From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from nausicaa2.coritel.it (host254-130-static.190-82-b.business.telecomitalia.it [82.190.130.254]) by ozlabs.org (Postfix) with ESMTP id F166BDDDF9 for ; Sat, 2 Feb 2008 03:19:20 +1100 (EST) Message-ID: <47A3473D.6040505@coritel.it> Date: Fri, 01 Feb 2008 17:22:21 +0100 From: Marco Stornelli MIME-Version: 1.0 To: Sergei Shtylyov Subject: Re: External Interrupt References: <47A2D424.3090903@coritel.it> <47A32C9C.3090709@ru.mvista.com> <47A3379F.2090004@coritel.it> <47A33C59.4020509@ru.mvista.com> In-Reply-To: <47A33C59.4020509@ru.mvista.com> Content-Type: text/plain; charset=us-ascii Cc: Linuxppc Embedded Mailing List List-Id: Linux on Embedded PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sergei Shtylyov ha scritto: > Marco Stornelli wrote: > >>>> I used the linux kernel 2.6.10 with a processor MPC8548E. I wrote a >>>> driver for a device connected with the local bus. This device has an >>>> external interrupt. In the local bus driver I have used the macro >>>> MPC85xx_IRQ_EXT to get the interrupt number and pass it to the >>>> driver and after that register the ISR. Now with a kernel 2.6.21 this >>>> macro isn't available because in the header file irq.h there is the >>>> option CONFIG_PPC_MERGE that disable those options. I think this >>>> problem is related to the migration of ppc code towards powerpc. I >>>> know that now there is the new device tree source file where I can add >>>> a device and its interrupt number but I think in this file I should >>>> describe only the platform device, and this device is not a platform >>>> device. > >>> How comes that it's not platform device if it hangs off the local bus? > >> In SoC system generally the platform device are (more or less) the >> microprocessor controller like i2c, pci, local bus itself and so on. > > That's SoC devices but the notion of the platform device is not limited to > SoC device only, rather to all the on-board devices. > >> I think is like if you consider a PCI board a platform device only because >> it hangs off the PCI link. > > No. PCI devices are detectable/configurable by kernel -- even if they are > on-board chips, they can be found by PCI bus scan (and finally presented as > the device nodes as well), while local bus devices (most probably) not. An > example (that comes to mind) of a device hanging off the local bus and yet > described by the device tree are the flash chips. Yes you are right. The local bus is like i2c, but I've never seen a device connected with i2c and described with a sub-node of i2c node in the dts file, however I think it's only philosophy, the most important thing is that there is a way to get the irq number :) Thanks for your response. Regards. Marco > > WBR, Sergei > > _______________________________________________ > Linuxppc-embedded mailing list > Linuxppc-embedded@ozlabs.org > https://ozlabs.org/mailman/listinfo/linuxppc-embedded >