Change logs for kernel-package source package in Feisty

  • kernel-package (10.065ubuntu6) feisty-proposed; urgency=low
    
      * Launchpad: #125816, Fix string search expressions to find exact tokens.
      Many string search expressions in this package could incorrectly match
      on the wrong string. Consider these 2 lines in /etc/kernel-img.conf:
    
        postinst_hook = /sbin/update-grub
        header_postinst_hook = /etc/kernel/postinst.d/dkms
    
      The original string search expression
        $postinst_hook = "$1" if /postinst_hook\s*=\s*(\S+)/ig;
      would match incorrectly on the last instance of 'postinst_hook' (header_postinst_hook),
      thereby running the wrong post-installation script.
    
      The new string search expression requires an exact match.
    
     -- Tim Gardner <email address hidden>   Tue, 17 Jul 2007 09:00:27 -0600
  • kernel-package (10.065ubuntu5) feisty; urgency=low
    
      * Touch '/var/run/do-not-hibernate' instead of '/var/run/do-not-suspend'
        to indicate to the power-management layer that the kernel that is likely
        to be used following a hibernate/resume-from-hibernate cycle has changed.
        Re-closes: (Launchpad: #14908)
    
     -- Paul Sladen <email address hidden>   Tue, 06 Mar 2007 14:12:24 +0000
  • kernel-package (10.065ubuntu4) feisty; urgency=low
    
      * Disable CRASH_DUMP kernel, now that we support relocatable ones.
    
     -- Ben Collins <email address hidden>   Fri,  2 Mar 2007 14:12:57 -0500
  • kernel-package (10.065ubuntu3) feisty; urgency=low
    
      * Fix shell-in-perl-script stupido.
    
     -- Ben Collins <email address hidden>   Thu,  4 Jan 2007 17:00:38 -0500
  • kernel-package (10.065ubuntu2) feisty; urgency=low
    
      * Touch file to show that kernel was upgraded, so suspend/resume scripts
        can recognize to not allow this: /var/run/do-not-suspend
    
     -- Ben Collins <email address hidden>   Tue,  2 Jan 2007 23:00:11 -0500
  • kernel-package (10.065ubuntu1) feisty; urgency=low
    
      * Merge from debian unstable.
    
      * Patches remaining:
        - kdump kernel type addition. This will probably get backed out once we
          can move to using stock kernel for kdump, with relocation.
        - update-initramfs usage
        - update-notifier usage
        - Disable some debconf stuff.
    
    kernel-package (10.065) unstable; urgency=low
    
      * Bug fix: "kernel-package: dependency on newer coreutils missing",
        thanks to Norbert Tretkowski. Added. As far as I can tell, readlink -m
        was present in coreutils 5.2.1, uploaded on 14 July 2004, over two
        years ago. However, updated the control file for kernel images.
                                                               (Closes: #396077).
      * Bug fix: "kernel-package: fails when a kernel-patch adds a config var
        without a default value".  Well. It is partially due to kbuild
        helpfully reconfiguring the kernel whenever it thinks it is needed.
        thanks to Yann Dirson                                   (Closes: #396122).
    
    kernel-package (10.064) unstable; urgency=low
    
      * Also remove initrd .bak backups
      * Bug fix: "leaves .bak files around after kernel purge, and chokes on
        them", thanks to martin f krafft. Well, k-p never generated these
        files, and does not actually choke on them. But I suppose we can clean
        up after update-mkinitramfs or whatever :)(Closes: #395323).
      * Take another crack at handling the ever friendly Kbuild now asking to
        update the .config on STDOUT, not STDERR, and bailing when it
        discovers output is redirected.
    
    kernel-package (10.063) unstable; urgency=low
    
      * Bug fix: "kernel-package: Generated postinst fails: Missing Required
        paramater 'Old'", thanks to Daniel Jacobowitz. I cleaned up the
        postinst as while I was at it, going with readlink -q -m instead of
        the built in version.                                (Closes: #391643).
      * Bug fix: "[powerpc] initramfs creation fails with 2.6.18 during
        install", thanks to Meelis Roos. This was fixed by the changes above.
                                                             (Closes: #394679).
      * Bug fix: "kernel-package: No need for any mentioning of bin86", thanks
        to Adrian Bunk                                       (Closes: #391677).
      * Bug fix: "postinst_hook runs after bootloader", thanks to Martin Buck
                                                             (Closes: #391436).
      * Bug fix: "kernel-package: A Debian revision is mandatory, but none was
        provided", thanks to Marc Haber. Hmm. We were not passing
        DEBIAN_REVISION for the clean target, and not doing any sanity checks.
        We still don't do sanity checks on clean, but we do pass along the
        provided debian revision, so warning is no longer triggered.
                                                             (Closes: #391433).
    
    kernel-package (10.062) unstable; urgency=low
    
      * Forgot to document KPKG_DEBUG introduced in the last upload.
    
    kernel-package (10.061) unstable; urgency=medium
    
      * Bug fix: "kernel-package: kernel package doesn't find $(ARCH) running
        make-kpkg debian", thanks to Elimar Riesebieter. Fix it for real this
        time. Fixes a grave bug.                              (Closes: #390487).
      * Bug fix: "Easing debugging by outputting commands as these are run",
        thanks to Loïc Minier.                                (Closes: #390222).
    
    kernel-package (10.060) unstable; urgency=low
    
      *  Only pass ARCH if KERNEL_ARCH is not empty.
    
    kernel-package (10.059) unstable; urgency=low
    
      * Bug fix: "kernel-package: Custom .config ignored, m-k clean generates
        errors on 2.6.18", thanks to Anthony L. Awtrey. Add an dummy call to
        make, if there are problems in the .config, kbuild rewrites .config,
        and the informational message messes up the VERSION variable. If the
        variable TEST did get a mesage about .config being written, pass it
        on. The problem is that any call to make seems to make Kbuild fix
        issues in .config -- nothing we can do there.            (Closes: #390012).
      * Bug fix: "Misleading information in postinst", thanks to Julien Danjou
                                                                 (Closes: #389697).
      * Bug fix: "kernel-package: link_in_boot produces symlinks that are
        insufficiently relative", thanks to Ross Boylan. Fix a typo in
        determining the relative path for images. Also, in several places,
        handle a special case when the source and destination directories are
        identical, and use a relative link there.  This might help the folks
        using a separate /boot and link_in_boot options.         (Closes: #389678).
      * Bug fix: "Should always pass ARCH= to upstream Makefile", thanks to
        Loïc Minier                                              (Closes: #390182).
    
    kernel-package (10.058) unstable; urgency=medium
    
      * Bug fix: "kernel-package: Fails to show warning when upgrading running
        kernel", thanks to Frans Pop                            (Closes: #389230).
      * Bug fix: "depmod: command not found; tar: -: file name read contains
        nul character", thanks to shaulka@012.net.il            (Closes: #385583).
    
    kernel-package (10.057) unstable; urgency=low
    
      *
      * Bug fix: "kernel-package: make-kpkg clean takes forever on a git
        repo", thanks to Will Dyson                           (Closes: #356765).
      * Bug fix: "Patch to add --verbose =&gt; make V=1", thanks to Philippe
        Teuwen                                                 (Closes: #387346).
    
    kernel-package (10.056) unstable; urgency=low
    
      * Bug fix: "postrm_hook is not run", thanks to Michael Biebl. Gak. A
        thinko made the postrm still test to see if the script was executable
        before looking for it in the path.                     (Closes: #387286).
    
    kernel-package (10.055) unstable; urgency=low
    
      * Bug fix: "kernel-package: failure with 2.6.18-rc4 (wireless-dev
        tree)", thanks to Marcus Better. Starting with 2.6.18, you can't run
        make at the top level like we do to determine the version number
        without having a .config file. But it should still be possible to run
        clean -- and fail for other initial operations, since we do not want
        to create ./debian if we can't detect the version number accurately.
                                                             (Closes: #385329).
      * Bug fix: "kernel-package: please build md5sums for kernel-packages",
        thanks to Mario 'BitKoenig' Holbe. Thanks to Roland Stigge for all
        the heavy lifting. I moved the macro definition to a file a location
        consistent with my other packages, and allowed for the caller to add
        locations that should be excluded from the MD5sums listing.
                                                             (Closes: #261019).
      * Add a variable called warn_reboot for kernel-img.conf; if set to no,
        it shall suppress warning about needing to reboot if installing a
        kernel image of the same versions as the kernel which is running.
        Some users want enough rope ...
      * Create md5sums for this package as well.
    
    kernel-package (10.054) unstable; urgency=low
    
      * Bug fix: "Please allow ~ in sanity check", thanks to Norbert
        Tretkowski. Second cut.                             (Closes: #383751).
    
    kernel-package (10.053) unstable; urgency=low
    
      * Bug fix: "Please allow ~ in sanity check", thanks to Norbert
        Tretkowski                                           (Closes: #383751).
      * Bug fix: "initramfs generator reversed order dep", thanks to
        maximilian attems                                    (Closes: #383605).
      * Make the question about whether or not to run the default bootloader a
        shared question. At this point, we still ask the question every time,
        though I am open to using the stored value -- but that would mean that
        if the admin has changed their mind since the question whether to run
        lilo or grub was asked, we would do the wrong thing, perhaps making
        the machine unbootable -- which is why the default is still to ask the
        human.
      * Bug fix: "Package changelog for generated kernel packages uses
        kernel-package changelog", thanks to Andrew Meaden. Truncated, and
        repaced with a stub.                                 (Closes: #382743).
      * Bug fix: "kernel-package - please remove config target check from
        make-kpkg", thanks to Bastian Blank. The check has been disabled for
        official kernel images, but not for the general upstream sources.
                                                             (Closes: #381828).
    
    kernel-package (10.052) unstable; urgency=low
    
      * Bug fix: "kernel-package: syntax error in pkg/image/postinst", thanks
        to Kiko Piris                                         (Closes: #381827).
    
    kernel-package (10.051) unstable; urgency=low
    
      * Bug fix: "kernel-package: When overwriting modules, bad suggestion on
        moving modules dir", thanks to Dylan Thurston       (Closes: #379434).
      * With this release of kernel package, changes have been made to the
        specification and behaviour of hook scripts for the kernel image,
        source, and header packages. Now relative paths can be used to specify
        the names of scripts, as long as these scripts live in a "safe" path,
        namely, /bin, /sbin, /usr/bin, /usr/sbin. Also, the sanity checking
        has gotten more rigorous, failure to find a hook script or failure for
        the script to execute correctly is now a fatal error.
      * Bug fix: "forces absolute paths in /etc/kernel-img.conf *_hook
        variables", thanks to Robert Millan. I elected not to use the patch
        provided, since I did not want to blindly trust the PATH variable, and
        wanted also to change the ignoring of the return value of the
        scripts.                                               (Closes: #380338).
      * Bug fix: "linux-image-2.6.16-2-k7-smp: upgrade re-ran lilo", thanks to
        Ross BoylanThe default behaviour of the postinst script has been to
        run the built in default bootloader (typically lilo) -- set at compile
        time, based on the arch. The installed default for Debian, however,
        has since moved on to be grub. However, users who are transitioning to
        grub may not have set the configuration variable in
        /etc/kernel-img.conf --and may still have lilo installed. The trick is
        to figure out whether they really want to run lilo. So, if there is no
        /etc/lilo.conf, we run nothing. If the user has explicitly asked to
        run lilo, we rin lilo. Or if there is no grub, or if there is no
        postinst hook script, we run lilo. Or else, we ask the user.
                                                                 (Closes: #380292).
    
    kernel-package (10.050) unstable; urgency=low
    
      * Bug fix: "kernel-package: make-kpkg "dirties" git kernel", thanks to
        Rob Sims.  Make sure that make-kpkg clean undoes the changes made in
        ./scripts. Also, added to the PROBLEMS file the following issue: When
        run in a git sourced tree, scripts/setlocalversion checks to see if
        the tree is the trunk, if not, it adds version information to the
        kernel version. It also checks to see if there are uncommited changes,
        if so, it adds the string "-dirty" to the version. Since we change a
        couple of files in ./scripts/, this falls afoul of the make-kpkg
        checks, since the version number is now different from what make-kpkg
        knows it to be. The workaround is to modify the setlocalversion
        script.                                                (Closes: #375559).
      * Bug fix: "kernel-package: Phony UTS_VERSION string mismatch", thanks
        to Daniel Werner. Actually, the version number contained a +, which is
        also a regexp meta-character, so the regexp match was getting confused.
                                                               (Closes: #378257).
      * Bug fix: "clone targets to support newer default linux prefix", thanks
        to maximilian attems                                   (Closes: #374707).
      * Bug fix: "yaboot: Incomplete message when removing kernel-package",
        thanks to Helge Kreutzmann                             (Closes: #364336).
      * Bug fix: "debian.README.gz: wrong (old) address for Linus", thanks to
        Lars Wirzenius                                         (Closes: #376466).
      * Bug fix: "kernel-package: strange behavior of bash_completion", thanks
        to Matthias Heinz Thanks to Rob Sims for the fix.      (Closes: #323668).
      * Bug fix: "kernel-package: kernel_image target fails with kernel
        2.6.18-rc1", thanks to Kevin Baradon. The fix thanks to Harald Dunkel
                                                               (Closes: #377267).
    
     -- Ben Collins <email address hidden>   Mon, 06 Nov 2006 04:15:16 +0000
  • kernel-package (10.049ubuntu5) edgy; urgency=low
    
      * Detect UTS_RELEASE_HEADER for 2.6.18 allows users to compile
        the 2.6.18 kernel.
    
      * Also this patch is pre-emptive for edgy+1 kernel builds.
    
     -- Chuck Short <email address hidden>   Sun,  1 Oct 2006 21:27:56 -0400