Tags: oxidecomputer/hubris
Tags
Introduce `cosmo` and `all-sp` release target With the addition of `cosmo`, it's gotten a bit unwieldy to manage four potential SP target: `gimlet`, `cosmo`, `psc`, `sidecar`. Introduce the `all-sp` target to be able to build all targets at once. We also still have the `cosmo` target for backwards compatibility.
Use `FailServer` in gimlet host flash Similar to cosmo, use a fail server to always return an error when the chip ID or capacity is incorrect.
Stop mentioing `A1` state outside the sequencer `A1` is supposed to be a transitory state from `A2` onto `A0`. Issue #1267 flagged that we were never broadcasting this information but also made the (modest) suggestion that we could also just remove it. It turns out, nobody is attempting to do anything significant when the system is in A1. Remove this from the public API to make everything a little easier on the clients. The sequencer still tracks A1 internally.
thermal: remove set_fan_pwm (#2063) I'm removing `set_fan_pwm` from the `thermal` task interface because its existence no longer serves a purpose. Fans can either be in `auto` control mode, where the thermal model decides the PWM, or `manual` control mode, which effectively hardcodes the PWM to the `initial_pwm` argument of `set_mode_manual`. I say "effectively hardcodes" because it calls the [`maintain_pwm`](https://github.com/oxidecomputer/hubris/blob/master/task/thermal/src/main.rs#L333) function each loop iteration. This means that when `set_fan_pwm` was setting the PWM value, it would just be overwritten the next iteration. In addition to that, we just don't control single fans really, so the fact that `set_fan_pwm` required a single fan to be specified just doesn't match reality any more. An alternative path here would be to remove the `index` argument from `set_fan_pwm` and change its function to update the `last_pwm` value read by `maintain_pwm` each iteration.
thermal: remove set_fan_pwm (#2063) I'm removing `set_fan_pwm` from the `thermal` task interface because its existence no longer serves a purpose. Fans can either be in `auto` control mode, where the thermal model decides the PWM, or `manual` control mode, which effectively hardcodes the PWM to the `initial_pwm` argument of `set_mode_manual`. I say "effectively hardcodes" because it calls the [`maintain_pwm`](https://github.com/oxidecomputer/hubris/blob/master/task/thermal/src/main.rs#L333) function each loop iteration. This means that when `set_fan_pwm` was setting the PWM value, it would just be overwritten the next iteration. In addition to that, we just don't control single fans really, so the fact that `set_fan_pwm` required a single fan to be specified just doesn't match reality any more. An alternative path here would be to remove the `index` argument from `set_fan_pwm` and change its function to update the `last_pwm` value read by `maintain_pwm` each iteration.
thermal: remove set_fan_pwm (#2063) I'm removing `set_fan_pwm` from the `thermal` task interface because its existence no longer serves a purpose. Fans can either be in `auto` control mode, where the thermal model decides the PWM, or `manual` control mode, which effectively hardcodes the PWM to the `initial_pwm` argument of `set_mode_manual`. I say "effectively hardcodes" because it calls the [`maintain_pwm`](https://github.com/oxidecomputer/hubris/blob/master/task/thermal/src/main.rs#L333) function each loop iteration. This means that when `set_fan_pwm` was setting the PWM value, it would just be overwritten the next iteration. In addition to that, we just don't control single fans really, so the fact that `set_fan_pwm` required a single fan to be specified just doesn't match reality any more. An alternative path here would be to remove the `index` argument from `set_fan_pwm` and change its function to update the `last_pwm` value read by `maintain_pwm` each iteration.
PreviousNext