On Saturday 28 May 2005 10:04, Alan Cox wrote:
> On Sad, 2005-05-28 at 14:57, Parag Warudkar wrote:
> > This current problem of Hang-On-Boot if USB drive is attached does not
> > happen with Windows - so it is some sort of additional (unnecessary?)
> > thing which Linux does and the BIOS doesn't like.  (Like re-enabling the
> > controller even if BIOS has already enabled it or some such.)
>
> Provide dmesg output and we might be able to guess. The first obvious
> candidate would be the BIOS refusing to do a handover if it booted from
> USB disk.

In the hang case, the machine is booting from internal HDD not USB disk, the 
USB disk is just attached and powered on during boot. As I said I have 
tracked the hang down to pci_enable_device() called by usb_hcd_pci_probe(). 
If the USB disk is attached and powered on, the usb_hcd_pci_probe()'s call to 
pci_enable_device() for the controller leads to this 2 minute hang. (Which 
cures itself after that). 

There are two cases in which the hang does not occur -
1) If I remove the pci_enable_device() call load of ohci-hcd doesn't hang even 
if the USB disk was attached/on.
and 
2) If I attach the USB HDD _After_ the load of ohci-hcd is done (i.e. after 
the call to pci_enable_device())

dmesg is perfectly normal, not even timestamp differences before and after 
call to pci_enable_device - since the machine is completely hung for that 
period - not even the clock is ticking?

Parag
-- 
The memory management in Windows 95 can be used to frighten small children.