Re: [PATCH RFC NOT TESTED 0/2] PCI: dra7xx: Try to clean up dra7xx_pcie_cpu_addr_fixup()

[Manivannan Sadhasivam]

On Mon, Mar 17, 2025 at 02:45:39PM -0500, Bjorn Helgaas wrote:
> On Tue, Mar 18, 2025 at 12:14:27AM +0530, Manivannan Sadhasivam wrote:
> > On Mon, Mar 17, 2025 at 12:30:08PM -0500, Bjorn Helgaas wrote:
> > > On Thu, Mar 13, 2025 at 11:35:21AM +0530, Manivannan Sadhasivam wrote:
> > > > On Wed, Mar 05, 2025 at 11:20:21AM -0500, Frank Li wrote:
> > > > > This patches basic on
> > > > > https://lore.kernel.org/imx/20250128-pci_fixup_addr-v9-0-3c4bb506f665@xxxxxxx/
> > > > > 
> > > > > I have not hardware to test.
> > > > > 
> > > > > Look for driver owner, who help test this and start move
> > > > > forward to remove cpu_addr_fixup() work.
> > > > 
> > > > If you remove cpu_addr_fixup() callback, it will break backwards
> > > > compatibility with old DTs.
> > > 
> > > Do you have any pointers to DTs that will be broken?  Or to
> > > commits where they were fixed?
> > 
> > Any patch that fixes issues in DT and then makes the required
> > changes in the driver without accounting for the old DTs will break
> > backwards compatibility.
> 
> Right, I guess the rule is that if we have patches that fix DT issues,
> we should apply them as soon as possible.
> 

Right, and those patches should not break old DTs.

> And later if we ever have confidence that unfixed DTs no longer exist
> (or if we can identify and work around them in the kernel), we can
> remove the .cpu_addr_fixup().
> 

Yeah. Unfortunately, we do not have a fixed deadline or process. Just like
supporting the legacy broken hw, we have to keep supporting the old DTs for some
time and then get rid of them.

- Mani

-- 
மணிவண்ணன் சதாசிவம்