Amanda::Constants::AMANDA_COMPONENTS having 'server' as one of it's
values. Instead of conditionally trying to eval in Amanda::XferServer,
do it unconditionally. The effect is (nearly) the same.
Since 2.0:
* Fix scrub abort on files/disks > 2G in size (issue 19)
This was introduced in 2.5.0 and the failure is quick and obvious.
* Fix loss of precision in disk size approximation (issue 18)
* Add -t option to disable creation of thread to generate randomness
in parallel with I/O.
* Prepare source for a libscrub api (work in progress)
* Cosmetic changes to man page and help output.
* Generate randomness and do I/O in parallel [Richard B. Hill]
* Replace O_SYNC with O_DIRECT+fsync+posix_fadvise [H. Peter Anvin]
This produces a significant performance improvement, at least on Linux.
* Add support for RDRAND instr. [H. Peter Anvin]
* Use <stdbool.h> if the system has it [H. Peter Anvin]
* Fix incorrect assertion in scrub --dirent handling (issue 9) [Peter Hyman]
* Clean up tmp files after scrub --freespace (issue 9) [Peter Hyman]
* Allow scrub --pattern=custom="string" (issue 11) [Peter Hyman]
Allow C style numerical escapes in string: \nnn (octal) and \xnn (hex).
This is a single-pass sequence with a custom pattern.
* Change default buffer size from 1m to 4m [H. Peter Anvin]
* Avoid closing and re-opening /dev/urandom [H. Peter Anvin]
* Fix bug in pfitzner7 pattern def and add fillff, fillzero
methods (issue 8) [Peter Hyman]
* Change dod method to perform random pass first and skip extra pass.
* Add --help option that lists available scrub methods.
* Make dist tarball work with rpmbuild -ta.
* Added usarmy, pfitzner7, and pfitzner33 methods.
* Added -L option to not follow symbolic links [Timothy Boronczyk]
* Treat a zero return from write(2) as an error.
* New patterns:
- random2: two random passes
- schneier (one 0x00 pass, one 0xff pass, and five random passes) [Joe]
* Add --pattern random (one random pass) mode.
* Portability fixes for Slackware [Timothy Boronczyk]
* Fix bug where scrub --remove on a zero length file would abort
without removing the file. Issue a warning and remove it.
* Fix a bug where some data from the original file could appear in
the first disk block after a verified scrub when running without
the -S (--no-signature) option.
changes since 0.9.6:
==New major user-visible features==
* Warn if all blocks were skipped while writing.
* Fix evil twins of Macronix MX25L1605, MX25L3205, MX25L6405.
* Exit if there are unused programmer parameters.
=== Related to programmers ===
* internal:
** AMD Geode on OpenBSD
** Add safe support for the integrated micro controller (IMC) in AMD chipsets.
** Add full support for VX800/VX820, VX855/VX875 and VX900 (both SPI and LPC).
** Various new untested board enables (please don't run away without testing our work).
* serprog
** Add support for setting the SPI frequency.
** Add opcode to control the programmer's output drivers.
** Enable serprog on Windows.
* buspirate_spi
** Improvements to support different firmware versions safely.
** Support the new fast SPI mode present in Bus Pirate firmware v5.5 or newer.
** Add user interface to activate the Bus Pirate pull-up resistors.
* ftdi_spi
** Allow to select FTDI device by serial number.
** Add support for all 4 possible channels.
** Add support for FT232H.
* pony_spi: Add support for AJAWe.
* Add support for Realtek RTL8169.
* Add Altera USB-Blaster SPI programmer.
* dediprog:
** Add a "device" parameter to support multiple devices.
** Adds a programmer parameter 'spispeed'.
===New or refined chips===
* AMIC A25L(Q) series
* Atmel AT26DF041
* Atmel AT49(H)F010, AT49F080 and AT49F080T
* Atmel AT25F series (AT25F512, AT25F512A, AT25F512B, AT25F1024, AT25F1024A, AT25F2048, AT25F4096)
* Eon EN25F64
* Eon EN25S series
* More Eon EN25QH chips
* More GigaDevice GD25* devices
* Intel S33 series (QB25FxxxS33x8)
* Macronix MX25U1635E, MX25U3235E/F and MX25U6435E/F
* Micron (Numonyx) N25Q series
* Micron (ST) M45PE series
* Micron (ST) M25PX80
* Nantronics N25 series
* PMC Pm25LV and Pm25LD series
* Sanyo LE25FW series
* Spansion S25FL2 and S25FL...S series
* SST SST25WF512, SST25WF010, SST25WF020 and SST25WF040
* Winbond W39F010/W39L010/W39L020
changes since 0.9.5:
==User interface changes==
* The -p/--programmer parameter is now mandatory for all operations that access the programmer/chip.
*: Previously there was a default programmer selected at build time. This was convenient for most users, but created a danger for others that increased with the number of platforms and programmers supported, namely that they get used to a default programmer that later changes when they use another binary. It is very easy to brick a mainboard in case you do not notice you are writing its firmware instead of a device attached to an external programmer. The disadvantage is of course that many users have now to add '''-p internal''' to the command line when they want to do ''the obvious'' and flash their mainboards.
* It is now possible to let flashrom create log files directly (by using the new -o/--output option).
* flashrom now checks for duplicate -i/--image arguments (which are used together with layout files).
==New major user-visible features==
* Programmer support:
** new driver ''ponyprog'' for hardware similar to SI Prog
** native SPI AAI write support to the Dediprog SF100 driver
* Support for new chips:
** Atmel AT49F040
** Eon EN25QH32
** Eon EN29LV640B
** GigaDevice GD25QXX
** Numonyx N25Q064
** PMC Pm39LV512
==Infrastructural improvements and fixes==
* The dummy programmer got support for SFDP and now has a dedicated status register.
* Various minor refactorings of SPI-related code.
* The code generating the list of supported hardware for the wiki has been improved.
* Many exit() calls have been replaced by proper error handling.
* I/O permissions are now released automatically on shutdown.
* Preliminary work on Winbond W836xx and ITE IT8707F/IT8710F detection.
= Minor versions =
flashrom 0.9.6 had a small bug regarding the version string, which has been fixed in 0.9.6.1.
No upstream change summary, but the old version no longer worked with
modern coreboot.
Distfile created from a full clone of the coreboot tree with below script;
#!/bin/sh
cbver=`git describe`
cbobj=`git ls-tree origin/master util/nvramtool|awk '{print $3}'|cut -c 1-7`
distname=nvramtool-"$cbver"-"$cbobj"
git archive -o /tmp/"$distname".tar.gz --prefix="$distname"/ "$cbobj"
upstream changes:
Enhancements in v5.01:
New Features
Added support for up to 2 TB of RAM on X64 CPUs
Added experimental SMT support up to 32 cores (Press F2 to enable at startup)
Added complete detection for memory controllers
Added Motherboard Manufacturer & Model reporting
Added CPU temperature reporting
Added enhanced Fail Safe Mode (Press F1 at startup)
Added support for Intel "Sandy Bridge-E" CPUs
Added support for Intel "Ivy Bridge" CPUs
Added preliminary support for Intel "Haswell" CPUs (Core 4th Gen)
Added preliminary support for Intel "Haswell-ULT" CPUs
Added support for AMD "Kabini" (K16) CPUs
Added support for AMD "Bulldozer" CPUs
Added support for AMD "Trinity" CPUs
Added support for AMD E-/C-/G-/Z- "Bobcat" CPUs
Added support for Intel Atom "Pineview" CPUs
Added support for Intel Atom "Cedar Trail" CPUs
Added SPD detection on most AMD Chipsets
Bug Fixes
Enforced Coreboot support
Optimized run time for faster memory error detection
Rewriten lots of memory timings detection cod
Corrected bugs, bugs and more bugs (some could remain)
1.2.1 - 2013-11-25
------------------
BUG FIXES
* #348: [Windows XP] fixed "ImportError: DLL load failed" occurring on module
import.
* #425: [Solaris] crash on import due to failure at determining BOOT_TIME.
* #443: [Linux] can't set CPU affinity on systems with more than 64 cores.
1.2.0 - 2013-11-20
------------------
ENHANCEMENTS
* #439: assume os.getpid() if no argument is passed to psutil.Process
constructor.
* #440: new psutil.wait_procs() utility function which waits for multiple
processes to terminate.
BUG FIXES
* #348: [Windows XP/Vista] fixed "ImportError: DLL load failed" occurring on
module import.
1.1.3 - 2013-11-07
------------------
BUG FIXES
* #442: [Linux] psutil won't compile on certain version of Linux because of
missing prlimit(2) syscall.
Some changes:
* Many new modules, sorted in the following subdirs:
cloud commands database files internal inventory messaging
monitoring net_infrastructure network notification packaging
source_control system utilities web_infrastructure
* accelerated mode: An enhanced fireball mode that requires zero
bootstrapping and fewer requirements plus adds capabilities like
sudo commands
* Connection default is now "smart", which discovers if the system
openssh can support ControlPersist, and uses it if so, if not
falls back to paramiko
* external inventory scripts may now return host variables in one
pass, which allows them to be much more efficient for large numbers
of hosts (AWS!)
* "~" now expanded on each component of configured plugin paths
* Countless feature enhancements and bugfixes
For a full list, see https://github.com/ansible/ansible/blob/devel/CHANGELOG.md
All:
- New autoconf tests for:
- NFSv4 ACL support
- issetugid()
- lpathconf()
- utimens()
- futimens()
- lutimens()
- The rules in the Schily Makefilesystem have been restructured in order
to allow to avoid problems on platforms like FreeBSD and Mac OS X:
FreeBSD and Mac OS X banned the not-free-enough (because GPLd)
software to /usr/local, forcing us to add -I/usr/local/include
and -L/usr/local/lib. Unfortunately, /usr/local/include frequently
carries a _very_ outdated and thus wrong copy of "cdda_paranoia.h"
which is more than 10 years old and definitely incompatible with
dynamic linking on Mac OS X. This defective copy was first in the
search path and prevented compilation.
We now have a new macro: DEFOSINCDIRS= that grants to add include
directories to the end of the search PATH to allow us to find the
correct "cdda_paranoia.h" first.
Please test and report in case of problems.
- gmake and SunPro make include COMPILE.c with wrong content in their
built-in rules. We now clear this macro un RULES/rules.top. It is
still possible to provice a modified version from command line
or from the environment.
- include/schily/stdio.h was reordered, as it prevented compilation
of the "bsh" on Linux. This was a problem recently introduced,
when we added #ifndef NO_SCHILY_STDIO_H
- Fixed autoconf typo HAVE_MKNODKAT -> HAVE_MKNODAT
- include/schily/intcvt.h is now self contained.
- include/schily/windows.h now includes the same type workaround
for MINGW32 as fir the Microsoft C compiler, as there are the same
autoconf detection problems.
- Addded missing include/schily/err_type.h
- New file include/schily/shedit.h
Libschily:
- New files for libschily:
at-base.c generic implementation for *at() functions.
fchownat.c fchownat()
fdopendir.c fdopendir()
fstatat.c fstatat()
futimens.c futimens()
futimesat.c futimesat() Solaris specific old for utimensat()
lutimens.c lutimens()
linkat.c linkat()
mkdirat.c mkdirat()
mkfifo.c mkfifo()
mkfifoat.c mkfifoat()
mknodat.c mknodat()
readlinkat.c readlinkat()
renameat.c renameat()
symlinkat.c symlinkat()
unlinkat.c unlinkat()
utimens.c utimens()
utimensat.c utimensat()
at-base.c The base code for all single fd *at() functions.
at-base2.c The base code for all double fd *at() functions.
contain emulations for system interfaces introduced in summer 2001 by Sun
and with POSIX.1-2008.
- New functions in libschily:
absfpath()
resolvefpath()
allow to better control the behavior using flags.
- The linker map file for libschily now adds some forgotten functions.
- libschily/strstr.c and libschily/wcsstr.c now include a
CDDL License hint
Libfind:
- Libfind now allows to check for -type P (Solaris event port).
A missing "case 'P':" was added to the parser.
Cdrecord:
- Trying to avoid to confuse users
on Solaris and Linux where a fine grained privilege implementation
exists. Before, fine grained privileges have been given up after
initializing the program and this may result in a root user that
is treated as a normal user that cannot write into directories
without global write permission. The named programs now no longer
give up privileges in case they have been called with the needed
privileges already and thus the programs cannot be used for
privilege escalations anymore.
Cdda2wav (Maintained/enhanced by Jörg Schilling, originated by Heiko Eißfeldt heiko@hexco.de):
- Trying to avoid to confuse users
on Solaris and Linux where a fine grained privilege implementation
exists. Before, fine grained privileges have been given up after
initializing the program and this may result in a root user that
is treated as a normal user that cannot write into directories
without global write permission. The named programs now no longer
give up privileges in case they have been called with the needed
privileges already and thus the programs cannot be used for
privilege escalations anymore.
- Cddda2wav now flushes stderr before asking for a specific cddb
entry index. It seems that on Linux stderr may not be unbuffered
as expected.
- Cdda2wav moved the option parsing code into a separate function
gargs().
Readcd:
- Trying to avoid to confuse users
on Solaris and Linux where a fine grained privilege implementation
exists. Before, fine grained privileges have been given up after
initializing the program and this may result in a root user that
is treated as a normal user that cannot write into directories
without global write permission. The named programs now no longer
give up privileges in case they have been called with the needed
privileges already and thus the programs cannot be used for
privilege escalations anymore.
* Added new option '-l, --logfile-size'.
* Added new option '-w, --ignore-write-errors'.
* Option '--fill' has been renamed to '--fill-mode'.
* Option '--generate-logfile' has been renamed to '--generate-mode'.
* Added option '--sector-size' as a synonym of '--block-size'.
* Added option '--retries' as a synonym of '--max-retries'.
* Added option '--size' as a synonym of '--max-size'.
* rescuebook.cc: Trimming is now done from both edges of each
non-trimmed block. Largest blocks are trimmed first.
* rescuebook.cc: Largest blocks are now split first until
logfile reaches '--logfile-size' entries.
* logbook.cc (extend_sblock_vector, truncate_vector): Terminate
if truncation would discard finished blocks.
* rescuebook.cc: Mark failed blocks with 1 sector as bad-sector.
* logbook.cc (extend_sblock_vector): Remove last block of
logfile if it starts at isize and is not marked as finished.
* Documented that 'direct disc access' only reads whole sectors.
Also fix problem where by k3b wont rip the first track when it is not
alphabetically the first track to be ripped.
Fixes PR pkg/48369 and pkg/48371. Bump PKGREVISION
D-Bus 1.6.18 (2013-11-01)
==
The “sunflower plains” release.
Fixes:
• path_namespace='/' in match rules incorrectly matched nothing; it
now matches everything. (fd.o #70799, Simon McVittie)
1.1.2 - 2013-10-22
------------------
BUG FIXES
* #442: [Linux] psutil won't compile on Debian 6.0 because of missing
prlimit(2) syscall.
1.1.1 - 2013-10-08
------------------
BUG FIXES
* #438: [Linux] psutil won't compile on kernels < 2.6.36 due to missing
prlimit(2) syscall.
1.1.0 - 2013-09-28
------------------
ENHANCEMENTS
* #410: tar.gz and windows binary files are now hosted on PYPI.
* #412: [Linux] get/set process resource limits.
* #415: [Windows] Process.get_children() is an order of magnitude faster.
* #426: [Windows] Process.name is an order of magnitude faster.
* #431: [UNIX] Process.name is slightly faster because it unnecessarily
retrieved also process cmdline.
BUG FIXES
* #391: [Windows] psutil.cpu_times_percent() returns negative percentages.
* #408: STATUS_* and CONN_* constants don't properly serialize on JSON.
* #411: [Windows] examples/disk_usage.py may pop-up a GUI error.
* #413: [Windows] Process.get_memory_info() leaks memory.
* #414: [Windows] Process.exe on Windows XP may raise ERROR_INVALID_PARAMETER.
* #416: psutil.disk_usage() doesn't work well with unicode path names.
* #430: [Linux] process IO counters report wrong number of r/w syscalls.
* #435: [Linux] psutil.net_io_counters() might report erreneous NIC names.
* #436: [Linux] psutil.net_io_counters() reports a wrong 'dropin' value.
API CHANGES
* #408: STATUS_* and CONN_* constants (returned by Process' status() and
get_connections() methods respectively) have been turned from constant
objects to plain Python strings.
