On 21/01/25 00:58, Mark Brown wrote: > On Mon, Jan 20, 2025 at 12:47:18PM -0600, Mario Limonciello wrote: >> On 1/20/2025 12:39, Mark Brown wrote: >>> That does feel like quirks and new features rather than a completely >>> distinct IP. >> I see it as different forms of tech debt. Either you keep track of which >> features the 62 vs 70 hardware supports by different drivers or add logic in >> all the relevant functions(). >> The former increases LoC but reduces risk for mistake (IE avoid oops, I >> forgot this is only supported on 70+ when adding new features). > Until someone fixes a bug or does some subsystem wide cleanup which > affects both copies of the code (perhaps that already happened since the > code was copied!). There's a reason why this is the general kernel > style. > >> Changing code that affects a lot of hardware means a lot more testing too. >> Perhaps after Vijendar's series lands he can split up some of the purely >> duplicated functions into helpers or callbacks and arrange all that testing? > Well, it was getting a new spin anyway for the bits that didn't have the > serial numbers filed off. Will drop code duplication and come up with new patch series.
