being able to use the strata as staging filler. The assemble-stratum script should be able to provide this stratum This requires not removing the built chunks from the artifact cache any more, and preferably not moving them around. Removing them should no longer be required, since cache keys have become more advanced since it used to reuse chunks build in prior stages. If it reuses them now, they should actually be usable.
* Use the artifacts/ subdirectory of the cache for dumping build times.Jannis Pohlmann2012-05-101-3/+3
|
* Use the new {x86_64-generic,imx53}-bootstrap-pass3 strata for pass 3.Jannis Pohlmann2012-05-091-8/+4
| | | | | | This should fix our "nasm not found when building syslinux" problems. The new strata also bundle all the desired results of pass 3 together for each platform we support bootstrapping on.
* Revert "bootstrap: disable dumping of build times"Jannis Pohlmann2012-05-091-3/+3
| | | | | | | This reverts commit 600218313a655815d902afad40c1eff5c45a20fd. I think we should now be able to log build times again, as the .meta files are being created again.
* Use baserock:morphs for what we build in pass 3 of baserock-bootstrap.Jannis Pohlmann2012-05-081-3/+3
|
* Use 'baserock' repo alias for baserock:morphs in baserock-bootstrap.Jannis Pohlmann2012-05-081-7/+7
|
* baserock-bootstrap: attempt to fix staging fillerRichard Maw2012-05-041-1/+1
| | | | --option=path*foo won't be glob expanded, but --option path*foo will
* baserock-bootstrap: fix path for staging fillerRichard Maw2012-05-041-7/+5
| | | | | | | The changes to the cache directory layout meant that this failed to find the strata. This also fixes it extracting both pass1 and pass2's strata.
* Make baserock-bootstrap use imx53/x86_64-generic BSP strata.Jannis Pohlmann2012-05-011-2/+4
|
* bootstrap: fix another non-sbin PATHRichard Maw2012-04-271-1/+1
|
* bootstrap: don't fail on /tree/{sys,proc} unmountRichard Maw2012-04-271-6/+6
|
* bootstrap: include sbin in path for chrootRichard Maw2012-04-271-1/+1
| | | | | | chroot is used in morph, an absolute path would cause problems later in the build, when /tools is the prefix, so PATH needs to be able to find chroot, so the usual sbin needs to be included.
* bootstrap: disable dumping of build timesRichard Maw2012-04-241-3/+3
| | | | | The required metadata is not written to the cache any more. This should be re-implemented.
* Dump build times of all passes in baserock-bootstrap.Jannis Pohlmann2012-04-121-0/+84
| | | | | | This commit renames extract-build-times.py to dump-build-times and makes use of it after every pass in baserock-bootstrap. This will add the build times of all chunks to the bootstrap log.
* Include /tools/{lib,lib64} in LD_LIBRARY_PATH for pass 2.Jannis Pohlmann2012-04-051-0/+1
|
* Pass CFLAGS=-O2 to everything from outside.Jannis Pohlmann2012-03-271-2/+2
| | | | | | At some point we will want to have a concept of a target architecture that includes default compiler flags etc. but for now, --target-cflags will do.
* Use -mfloat-abi=softfp for now until we have GCC > 4.4 in the chroot.Jannis Pohlmann2012-03-271-1/+1
| | | | | | | | | | | The combination -mfpu=vfpv3-d16 =mfloat-abi=hard is only supported from GCC > 4.5, according to http://wiki.debian.org/ArmHardFloatPort/VfpComparison If we want hard float we'll have to switch to a more recent version of GCC installed in the squeeze chroot we run baserock-bootstrap in. Upstream GCC 4.5 supports it, as does the CodeSourcery 2009q1 compiler.
* Add --target-cflags option; set arch/fp in baserock-bootstrap on ARM.Jannis Pohlmann2012-03-261-3/+8
| | | | | | | | The --target-cflags option is really just a temporary solution. At some point we will want to have a proper target architecture/device description that morph uses to define compiler flags etc. For now, I hope this will do.
* Remove unused code in baserock-bootstrap.Jannis Pohlmann2012-03-211-21/+0
|
* Set --no-ccache for morph in pass 1.Jannis Pohlmann2012-03-201-0/+1
| | | | It might be the cause of weird internal binutils and gcc errors.
* baserock-bootstrap: switch from topic branchRichard Maw2012-03-191-2/+2
|
* Merge remote branch 'origin/master' into rm/morph-pass1Richard Maw2012-03-191-6/+2
|\
| * Use the bootstrap-pass2.morph stratum for pass 2.Jannis Pohlmann2012-03-161-7/+3
| |
* | bootstrap: fix lfs substitutionRichard Maw2012-03-191-1/+1
| | | | | | | | | | bash variable expansion behaves a little differently to sed the final section's / are not treated specially
* | bootstrap: convert double slash to singleRichard Maw2012-03-191-0/+1
| | | | | | | | | | | | | | | | pwd may be just /, at which point LFS begins with // Python's distutils fails to compile when --prefix is //tree/tools and --root is given as it installs to //tree/tools but tries to compile files in --root. This works without the double slash, so fix it
* | Revert "bootstrap: fix path beginning with //"Richard Maw2012-03-191-1/+1
| | | | | | | | | | This reverts commit e522e8d4798dec4092fabaeb89b532e970c3296f. pwd does not always end with /
* | bootstrap: fix path beginning with //Richard Maw2012-03-191-1/+1
| | | | | | | | | | | | | | This strangely breaks python's distutils so setup.py fails. if --prefix is //tree/tools then it ignores --root when it copies the files, however when it later goes to compile the scripts it does use --root so it fails to find the files.
* | bootstrap: fix path for pass 1Richard Maw2012-03-161-2/+0
| | | | | | | | | | Pass 1 should have $tools first in the path, otherwise the wrong compiler will be used
* | Merge remote branch 'origin/master' into rm/morph-pass1Richard Maw2012-03-161-0/+1
|\ \ | |/
| * tar: add -h option to extractionRichard Maw2012-03-161-2/+2
| | | | | | | | | | | | This should stop tar from replacing symbolic links with directories and instead follow the symbolic links, extracting the files into the linked directories
* | Merge branch 'master' into rm/morph-pass1-mergeRichard Maw2012-03-161-14/+40
|\ \ | |/
| * Move pass1_file to the top right after pass1_zlib.Jannis Pohlmann2012-03-151-2/+2
| | | | | | | | | | | | | | | | | | | | | | This fixes a strange issue where the file binary will segfault in __libc_start_main() because it uses the host runtime dynamic linker and not the on in /tree/tools. The host's ld-linux comes from eglibc 2.12 and the rtld_global_ro struct in 2.12 has a different size than the one in eglibc 2.15. This, the _rtld_global_ro.dl_audit pointer gets corrupted and __libc_start_main() segfaults. This is miraculously fixed by building file immediately after zlib.
| * Only create the libgcc_eh.a link if the system doesn't have that file.Jannis Pohlmann2012-03-151-1/+4
| |
| * Add ARM include files to the headers to be modified.Jannis Pohlmann2012-03-151-1/+2
| | | | | | | | | | | | | | | | This was mentioned on http://www.linuxfromscratch.org/hints/downloads/files/lfs70armhint.txt as one of the things that need to be changed to build LFS 7.0 on ARM.
| * Add switch to set gnueabi LFS_TARGET on ARM.Jannis Pohlmann2012-03-151-1/+6
| |
| * Measure the time it takes to unpack things.Jannis Pohlmann2012-03-141-1/+1
| |
| * Run all builds/downloads in "time" to measure how long they take.Jannis Pohlmann2012-03-141-47/+47
| |
| * Add pass1_ccache() after zlib.Jannis Pohlmann2012-03-131-1/+24
| |
* | bootstrap: fix typoRichard Maw2012-03-151-1/+1
| | | | | | | | that meant that options weren't being passed
* | bootstrap: set tempdir and install flexRichard Maw2012-03-151-8/+10
| | | | | | | | | | | | | | flex is needed in binutils tempdir needs to be set as binutils' install script expects the files to be in an absolute path also use $LFS in paths
* | baserock-bootstrap: disable distcc for buildRichard Maw2012-03-151-0/+1
| | | | | | | | | | | | binutils ./configure fails because distcc is not found in the debian squeeze chroot. I am not sure how it is trying to invoke distcc, but we probably don't want it in pass1 anyway
* | bootstrap: more passes and keep cacheRichard Maw2012-03-151-6/+9
| | | | | | | | | | | | baserock-bootstrap: set cachedir so that the git cache can be reused in pass 2 onwards run-bootstrap-in-chroot: add more passes
View Lorry Controller Status