On 2023/5/11 4:22, Conor Dooley wrote: > On Wed, May 10, 2023 at 05:21:21PM +0800, Walker Chen wrote: >> On 2023/5/10 16:33, Emil Renner Berthing wrote: >> > On Tue, 9 May 2023 at 20:05, Conor Dooley <conor@xxxxxxxxxx> wrote: >> >> >> >> On Tue, May 09, 2023 at 08:52:48PM +0800, Walker Chen wrote: >> >> > Hi Conor/Emil, >> >> > >> >> > DT overlay is used to describe combinations of VF2 and hat. >> >> > Do you have any comments on this patch ? >> >> >> >> I seem to recall that he said at the linux-riscv sync-up call that we >> >> have* that he was not in favour of overlays for hats like this. >> >> I'll let him confirm that though, I might very well be misinterpreting or >> >> misremembering what he said. >> > >> > What probably meant was that I didn't want a bunch of different device >> > trees for each combination board * hat. An overlay makes a lot more >> > sense. However, looking through the kernel tree there is a surprising >> > lack of overlays for hats committed already, so I suspect there is >> > some sort of policy around overlays already in place. > >> About the specific usage of overlay for this board + hat, referenced to the following example in kernel: >> linux/arch/arm64/boot/dts/freescale/imx8mm-venice-gw73xx-0x-imx219.dtso >> That board is connected with imx219 sensor via mipi_csi interface. That patch was accepted in 2022. > > Hmm, so spoke to Emil again today about it. In the interest of being > fair, I did go looking at that particular board & looked through their > documentation for more information on why there are overlays. > They do actually sell the bits required to use the overlays, based on > what I saw in their datasheet for the board & wiki. That said, what is > done for one arm64 platform does not necessarily apply elsewhere ;) > > I'm not against allowing in-tree overlays for hats/capes/daughter-boards > that come bundled with a board, but accepting ones for a hat that > someone decided to use theoretically has no limit! The "someone" in this > case might be a StarFive developer, but it could be any random one of > your customers next! > We've got to draw a line somewhere, so my answer to the overlay *in this > case* is a no. Sorry. > When you submit your next version, please drop the overlay from this > patch. I'm trying to understand what you mean. so the conclusion is that I need to drop the file 'jh7110-starfive-visionfive-2-wm8960.dtso' from this patch. When I submit the next version, just keep the TDM node and the pins configuration for TDM in patch [3/3], need to drop the wm8960 stuff. Right ? Best regards, Walker
