Iterating through all the processes in a module

Iterating through all the processes in a module

[Arokux B.]

Hi,

(for learning purposes) I would like to iterate through all the tasks
in a module and output different information about them. For this task
I need to lock the list of all tasks (need I?). I've seen some example
in the kernel code which lock tasklist_lock. However this symbol
cannot be used by modules. Its export was removed by
c59923a15c12d2b3597af913bf234a0ef264a38b commit.

Is there any other way I can lock the list of tasks then?

Thanks

Iterating through all the processes in a module

[Greg KH]

On Tue, Feb 07, 2012 at 06:45:24PM +0100, Arokux B. wrote:
> Hi,
> 
> (for learning purposes) I would like to iterate through all the tasks
> in a module and output different information about them. For this task
> I need to lock the list of all tasks (need I?). I've seen some example
> in the kernel code which lock tasklist_lock. However this symbol
> cannot be used by modules. Its export was removed by
> c59923a15c12d2b3597af913bf234a0ef264a38b commit.
> 
> Is there any other way I can lock the list of tasks then?

Don't do such a foolish thing?  :)

Seriously, don't make your code a module, just build it into the kernel.

greg k-h

Iterating through all the processes in a module

[Arokux B.]

Dear Greg,

thank you very much for your quick reply.

Having my code as a module I can trigger its execution (load a module)
and disable it (unload a module). How can I achieve this if the code
is inside the kernel? One possibility I see is adding an entry in the
procfs.

Regards

On Tue, Feb 7, 2012 at 8:17 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Tue, Feb 07, 2012 at 06:45:24PM +0100, Arokux B. wrote:
>> Hi,
>>
>> (for learning purposes) I would like to iterate through all the tasks
>> in a module and output different information about them. For this task
>> I need to lock the list of all tasks (need I?). I've seen some example
>> in the kernel code which lock tasklist_lock. However this symbol
>> cannot be used by modules. Its export was removed by
>> c59923a15c12d2b3597af913bf234a0ef264a38b commit.
>>
>> Is there any other way I can lock the list of tasks then?
>
> Don't do such a foolish thing? ?:)
>
> Seriously, don't make your code a module, just build it into the kernel.
>
> greg k-h

Iterating through all the processes in a module

[Greg KH]

A: No.
Q: Should I include quotations after my reply?

http://daringfireball.net/2007/07/on_top

On Wed, Feb 22, 2012 at 02:07:22PM +0100, Arokux B. wrote:
> Dear Greg,
> 
> thank you very much for your quick reply.
> 
> Having my code as a module I can trigger its execution (load a module)
> and disable it (unload a module). How can I achieve this if the code
> is inside the kernel? One possibility I see is adding an entry in the
> procfs.

I don't see what procfs has to do with this, please explain.

And as you said this was for "learning purposes", then just build it
into the kernel, there's no need for it to be a module.

good luck,

greg k-h

Iterating through all the processes in a module

[Srivatsa Bhat]

On Wed, Feb 22, 2012 at 8:23 PM, Greg KH <greg@kroah.com> wrote:

>
> A: No.
> Q: Should I include quotations after my reply?
>
> http://daringfireball.net/2007/07/on_top
>
> On Wed, Feb 22, 2012 at 02:07:22PM +0100, Arokux B. wrote:
> > Dear Greg,
> >
> > thank you very much for your quick reply.
> >
> > Having my code as a module I can trigger its execution (load a module)
> > and disable it (unload a module). How can I achieve this if the code
> > is inside the kernel? One possibility I see is adding an entry in the
> > procfs.
>
> I don't see what procfs has to do with this, please explain.
>
> And as you said this was for "learning purposes", then just build it
> into the kernel, there's no need for it to be a module.
>
>
Heh, you can just put back the relevant parts of what commit
c59923a15c12d2b3597af913bf234a0ef264a38b removed and then do what
you want as a module, if it is for learning purposes :-)

Regards,
Srivatsa S. Bhat
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20120223/52b52959/attachment.html 

Iterating through all the processes in a module

[Ezequiel García]

Hi Arokux,

On 2/22/12, Arokux B. <arokux@gmail.com> wrote:
> Having my code as a module I can trigger its execution (load a module)
> and disable it (unload a module). How can I achieve this if the code
> is inside the kernel? One possibility I see is adding an entry in the
> procfs.

I think you want to have the kernel execute some code of yours just for learning
purposes right?
So, your question could be put like this:
Where are the "doors" for me -userspace- to enter to the kernel?

In that case, I believe the "canonical" answer is: system calls.
If you have a copy of Love's book you could look there, or you could
just search through the
code and look how system calls are implemented and add own of your own.
Perhaps you could even modify one. You can search all syscalls definitions with:

grep "SYSCALL_DEFINE" `find . -name "*.c"`

Actually, that's the path I would take if I were to try this. You can
pick a simple syscall
like "getcwd", defined in fs/dcache.c and just add my test code there. That way
every time you call pwd in your shell, you get your code called (you
can check with
strace).

Hope this helps,
Ezequiel.

Iterating through all the processes in a module

[Javier Martinez Canillas]

2012/2/22 Ezequiel Garc?a <elezegarcia@gmail.com>:
> Hi Arokux,
>
> On 2/22/12, Arokux B. <arokux@gmail.com> wrote:
>> Having my code as a module I can trigger its execution (load a module)
>> and disable it (unload a module). How can I achieve this if the code
>> is inside the kernel? One possibility I see is adding an entry in the
>> procfs.
>
> I think you want to have the kernel execute some code of yours just for learning
> purposes right?
> So, your question could be put like this:
> Where are the "doors" for me -userspace- to enter to the kernel?
>
> In that case, I believe the "canonical" answer is: system calls.
> If you have a copy of Love's book you could look there, or you could
> just search through the
> code and look how system calls are implemented and add own of your own.
> Perhaps you could even modify one. You can search all syscalls definitions with:
>
> grep "SYSCALL_DEFINE" `find . -name "*.c"`
>
> Actually, that's the path I would take if I were to try this. You can
> pick a simple syscall
> like "getcwd", defined in fs/dcache.c and just add my test code there. That way
> every time you call pwd in your shell, you get your code called (you
> can check with
> strace).
>
> Hope this helps,
> Ezequiel.
>

This is also a good summary of Kernel space / User space interfaces

http://people.ee.ethz.ch/~arkeller/linux/kernel_user_space_howto.html

Hope it helps,

-- 
Javier Mart?nez Canillas
(+34) 682 39 81 69
Barcelona, Spain

Iterating through all the processes in a module

[Arokux B.]

Hi all,

thank you very much guys. Your answers helped me to realize what I
really wanted to do.

Best regards,
Arokux