Hi Jac,

Le Thursday 26 March 2009 16:37:53 Jac Goudsmit, vous avez écrit :
> There's another problem that the RDC GPIO code appears to be a little
> device-specific: I may be wrong but it looks like the current code
> specifies a DMZ LED on the GPIO and nothing else. This may make sense for
> the Airlink and for the other RDC-based routers that are in the tree and
> happen to have an LED there, but for the Linksys NAS200 (which I'm working
> on) the GPIO situation is of course completely different.

Right, the code is very device-centric for now but it is a bit hard to runtime 
detect RDC boards, at least I did not came up with any other idea than 
probing for different flash sizes and the presence of some PCI devices.

>
> By the way, I managed to get a Gentoo (2.6.27-r8) kernel to run on the
> NAS200 (see http://www.linksysinfo.org/forums/showthread.php?t=60923).
> Evidently it's possible to bring up a mainline kernel (albeit with Gentoo
> patches) on an RDC; I just haven't been able to build a working one from
> OpenWRT (the SATA controller isn't recognized). But that's just a matter of
> time: I'm trying hard to find some time to learn the build system.
>
> To me, the GPIO on the RDC is less important for now, but it will become
> more important once I get OpenWRT to run. So any work in that corner of the
> source tree will be appreciated. :-)

I will work on it soon, now that I got an RDC eval board it is much easier.

>
> ===Jac
>
> --- On Thu, 3/26/09, Florian Fainelli <[email protected]> wrote:
> > From: Florian Fainelli <[email protected]>
> > Subject: Re: [OpenWrt-Devel] RDC GPIO broken?
> > To: "bifferos" <[email protected]>
> > Cc: [email protected]
> > Date: Thursday, March 26, 2009, 4:24 AM
> >
> > -----Inline Attachment Follows-----
> >
> > Le Thursday 26 March 2009 12:19:50
> >
> > bifferos, vous avez écrit :
> > > OK, thanks for clarifying that.  I guess this
> >
> > means you will not be
> >
> > > implementing RDC GPIOLIB support any time soon. 
> >
> > I will look into this
> >
> > > if others are interested.
> >
> > Getting GPIOs to work is not my primary concern since right
> > now, the mainline
> > kernel does not probably boot correctly on any RDC 321x SoC
> > due to no way to
> > set the tick rate properly with everything that it implies
> > on the other
> > peripherals.
> >
> > GPIOlib or something else, I do not really care as long as
> > it works :)
> >
> > > --- On Thu, 26/3/09, Florian Fainelli <[email protected]>
> >
> > wrote:
> > > > From: Florian Fainelli <[email protected]>
> > > > Subject: Re: [OpenWrt-Devel] RDC GPIO broken?
> > > > To: [email protected]
> > > > Cc: "bifferos" <[email protected]>
> > > > Date: Thursday, 26 March, 2009, 9:08 AM
> > > >
> > > > -----Inline Attachment Follows-----
> > > >
> > > > Hi Simon,
> > > >
> > > > Le Thursday 26 March 2009 01:10:25 bifferos, vous
> >
> > avez
> >
> > > > écrit :
> > > > > I'm not very familiar with the kernel
> >
> > Kconfig/makefile
> >
> > > > layout, but it seems
> > > >
> > > > > the mach-rdc321x stuff, namely gpio.c and
> >
> > platform.c
> >
> > > > in
> > > >
> > > > > arch/x86/mach-rdc321x/ don't get compiled. 
> >
> > I
> >
> > > > couldn't work out if there
> > > >
> > > > > was some kernel option needed to enable
> >
> > them, but
> >
> > > > judging from the other
> > > >
> > > > > x86 architectures I think the above patch is
> >
> > what's
> >
> > > > needed.
> > > >
> > > > > I'm baffled as to why this would be missing
> >
> > - it seems
> >
> > > > this is broken all
> > > >
> > > > > the way from 2.6.27.5 -> 2.6.28.8.  I
> >
> > guess
> >
> > > > nobody uses GPIO on RDC
> > > >
> > > > > platforms?
> > > >
> > > > The x86 maintainers wanted me to convert RDC to
> >
> > the x86
> >
> > > > generic platform which
> > > > I could not do in time, therefore I agreed on
> >
> > letting it
> >
> > > > that way and not
> > > > compile the platform code. At the time they asked
> >
> > me to do
> >
> > > > this I only had an
> > > > Airlink A525W which has a stupid limitation on
> >
> > the kernel
> >
> > > > size and I could
> > > > not get a 2.6.25 kernel to boot on it.
> > > >
> > > > So no, this is not what was expected, but
> >
> > provided that I
> >
> > > > did not have time to
> > > > fix this, this is what happened. Feel free to
> >
> > submit the
> >
> > > > necessary patches
> > > > mainline to avoid maintaining local patches any
> >
> > longer.
> >
> > > > Small hints :
> > > >
> > > > - RDC has a different PIT_TICK_RATE value than
> >
> > what the x86
> >
> > > > generic code
> > > > expects, there is no infrastructure to runtime
> >
> > change the
> >
> > > > tick rate
> > > > - runtime detection should be done reading the
> >
> > presence of
> >
> > > > a R6040 Ethernet
> > > > MAC on the PCI bus since RDC provides no CPUID
> >
> > thus is
> >
> > > > always detected as an
> > > > unknown 486 CPU
> > > > --
> > > > Best regards, Florian Fainelli
> > > > Email : [email protected]
> > > > http://openwrt.org
> > > > -------------------------------
> >
> > --
> > Best regards, Florian Fainelli
> > Email : [email protected]
> > http://openwrt.org
> > -------------------------------
> >
> > -----Inline Attachment Follows-----
> >
> > _______________________________________________
> > openwrt-devel mailing list
> > [email protected]
> > http://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
>
> _______________________________________________
> openwrt-devel mailing list
> [email protected]
> http://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel



-- 
Best regards, Florian Fainelli
Email : [email protected]
http://openwrt.org
-------------------------------

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
openwrt-devel mailing list
[email protected]
http://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel

Reply via email to