From mboxrd@z Thu Jan  1 00:00:00 1970
From: horms@verge.net.au (Simon Horman)
Date: Thu, 17 Dec 2009 11:06:20 +1100
Subject: [Fwd: [PATCH] PCI Domains Support]
In-Reply-To: <4B28EB0F.3070409@ms1.techarea.org>
References: <4B28EB0F.3070409@ms1.techarea.org>
Message-ID: <20091217000620.GC22449@verge.net.au>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On Wed, Dec 16, 2009 at 10:13:35PM +0800, Richard Liu wrote:
> Hi, all:
> 
>     Seems ARM don's support PCI Domains function now.
> Here is a tiny patch for PCI Domains support
> 
>     It's work on my platform,
> There are two PCIe host controller on my platform, and both bus number
> should be 0
> So, use PCI Domain to control both PCIe host controller is better choice.
> 
> Here is platform PCIe sample code,
> static struct hw_pci cxxxxxx_pcie0 __initdata = {
>         .swizzle = pci_std_swizzle,
>         .map_irq = cxxxxxx_pcie0_map_irq,
>         .nr_controllers = 1,
>         .nr_domains = 0,
>         .setup = cxxxxxx_pci_setup,
>         .scan = cxxxxxx_pci_scan_bus,
>         .preinit = cxxxxxx_pci_preinit,
>         .postinit = cxxxxxx_pcie0_postinit,
> };
> static struct hw_pci cxxxxxx_pcie1 __initdata = {
>         .swizzle = pci_std_swizzle,
>         .map_irq = cxxxxxx_pcie1_map_irq,
>         .nr_controllers = 1,
>         .nr_domains = 1,
>         .setup = cxxxxxx_pci_setup,
>         .scan = cxxxxxx_pci_scan_bus,
>         .postinit = cxxxxxx_pcie1_postinit,
> };
> 
> static int __init cxxxxxx_pci_init(void)
> {
>         pci_common_init(&cxxxxxx_pcie0);
>         pci_common_init(&cxxxxxx_pcie1);
>         return 0;
> }
> 
> 
> --
> diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
> index 79a3074..33a49ba 100644
> --- a/arch/arm/Kconfig
> +++ b/arch/arm/Kconfig
> @@ -865,6 +865,10 @@ config PCI
>        your box. Other bus systems are ISA, EISA, MicroChannel (MCA) or
>        VESA. If you have PCI, say Y, otherwise N.
>  
> +config PCI_DOMAINS
> +        def_bool y
> +        depends on PCI
> +

The PCI_DOMAINS config variable seems unnecessary to me.

>  config PCI_SYSCALL
>      def_bool PCI
>  
> diff --git a/arch/arm/include/asm/mach/pci.h
> b/arch/arm/include/asm/mach/pci.h
> index a38bdc7..f099067 100644
> --- a/arch/arm/include/asm/mach/pci.h
> +++ b/arch/arm/include/asm/mach/pci.h
> @@ -20,6 +20,9 @@ struct hw_pci {
>      void        (*postinit)(void);
>      u8        (*swizzle)(struct pci_dev *dev, u8 *pin);
>      int        (*map_irq)(struct pci_dev *dev, u8 slot, u8 pin);
> +#ifdef CONFIG_PCI_DOMAINS
> +    int        nr_domains;
> +#endif
>  };
>  
>  /*
> @@ -37,8 +40,12 @@ struct pci_sys_data {
>                      /* IRQ mapping                */
>      int        (*map_irq)(struct pci_dev *, u8, u8);
>      struct hw_pci    *hw;
> +#ifdef CONFIG_PCI_DOMAINS
> +    int        domain;
> +#endif
>  };
>  
> +
>  /*
>   * This is the standard PCI-PCI bridge swizzling algorithm.
>   */
> diff --git a/arch/arm/include/asm/pci.h b/arch/arm/include/asm/pci.h
> index 0abf386..57ce5bd 100644
> --- a/arch/arm/include/asm/pci.h
> +++ b/arch/arm/include/asm/pci.h
> @@ -25,6 +25,11 @@ static inline void pcibios_penalize_isa_irq(int irq,
> int active)
>      /* We don't do dynamic PCI IRQ allocation */
>  }
>  
> +#ifdef CONFIG_PCI_DOMAINS
> +int pci_domain_nr(struct pci_bus *bus);
> +int pci_proc_domain(struct pci_bus *bus);
> +#endif
> +
>  /*
>   * The PCI address space does equal the physical memory address space.
>   * The networking and block device layers use this boolean for bounce
> diff --git a/arch/arm/kernel/bios32.c b/arch/arm/kernel/bios32.c
> index 8096819..b364ca8 100644
> --- a/arch/arm/kernel/bios32.c
> +++ b/arch/arm/kernel/bios32.c
> @@ -531,6 +531,7 @@ static void __init pcibios_init_hw(struct hw_pci *hw)
>          sys->busnr   = busnr;
>          sys->swizzle = hw->swizzle;
>          sys->map_irq = hw->map_irq;
> +        sys->domain  = hw->nr_domains;
>          sys->resource[0] = &ioport_resource;
>          sys->resource[1] = &iomem_resource;
>  
> @@ -694,3 +695,20 @@ int pci_mmap_page_range(struct pci_dev *dev, struct
> vm_area_struct *vma,
>  
>      return 0;
>  }
> +#ifdef CONFIG_PCI_DOMAINS
> +int pci_domain_nr(struct pci_bus *bus)
> +{
> +
> +        //struct pci_sysdata *sd = bus->sysdata;

	Should the line above should be removed?
	It seems to be noise.

> +        struct pci_sys_data *sd = bus->sysdata;
> +        return sd->domain;
> +
> +}
> +EXPORT_SYMBOL(pci_domain_nr);
> +
> +int pci_proc_domain(struct pci_bus *bus)
> +{
> +        return pci_domain_nr(bus);
> +}
> +EXPORT_SYMBOL(pci_proc_domain);
> +#endif
> 
> 
> 
> 
> 
> _______________________________________________
> linux-arm-kernel mailing list
> linux-arm-kernel at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel