Re: suspend for raid devices?

[Steven Bellovin <smb%cs.columbia.edu@localhost>]

On Oct 15, 2009, at 10:04 PM, Malcolm Herbert wrote:

> For portability and DR reasons I have set up my laptops so that the
> internal disk is under RAIDFrame management as one half of a broken
> mirror - this also allows me to boot my wife's laptop with the
> still-beating HDD from mine in a USB caddy with no ill-effects if I  
> have
> to ...
>
> This works very well in most situations[1], however the version of
> -current I had at the time (5.99.15 from mid-July) did not allow me to
> suspend the laptop as the raid device lacked power management.  dmesg
> has this to say:
>
> |# zzz
> |Suspending system...
> |# dmesg | tail
> :
> :
> |acpi0: entering state 3
> |Devices without power management support: raid0
> |acpi0: aborting suspend
>
> I'm not sure what support is needed, however I was wondering whether  
> it
> was available now in a current kernel, or if not, how likely would  
> it be
> to be implemented?


Given that raid0 is a metadevice for real devices that do have power  
management, I suspect that a null routine will suffice.  If I'm  
remembering and reading the code directly, see the calls to  
pmf_device_register and pmf_device_deregister in usb/ugen.c (to pick  
one pseudo-random example).

                --Steve Bellovin, http://www.cs.columbia.edu/~smb