On 13/04/2024 12:56, Lukas Wunner wrote: > On Sat, Apr 13, 2024 at 12:53:25PM +0200, Krzysztof Kozlowski wrote: >> On 13/04/2024 12:51, Lukas Wunner wrote: >>> The binding requires two entries in the compatible string used in the DT, >>> the chip name followed by the generic string: >>> >>> items: >>> - enum: >>> - infineon,slb9673 >>> - nuvoton,npct75x >>> - const: tcg,tpm-tis-i2c >>> >>> This allows us to deal with device-specific quirks, should they pop up >>> (e.g. special timing requirements, hardware bugs). We don't know in >>> advance if they will be discovered, but if they are, it's cumbersome >>> to determine after the fact which products (and thus DTs) are affected. >>> So having the name of the actual chip used on the board has value. >> >> So you say devices are compatible. Then the second patch is wrong. >> >> I cannot respond to it, though... so NAK-here-for-second-patch. > > I disagree. It's ugly to have inconsistencies between the DT bindings > and the driver. So I think patch [1/2] in this series is fine. You are mixing different things. This patchset creates inconsistency. You even refer here to bindings while we discuss the driver... Why this one driver is different than all other Linux drivers? Why do you keep pushing here entirely different behavior? The devices are compatible, so growing match table is both redundant and confusing. That's everywhere. TPM is not different. Best regards, Krzysztof
