Re: SS_CAP_STATIC_MAP and I/O resources manipulation in ide-cs and PCMCIA core

Russell King <[email protected]> Thu, 13 Mar 2003 18:58:20 +0000
Newsgroups gmane.linux.pcmcia.devel
Message-ID <[email protected]>
On Thu, Mar 13, 2003 at 10:44:19AM -0800, David Hinds wrote:
> On Thu, Mar 13, 2003 at 05:31:51PM +0000, Russell King wrote:
> > Linus has also expressed an interest in making PCMCIA allocate resources
> > for regions it uses, but not marking them busy.
> 
> How do you do that?  I argued with Linus about that a few years ago
> but he did not want the resource management system to have that
> capability.  Has it been added??

It's been in there since 2.3.xx, used extensively by PCI.

> For non-static controllers, ide-cs jumps through some hoops to make
> resource management seem to mostly work.  The PCMCIA core wants to
> manage resources because only it knows when resources are actually
> committed to a piece of hardware; anything else invites conflicts with
> any other subsystem that needs to dynamically allocate resources.  But
> the IDE driver also wants to do conflict checking and claim resources
> for itself.  To resolve the conflicting claims, the core allocates the
> resource, and ide-cs frees it just prior to calling register_ide() so
> that the IDE driver won't complain.  Then, when the card is released,
> ide-cs allocates the resource back so that the PCMCIA core can free
> it.

This sounds similar to the way serial handles this at present as
well.

> Yes this sucks, but a simple alternative doesn't seem to present
> itself.  We could modify the IDE driver so that when a device is
> registered at a specific location, it skips its checks for resource
> conflicts.  Or add a flag so that the ide-cs driver can tell the core
> to not actually register resources with the kernel in this case.

At present, I'm not that concerned about 2.4 tbh.  Yes, this sounds like
a hard thing to say, but my 2.4 work has already dropped behind where it
should be.  Either myself or Dominik will eventually get around to
sorting this out in 2.5; I suspect it'll be me, since I have the
statically mapped socket hardware here.

-- 
Russell King ([email protected])                The developer of ARM Linux
             http://www.arm.linux.org.uk/personal/aboutme.html



-------------------------------------------------------
This SF.net email is sponsored by:Crypto Challenge is now open! 
Get cracking and register here for some mind boggling fun and 
the chance of winning an Apple iPod:
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en