| Message ID | 20221121015451.2471196-1-bero@baylibre.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Remove the pins-are-numbered DT property | expand |
Hi Bernhard, On 21/11/2022 02:54, Bernhard Rosenkränzer wrote: > During the review of my MT8365 support patchset > (https://lore.kernel.org/linux-mediatek/20221117210356.3178578-1-bero@baylibre.com/), > the issue of the "pins-are-numbered" DeviceTree property has come up. > > This property is unique to Mediatek MT65xx and STM32 pinctrls, and > doesn't seem to serve any purpose (both the Mediatek and STM32 drivers > simply refuse to deal with a device unless pins-are-numbered is set to > true). > > There is no other use of this property in the kernel or in other projects > using DeviceTrees (checked u-boot and FreeBSD -- in both of those, the > flag is present in Mediatek and STM devicetrees, but not used anywhere). > > There is also no known use in userspace (in fact, a userland application > relying on the property would be broken because it would get true on > any Mediatek or STM chipset and false on all others, even though other > chipsets use numbered pins). > > This patchset removes all uses of pins-are-numbered. > > My personal preference is to add a summary of the files touched by the series in the cover letter (the tools will do that for you). This allows maintainers to easier understand if they have to look deeper into the series or can ignore it. No need to send again, just saying for the future. Regards, Matthias
On 21/11/2022 13:31, Matthias Brugger wrote: > Hi Bernhard, > > On 21/11/2022 02:54, Bernhard Rosenkränzer wrote: >> During the review of my MT8365 support patchset >> (https://lore.kernel.org/linux-mediatek/20221117210356.3178578-1-bero@baylibre.com/), >> the issue of the "pins-are-numbered" DeviceTree property has come up. >> >> This property is unique to Mediatek MT65xx and STM32 pinctrls, and >> doesn't seem to serve any purpose (both the Mediatek and STM32 drivers >> simply refuse to deal with a device unless pins-are-numbered is set to >> true). >> >> There is no other use of this property in the kernel or in other projects >> using DeviceTrees (checked u-boot and FreeBSD -- in both of those, the >> flag is present in Mediatek and STM devicetrees, but not used anywhere). >> >> There is also no known use in userspace (in fact, a userland application >> relying on the property would be broken because it would get true on >> any Mediatek or STM chipset and false on all others, even though other >> chipsets use numbered pins). >> >> This patchset removes all uses of pins-are-numbered. >> >> > > My personal preference is to add a summary of the files touched by the series in > the cover letter (the tools will do that for you). This allows maintainers to > easier understand if they have to look deeper into the series or can ignore it. > > No need to send again, just saying for the future. Yep. And git format-patch does it automatically in both common cases - writing cover letter manually or taking it from branch description (which is absolutely cool, IMHO). Best regards, Krzysztof