Re: Patch: linux-2.5.42/kernel/sys.c - warm reboot should not suspend devices

[Eric W. Biederman]

Eric Blade <[e-mail]> writes:

> Here is where I realize that I've forgotten to CC: the list on all the
> traffic that I keep sending between Eric and Adam.  D'oh.
> 
> 
> On Sun, 2002-10-13 at 20:07, Eric W. Biederman wrote:
> > > 
> > > 	However, I'm not trying to quash what you want to discuss.
> > > I'd be interested in hearing about clarifications and perhaps
> > > extensions of the struct device_driver methods, which I think is what
> > > you're getting at, perhaps here or on linux-hotplug.  It's just that,
> > > for this thread, I'm trying to focus on my patch that eliminates the
> > > software suspend on reboot (pros and cons, alternatives to it, etc.).
> > 
> > The 2.5.41 variant is below.  The bug is reusing the old enumeration value
> > as was previously mentioned.
> > 
> 
> I tried to submit a fix to this, but the only response I've gotten back
> is that it failed to apply.  My original patch excluded the bit from
> device.h that added a new state to the enumeration, and when it got into
> the tree, it got into the tree using the current states that were
> available.  My bad.  
> 
> Eric has already indicated earlier, that Adam's issue is, however, not
> with the changes to drivers/base/power.c but to the changes to the IDE
> driver.  

Correct.  But when people try to use power management they will see
the bug.  SUSPEND_POWER_DOWN (a request for the driver to remove power
from the device) is a very different state from SUSPEND_DETACH (a
request to disassociate the driver from the device).  I am not fully
convinced the two cases should be merged.

SUSPEND_POWER_DOWN as defined should actually cause the behavior Adam
was seeing.  Which made it doubly interesting.

Will you submit a patch detangling this mess?  I believe Adams
only problem was that it did not obviously fix his problem.

Eric