Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Commits on Apr 25, 2015
  1. Sync zoneinfo database with tzdata2015d from…

    Sascha Wildner authored
    Changes affecting future time stamps
    * Egypt's spring-forward transition is at 24:00 on April's last Thursday,
        not 00:00 on April's last Friday.  2015's transition will therefore
        be on Thursday, April 30 at 24:00, not Friday, April 24 at 00:00.
        Similar fixes apply to 2026, 2037, 2043, etc.
        (Thanks to Steffen Thorsen.)
    * Egypt will not observe DST in 2015 and will consider canceling it
        permanently.  For now, assume no DST indefinitely.
        (Thanks to Ahmed Nazmy and Tim Parenti.)
    Changes affecting past time stamps
    * The following changes affect some pre-1991 Chile-related time stamps
        in America/Santiago, Antarctica/Palmer, and Pacific/Easter.
          The 1910 transition was January 10, not January 1.
          The 1918 transition was September 10, not September 1.
          The UTC-4 time observed from 1932 to 1942 is now considered to be
          standard time, not year-round DST.
          Santiago observed DST (UTC-3) from 1946-07-15 through 1946-08-31,
          then reverted to standard time, then switched its time zone to
          UTC-5 on 1947-04-01.
          Assume transitions before 1968 were at 00:00, since we have no data
          saying otherwise.
          The spring 1988 transition was 1988-10-09, not 1988-10-02.
          The fall 1990 transition was 1990-03-11, not 1990-03-18.
          Assume no UTC offset change for Pacific/Easter on 1890-01-01,
          and omit all transitions on Pacific/Easter from 1942 through 1946
          since we have no data suggesting that they existed.
        One more zone has been turned into a link, as it differed
        from an existing zone only for older time stamps.  As usual,
        this change affects UTC offsets in pre-1970 time stamps only.
        The affected zone is America/Montreal.
    * America/Whitehorse switched from UTC-9 to UTC-8 on 1967-05-28, not
        1966-07-01.  Also, Yukon's time zone history is documented better.
        (Thanks to Brian Inglis and Dennis Ferguson.)
    Change affecting past and future time zone abbreviations
    * The abbreviations for Hawaii-Aleutian standard and daylight times
        have been changed from HAST/HADT to HST/HDT, as per US Government
        Printing Office style.  This affects only America/Adak since 1983,
        as America/Honolulu was already using the new style.
Commits on Apr 24, 2015
  1. dntpd - Fix va_list handling bug crash

    Matthew Dillon authored
    * Later versions of gcc manipulate va_lists in ways that prevent them from
      being reused, use va_copy() to make a copy in the log code which uses
      the va_list twice.
    * Fixes a seg-fault on startup.
  2. hammer - Allow NFS exports of slaves

    Matthew Dillon authored
    * Allows a SLAVE PFS to be exported.
    * WARNING: Clients who mount a SLAVE PFS from a server will only mount a
      snapshot.  The client must unmount and remount the slave to resynchronize
      it.  Also, if the server wipes the history pertaining the snapshot the
      client mounted, the client's mount will become corrupt.
  3. sound - Get sound working on alternative Acer C720p

    Matthew Dillon authored
    * This is a bit of a hack but one of our users had a C720p which had
      a HP vendor id for its chips.  It is possible that this was due to
      a BIOS mixup by Google.
    * Add hacks for it so sound works on his device.
Commits on Apr 23, 2015
  1. wpa_supplicant - CVE-2015-1863 (dead code in DFly so not an issue)

    Matthew Dillon authored
    * Correct buffer overflow from CVE-2015-1863.
    * Note that this code is NOT compiled into DragonFly's wpa_supplicant anyway,
      so the security advisory is not applicable to DragonFly.  Fix the code
      anyway so it isn't accidently enabled with the security hole before the
      next vendor update.
  2. sys/vfs/hammer: Make comments up-to-date

    Tomohiro Kusumi authored
    - There was no 'if (dip)' when the .. comment was added in 5a930e6,
      but ea434b6 needed 'if (dip)' and was added.
  3. Revert "sbin/hammer: Print key_end.lo instead of key_beg.lo on reblock"

    Tomohiro Kusumi authored
    This reverts commit d58869b.
    - According to the hammer rebalance (cmd_rebalance.c) which has
      a similar printf, this printf was probably intended to show
      localization type('s lower limit) when btree iteration starts,
      rather than localization types to look for during the iteration.
      Then it was right to print 0x0 (key_beg.localization) for %04x.
  4. emx: Add errata workaround for multiple TX queues

    Sepherosa Ziehau authored
    This makes multiple TX queues work on 82574.
    Information-from: Sean Bruno <>
    Tested-with: 82571, 82574, I217
Commits on Apr 22, 2015
  1. Set gcc50 as the primary compiler, gcc47 as secondary

    John Marino authored
    Per request, rather than define NO_GCC47 to prevent the building of the
    secondary compiler, the new variable is called NO_ALTCOMPILER.  This
    will avoid some disruption in the future where the actual version is not
    important, only the desire to avoid the secondary compiler.
  2. gcc50: Update to release version 5.1.0 => 5.1.1

    John Marino authored
    Immediately after the release was tagged, another commit was added that
    changed the version to 5.1.1.  Other than the version differing by 0.0.1,
    this version is identical to what the GCC just released.
  3. Merge branch 'vendor/GCC50'

    John Marino authored
  4. Update gcc-50 to SVN version 222321 (gcc-5-branch)

    John Marino authored
    This is equivalent to the GCC 5.1 release plus 1 commit that changes
    the version from 5.1.0 to 5.1.1 to reopen the release branch.
    Last Changed Date: 2015-04-22 12:45:39 +0200 (Wed, 22 Apr 2015)
Commits on Apr 21, 2015
  1. sbin/hammer: Print key_end.lo instead of key_beg.lo on reblock

    Tomohiro Kusumi authored
    - Print reblock.key_end.localization when reblock starts instead of
    - hammer reblock printing 0000 while reblock-inodes printing 0001
      and reblock-data|dirs printing 0002 seems strange because reblock
      covers all localization types ranging from MIN(0000) to MAX(ffff)
      on btree iteration (which obviously includes existing two types
      for inode and misc). Printing end localization should make better
      sense here.
  2. sbin/hammer: Print big block size on reblock

    Tomohiro Kusumi authored
    - Print free_level/big_block_size when reblock starts instead of
      just free_level.
    - The meaning of 'fill_percentage' arg is easier to understand
      this way because the big block size is not obvious for most users
      while this arg is about selecting which big block to defrag.
  3. sys/vfs/hammer: Cleanup cursor initialization code on reblock

    Tomohiro Kusumi authored
    - Just make things a bit more clear.
    - The rule is the ioctl caller sets localization type to reblock,
      and the ioctl code adds up ip localization to initialize cursor.
  4. sys/vfs/hammer: Make hammer_blockmap_getfree() clear *curp on error case

    Tomohiro Kusumi authored
    - Set *curp to either 0 or 1 regardless of *errorp.
      The previous return(0); case does this then all should do this.
  5. sys/vfs/hammer: Fix comment

    Tomohiro Kusumi authored
    - Sync a comment with what's written in reblock_usage().
  6. sys/vfs/hammer: Cleanup sanity checks

    Tomohiro Kusumi authored
    - Move sanity checks to the beginning of the function.
    - Check 'free_level > HAMMER_BIGBLOCK_SIZE'.
      free_level is somewhere between 0 and 8MB (inclusive).
Commits on Apr 20, 2015
  1. sbin/hammer: Remove irrelevant if(S_ISLNK()) case

    Tomohiro Kusumi authored
    - Remove if(S_ISLNK()) case for hammer snap since it never matches.
    - Also note it shouldn't be matching S_ISLNK() in the first place.
      If it does match hammer snap|snaplo|snapq commands will remove
      a symlink to the existing snapshot and relink to the new snapshot
      which results losing a reference to the old snapshot (although
      the old snapshot is still accessible as HAMMER's snapshot is just
      a pointer to the filesystem as of certain tid that is resolved
      when accessed).
Commits on Apr 19, 2015
  1. sbin/hammer: Add missing snapshots related free(3)

    Tomohiro Kusumi authored
    - Call free(3) after asprintf(3)
  2. sys/vfs/hammer: Fix off-by-one for snapshot index

    Tomohiro Kusumi authored
    - The index starts from 0 so these ioctls should test 'index >= count'.
  3. sbin/hammer: Remove redundant exit(3)

    Tomohiro Kusumi authored
    - snapshot_usage(1); does exit(1); as the comment implies.
Commits on Apr 18, 2015
  1. sshlockout - minor cleanup

    Matthew Dillon authored
    * If ac >= 2 then av[1] will not be NULL.
  2. sound - Fix default & unit specs for /dev/dsp and /dev/mixer (2)

    Matthew Dillon authored
    * Cleanup a debugging kprintf().
    * Cleanup incorrect dev reference and remove some unnecessary locking.
  3. sbin/hammer: Make hammer mirror-dump print localization

    Tomohiro Kusumi authored
    - Add printf for localiazation in addition to other
      struct hammer_base_elm fields.
  4. sys/vfs/hammer: Try to fix hammer_ioc_pfs_iterate() [2/2]

    Tomohiro Kusumi authored
    - The ioctl HAMMERIOC_PFS_ITERATE added in commit 29d31c2 has
      several design issues. This commit and the previous try to fix
      them. Since it's been exposed to userspace, some of them are
      left as it is with comments.
    - Move hammer_ioc_pfs_iterate() to sys/vfs/hammer/hammer_pfs.c
      and rename it to hammer_ioc_iterate_pseudofs().
    - struct hammer_ioc_pfs_iterate was not necessary. The existing
      struct hammer_ioc_pseudofs_rw which is being used by all other
      pfs ioctls has everything HAMMERIOC_PFS_ITERATE needs given
      that hammer_ioc_pfs_iterate::pos is just used as a pfs id.
      Using the unique struct requires hammer_ioc_iterate_pseudofs()
      extra conversion in order to make this ioctl behave like other
      pfs ioctls using autodetect.
  5. sys/vfs/hammer: Try to fix hammer_ioc_pfs_iterate() [1/2]

    Tomohiro Kusumi authored
    - The ioctl HAMMERIOC_PFS_ITERATE added in commit 29d31c2 has
      several design issues. This commit and the next one try to fix
      them. Since it's been exposed to userspace, some of them are
      left as it is with comments.
    - The name hammer_ioc_pfs_iterate::pos (pi->pos) is misleading.
      It should have been 'pfs_id'. Users will have no idea what
      the 'pos' is supposed to be.
    - pi->pos is unsigned.
    - Don't shift pi->pos (Don't ip localize pi->pos). Copy it to a
      local variable and then shift that.
    - cursor.key_end fields are necessary only when iterating btree.
      This function has 'iterate' in its name but all it does is
      a single btree lookup.
    - Don't |= HAMMER_PFSD_DELETED with pi->head.flags. pi->head.flags
      is for HAMMER_IOC_XXX macros, but not for ondisk pfs status
      (it was lucky PFSD_DELETED does not conflict with those values).
      Users get pfs status from>pfsd.mirror_flags copied
      to userspace. libhammer code that is based on this wrong flag
      checking is fixed accordingly.
    - Check if pi->ondisk is allocated.
    - Copy ondisk pfs data regardless of ondisk pfs delete flag bit.
      The delete flag doesn't mean the data is gone.
  6. sbin/hammer: Add missing free(3) and relpfs()

    Tomohiro Kusumi authored
    - Add missing free(3) for mirror buffer and relpfs() to release pfs.
  7. sbin/hammer: Add missing free(3)

    Tomohiro Kusumi authored
    - Add missing free(3) for a return value of read_mrecord().
  8. sbin/hammer: Remove unnecessary strndup(3)

    Tomohiro Kusumi authored
    - user string is relative to host string, not filesystem string.
  9. sys/vfs/hammer: Add comments on ip localization of @@PFS symlink

    Tomohiro Kusumi authored
    - The userspace hammer command shows "You are attempting to access
      a PFS softlink from a PFS..." and exit(1) whenever one tries to
      make a PFS from non root fs directory.
    - This restriction by userspace ensures @@PFS symlinks always belong
      to the root fs, in other words ip localization of those symlinks
      are always 0. HAMMER's kernel code doesn't impose this restriction
      on pfs, but userspace does.
Commits on Apr 17, 2015
  1. gcc50: Upgrade to GCC 5.1 release candidate plus

    John Marino authored
    while here:
      * remove 4+ Mb of unused texi files from contrib
      * update the 3 man pages from Release candidate
      * Change makefiles to not take man pages from contrib directory
    After thinking about it, I've decided to continue to update GCC from
    Subversion sources rather than limit updates to designated releases.  This
    is the reason the man pages are not taken from contrib; they are only
    present in actual releases and not the SVN repository.  The plan is to
    only update the man pages when GCC minor version is released.
    This is the first (and likely only) release candidate for GCC5.  It will
    be known as gcc 5.1.0, but CCVER will remain at "gcc50" regardless if the
    version of gcc is 5.1.1 or 5.4.2.  This includes changes up to SVN 222168
    (April 17).
  2. Merge branch 'vendor/GCC50'

    John Marino authored
  3. Update gcc-50 to SVN version 222168 (gcc-5-branch)

    John Marino authored
    Last Changed Date: 2015-04-17 09:45:46 +0200 (Fri, 17 Apr 2015)
Commits on Apr 16, 2015
  1. sysperf - Adjust syscall tests

    Matthew Dillon authored
    * Remove getuid_msg.c, no longer used.
    * Adjust syscall1.c to be more verbose.
    * Refactor syscall2.c to test on all threads in a somewhat more
      sophisticated manner.  The test expects the scheduler to move
      the threads to their own cpus (and it should).
Something went wrong with that request. Please try again.