Using write-combine is crucial for performance of PCI devices where
significant amounts of transactions go over PCI BARs.
arm64 supports write-combine PCI mappings, so the appropriate define
has been added which will expose write-combine mappings under sysfs
for prefetchable PCI resources.
Signed-off-by: Clint Sbisa <[email protected]>
---
arch/arm64/include/asm/pci.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/arm64/include/asm/pci.h b/arch/arm64/include/asm/pci.h
index 70b323cf8300..b33ca260e3c9 100644
--- a/arch/arm64/include/asm/pci.h
+++ b/arch/arm64/include/asm/pci.h
@@ -17,6 +17,7 @@
#define pcibios_assign_all_busses() \
(pci_has_flag(PCI_REASSIGN_ALL_BUS))
+#define arch_can_pci_mmap_wc() 1
#define ARCH_GENERIC_PCI_MMAP_RESOURCE 1
extern int isa_dma_bridge_buggy;
--
2.23.3
On Fri, Aug 21, 2020 at 03:51:54PM +0000, Clint Sbisa wrote:
> Using write-combine is crucial for performance of PCI devices where
> significant amounts of transactions go over PCI BARs.
>
> arm64 supports write-combine PCI mappings, so the appropriate define
> has been added which will expose write-combine mappings under sysfs
> for prefetchable PCI resources.
>
> Signed-off-by: Clint Sbisa <[email protected]>
> ---
> arch/arm64/include/asm/pci.h | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/arch/arm64/include/asm/pci.h b/arch/arm64/include/asm/pci.h
> index 70b323cf8300..b33ca260e3c9 100644
> --- a/arch/arm64/include/asm/pci.h
> +++ b/arch/arm64/include/asm/pci.h
> @@ -17,6 +17,7 @@
> #define pcibios_assign_all_busses() \
> (pci_has_flag(PCI_REASSIGN_ALL_BUS))
>
> +#define arch_can_pci_mmap_wc() 1
> #define ARCH_GENERIC_PCI_MMAP_RESOURCE 1
>
> extern int isa_dma_bridge_buggy;
> --
> 2.23.3
>
Ping. Let me know if any more info is needed.
Clint
On Fri, Aug 21, 2020 at 03:51:54PM +0000, Clint Sbisa wrote:
> Using write-combine is crucial for performance of PCI devices where
> significant amounts of transactions go over PCI BARs.
>
> arm64 supports write-combine PCI mappings, so the appropriate define
> has been added which will expose write-combine mappings under sysfs
> for prefetchable PCI resources.
>
> Signed-off-by: Clint Sbisa <[email protected]>
> ---
> arch/arm64/include/asm/pci.h | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/arch/arm64/include/asm/pci.h b/arch/arm64/include/asm/pci.h
> index 70b323cf8300..b33ca260e3c9 100644
> --- a/arch/arm64/include/asm/pci.h
> +++ b/arch/arm64/include/asm/pci.h
> @@ -17,6 +17,7 @@
> #define pcibios_assign_all_busses() \
> (pci_has_flag(PCI_REASSIGN_ALL_BUS))
>
> +#define arch_can_pci_mmap_wc() 1
> #define ARCH_GENERIC_PCI_MMAP_RESOURCE 1
>
> extern int isa_dma_bridge_buggy;
> --
> 2.23.3
>
Please disregard this submission, I'm resubmitting it to more appropriate
maintainers than what was suggested by the get_maintainer script.
Clint
On Tue, 2020-09-01 at 13:37 -0500, Bjorn Helgaas wrote:
> On Mon, Aug 31, 2020 at 03:18:27PM +0000, Clint Sbisa wrote:
> > Using write-combine is crucial for performance of PCI devices where
> > significant amounts of transactions go over PCI BARs.
> >
> > arm64 supports write-combine PCI mappings, so the appropriate
> > define
> > has been added which will expose write-combine mappings under sysfs
> > for prefetchable PCI resources.
> >
> > Signed-off-by: Clint Sbisa <[email protected]>
>
> Fine with me, I assume Will or Catalin will apply this.
Haha ! Client had sent it to them originally and I told him to resend
it to linux-pci, yourself and Lorenzo :-)
So the confusion is on me.
Will, Catalin, it's all yours. You should have the original patch in
your mbox already, otherwise:
https://patchwork.kernel.org/patch/11729875/
Cheers,
Ben.
On Wed, Sep 02, 2020 at 09:22:53AM +1000, Benjamin Herrenschmidt wrote:
> On Tue, 2020-09-01 at 13:37 -0500, Bjorn Helgaas wrote:
> > On Mon, Aug 31, 2020 at 03:18:27PM +0000, Clint Sbisa wrote:
> > > Using write-combine is crucial for performance of PCI devices where
> > > significant amounts of transactions go over PCI BARs.
> > >
> > > arm64 supports write-combine PCI mappings, so the appropriate
> > > define
> > > has been added which will expose write-combine mappings under sysfs
> > > for prefetchable PCI resources.
> > >
> > > Signed-off-by: Clint Sbisa <[email protected]>
> >
> > Fine with me, I assume Will or Catalin will apply this.
>
> Haha ! Client had sent it to them originally and I told him to resend
> it to linux-pci, yourself and Lorenzo :-)
>
> So the confusion is on me.
>
> Will, Catalin, it's all yours. You should have the original patch in
> your mbox already, otherwise:
>
> https://patchwork.kernel.org/patch/11729875/
Yup, it's not the radar. We don't usually start queuing stuff until -rc3, so
I'm working through the backlog this week. Would like an Ack from Lorenzo,
though.
Will