Thanks a lot for your explanation!
I think if this is an area for improvement..
 Maybe the modules itself are not structures "isolated" enough to
provide the load/unload features we want. And, like you said, seems to
be a really complicated procedure to do,  isolate a module, and give
it the features needed to handle the communications (needed too) with
the other kernel modules.
 I think the possibility to load/unload a kernel module "on the fly",
was a desired feature when the modules were created in the first
place. But seems like that feature is not "reality" in today's
environment. I mean, the complexity today, create almost the same
problem with the "modular" kernel, as the older one had.
Is that right?

 Leal.

2008/5/20 James Carlson <[EMAIL PROTECTED]>:
> Marcelo Leal writes:
>>  I just want to ask you about the installation of kernel patches on 
>> opensolaris/solaris...
>>  opensolaris(solaris) is the more modular kernel that i know about, and i 
>> think that is a really good thing (thinking about all the features that such 
>> configuration has, like updates). But, seems like the procedure to apply a 
>> update without shutdown the server, is not the "standard" way, because 
>> install a solaris kernel patch is much like install a windows service pack 
>> (you need to reboot the server and cross the fingers).
>>  I think opensolaris have not kernel updates, because here is the project 
>> site, and i don't know how the updates will be handled by OpenSolaris 
>> distribution, but here we have the engineering for the product... so, i ask: 
>> the modular kernel should not be for make more easy the modules' fix?
>
> You've asked a somewhat complicated question.
>
> First, yes, the kernel is modular.  And it certainly is possible to
> unload a kernel module and insert a replacement on a running system
> without rebooting.
>
> Not all modules can do this, though.  In order to unload a kernel
> module, you have to be able to shut down all references to that module
> -- delete all internal state, close all open descriptors, disable all
> callbacks and timers, and so on.  How you do that depends entirely on
> the module itself; there is no one mechanism.  Some modules are
> written such that their _fini(9E) entry point aways returns an error;
> they can never be unloaded, because the designer of the module didn't
> figure out how to make it unloadable.
>
> The other part of your question has to do with patches.  Patches are
> collections of sparse packages.  They are crafted as part of the
> update creation process.  The normal development process, though, does
> *NOT* create patches.  There are no patches for Nevada or
> OpenSolaris.  Instead, new full packages are built each time, and you
> must upgrade between builds.
>
> --
> James Carlson, Solaris Networking              <[EMAIL PROTECTED]>
> Sun Microsystems / 35 Network Drive        71.232W   Vox +1 781 442 2084
> MS UBUR02-212 / Burlington MA 01803-2757   42.496N   Fax +1 781 442 1677
>
>



-- 
pOSix rules
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code

Reply via email to