On 05/09/2024 15:23, Bryan O'Donoghue wrote: > On 27/08/2024 11:05, Dikshita Agarwal via B4 Relay wrote: >> From: Dikshita Agarwal <quic_dikshita@xxxxxxxxxxx> >> >> Implement runtime power management for iris including >> platform specific power on/off sequence. >> >> Signed-off-by: Dikshita Agarwal <quic_dikshita@xxxxxxxxxxx> > >> +int iris_hfi_pm_suspend(struct iris_core *core) >> +{ >> + int ret; >> + >> + if (!mutex_is_locked(&core->lock)) >> + return -EINVAL; >> + >> + if (core->state != IRIS_CORE_INIT) >> + return -EINVAL; > > Reiterating a previous point > > Are these checks realistic or defensive coding ? Well, this one: if (!mutex_is_locked(&core->lock)) is clear bug or someone is reinventing lockdep. >> + >> + if (!core->power_enabled) { >> + dev_err(core->dev, "power not enabled\n"); >> + return 0; >> + } > > Similarly is this a real check an error that can happen and if so how ? And here re-inventing runtime PM. Best regards, Krzysztof