Re: suspicious pci hang

[Alan Cox]

> 	x = inl(reg->bmsr);
> 
> The board is in a pci space.  I suspect the board simultaneously
> died some time around the upgrade or the perhaps the pci hardware
> implementation is not quite stable, or, maybe something changed in the
> kernel.
> 
> Yes, reg->bmsr contains a valid address (0x0000ec7c) and this value did
> not get "stomped on" since the PCI block was requested.  Further, I know

ec7c is in ram. That sounds like reg is a large object and reg = NULL