On 05/05/2023 14:19, Bharat Bhushan wrote: >>>>> Query to firmware required arm SMC call, to me that does not look >>>>> correct >>>> approach. Thought of using first one but that is already used and >>>> moving that is as same as this. >>>>> >>>>> Hardcoding to 63 will make it work on some SoCs but not all. >>>> >>>> But you know which one is started or is not. GTI_CWD_WDOG tells you this. >>> >>> On a given SoC, Firmware can reserve and/or use one or more timer for some >> other use case (customer use) and configure one of the timer as watchdog timer. >> Linux have to use the configured timer only and cannot decide by its own. >> >> Then the SoCs which have such firmware could use proposed property. >> Provide some rationale property description in your next version (adding >> necessary compatibles). > > Will add compatible as below: > > properties: > compatible: > enum: > - marvell,cn10k-wdt > - marvell,octeontx2-wdt And the rest? You said you have 10 SoCs. You made them not compatible with each other, so I assume on purpose and your driver will make use of it. > > And define this as optional property > marvell,wdt-timer-index: > $ref: /schemas/types.yaml#/definitions/uint32 > minimum: 0 > maximum: 63 > description: > An SoC have many timers (up to 64), firmware can reserve one or more timer > for some other use case and configures one of the global timer as watchdog > timer. Firmware will update this field with the timer number configured > as watchdog timer. Looks ok. Best regards, Krzysztof
