Re: Daemonize() should re-parent its caller

[Greg KH]

On Wed, Jan 23, 2002 at 10:54:44AM -0500, Alan Stern wrote:
> 
> If the parent is init or another user process, we can reasonably
> expect that the zombie will be reaped eventually.  But what if the
> parent is another kernel thread?  This situation arises in the USB
> mass-storage device driver, where the device manager and scsi
> error-handler threads are spawned (indirectly) by the khubd kernel
> thread.

What problem are you seeing with the khubd and USB mass-storage kernel
threads?  There is a patch in the most recent kernel versions that
slightly modifies the usb-storage kernel thread logic, supposedly to fix
a problem that people were having under very long scsi timeouts.

thanks,

greg k-h