On 25-03-04 07:29:46, Uwe Kleine-König wrote:
Hello,
On Mon, Mar 03, 2025 at 01:52:49PM +0200, Abel Vesa wrote:
The PWM allow configuring the PWM resolution from 8 bits PWM values up to 15 bits values, for the Hi-Res PWMs, and then either 6-bit or 9-bit for the normal PWMs. The current implementation loops through all possible resolutions (PWM sizes), for the PWM subtype, on top of the already existing process of determining the prediv, exponent and refclk.
The first and second issues are related to capping the computed PWM value.
I just took a very quick look. I'd say squash the first and second patch into a single one. Splitting a change that fixes the same issue in the two branches of an if condition has no benefit.
Actually, the first two patches fix different commits. The first patch fixes a commit that is only in linux-next for now, while the second patch fixes a commit that has been merged in 6.4.
So they need to be separate patches.
Other than that this patch set would also benefit from what I wrote in the review of the other patch you send: Please mention a request where the result becomes wrong. This considerably simplifies understanding your changes.
Sure. Will describe the 5ms vs 4.26ms period scenario. Hope that's OK.
Thanks Uwe
Thanks for reviewing, Abel