On Mon, 2006-04-17 at 19:11 -0700, Matt Helsley wrote:
> On Tue, 2006-04-18 at 10:28 +0900, MAEDA Naoaki wrote:
> > Matt Helsley wrote:
> > > On Mon, 2006-04-17 at 11:47 -0700, Chandra Seetharaman wrote:
> > >> The reason i made it a bool is we merged ckrm_iface to rcfs, and
> > >> ckrm_iface was using put_task_struct, which is not an exported symbol.
> > >>
> > >> May be we can add a new function to core, that takes a pid, instead of a
> > >> task data struture, to set the class that does the get/put_task_struct.
> > >>
> > >> Comments anybody ?
> > >
> > > I would rather see the options merged than add a function that takes a
> > > pid. I don't think merging RCFS with the CKRM option adds much (if any)
> > > overhead in the case that CKRM is unused, and keeping the interface as a
> > > separate option doesn't seem necessary. Also, RCFS will only appear when
> > > configfs gets mounted anyway.
IMO, object of building something as a module is not directly tied with
whether the feature is visible to the user or not, but to bring it in
only the user desires so. It makes sense if you consider distro releases
(they don't want to build everything into their generic kernel).
> >
> > That makes sense. But a flaw of this approach is it also forces configfs
> > statically linked to a vmlinux. Do you think it is acceptable?
I think, if any features _can_ be made available as a module, then they
should. And we should not be forcing some other stuff that we depend on
to be compiled in, when we certainly have the option of making ours also
a module.
> >
> > Thanks,
> > MAEDA Naoaki
>
> Ah. Right -- so it would still make sense to be able to build configfs
> and rcfs as a module. However, that said I still think there should only
> be one CONFIG option -- CONFIG_CKRM. However, make it a tristate and, if
> it's =m build rcfs as a module. If it's =y then staticly-link rcfs and
> configfs into the kernel. The point being that in either case rcfs
> should be built.
You mean compile-in rest of the ckrm and build rcfs as a module ? Does
the kernel build process allow that ?
If we are going to make rcfs to be available as a module, then we do
have to move put_task_struct() to the compiled-in part.
>
> Cheers,
> -Matt Helsley
>
--
----------------------------------------------------------------------
Chandra Seetharaman | Be careful what you choose....
- [EMAIL PROTECTED] | .......you may get it.
----------------------------------------------------------------------
-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
ckrm-tech mailing list
https://lists.sourceforge.net/lists/listinfo/ckrm-tech
