Re: [RFC] Autosuspend update

On Sat, 10 Feb 2007, Oliver Neukum wrote:

> > The problem is that after the device is suspended, we don't know whether
> > it was an autosuspend or the result of an external user command.  (It
> 
> If that is true then we cannot know whether to wake up on demand.
> I am afraid that state will have to be recorded.

No -- we always wake up on demand, regardless.

A "demand" can take either of two forms.  One is a remote wakeup request 
from the device itself.  Such a request will be sent only if remote wakeup 
was enabled at the time the device was suspended.  We don't need to 
remember the state; the device will remember it for us.

The other form is a software resume request bubbling up from one of the
device's children.  (Example: A hub with a mass-storage device connected
to it.  If the block layer wants to do I/O to the child device then the
hub has to be resumed.)  Such requests must always be carried out (as the
example shows).


> > > Secondly, how about exposing USB_IF_DEVICE_BUSY through sysfs?
> > 
> > It's constantly changing.  The value exposed wouldn't be reliable.  (Or 
> > for devices where it's not constantly changing, the value is always 0 and 
> > hence uninteresting.)
> 
> Then interesting direction is writing, not reading.

Why would anyone want to write it?  Especially since you can force the 
device to suspend or resume whenever you want, anyway.


> > > > +       if (value)
> > > > +               rc = usb_bus_type.suspend(dev, PMSG_SUSPEND);
> > > 
> > > Thirdly, doesn't this have security implications? You can shut down
> > > network interfaces and swap spaces this way.
> > 
> > Yes indeed -- just like many other sysfs interfaces, such as 
> > bConfigurationValue.  That also is capable of shutting down network 
> > interfaces and swap spaces.
> 
> True, but it means that the security implications are still different.
> You cannot rely on this interface to resume devices from user
> space. Therefore the delay interface should resume the device.

I don't follow your argument.  Why does the "delay" interface (more
properly, the "autosuspend" attribute) need to be able to resume devices?  
There are other ways to resume them from userspace.  Opening a device's
usbfs file will do it, for instance.  (Running lsusb resumes _every_ USB
device.)  Or simply trying to use the device, since it will resume on
demand.

Alan Stern


-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier.
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel