Re: [PATCH 3/3] PCI: endpoint: Document pci_epc_set_bar() caller ownership and lifetime rules

[Niklas Cassel]

On Sun, Feb 01, 2026 at 10:37:08PM +0100, Niklas Cassel wrote:
> 
> Considering that we probably want to support in place updates after all...
> 
> I guess we probably only need patch 1/3 in this series, plus another
> patch that makes sure that we call dw_pcie_ep_clear_ib_maps() unconditionally?
> 
> I still don't like that dw_pcie_ep_clear_ib_maps() will be called
> unconditionally, but I don't see any other way to support in place updates...

Perhaps just add a comment above the unconditonal call to
dw_pcie_ep_clear_ib_maps() which explains why the call is unconditional
(i.e. it has to be unconditional in order to support in place updates).


Kind regards,
Niklas