Messages in this thread | | | Subject | Re: [PATCH 2/3] PCI: uniphier-ep: Add support for non-legacy SoC | From | Kunihiko Hayashi <> | Date | Tue, 8 Feb 2022 20:11:45 +0900 |
| |
Hi Bjorn,
Thank you for your comment.
On 2022/02/05 1:06, Bjorn Helgaas wrote: > On Thu, Feb 03, 2022 at 11:36:25AM +0900, Kunihiko Hayashi wrote: >> Define SoC data that includes pci_epc_features and boolean 'is_legacy' >> to distinguish between legacy SoC (ex. Pro5) and non-legacy SoC. > > I'm not a big fan of "legacy" as a description because it conveys very > little information. You're adding support for a "non-legacy" SoC, so > now there are "legacy" ones and "non-legacy" ones. Next year there > will be another new SoC, and then there will be *two* kinds of > "legacy" ones that must be distinguished plus the next "non-legacy" > one.
Make sense. There will be multiple different "legacy", so it isn't desirable to distinguish SoCs by legacy and non-legacy.
> You mentioned "Pro5" as an example of "legacy," which is a good start. > Are there any others? If Pro5 is the only one, you could just use > "pro5" where you now use "legacy." I see. In consideration of the future, I'll define callback functions and prepare functions for each SoC. And the "gio" clocks/resets are distinguished by the flag "has_gio".
Thank you,
--- Best Regards Kunihiko Hayashi
| |