Does task_info support the TASK_EVENTS_INFO type of information?

2009-06-02 Thread Da Zheng

Hi,

I tried the code below:
 struct task_events_info taskevents;
 size_t tkcount = TASK_EVENTS_INFO_COUNT;
 error_t err;

 err = task_info (mach_task_self (), TASK_EVENTS_INFO,
  (task_info_t) taskevents, tkcount);
 assert_perror (err);

and I always get the error:
Unexpected error: (os/kern) invalid argument.

It seems to me that the Mach doesn't support TASK_EVENTS_INFO.
I read Mach 3 Kernel Interfaces and it seems that the original Mach only 
supports TASK_BASIC_INFO and TASK_THREAD_TIMES_INFO, but The GNU Mach 
Reference Manual mentions TASK_EVENTS_INFO. I assume that this feature 
is added in GNU Mach.


I use Debian/Hurd, and the version of Mach is GNU Mach 1.3.99. I wonder 
which version of GNU Mach supports TASK_EVENTS_INFO.


Best regards,
Zheng Da




Re: Does task_info support the TASK_EVENTS_INFO type of information?

2009-06-02 Thread Samuel Thibault
Da Zheng, le Tue 02 Jun 2009 16:24:27 +0800, a écrit :
 It seems to me that the Mach doesn't support TASK_EVENTS_INFO.

/usr/src/gnumach-1€ rgrep TASK_EVENTS_INFO .
./include/mach/task_info.h:#define  TASK_EVENTS_INFO2 /* various 
event counts */
./include/mach/task_info.h:#define  TASK_EVENTS_INFO_COUNT  \
./doc/mach.info-1:`TASK_EVENTS_INFO'
./doc/mach.info-1:  returned is `TASK_EVENTS_INFO_COUNT'.
./doc/mach.info-1: provided it as the TASK_INFO parameter for the 
`TASK_EVENTS_INFO'
./doc/mach.texi:@item TASK_EVENTS_INFO
./doc/mach.texi:@code{TASK_EVENTS_INFO_COUNT}.
./doc/mach.texi:@code{TASK_EVENTS_INFO} flavor of @code{task_info}.  It has the
/usr/src/gnumach-1€

so yes, it seems not supported by GNU Mach albeit documented.

Samuel




[patch #6844] fix a bug in init: initialize 'flags' in reboot_system

2009-06-02 Thread Zheng Da

URL:
  http://savannah.gnu.org/patch/?6844

 Summary: fix a bug in init: initialize 'flags' in
reboot_system
 Project: The GNU Hurd
Submitted by: zhengda
Submitted on: Tue 02 Jun 2009 08:47:27 AM GMT
Category: other Hurd
Priority: 5 - Normal
  Status: Works For Me
 Privacy: Public
 Assigned to: None
Originator Email: 
 Open/Closed: Open
 Discussion Lock: Any
 Planned Release: None
Wiki-like text discussion box: 

___

Details:

We should initialize 'flags' in reboot_system.
The only information we need to get from proc_getprocinfo is the state of the
process. 'flags = 0' is very enough to do the job and it can make
proc_getprocinfo less likely to fail.
The Mach in my system somehow doesn't support the TASK_EVENTS_INFO type of
information. Without the initialization of 'flags', it is possible that
proc_getprocinfo returns error, which results in the failure to shut down the
subhurd.



___

File Attachments:


---
Date: Tue 02 Jun 2009 08:47:27 AM GMT  Name: init.diff  Size: 492B   By:
zhengda

http://savannah.gnu.org/patch/download.php?file_id=18216

___

Reply to this item at:

  http://savannah.gnu.org/patch/?6844

___
  Message sent via/by Savannah
  http://savannah.gnu.org/





Re: Does task_info support the TASK_EVENTS_INFO type of information?

2009-06-02 Thread Da Zheng

Samuel Thibault wrote:

Da Zheng, le Tue 02 Jun 2009 16:24:27 +0800, a écrit :
  

It seems to me that the Mach doesn't support TASK_EVENTS_INFO.



/usr/src/gnumach-1€ rgrep TASK_EVENTS_INFO .
./include/mach/task_info.h:#define  TASK_EVENTS_INFO2 /* various 
event counts */
./include/mach/task_info.h:#define  TASK_EVENTS_INFO_COUNT  \
./doc/mach.info-1:`TASK_EVENTS_INFO'
./doc/mach.info-1:  returned is `TASK_EVENTS_INFO_COUNT'.
./doc/mach.info-1: provided it as the TASK_INFO parameter for the 
`TASK_EVENTS_INFO'
./doc/mach.texi:@item TASK_EVENTS_INFO
./doc/mach.texi:@code{TASK_EVENTS_INFO_COUNT}.
./doc/mach.texi:@code{TASK_EVENTS_INFO} flavor of @code{task_info}.  It has the
/usr/src/gnumach-1€

so yes, it seems not supported by GNU Mach albeit documented.

Samuel
  
However, TASK_EVENTS_INFO is used by Hurd in proc_getprocinfo. libps 
also tries to use it.

Is it a planned feature or it was supported by GNU Mach but was removed?
If it was removed, we should update the Hurd too.

Anyway, it causes some trouble when subhurd is shutdown.
I just provided a patch to make sure that proc_getprocinfo doesn't try 
to get the TASK_EVENTS_INFO type of information when the subhurd is 
shutdown.


Best regards,
Zheng Da




[b...@gnu.org: [Savannah-announce] Recover your repositories]

2009-06-02 Thread Thomas Schwinge
Hello!

Email is full-quoted and with my comments inlined.

| - Forwarded message from Sylvain Beucler b...@gnu.org -
| Date: Tue, 2 Jun 2009 09:16:55 +0200
| Subject: [Savannah-announce] Recover your repositories
| 
| The system is now completely functional.
| It's time to decide what to do with your data.
| (see first explanations at 
http://lists.gnu.org/archive/html/savannah-users/2009-05/msg00023.html)
| 
| 
| Please discuss and send questions at savannah-us...@gnu.org !
| 
| For import requests: see below, you need to send a support request via
| the web interface.
| 
| 
| 
| Basically your choices are:
| a) you can use the backup from April 29

This is good enough for us, because...

| b) you can use our latest (incomplete) backup from May 29
| c) you can import another backup you have
| d) you can wait a few days until we possibly recover latest data from the 
crashes disk (nothing is sure though). More details about this later.
| 
| 
| Here's how do it for the various VCSs:
| 
| * CVS:
| a) The version from 29 April in the 'backup-20090429' subdirectory.
| b) The version from 29 May in the 'backup-20090529-incomplete' subdirectory.
| c) Upload your backup in your download area (SFTP) for us to import
| 
| Access commands:
| cvs -d:pserver:anonym...@cvs.sv.gnu.org:/sources/backup-20090429/your_project 
co your_module
| cvs 
-d:pserver:anonym...@cvs.sv.gnu.org:/sources/backup-20090529-incomplete/your_project
 co your_module
| rsync rsync://cvs.sv.gnu.org/sources/backup-20090429/your_project/

... this version (backup-20090429) is sufficiently recent for us, as no
checkins to the CVS repository have been done after this date, due to the
(still) ongoing CVS to Git migration.  (By the way, I'm currently (a)
waiting for a clean-up script by Jim Meyering, and (b) tracking down,
with the help of its author, either a bug in the conversion program, or a
repository corruption in the gnumach module.)  Also, I diffed the
backup-20090429 rsync tree against my local rsync mirror and found it to
be in a correct state.

| rsync rsync://cvs.sv.gnu.org/sources/backup-20090529-incomplete/your_project/
| 
| IN ALL CASES: post a support request, authenticated (i.e. login first) at 
https://savannah.gnu.org/support/?func=additemgroup=administration with 
category RECOVERY: CVS/SVN. Requests send by mail or in the wrong category 
won't be processed.
| 
| If after 2 weeks we didn't receive an answer for your project, we'll 
reinstall the latest backup we have.
| 
| 
| * SVN:

[Snipped, as not relevant for us.]

| * Git:
| a) We installed the version from 29 April.
| b) You can attempt to recover from the partial 29 May backup: 
http://git.savannah.gnu.org/r/backup-20090529-incomplete/
| c) You can push a local version if you have one (git push --all; git push 
--tags)
| 
| You can import the repository directly, without our help.

... exactly! (thanks DVCS!); for the web pages repository I already
pushed the missing changes, and for the other repositories the same can
be done, and I'll take care of that as soon as I have the conversion
finished properly.  Neal, you can also simply push your viengoos and
viengoos-on-bare-meal branches back into the Viengoos Git repository.

| * Mercurial:

[Snipped, as not relevant for us.]

| * WebCVS:
| 
| We plan to either recover the data from disk, or failing that, manually 
commit the latest version at www.gnu.org and www.nongnu.org.

... we don't care, as we have all the history in our web pages' Git
repository and use the CVS one only as a transport mechanism to get the
files to the web server.


Regards,
 Thomas


signature.asc
Description: Digital signature


Re: [b...@gnu.org: [Savannah-announce] Recover your repositories]

2009-06-02 Thread Arne Babenhauserheide
On Tuesday, 2. June 2009 11:30:25 Thomas Schwinge wrote:
 | * Mercurial:

 [Snipped, as not relevant for us.]

Except the pyhurd code, but I just checked the remote repo and it works. We 
only need to push to get it up to date again. 

Best wishes, 
Arne

--- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- 
   - singing a part of the history of free software -
  http://infinite-hands.draketo.de


signature.asc
Description: This is a digitally signed message part.


A GNU/Hurd Roadmap dream

2009-06-02 Thread Arne Babenhauserheide
I just dreamed of a Roadmap the Hurd might have - a Roadmap which would be the 
perfect Roadmap for making the Hurd useful for me. 

I thought I'd share the dream with you. 

This is the Roadmap I dreamt of: 


1) Sound support: Now I can listen to my music (essential). 

2) Qemu image with X11 out of the box: Now I can do most of my general 
development work in the Hurd, including convenient websurfing. 

3) Solid hardware support: Now I can use a dual-boot on my main computer. 

4) Linux device driver wrapper: Now I don't really need Linux anymore - except 
for some applications. I can spend most of my computer time using the Hurd. 
And I can port some applications I need. 



Then I awoke and I thought and it would be great to add just one more item: 

5) Gentoo Portage on the GNU/Hurd: Now the work for maintaining the software 
bases shrinks massively and more software can be made available at far lower 
maintenance cost. 

This is not necessary, though. It also wasn't part of the dream. 

Best wishes, 
Arne

PS: A native installer wasn't part of the dream, either. I don't need it for 
using the Hurd. 

--- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- 
   - singing a part of the history of free software -
  http://infinite-hands.draketo.de


signature.asc
Description: This is a digitally signed message part.


Re: A GNU/Hurd Roadmap dream

2009-06-02 Thread Michael Banck
On Tue, Jun 02, 2009 at 10:22:45PM +0200, Arne Babenhauserheide wrote:
 This is the Roadmap I dreamt of: 

Sorry, but this is a wishlist, not a roadmap.


Michael