After two months since -rc1 - which is way too long, I know - there is no point in delaying 2.5-rc2 any longer, so here it is.
http://download.gna.org/xenomai/testing/xenomai-2.5-rc2.tar.bz2 Since -rc1, we mainly focused on solving multiple interrupt pipeline issues for x86, introducing optimizations in core arithmetic helpers, and eventually adding new supported platforms to the powerpc port. And as usual, fixes are all over the map. x86-wise(32/64), if you happen to run any I-pipe release in SMP mode, prior to Linux 2.6.28.9, and/or earlier than 2.2-07, you should definitely upgrade to a more recent version. Really. Regarding powerpc, we now officially support powerpc32/SMP, in addition to powerpc64/SMP we have been supporting for quite some time now. The first powerpc32/SMP platform Xenomai officially supports is Emerson's MVME7100, which is based on a dual core MPC8641D. There are things brewing in the ARM world, particularly an omap3 port, but this is not ready for prime time yet (ETA is -rc3). Stay tuned. Finally, Blackfin users may want to know that we will be tracking Mike Frysinger's tree from now on: git://git.kernel.org/pub/scm/linux/kernel/git/vapier/blackfin.git. On this architecture, the I-pipe bits should be on their way to 2.6.31 mainline already, so Blackfin is actually ahead of everyone else with respect to kernel support. Official support available with -rc2 is still based on 2.6.28.10 though, from the Blackfin project SVN/GIT mirror (see ksrc/arch/blackfin/patches/README for more). Btw, more information regarding which hardware Xenomai supports is available from: http://www.xenomai.org/index.php/Embedded_Device_Support http://www.xenomai.org/index.php/Other_Hardware_Support Oh, last point: 2.5-rc2 has a decent stability record among Xenomai maintainers, but as usual, let me emphasize the obvious: at the end of the day, only _you_ can make sure the upcoming 2.5 will work for you, and unless you feel insanely lucky these days or simply don't care about upgrading, you should not wait for 2.5.0 to perform that check. Here is the short log: Gilles Chanteperdrix (10): arith: Fix xnarch_divrem_billion not to use rthal_ullmul. arm: Adapt to preemptible context switch proposed by newer I-pipe patches arm: Fix user-space tsc on ARM arm: Improve decrementer user-space tsc emulation arm: Use PIC mute routines on the ARM platform misc: Remove SIGSHADOW debug message nucleus: Fix computation of heap overhead posix: Fix the POSIX skin to use the common time conversion. posix: Preserve the main thread scheduling parameters when shadowing it. rtdm: Reorder headers include directives. Jan Kiszka (6): wrappers: Add rthal_local_irq_disabled wrapper nucleus: Add missing header for isspace nucleus: Move xnarch_set_irq_affinity out of intrlock posix: Improve access checks in select rtdm: Avoid namespace pollution in RTDM_EXECUTE_ATOMICALLY rtdm: Instrument rtdm_lock_get for proper use Martin Shepherd (1): rtdm: Document the special meaning of returning -ENOSYS Philippe Gerum (45): arith: Fix signedness of divisor arith: Introduce xnarch_divrem_billion arith: Make optimized tsc<->ns conversion routines available to user-space. arith: Rename imuldiv_up to imuldiv_ceil arm: Update Adeos/arm support blackfin: Decrease some calibration values due to intrinsic latency reduction blackfin: Enable CORETMR interrupt when CONFIG_TICKSOURCE_GPTMR0 is set blackfin: Make sure to program the next core timer shot in a canonical way blackfin: Update Adeos/blackfin support build: Fix dist rule build: Fix host selection build: Fix patch selection with --enable-linux-build build: Upgrade to autoconf >= 2.62 doc: Update README.INSTALL doc: fix generate-doc rule doc: regenerate documentation native, posix: Warn about deprecated usage of Xenomai skins in kernel space nucleus: Decentralize PROC_FS-based support nucleus: Increase default generic stack pool size nucleus: Introduce /proc/xenomai/schedclasses nucleus: Make sure to always propagate the host tick in time. nucleus: Move arch-specific capability flags to the feature header nucleus: Remove misleading comment nucleus: Round up results of tsc -> timer ticks conversions when programming nucleus: Shorten uninterruptible section within shadow relaxing path nucleus: Tell Linux whenever we failed programming the next host tick event. posix: Fix build w/o PROC_FS support posix: fix handling of HAVE_PTHREAD_MUTEXATTR_SETPROTOCOL posix: fix pointer signedness in __wrap_clock_gettime() powerpc: Add calibration default for Emerson's MVME7100 SBC (MPC8641D) powerpc: Add calibration default for MPC8272-ADS powerpc: Add calibration default for TQM8560 powerpc: Add calibration info for 440EP Yosemite powerpc: Do not restore Altivec and SPE regs for kernel threads powerpc: Fix calibration values for known 85xx-based SBCs powerpc: Introduce PIC mute support, initially for powerpc. powerpc: Remove obsolete kernel support powerpc: Update Adeos/powerpc support powerpc: make thread_info reference depend on CONFIG_IPIPE_UNMASKED_CONTEXT_SWITCH powerpc: upgrade I-pipe/2.4 support to 2.4.25-ppc-2.0-01 rtdm: fix build for 2.4 testsuite/mutex-torture: make sure pthread_mutexattr_setprotocol() is available when building wrappers: Void wrap_proc_dir_entry_owner() argument if unused x86: upgrade I-pipe support to 2.6.29.5-x86-2.4-01 Xenomai v2.5-rc2 Stefan Kisdaroczi (2): build: Fix Debian package build rtcan: Disambiguate PCI device id. for IXXAT-PCI boards Vladimir Zapolskiy (1): wrappers: Made conditional inclusion of owner field in struct proc_dir_entry due to changes in 2.6.30. Wolfgang Grandegger (2): rtcan: ixxat-pci: fix problems with device probing rtcan: mscan: fix build problems with recent kernel versions -- Philippe. _______________________________________________ Xenomai-help mailing list [email protected] https://mail.gna.org/listinfo/xenomai-help
