[PATCH] ARM/PCI: move align_resource function pointer into pci_host_bridge structure

Gabriele Paoloni gabriele.paoloni at huawei.com
Wed Nov 11 02:09:45 PST 2015


> -----Original Message-----
> From: Arnd Bergmann [mailto:arnd at arndb.de]
> Sent: Wednesday, November 11, 2015 4:48 PM
> To: linux-arm-kernel at lists.infradead.org
> Cc: Gabriele Paoloni; linux-pci at vger.kernel.org;
> lorenzo.pieralisi at arm.com; linux at arm.linux.org.uk; xuwei (O); Wangzhou
> (B); bhelgaas at google.com; Liguozhu (Kenneth)
> Subject: Re: [PATCH] ARM/PCI: move align_resource function pointer into
> pci_host_bridge structure
> 
> On Wednesday 11 November 2015 09:12:25 Gabriele Paoloni wrote:
> > From: gabpao01 <gabriele.paoloni at huawei.com>
> 
> Regardless of this patch, you might want to update your global git
> config
> to use the proper name for your commits.

Sorry, I am travelling and moved to another PC...I didn't notice
the wrong name, my bad.

> 
> > commit b3a72384fe29 ("ARM/PCI: Replace
> > pci_sys_data->align_resource with global function pointer") has
> > introduced a global function pointer that makes the ARM specific
> > code not portable and broken in case any platform have a two
> > different HW IPs for the PCIe host bridge controller.
> > This patch moves align_resource function pointer into
> > pci_host_bridge structure so that the code is now suitable to be
> > reworked as we want to get rid of hw_pci structure (the host bridge
> > drivers can just set the align function pointer in the
> > pci_host_bridge structure) and there is no more broken code for a
> > SoC with two HW IPs.
> >
> > Signed-off-by: Gabriele Paoloni <gabriele.paoloni at huawei.com>
> 
> Thanks for doing this, that addresses my main concerns, and is probably
> small enough to still get it into 4.4 if Bjorn agrees with the general
> idea of putting more function pointers into struct pci_host_bridge.
> 
> As a follow-on to this, we can now try to get rid of
> pcibios_align_resource
> by moving the existing per-architecture implementations in there as
> well,
> with a reasonable default for the common case.
> 
> 	Arnd



More information about the linux-arm-kernel mailing list