Re: [RFC] Debugging posted PCI writes

[Grant Grundler]

Matthew Wilcox wrote:
...
> Perhaps a timer that writes the queue every second would help?  It may
> be considered unreasonable for writes to be delayed _indefinitely_.

it is. that's what I was warning you about. need a timer to push
writes out every jiffie or so.

overall this is a great idea that was discussed at OLS/kernel summit.

grant