Brad Peters (1):
      When executed, serv_config is causing a partition hang.  I was able to
         track the cause of the hang down to a single query against PHYP via
         RTAS.  Attempting to access 'proc-cuod-card-info', the Processor CUoD
         Capacity Card Info, seems to no longer be supported.  I've removed
         this inquiry from the array.

Brian King (2):
      The OpenFirmware binding for Fibre Channel devices permits any leading
         zeros in the LUN field to be stripped, but does not require it. Fix
         ofpathname to handle this case when translating OF pathname to logical
         device name.
      According to PAPR, we need to call activate-firmware after a migration,
         which we weren't doing. Add this call after updating the device tree.
         Add another device tree update call after it to ensure the SLB size is
         correctly updated in the kernel.

Nathan Fontenot (29):
      Initial commit for the new combined powerpc-utils package from the
         previous powerpc-utils and p;owerpc-utils-papr packages.
      Ignore hotplug power-off failures when the devie tree property
         "/proc/device-tree/ibm,ignore-hp-po-fails-for-dlpar" exists.
      Return RC_ALREADY_OWN when trying to add a PHB that is already owned by
         the partition.
      There are instances where goto_dir() will incorrectly return success.  An
         updated find_dir allows us to check the result before trying to cd to
         the directory we want.
      Update the find command in get_vfc_wwpn() to handle being passed a
         symbolic link.  Previously this lead to an infinite loop due to the
         possible circular nature of sysfs.
      Change the name format for vio devices and correct indentation.
      When disabling SMT for a processor, make sure that we leave the first
         thread of the processor enabled and disable all the others.
      Remove unneeded spaces to correct previous patch for find_dir()
      Update Meta data
      The sysfs_dev_path member of the dr_node struct is a char[] and needs to
         be treated as such.  this fixes a bug where the sysfs_dev_path is
         treated as a char * and returning invalid info when the buffer is
         actually empty.
      Cpu add failures need to return 1 instead of -1.  A return code of -1
         causes the IVM to think there has been a really bad error and the IVM
         will wait 15 minutes prior to re-trying the operation.
      Logical device name to open firmware path translation for fibre-channel
         devices can fail on systems where the fc_remote_ports file in sysfs
         has a slightly different name.  On some systems this file appears as
         fc_remote_ports:rport-X:Y-Z.  This patch corrects the lookup of this
         file.
      The amsvis command will fail if vio devices have addresses greater than
         30000009 due to a problem with hexidecimal values attempting to be
         sorted as long integers.  This patch changes the sort to use string
         comparisions.
      I'm removing the __date__, __version__, and __source fields from each of
         the files.  They were filled in by CVS keyword expansion and won't be
         updated by GIT.
      s patch adds an additional output message when drmgr fails to remove all
         of the requested lmbs in a drmgr memory remove request.  This
         additional message is meant just to help clarify that the memory
         remove operations didn't fail, but was unable to completely fulfill
         the request.  This update came as a request from testers.
      When removing nodes from the device tree in /proc the recusrsive routine
         can fail with a bad pointer to the DIR struct.  The DIR struct should
         be rewound after each remove to avoid this.
      The /usr/sbin/rsct/bin/refrsrc command forces a refresh of the HMC/IVM
         and RMC status after a migration occurs.  We should do this from the
         drmgr command instead of assuming servicelog is installed.  A failure
         to invoke this command can cause additional migrations to be delayed
         for up to 10 minutes while the framework refreshes itself.
      This is really more of a re-write than an update.  The ppc64_cpu command
         is being ported from Perl to C to support new functionality.  This
         update introduces two new pieces of functionality to the command.
      This patch contains all of the additional updates needed to support the
         new version of ppc64_cpu, removal of old script and Makefile updates.
      Changelog for v1.2.0
      The ppc64_cpu command needs to properly turn off the smt snooze delay.
         The current code does not handle the option 'off', it should as this
         is regarded as a delay of -1.
      When setting or retrieving system wide attributes such as the
         smt-snooze-delay or dscr we should only be concerned with online
         cpus/threads.  Offline cpus/threads will not have these attributes in
         their sysfs directory.
      This is some minor code cleanups, no functional changes.
      When making changes to increase the smt state, the cpus that are brought
         online come online with the same smt-snooze-delay they had when they
         were taken offline.  This leads to a mix of smt-snooze-delay states
         for a system.
      This patch updates the retrieval of a system attribute, such as
         smt-snooze-delay or dscr, to handle a valid return value of -1.
      The current bounds checking only validates that the specified smt setting
         is not zero and less than or equal to the threads per cpu on the
         system.  This causes smt settings such as 4000000000 to be converted
         to some negative number and thus odd smt settings ensue.  This patch
         ensures checks to make the smt setting specified is not less than or
         equal to zero.
      Updates and prep for powerpc-utils-1.2.1 release
      Remove the amsvis command from powerpc-utils.
      Re-add the powerpc-utils.spec file.

Robert Jennings (4):
      Add HEA ports as a new connector type to lsslot.  Ports will be sorted on
         location code, not name, so the output might look as follows (debug
         output included).
      Currently, print_node_list will never print the children of a node. Just
         correcting a small bit of the logic.
      The logic in get_node_by_name() currently does not return when it finds a
         child with the name being search for.  This would mean that a command
         such as `drmgr -Q -c port -s "Port 1" -w 0 -d 3` would always return: 
          drmgr: Port 1 not owned by partition
      DR commands shouldn't take long to complete, but having both an entry and
         exit timestamp eliminates the guessing and can point people in the
         correct direction while they debug issues.

Santiago Leon (1):
      Introduce the lsdevinfo utility.  This command is used to provide the HMC
         or IVM with name information for virtual devices so they can be
         matched against the VIOS names.

