Hi Laurent, Thank you for taking the time. Two of the three points came from the v1 posting, one of them is a real miss on my side, and there is a fourth thing I should have found before posting at all -- I get to that below.
The message you replied to is the 09-15 posting. v3 went out on 09-16: <[email protected]> and it rewrote the commit message to give the rationale you are missing: why "ti,tpd12s016" can only be a fallback of "ti,tpd12s015" (the driver matches on "ti,tpd12s015" alone, so a board using the 016 has to list both), and why gpios has to take three to five entries. The v2 version of that description had capped gpios at three, which broke dtbs_check for omap5-igep0050.dts -- that board passes the three control/HPD lines plus the DDC SCL and SDA lines. Rob Herring gave v3 his Reviewed-by on 09-24. The Assisted-by tag is a fair hit, and it applies to v3 as well: v3 does not carry one either. That is my oversight and I will add it. Your third point also stands for v3. The conversion and the gpios semantics change are still one patch. Splitting them is right. Now the part I should have caught myself. This conversion is not new work. Eduard Bostina posted the same conversion on 2026-07-30: <[email protected]> That series makes the same two binding changes -- the "ti,tpd12s016" fallback compatible and gpios up to five entries -- and its 2/2 renames the same tpd12s015 endpoint nodes that mine does. Rob reviewed it on 08-03. It did not land: there is no ti,tpd12s015.yaml in mainline, v6.18, v6.17 or linux-next, and the .txt is still in all of them. I did not find it before posting mine, and I have cc'd Eduard here. I would rather ask than add a fourth revision to the pile. Which would you prefer: - a v4 of mine, split into a pure conversion plus a separate semantics patch, or - for me to drop mine and leave this to Eduard's series? Either way I will not repost the current form. If there is a reason the Bostina series stalled that I should not repeat, I would be glad to hear it. Thanks, Chaosheng Qu
