The out-of-tree DTS in question was developed quickly and lazily for getting CAN working on a prototype, where the MCU's built-in CAN had not been routed correctly (so I added an MCP251x via SPI for testing). It was a one-off fix for a problem that no longer exists, for hardware which no longer exists. Probably not the best approach to fix the problem either. I shared the DTS to help others at the time who might be using the same part with similar kernel versions (~4.9?) and out-of-tree mcp2515 driver. There is an driver available in-tree, with documentation of DTS properties at linux://Documentation/devicetree/bindings/net/can/microchip,mcp251x.txt I don't care about that out-of-tree MCP2515 DTS any more, the kernel developers certainly shouldn't care about it, and neither should anyone else. -- Mark Kuckian Cowan On Wed, 13 Nov 2019 at 11:19, andriy.shevchenko@xxxxxxxxx <andriy.shevchenko@xxxxxxxxx> wrote: > > On Tue, Nov 12, 2019 at 01:15:26PM -0800, Drew Fustini wrote: > > On Tue, Nov 12, 2019 at 1:01 PM Drew Fustini <pdp7pdp7@xxxxxxxxx> wrote: > > > I based the overlay I was using one a MCP2515 overlay for SPI0 on > > > BeagleBone Black: > > > https://github.com/battlesnake/beaglebone-spi0-mcp2515/blob/master/MCP2515.dts > > > > > > I'm not sure where those extra "mcp251x," properties come from since > > > the driver does not appear to use them at all. > > > > I searched for "mcp251x,oscillator-frequency" on google and found this > > old patch by Alessandro Zummo <a.zummo@xxxxxxxxxxxx>: > > https://github.com/beagleboard/meta-beagleboard/blob/master/common-bsp/recipes-kernel/linux/linux-mainline-3.8/net/0006-mcp251x-add-device-tree-support.patch > > > > It does read some of those properties: > > + of_property_read_u32(np, "mcp251x,oscillator-frequency", &tmp); > > + pd->stay_awake = of_property_read_bool(np, "mcp251x,stay-awake"); > > + pd->enable_clkout = of_property_read_bool(np, "mcp251x,enable-clkout"); > > > Maybe there are people using this out of tree patch still and > > therefore including those properties like Mark K Cowan did in: > > https://github.com/beagleboard/bb.org-overlays/blob/master/src/arm/MCP2515-SPI0.dts > > First of all, there are rules (which are getting stricter all the time) about > property names and use in DT. Second, no one cares about out-of-tree > customizations. In kernel we consider no regression made if we introduce > properties that are not compatible with no-one-knows DTSes (it didn't work > with older kernel, it continue not working with new — status quo). > > -- > With Best Regards, > Andy Shevchenko > >
