On 08/06/2013 12:39 PM, Alan Stern wrote: > On Tue, 6 Aug 2013, Shuah Khan wrote: > >> sl811h_suspend() seems to be the odd routine in the way it handles the >> PM_EVENT_PRETHAW state. It treats it same as PM_EVENT_SUSPEND and >> PM_EVENT_HIBERNATE. All other uses I could find treat it same as >> PM_EVENT_FREEZE and PM_EVENT_QUIESCE. Makes sense since PM_EVENT_PRETHAW >> is PM_EVENT_QUIESCE. >> >> #define PM_EVENT_PRETHAW PM_EVENT_QUIESCE >> >> Reference: Commit 185849991d592497e43bcd264c6152af1261ffe2 introduced >> PM_EVENT_PRETHAW state to sl811h_suspend(). >> >> Couple of questions? >> >> - Why does sl811h_suspend() treat PM_EVENT_PRETHAW different from >> PM_EVENT_FREEZE? > > Because with FREEZE, the driver wants to retain the current bus state. > With PRETHAW, it doesn't care about retaining the bus state. > > At least, that's how it looks to me. The original author of this > driver died a couple of years ago. I don't know if anyone is using it > any more. > >> There is no problem with this code as such, since state is passed in. >> However, this usage conflicts with the rest of the usages and the way >> pm_op() routine maps PM_EVENT_PRETHAW/PM_EVENT_QUIESCE to freeze() pm_ops. >> >> case PM_EVENT_FREEZE: >> case PM_EVENT_QUIESCE: >> return ops->freeze; >> >> Assuming the handling PM_EVENT_PRETHAW is correct in this routine, what >> would be the right mapping for this legacy handling to dev_pm_ops? > > It depends on the driver; there is no one answer. > > Alan Stern > >
With the dev_pm_ops model, drivers have to provide interfaces for each one of these states. In this case, there will be a conflict since pm_op() treats this state as freeze where as the driver wants to do treat it as a suspend/hibernate. In the case of legacy pm_ops, state is passed in as a parameter and driver could take special action if need be, based on the state, however in dev_pm_ops model, state is not passed in. Instead it is handled with state specific pm_ops interfaces. For example, if this driver were to be converted to dev_pm_ops, it would require a freeze interface which will call sl811h_bus_suspend(). Once that is done, PM_EVENT_PRETHAW will be mapped to freeze() ops and sl811h_bus_suspend() will be called instead of port_power(sl811, 0); What I am getting at is, there is no provision to handle the special case for PM_EVENT_PRETHAW like in the case of this driver when using dev_pm_ops. -- Shuah Shuah Khan, Linux Kernel Developer - Open Source Group Samsung Research America (Silicon Valley) [email protected] | (970) 672-0658 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
