Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Commits on Feb 5, 2009
  1. Release Engineering: Update master to 2.3.0-DEVELOPMENT

    Matthew Dillon authored
Commits on Feb 3, 2009
  1. mount(8): Start sentence on new line.

    Sascha Wildner authored
  2. Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git…

    Matthew Dillon authored
    …/dragonfly
  3. Suppress warnings if /etc/fstab is missing when generating output for

    Matthew Dillon authored
    mount -p.  Generate reasonable defaults to fill in missing pieces.
    
    Submitted-by: Stathis Kamperis <ekamperi@gmail.com> added the comment:
  4. Add a vcount(9) manual page.

    Sascha Wildner authored
    Submitted-by:  Stathis Kamperis <ekamperi@gmail.com>
    Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1266>
  5. Silence 3 warnings in libc.

    Sascha Wildner authored
Commits on Feb 1, 2009
  1. Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git…

    Matthew Dillon authored
    …/dragonfly
  2. Comment out the session line for PAM's xdm config. This works around

    Matthew Dillon authored
    a widely known bug in xdm where a PAM callback can cause xdm to access
    a bogus widget pointer after the widget has been destroyed.
    
    With-help-from: Sephe
  3. Make XPT_SCSI_IO path quiet again.

    Hasso Tepper authored
    Now when problems with SATA ATAPI devices are resolved, it's not
    helpful any more.
  4. HAMMER VFS - Fix I/O invalidation collision assertion

    Matthew Dillon authored
    Fix more issues with the hammer_reserve subsystem.
    
    * Fix a broken base_offset calculation which had reversed the zone
      and offset fields in a HAMMER_ZONE_ENCODE() call.  The broken
      base_offset caused hammer_del_buffers() to scan the wrong offset
      range, leaving conflicting buffer cache buffers intact.
    
      The resulting conflict caused a KKASSERT(LIST_FIRST(&bp->b_dep) == NULL)
      to fail later on.
    
    * Hold a reference through a potentially blocking operation in
      hammer_blockmap_alloc().
    
    * Record the proper zone in a hammer_reserve() structure created via
      hammer_blockmap_free().  Without this the I/O invalidation code
      cannot locate conflicting hammer_buffer structures.
    
    * hammer_reserve_setdelay_offset() was not actually placing the
      hammer_reserve() structure in the delay queue, allowing it to
      be disposed of too early.
Commits on Jan 31, 2009
  1. make(1) tests: Remove remaining CVS IDs from the tests to make them r…

    Sascha Wildner authored
    …un again.
  2. Remove some unexpanded $DragonFly$ IDs in our tree.

    Sascha Wildner authored
  3. rpc_svc_calls(3): Remove __svc_getcallercreds() documentation.

    Sascha Wildner authored
    FreeBSD removed the macro but forgot to adjust the manpage.
  4. make.conf: Bring back WARNS_WERROR to build with -Werror if WARNS is …

    Sascha Wildner authored
    …set.
    
    In-discussion-with: pavalos
  5. Typo fixing round.

    Sascha Wildner authored
  6. installer: Fix reading disk descriptions when booting verbose.

    Sascha Wildner authored
    This fixes issue1256.
    
    Reported-by: corecode
  7. installer: Remove .gitignore files.

    Sascha Wildner authored
  8. manpages: Split DIAGNOSTICS into EXIT STATUS and DIAGNOSTICS.

    Sascha Wildner authored
    DIAGNOSTICS shall have information about diagnostic messages while
    exit code information (.Ex macro) goes to EXIT STATUS.
    
    Inspired-by: FreeBSD
  9. syncache_socket(): fix abort path by calling pru_abort directly

    Sepherosa Ziehau authored
    On syncache_socket() abort path, the inp's {fport,lport,faddr,laddr}
    is either incomplete or half destroyed, which causes tcp_soport()
    mapping the socket to a wrong CPU.  However, in syncache code, inp
    is only accessed or created by its owner CPU, i.e. we are already on
    the inp's owner CPU during the socket aborting.
    
    Add soabort_oncpu(), which calls pru_abort directly.  Use it on
    syncache_socket() abort path.
Commits on Jan 30, 2009
  1. Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git…

    Matthew Dillon authored
    …/dragonfly
  2. Fix a string parsing bug when trying to run hammer pfs-master on a

    Matthew Dillon authored
    subdirectory of the root directory, e.g. "hammer pfs-master /fubar".
    
    Reported-by: "Simon 'corecode' Schubert" <corecode@fs.ei.tum.de>
  3. Put the repo version in the ISO's volid.

    Sascha Wildner authored
  4. Fix capital letter in the middle of a sentence.

    Peter Avalos authored
    Submitted-by:  Stathis Kamperis <ekamperi@gmail.com>
Commits on Jan 29, 2009
  1. jail(8): Fix path in the example.

    Sascha Wildner authored
  2. @corecode
  3. @corecode

    nata: process interrupt even if busmaster active bit is set

    corecode authored
    The nata driver previously assumed that an interrupt which has the
    interrupt and active bits set in the busmaster status register is not a
    valid/completed interrupt.  In these situations if would return without
    acknowledging the irq, leading to an interrupt livelock situation.
    
    This condition of INTERRUPT|ACTIVE occured on two different machines
    both equipped with different versions of an ICH SATA controller, when
    accessing an attached SATA ATAPI (DVD) drive by TSScorp (Samsung
    OEM/bulk).  Specifically, natapicam seems to trigger this condition on
    attach.  It seems that the drive is unhappy with the INQUIRY command
    sent by natapicam.  Related artifacts appear when trying to access a
    video DVD (possibly due to missing region/css key).
    
    Under the hood it seems that the drive reports an error condition
    (visible in the ALT/STATUS register), but does not clear the ACTIVE bit
    of the busmaster status.  As a result the nata driver could be waiting
    for the completion of the transfer, while the SATA busmaster would keep
    the interrupt line asserted.
    
    We work around this issue by following the precedent of other operating
    systems and ignoring the ACTIVE bit if the INTERRUPT bit is set.
    Specifically, we wait a little bit to accomodate for possibly faulty
    hardware which might assert the irq before the transfer is completely
    finished.  Nevertheless, we do not reject interrupt proecessing if the
    INTERRUPT bit is set in the busmaster status register.
  4. nrelease: Switch to building pkgsrc packages from source.

    Sascha Wildner authored
    The pkgsrc tree that shall be used must be specified via PKGSRC_PATH.
    
    So far the LiveCD build has been successfully tested. The LiveDVD might
    still need some love, but I'm committing this now so work on the release
    can begin.
    
    Still on the TODO is using binary packages optionally.
    
    Requested-by: dillon
Commits on Jan 28, 2009
  1. Fix format order.

    Sascha Wildner authored
    Reported-by: corecode
Commits on Jan 27, 2009
  1. Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git…

    Matthew Dillon authored
    …/dragonfly
  2. HAMMER VFS - Add missing bqrelse() in last patch.

    Matthew Dillon authored
    This fixes a deadlock introduced by the last patch due to a missing
    bqrelse() call.
Something went wrong with that request. Please try again.