[Xenomai-core] [announce] Xenomai v2.5.4

[Xenomai-core] [announce] Xenomai v2.5.4

[Gilles Chanteperdrix]

Hi,

you will find Xenomai 2.5.4, aka "Sleep Walk", at the usual place:
http://download.gna.org/xenomai/stable/xenomai-2.5.4.tar.bz2

It contains the usual amount of bug fixes and I-pipe patches update, as 
well as a new piece of nice code: the so-called mayday support, which 
should allow Xenomai watchdog to work less brutally (namely, avoid 
killing the tasks occupying too much CPU, and send them a signal), and 
be the base for a new version of user-space signals in the 2.6 branch.

Alex also pushed a lot of improvements for the analogy support.

So, it seems not everybody has been lazy like me for this last version.

The complete shortlog follows:
Alexis Berlemont (48):
      analogy: change the context's role (broken)
      analogy: the buffer structure is now the central field of a4l_context (bro
ken)
      analogy: the subdevice structure got a new status field (broken)
      analogy: the transfer structure is left with a minimal role (broken)
      analogy: first draft of buffer initialization functions (broken)
      analogy: adapt open, r/w, select and ioctl functions (broken)
      analogy: adapt a4l_set_dev() after a4l_context's overhaul (broken)
      analogy: update a4l_set_dev() declaration (broken)
      analogy: update comments on a4l_context (broken)
      analogy: changes related with subdevice's status field (broken)
      analogy: replace transfer setup functions with buffer setup ones (broken)
      analogy: update cancel functions (broken)
      analogy: rewrite the cancel ioctl handler (broken)
      analogy: fix bulk flag declaration in buffer.h (broken)
      analogy: update a4l_read and a4l_write (broken)
      analogy: update all a4l_buf_* functions (broken)
      analogy: last updates in the buffer part (broken)
      analogy: cosmetic changes (broken)
      analogy: declare the reserve / release functions at the subd level (broken
)
      analogy: update a4l_get_minor function (broken)
      analogy: update a4l_set_dev and remove useless info traces (broken)
      analogy: use rtdm_context_to_private (broken)
      analogy: minor fix in the subdevice structure declaration
      analogy: add some helper macros to test the subdevice's characteristics
      analogy: remove useless functions in the subdevice part
      analogy: fix the buffer syscalls (ioctl + r/w) after buffer review (broken)
      analogy: fix the declaration of the structure a4l_context (broken)
      analogy: fix compilation issues and review the mmap ioctl handler (broken)
      analogy: cosmetic change (broken)
      analogy: fix buffer's compilation issues (broken)
      analogy: prettify some subdevice tests (broken)
      analogy: [pcimio] fix a huge hack in the mite initialization (broken)
      analogy: fix the last compilation problems
      analogy: fix a missing setting of the buf field in subdevice (broken)
      analogy: fix the subdevice status management
      analogy: fix buffer initialization/cleanup calls at open/close times
      analogy: [loop] add a debug trace when trigger is called
      analogy: fix test of subdevice status in a4l_write
      analogy: [fake - loop] remove volatile keywords
      analogy: add a detail in a4l_close doxygen doc
      analogy: add an arbitrary sleep in cmd_write before closing the device
      analogy: [ni_pcimio] really minor changes
      analogy: [ni_pcimio] add the missing allocation of the digital ring
      analogy: [ni_pcimio] fix timeout value in digital trigger
      analogy: remove a4l_subd_is_busy calls in analogy core
      analogy: remove calls of a4l_release/reserve_subd in the core
      analogy: remove some tests which become with the buffer overhaul
      analogy: fix a bug in a4l_fill_desc() when called on an idle device

Gilles Chanteperdrix (18):
      arm: fix VFP handling in the SMP case
      arm: get the nodiv_llimd code to compile in thumb mode
      Merge commit 'rpm' into pending
      Merge branch 'pending'
      rtcan: add missing PCI IDs for old kernels
      Merge commit 'rpm/for-upstream' into build-test
      Update autotools files
      arm: clarify the patches README with regard to vendor-specifi branches
      Merge commit 'analogy' into pending
      testsuite: adapt "run" scripts to the --with-testdir option
      native: add cancellation points
      arm: upgrade adeos patches to 2.6.30-1.15-02, 2.6.31-1.16-02, 2.6.33-1.17-
02
      Add IMX51 patch
      build: bootstrap
      compat: add missing PCI ID for 2.4 kernels
      sched: avoid infinite reschedule loops
      build: bump version number
      doc: regenerate

Pavel Cheblakov (1):
      rtcan: add rtcan_plx_pci driver

Philippe Gerum (46):
      generic/hal: remove dead code
      powerpc/hal: test XENO_HW_UNLOCKED_SWITCH instead of IPIPE_WANT_PREEMPTIBL
E_SWITCH
      powerpc: use hardened switch_mm() support when available
      powerpc: upgrade I-pipe support to 2.6.33.4-powerpc-2.9-01
      nucleus: prevent stack overflow w/ unlocked context switch
      scripts: fix --enable-debug help string
      posix: fix build w/o CONFIG_PROC_FS
      nucleus/heap: fix build w/o CONFIG_PROC_FS
      x86: upgrade I-pipe support to 2.6.32.13-x86-2.6-04
      x86: update fpu state accessor for 2.6.35+
      x86: upgrade I-pipe support to 2.6.34-x86-2.6-04
      powerpc: upgrade I-pipe support to 2.6.34-powerpc-2.9-01
      uitron: fix wait flag helper
      hal/generic: always track switch to foreign stack
      compat: provide linux/err.h wrapper to linux/fs.h
      nucleus: silence deprecated warnings on __builtin_expect for old compilers
      rtai: mark as deprecated
      skins: sanitize heap binding
      build: check for CONFIG_MMU in user-space
      nucleus: fix heap mapping for nommu
      wrappers: no need to reserve pages passed to vm_insert_page
      x86: upgrade I-pipe support to 2.6.32.15-x86-2.7-00, 2.6.34-x86-2.7-00
      powerpc: upgrade I-pipe support to 2.6.33.5-powerpc-2.10-01, 2.6.34-powerpc-2.10-00
      x86: upgrade I-pipe support to 2.6.32.15-x86-2.7-00, 2.6.34-x86-2.7-00
      arm: upgrade I-pipe support to 2.6.33-arm-1.17-00
      blackfin: upgrade I-pipe support to 2.6.34-blackfin-1.14-00
      nucleus: introduce generic bits for MAYDAY support
      powerpc: enable MAYDAY support
      x86: enable MAYDAY support
      arm: enable MAYDAY support
      blackfin: enable MAYDAY support
      nios2: enable MAYDAY support
      arm: upgrade I-pipe support to 2.6.33-arm-1.17-01
      blackfin: upgrade I-pipe support to 2.6.34-blackfin-1.14-02
      powerpc: upgrade I-pipe support to 2.6.33.5-powerpc-2.10-03, 2.6.34-powerpc-2.10-02
      x86: upgrade I-pipe support to 2.6.32.15-x86-2.7-02, 2.6.34-x86-2.7-01
      common: warn early about missing /dev/rtheap
      nucleus: fix uninit variable
      nios2: upgrade I-pipe support to 2.6.30-nios2-1.2-00
      rtai: disable for MMU-less and 64bit builds
      rtdm: fix memory mapping services for nommu platforms
      powerpc: introduce HW_MATH_EMU symbol
      switchtest: bail out upon sem_wait() error
      nucleus: export xnshadow_call_mayday() as a service
      powerpc: add calibration values for mpc8xx
      powerpc: upgrade I-pipe support to 2.6.34-powerpc-2.10-03

Stefan Kisdaroczi (3):
      debian: switch to dpkg-source 3.0 (native) format
      debian: fix init script start/stop values
      x86: add Intel ICH10_1 to rthal_smi_pci_tbl

Tschaeche IT-Services (1):
      nucleus: enable tswitch callout for the root thread

Enjoy.

-- 
					    Gilles.

Re: [Xenomai-core] [announce] Xenomai v2.5.4

[Wolfgang Mauerer]

Gilles Chanteperdrix wrote:
> Hi,
> 
> you will find Xenomai 2.5.4, aka "Sleep Walk", at the usual place:
> http://download.gna.org/xenomai/stable/xenomai-2.5.4.tar.bz2
> 
> It contains the usual amount of bug fixes and I-pipe patches update, as 
> well as a new piece of nice code: the so-called mayday support, which 
> should allow Xenomai watchdog to work less brutally (namely, avoid 
> killing the tasks occupying too much CPU, and send them a signal), and 
> be the base for a new version of user-space signals in the 2.6 branch.
> 
> Alex also pushed a lot of improvements for the analogy support.
> 
> So, it seems not everybody has been lazy like me for this last version.

to satisfy my curiosity: What was the exact reason for not including
the hostrt stuff? After having spent quite some (obviously wasted)
effort to make the code compliant with the maintainer's requests, I
wonder what could have been missing.

Wolfgang
> 
> The complete shortlog follows:
> Alexis Berlemont (48):
>       analogy: change the context's role (broken)
>       analogy: the buffer structure is now the central field of a4l_context (bro
> ken)
>       analogy: the subdevice structure got a new status field (broken)
>       analogy: the transfer structure is left with a minimal role (broken)
>       analogy: first draft of buffer initialization functions (broken)
>       analogy: adapt open, r/w, select and ioctl functions (broken)
>       analogy: adapt a4l_set_dev() after a4l_context's overhaul (broken)
>       analogy: update a4l_set_dev() declaration (broken)
>       analogy: update comments on a4l_context (broken)
>       analogy: changes related with subdevice's status field (broken)
>       analogy: replace transfer setup functions with buffer setup ones (broken)
>       analogy: update cancel functions (broken)
>       analogy: rewrite the cancel ioctl handler (broken)
>       analogy: fix bulk flag declaration in buffer.h (broken)
>       analogy: update a4l_read and a4l_write (broken)
>       analogy: update all a4l_buf_* functions (broken)
>       analogy: last updates in the buffer part (broken)
>       analogy: cosmetic changes (broken)
>       analogy: declare the reserve / release functions at the subd level (broken
> )
>       analogy: update a4l_get_minor function (broken)
>       analogy: update a4l_set_dev and remove useless info traces (broken)
>       analogy: use rtdm_context_to_private (broken)
>       analogy: minor fix in the subdevice structure declaration
>       analogy: add some helper macros to test the subdevice's characteristics
>       analogy: remove useless functions in the subdevice part
>       analogy: fix the buffer syscalls (ioctl + r/w) after buffer review (broken)
>       analogy: fix the declaration of the structure a4l_context (broken)
>       analogy: fix compilation issues and review the mmap ioctl handler (broken)
>       analogy: cosmetic change (broken)
>       analogy: fix buffer's compilation issues (broken)
>       analogy: prettify some subdevice tests (broken)
>       analogy: [pcimio] fix a huge hack in the mite initialization (broken)
>       analogy: fix the last compilation problems
>       analogy: fix a missing setting of the buf field in subdevice (broken)
>       analogy: fix the subdevice status management
>       analogy: fix buffer initialization/cleanup calls at open/close times
>       analogy: [loop] add a debug trace when trigger is called
>       analogy: fix test of subdevice status in a4l_write
>       analogy: [fake - loop] remove volatile keywords
>       analogy: add a detail in a4l_close doxygen doc
>       analogy: add an arbitrary sleep in cmd_write before closing the device
>       analogy: [ni_pcimio] really minor changes
>       analogy: [ni_pcimio] add the missing allocation of the digital ring
>       analogy: [ni_pcimio] fix timeout value in digital trigger
>       analogy: remove a4l_subd_is_busy calls in analogy core
>       analogy: remove calls of a4l_release/reserve_subd in the core
>       analogy: remove some tests which become with the buffer overhaul
>       analogy: fix a bug in a4l_fill_desc() when called on an idle device
> 
> Gilles Chanteperdrix (18):
>       arm: fix VFP handling in the SMP case
>       arm: get the nodiv_llimd code to compile in thumb mode
>       Merge commit 'rpm' into pending
>       Merge branch 'pending'
>       rtcan: add missing PCI IDs for old kernels
>       Merge commit 'rpm/for-upstream' into build-test
>       Update autotools files
>       arm: clarify the patches README with regard to vendor-specifi branches
>       Merge commit 'analogy' into pending
>       testsuite: adapt "run" scripts to the --with-testdir option
>       native: add cancellation points
>       arm: upgrade adeos patches to 2.6.30-1.15-02, 2.6.31-1.16-02, 2.6.33-1.17-
> 02
>       Add IMX51 patch
>       build: bootstrap
>       compat: add missing PCI ID for 2.4 kernels
>       sched: avoid infinite reschedule loops
>       build: bump version number
>       doc: regenerate
> 
> Pavel Cheblakov (1):
>       rtcan: add rtcan_plx_pci driver
> 
> Philippe Gerum (46):
>       generic/hal: remove dead code
>       powerpc/hal: test XENO_HW_UNLOCKED_SWITCH instead of IPIPE_WANT_PREEMPTIBL
> E_SWITCH
>       powerpc: use hardened switch_mm() support when available
>       powerpc: upgrade I-pipe support to 2.6.33.4-powerpc-2.9-01
>       nucleus: prevent stack overflow w/ unlocked context switch
>       scripts: fix --enable-debug help string
>       posix: fix build w/o CONFIG_PROC_FS
>       nucleus/heap: fix build w/o CONFIG_PROC_FS
>       x86: upgrade I-pipe support to 2.6.32.13-x86-2.6-04
>       x86: update fpu state accessor for 2.6.35+
>       x86: upgrade I-pipe support to 2.6.34-x86-2.6-04
>       powerpc: upgrade I-pipe support to 2.6.34-powerpc-2.9-01
>       uitron: fix wait flag helper
>       hal/generic: always track switch to foreign stack
>       compat: provide linux/err.h wrapper to linux/fs.h
>       nucleus: silence deprecated warnings on __builtin_expect for old compilers
>       rtai: mark as deprecated
>       skins: sanitize heap binding
>       build: check for CONFIG_MMU in user-space
>       nucleus: fix heap mapping for nommu
>       wrappers: no need to reserve pages passed to vm_insert_page
>       x86: upgrade I-pipe support to 2.6.32.15-x86-2.7-00, 2.6.34-x86-2.7-00
>       powerpc: upgrade I-pipe support to 2.6.33.5-powerpc-2.10-01, 2.6.34-powerpc-2.10-00
>       x86: upgrade I-pipe support to 2.6.32.15-x86-2.7-00, 2.6.34-x86-2.7-00
>       arm: upgrade I-pipe support to 2.6.33-arm-1.17-00
>       blackfin: upgrade I-pipe support to 2.6.34-blackfin-1.14-00
>       nucleus: introduce generic bits for MAYDAY support
>       powerpc: enable MAYDAY support
>       x86: enable MAYDAY support
>       arm: enable MAYDAY support
>       blackfin: enable MAYDAY support
>       nios2: enable MAYDAY support
>       arm: upgrade I-pipe support to 2.6.33-arm-1.17-01
>       blackfin: upgrade I-pipe support to 2.6.34-blackfin-1.14-02
>       powerpc: upgrade I-pipe support to 2.6.33.5-powerpc-2.10-03, 2.6.34-powerpc-2.10-02
>       x86: upgrade I-pipe support to 2.6.32.15-x86-2.7-02, 2.6.34-x86-2.7-01
>       common: warn early about missing /dev/rtheap
>       nucleus: fix uninit variable
>       nios2: upgrade I-pipe support to 2.6.30-nios2-1.2-00
>       rtai: disable for MMU-less and 64bit builds
>       rtdm: fix memory mapping services for nommu platforms
>       powerpc: introduce HW_MATH_EMU symbol
>       switchtest: bail out upon sem_wait() error
>       nucleus: export xnshadow_call_mayday() as a service
>       powerpc: add calibration values for mpc8xx
>       powerpc: upgrade I-pipe support to 2.6.34-powerpc-2.10-03
> 
> Stefan Kisdaroczi (3):
>       debian: switch to dpkg-source 3.0 (native) format
>       debian: fix init script start/stop values
>       x86: add Intel ICH10_1 to rthal_smi_pci_tbl
> 
> Tschaeche IT-Services (1):
>       nucleus: enable tswitch callout for the root thread
> 
> Enjoy.
> 

Re: [Xenomai-core] [announce] Xenomai v2.5.4

[Gilles Chanteperdrix]

Wolfgang Mauerer wrote:
> Gilles Chanteperdrix wrote:
>> Hi,
>>
>> you will find Xenomai 2.5.4, aka "Sleep Walk", at the usual place:
>> http://download.gna.org/xenomai/stable/xenomai-2.5.4.tar.bz2
>>
>> It contains the usual amount of bug fixes and I-pipe patches update, as 
>> well as a new piece of nice code: the so-called mayday support, which 
>> should allow Xenomai watchdog to work less brutally (namely, avoid 
>> killing the tasks occupying too much CPU, and send them a signal), and 
>> be the base for a new version of user-space signals in the 2.6 branch.
>>
>> Alex also pushed a lot of improvements for the analogy support.
>>
>> So, it seems not everybody has been lazy like me for this last version.
> 
> to satisfy my curiosity: What was the exact reason for not including
> the hostrt stuff? After having spent quite some (obviously wasted)
> effort to make the code compliant with the maintainer's requests, I
> wonder what could have been missing.

I was Ok with the Xenomai part, but we needed Philippe to review the
Adeos part, and release new versions of the patches, and having delayed
the 2.5.4 release for some time, and did not feel like delaying it much
longer.

Besides, we have started the 2.6 branch, and such a big change makes
perfect material for this new branch.

Your patches are not wasted, they have just been put on hold.

-- 
					    Gilles.

Re: [Xenomai-help] [Xenomai-core] [announce] Xenomai v2.5.4

[Peter Soetens]

On Monday 02 August 2010 23:46:51 Gilles Chanteperdrix wrote:
> Hi,
> 
> you will find Xenomai 2.5.4, aka "Sleep Walk", at the usual place:
> http://download.gna.org/xenomai/stable/xenomai-2.5.4.tar.bz2

There's an rtcan patch I submitted on July 9, 2010 ( IXXAT SJA_MOD register 
refuses to set up dual channel cards) missing in this release. Must it first be 
picked up by Wolfgang before it gets into the release ?

Peter

Re: [Xenomai-help] [Xenomai-core] [announce] Xenomai v2.5.4

[Gilles Chanteperdrix]

Peter Soetens wrote:
> On Monday 02 August 2010 23:46:51 Gilles Chanteperdrix wrote:
>> Hi,
>>
>> you will find Xenomai 2.5.4, aka "Sleep Walk", at the usual place:
>> http://download.gna.org/xenomai/stable/xenomai-2.5.4.tar.bz2
> 
> There's an rtcan patch I submitted on July 9, 2010 ( IXXAT SJA_MOD register 
> refuses to set up dual channel cards) missing in this release. Must it first be 
> picked up by Wolfgang before it gets into the release ?

Or maybe simply acked... But arguably, it was simple enough to be merged
directly. I simply missed it, sorry.

-- 
					    Gilles.

Re: [Xenomai-help] [Xenomai-core] [announce] Xenomai v2.5.4

[Wolfgang Grandegger]

On 08/11/2010 02:23 PM, Gilles Chanteperdrix wrote:
> Peter Soetens wrote:
>> On Monday 02 August 2010 23:46:51 Gilles Chanteperdrix wrote:
>>> Hi,
>>>
>>> you will find Xenomai 2.5.4, aka "Sleep Walk", at the usual place:
>>> http://download.gna.org/xenomai/stable/xenomai-2.5.4.tar.bz2
>>
>> There's an rtcan patch I submitted on July 9, 2010 ( IXXAT SJA_MOD register 
>> refuses to set up dual channel cards) missing in this release. Must it first be 
>> picked up by Wolfgang before it gets into the release ?
> 
> Or maybe simply acked... But arguably, it was simple enough to be merged
> directly. I simply missed it, sorry.

And I forgot to take care as well :-(.

Wolfgang.