Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Make ./check check for long lines (with excptions); fix long lines | Lars Wirzenius | 2012-02-27 | 1 | -2/+0 |
| | |||||
* | scripts: don't skip if not snapshotting | Richard Maw | 2012-02-27 | 1 | -2/+2 |
| | | | | | | | | | | | | run-bootstrap-in-chroot would not build a stage if it had a snapshot for the next stage. If the same directory had previously run while snapshotting then it would skip earlier stages without building them as it used to be able to extract that tarball. This fails as stages depend on earlier stages. This change is so that it will build if it is not snapshotting or if the stage has not been snapshotted. | ||||
* | scripts: allow running without snapshots | Richard Maw | 2012-02-27 | 1 | -4/+4 |
| | | | | | | | | Set the #! to bash, as it uses bashisms Everywhere where it would try to run tar has a check for $snapshot Allows the snapshot variable to be set in the environment for the script to use that, it seemed easier than adding an option Use run_pass for stage 3 since it can have snapshotting disabled | ||||
* | Export TMPDIR to avoid confusion. | Jannis Pohlmann | 2012-02-21 | 1 | -1/+1 |
| | | | | | | If the TMPDIR variable is not yet set, then setting it without exporting will not make it available to child processes. Thus, let's just make sure we export it and everyone will be happy. | ||||
* | Clear the TMPDIR variable when entering the chroot. | Jannis Pohlmann | 2012-02-21 | 1 | -0/+4 |
| | |||||
* | Add tar 1.26 from code.liw.fi to the squeeze chroot for bootstrapping | Lars Wirzenius | 2012-02-20 | 1 | -0/+10 |
| | |||||
* | Update morph and related stuff after unpacking a snapshot | Lars Wirzenius | 2012-02-16 | 1 | -9/+11 |
| | |||||
* | Skip unpacking earlier passes when later snapshots exist | Lars Wirzenius | 2012-02-16 | 1 | -7/+30 |
| | |||||
* | Simplify bootstrap dir tree; split pass2 into two parts | Lars Wirzenius | 2012-02-16 | 1 | -26/+23 |
| | |||||
* | Actually make it compressed | Lars Wirzenius | 2012-02-15 | 1 | -1/+1 |
| | |||||
* | Make chroot snapshot also compressed | Lars Wirzenius | 2012-02-15 | 1 | -3/+4 |
| | |||||
* | Compress pass1 and pass2 snapshots | Lars Wirzenius | 2012-02-15 | 1 | -6/+8 |
| | |||||
* | When unmounting proc, sys, also unmount them from the inner chroot | Lars Wirzenius | 2012-02-14 | 1 | -0/+2 |
| | |||||
* | Make run-bootstrap-in-chroot snapshot pass2 | Lars Wirzenius | 2012-02-14 | 1 | -4/+20 |
| | | | | | | This changes the meaning of $1 for baserock-bootstrap: $1 now specifies which pass to run. When building passX, X > 1, the previous passes are assumed to have been built successfully. | ||||
* | scripts: use EXIT traps to unmount | Richard Maw | 2012-02-14 | 1 | -2/+2 |
| | | | | | | | | | | | Jenkins failed to build because proc and sys were still mounted after a forced termination of a build. The exit traps will be run when the shell exits, so there are fewer code paths that can avoid unmounting EXIT is not the only signal that can be used INT and TERM will be executed when Interrupted (by ^C) or Terminated (by kill) | ||||
* | Make run-bootstrap-in-chroot unless required env vars are set | Lars Wirzenius | 2012-02-14 | 1 | -4/+2 |
| | | | | | | | This removes the previous action which tried to be interactive, and let the user continue the build anyway, but since the build will fail in that case, it's better to not let the user continue. Also simplifies things in general. | ||||
* | Make the parameter to run-bootstrap-in-chroot not be optional | Lars Wirzenius | 2012-02-13 | 1 | -7/+7 |
| | | | | | It is confusing to have a default, especially since it doesn't actually work anymore. | ||||
* | Always (re-)create the do-squeeze-chroot script | Lars Wirzenius | 2012-02-13 | 1 | -4/+2 |
| | |||||
* | Copy current working directory (= morph repo) into chroot | Lars Wirzenius | 2012-02-09 | 1 | -2/+2 |
| | | | | | | The bootstrapping can, for now, only be run from the git repo anyway, so we might as well copy just that into the bootstrap chroot. Everything else will be accessed via git from the git server. | ||||
* | Copy only the morph repository into the bootstrap chroot | Lars Wirzenius | 2012-02-09 | 1 | -1/+2 |
| | |||||
* | Make bootstrap build use git bundles | Lars Wirzenius | 2012-02-09 | 1 | -0/+1 |
| | | | | | | Also, run morph update-gits in loop until it succeeds. Gitorious or git or the office network are unreliable enough that otherwise the build pretty much is guaranteed to fail during the initial cloning and fetching. | ||||
* | Make caller specify $GITS for the location (dir) where git repos are | Lars Wirzenius | 2012-02-01 | 1 | -1/+2 |
| | | | | | | The original code was using $HOME, which gets expanded to /root under sudo, and that's not useful. This avoid hardcoding paths into the script. | ||||
* | run-bootstrap-in-chroot: make a script and run that to chroot | Richard Maw | 2012-01-30 | 1 | -27/+19 |
| | | | | | | | | | | It is often useful to be able to chroot into the system it is easier to investigate build problems there So run-bootstrap-in-chroot will write a script to do this which can then be used to enter the squeeze chroot with ./do-squeeze-chroot bash or enter the baserock chroot with ./do-squeeze-chroot ./do-chroot | ||||
* | run-bootstrap-in-chroot: Take mirrors from environment variables | Richard Maw | 2012-01-27 | 1 | -5/+20 |
| | |||||
* | run-bootstrap-in-chroot: use backticks instead of $() | Richard Maw | 2012-01-26 | 1 | -1/+1 |
| | | | | vim syntax highlighting thinks that this is not /bin/sh portable | ||||
* | scripts: use true/false instead of yes/no | Richard Maw | 2012-01-26 | 1 | -8/+8 |
| | |||||
* | Actually run pass2 | Lars Wirzenius | 2012-01-20 | 1 | -3/+3 |
| | | | | Also, unmount virtual filesystems using the shell function for it. | ||||
* | Fix snapshotting logic and git repo updating | Lars Wirzenius | 2012-01-20 | 1 | -42/+25 |
| | |||||
* | Keep proc, sys mounted only during chroot runs | Lars Wirzenius | 2012-01-20 | 1 | -13/+26 |
| | |||||
* | Do not put contents of proc, sys in snapshot tarball | Lars Wirzenius | 2012-01-20 | 1 | -1/+5 |
| | | | | Do, however, put the actual dirs in the tarball. | ||||
* | Fix creation of pass1 snapshot tarball | Lars Wirzenius | 2012-01-20 | 1 | -1/+1 |
| | |||||
* | Fix sh script line continuation indentation bug | Lars Wirzenius | 2012-01-20 | 1 | -3/+3 |
| | |||||
* | Change snapshot handling to be more flexible | Lars Wirzenius | 2012-01-20 | 1 | -7/+58 |
| | | | | Also, do snapshot of the debootstrapped thing too. | ||||
* | Improve error recovery and cleanup the chroot dir at startup | Lars Wirzenius | 2012-01-19 | 1 | -0/+3 |
| | |||||
* | add missing tools to the squeeze chroot so a system image can be built | Lars Wirzenius | 2012-01-19 | 1 | -1/+1 |
| | |||||
* | make run-bootstrap-in-chroot build the whole bootstrap, not just pass1 | Lars Wirzenius | 2012-01-19 | 1 | -2/+9 |
| | |||||
* | Avoid bc as a dependency for baserock-bootstrap | Lars Wirzenius | 2012-01-17 | 1 | -1/+1 |
| | | | | | | | | | | We want to avoid having to have bc in the bootstrap environment, since it's yet another thing to install there, and to build in the bootstrap process. bc is only used to compute the number of parallel make jobs. We can approximate the answer easily with expr (but need integer arithmetic so the result is slightly different). | ||||
* | remove copying of LFS tarballs (it only ever worked for me) | Lars Wirzenius | 2012-01-13 | 1 | -1/+0 |
| | |||||
* | add script to run baserock-bootstrap in a newly created clean chroot | Lars Wirzenius | 2012-01-13 | 1 | -0/+37 |