[Xenomai] mlockall and dlopenskins?

[Xenomai] mlockall and dlopenskins?

[Yogi A. Patel]

Hi -

I used to provide --enable-posix-auto-mlockall --enable-dlopen-skins as 
options to the configure script when compiling user space support. 
However in 2.6.3 I get a warning saying those are unrecognized options.

Now. I was looking online at the documentation and found the following:

/Xenomai libraries only invoked mlockall if a special option was passed 
to the configure script when compiling Xenomai user-space support. So, 
applications which did not want to depend on this configuration had to 
call mlockall by themselves, before using any Xenomai service, by using://
//
//mlockall(MCL_CURRENT | MCL_FUTURE);/

This is slightly vague and hard to understand. The warnings make it 
obvious that the enable-posix-auto-mlockall call does nothing when used 
during compile. Does this mean that in 2.6.3 and above mlockall is 
included in user space libraries by default (requring no changes to our 
software besides removing that flag from the compile)?

And what about enable-dlopen-skins? I can't find any documentation 
online on that changing in 2.6.3.

Thanks!

-- 
Yogi

Re: [Xenomai] mlockall and dlopenskins?

[Gilles Chanteperdrix]

On 09/07/2014 06:21 PM, Yogi A. Patel wrote:
> Hi -
> 
> I used to provide --enable-posix-auto-mlockall --enable-dlopen-skins as 
> options to the configure script when compiling user space support. 
> However in 2.6.3 I get a warning saying those are unrecognized options.
> 
> Now. I was looking online at the documentation and found the following:
> 

The title of the section is:

BEFORE VERSION 2.6.3

> /Xenomai libraries only invoked mlockall if a special option was passed 
> to the configure script when compiling Xenomai user-space support. So, 
> applications which did not want to depend on this configuration had to 
> call mlockall by themselves, before using any Xenomai service, by using://
> //
> //mlockall(MCL_CURRENT | MCL_FUTURE);/


The section named:

MLOCKALL AND STACK SIZE

says:
As part of its initialization, Xenomai libraries call mlockall to commit
and lock the whole application memory.

How can that be made more clear?

-- 
                                                                Gilles.

Re: [Xenomai] mlockall and dlopenskins?

[Yogi A. Patel]

On 09/07/2014 12:37 PM, Gilles Chanteperdrix wrote:
> On 09/07/2014 06:21 PM, Yogi A. Patel wrote:
> The section named: MLOCKALL AND STACK SIZE says: As part of its 
> initialization, Xenomai libraries call mlockall to commit and lock the 
> whole application memory. How can that be made more clear? 

It might be clearer if that statement was in or referenced to by the 
section titled "BEFORE VERSION 2.6.3".

What about the --enable-dlopen-skins warning? Are all skins compiled by 
default now? I'm not experiencing any problems so far, but I would 
rather understand what's going on than accept it blindly.

Thanks, Gilles.

Re: [Xenomai] mlockall and dlopenskins?

[Gilles Chanteperdrix]

On 09/07/2014 06:39 PM, Yogi A. Patel wrote:
> On 09/07/2014 12:37 PM, Gilles Chanteperdrix wrote:
>> On 09/07/2014 06:21 PM, Yogi A. Patel wrote:
>> The section named: MLOCKALL AND STACK SIZE says: As part of its 
>> initialization, Xenomai libraries call mlockall to commit and lock the 
>> whole application memory. How can that be made more clear? 
> 
> It might be clearer if that statement was in or referenced to by the 
> section titled "BEFORE VERSION 2.6.3".
> 
> What about the --enable-dlopen-skins warning? Are all skins compiled by 
> default now? I'm not experiencing any problems so far, but I would 
> rather understand what's going on than accept it blindly.

--enable-dlopen-skins was a bad workaround for a stupid problem which
has been fixed in 2.6.3.

-- 
                                                                Gilles.

Re: [Xenomai] mlockall and dlopenskins?

[Yogi A. Patel]

On 09/07/2014 12:41 PM, Gilles Chanteperdrix wrote:
> On 09/07/2014 06:39 PM, Yogi A. Patel wrote:
>> On 09/07/2014 12:37 PM, Gilles Chanteperdrix wrote:
>>> On 09/07/2014 06:21 PM, Yogi A. Patel wrote:
>>> The section named: MLOCKALL AND STACK SIZE says: As part of its
>>> initialization, Xenomai libraries call mlockall to commit and lock the
>>> whole application memory. How can that be made more clear?
>> It might be clearer if that statement was in or referenced to by the
>> section titled "BEFORE VERSION 2.6.3".
>>
>> What about the --enable-dlopen-skins warning? Are all skins compiled by
>> default now? I'm not experiencing any problems so far, but I would
>> rather understand what's going on than accept it blindly.
> --enable-dlopen-skins was a bad workaround for a stupid problem which
> has been fixed in 2.6.3.
So all skins are compiled by default, is that correct?

Re: [Xenomai] mlockall and dlopenskins?

[Gilles Chanteperdrix]

On 09/07/2014 06:39 PM, Yogi A. Patel wrote:
> On 09/07/2014 12:37 PM, Gilles Chanteperdrix wrote:
>> On 09/07/2014 06:21 PM, Yogi A. Patel wrote:
>> The section named: MLOCKALL AND STACK SIZE says: As part of its 
>> initialization, Xenomai libraries call mlockall to commit and lock the 
>> whole application memory. How can that be made more clear? 
> 
> It might be clearer if that statement was in or referenced to by the 
> section titled "BEFORE VERSION 2.6.3".

The documentation is meant to be used the other way around: you read the
mlockall section, and find a link to what happened before 2.6.3.

If you are interested in the exhaustive changes between versions, the
information is available in the git logs and release announcements.

For instance, version 2.6.3 announcement, here:
http://www.xenomai.org/pipermail/xenomai/2013-October/029289.html
says:

Jan Kiszka (29):
      Invoke mlockall on every skin library initialization
      posix: Control auto-shadowing via environment variable
      Remove mlockall alert handler
      Drop --enable-dlopen, disable optimized TLS support by default
      posix: Skip auto-shadowing if current thread is already shadowed

And I find that the title of the section makes it pretty clear that this
information concerns what happened before version 2.6.3.

-- 
                                                                Gilles.

Re: [Xenomai] mlockall and dlopenskins?

[Gilles Chanteperdrix]

On 09/07/2014 06:44 PM, Yogi A. Patel wrote:
> On 09/07/2014 12:41 PM, Gilles Chanteperdrix wrote:
>> On 09/07/2014 06:39 PM, Yogi A. Patel wrote:
>>> On 09/07/2014 12:37 PM, Gilles Chanteperdrix wrote:
>>>> On 09/07/2014 06:21 PM, Yogi A. Patel wrote:
>>>> The section named: MLOCKALL AND STACK SIZE says: As part of its
>>>> initialization, Xenomai libraries call mlockall to commit and lock the
>>>> whole application memory. How can that be made more clear?
>>> It might be clearer if that statement was in or referenced to by the
>>> section titled "BEFORE VERSION 2.6.3".
>>>
>>> What about the --enable-dlopen-skins warning? Are all skins compiled by
>>> default now? I'm not experiencing any problems so far, but I would
>>> rather understand what's going on than accept it blindly.
>> --enable-dlopen-skins was a bad workaround for a stupid problem which
>> has been fixed in 2.6.3.
> So all skins are compiled by default, is that correct?
> 
All skins have always been compiled by default, this has nothing to do
with that option.

-- 
                                                                Gilles.

Re: [Xenomai] mlockall and dlopenskins?

[Yogi A. Patel]

On 09/07/2014 12:46 PM, Gilles Chanteperdrix wrote:
> On 09/07/2014 06:39 PM, Yogi A. Patel wrote:
>> On 09/07/2014 12:37 PM, Gilles Chanteperdrix wrote:
>>> On 09/07/2014 06:21 PM, Yogi A. Patel wrote:
>>> The section named: MLOCKALL AND STACK SIZE says: As part of its
>>> initialization, Xenomai libraries call mlockall to commit and lock the
>>> whole application memory. How can that be made more clear?
>> It might be clearer if that statement was in or referenced to by the
>> section titled "BEFORE VERSION 2.6.3".
> The documentation is meant to be used the other way around: you read the
> mlockall section, and find a link to what happened before 2.6.3.
>
> If you are interested in the exhaustive changes between versions, the
> information is available in the git logs and release announcements.
>
> For instance, version 2.6.3 announcement, here:
> http://www.xenomai.org/pipermail/xenomai/2013-October/029289.html
> says:
>
> Jan Kiszka (29):
>        Invoke mlockall on every skin library initialization
>        posix: Control auto-shadowing via environment variable
>        Remove mlockall alert handler
>        Drop --enable-dlopen, disable optimized TLS support by default
>        posix: Skip auto-shadowing if current thread is already shadowed
>
> And I find that the title of the section makes it pretty clear that this
> information concerns what happened before version 2.6.3.
Thanks for the info and pointing me to the resources, Gilles,