Re: pci_pool_free from IRQ

[Alan Cox]

> This sure makes life difficult. Device removal events can be called
> from interrupt context according to Documentation/pci.txt. This is
> certainly a place where one might want to call pci_consistent_free.

None of our device code supports interrupt based device removal. In fact
many drivers use vmalloc directly so will hit the same problem the
pci_consistent_free hits on the ARM.

I suspect we should fix the documentation (and if need be the code) to reflect
the fact that you have to be completely out of your tree to handle device 
removal in the irq handler