qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Qemu-devel] Re: [RFC][PATCH] PCI: fix pci_to_cpu_addr() issue


From: Isaku Yamahata
Subject: [Qemu-devel] Re: [RFC][PATCH] PCI: fix pci_to_cpu_addr() issue
Date: Wed, 30 Jun 2010 22:38:11 +0900
User-agent: Mutt/1.5.19 (2009-01-05)

Can you elaborate on how pci bus is mapped into local bus?
Is there specification publicly available? Google didn't tell me.


On Wed, Jun 30, 2010 at 06:39:53PM +0800, Huacai Chen wrote:
> It seems like software may both use CPU address or PCI address to access a PCI
> device. For example, Bonito north bridge map PCI memory space at 0x10000000 ~ 
> 0x1C000000. PMON code use 0x00000000 ~ 0x0C000000, but Linux kernel code use 
> 0x10000000 ~ 0x1C000000 to access devices. If set pci_mem_base to 0, PMON 
> can't
> work, but if set pci_mem_base to 0x10000000, Linux can't access PCI. So I make
> this patch to make both cases works.
> 
> However, I don't know whether the modification will break other archs, so
> request for comments here.
> 
> Signed-off-by: Huacai Chen <address@hidden>
> ---
>  hw/pci.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/hw/pci.c b/hw/pci.c
> index 7787005..50e3572 100644
> --- a/hw/pci.c
> +++ b/hw/pci.c
> @@ -672,7 +672,7 @@ PCIDevice *pci_register_device(PCIBus *bus, const char 
> *name,
>  static target_phys_addr_t pci_to_cpu_addr(PCIBus *bus,
>                                            target_phys_addr_t addr)
>  {
> -    return addr + bus->mem_base;
> +    return addr | bus->mem_base;
>  }
>  
>  static void pci_unregister_io_regions(PCIDevice *pci_dev)
> -- 
> 1.7.0.4
> 

-- 
yamahata



reply via email to

[Prev in Thread] Current Thread [Next in Thread]