Re: Regression caused by "PCI: Shrink decoding-disabled window while sizing BARs"

[Thomas Petazzoni <thomas.petazzoni@free-electrons.com>]

Hello,

On Tue, 18 Nov 2014 10:58:43 +0100, Thomas Petazzoni wrote:

> Another issue (possibly related, or not), is the following compile time
> warning:
> 
> /home/thomas/projets/linux-2.6/drivers/pci/probe.c: In function ‘pci_read_bridge_mmio_pref’:
> /home/thomas/projets/linux-2.6/drivers/pci/probe.c:430:5: warning: left shift count >= width of type
>      base |= ((dma_addr_t) mem_base_hi) << 32;
>      ^
> /home/thomas/projets/linux-2.6/drivers/pci/probe.c:431:5: warning: left shift count >= width of type
>      limit |= ((dma_addr_t) mem_limit_hi) << 32;
>      ^
> 
> This happens because dma_addr_t is 32 bits on ARM platforms, which
> isn't properly taken into account by the commit that introduces support
> for 64 bits DMA addresses.

Not sure about this one: the commit introducing the issue is in
next-20141117, but not in the current pci/next of Bjorn. Was it
reverted? Am I looking at the wrong place?

Best regards,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com