NFTL cleanup

[David Woodhouse]

I've realised that we're not permitted to use a semaphore, or call any of 
the MTD access functions, from the nftl_request() function.

The nftl_request() function isn't permitted to sleep.

Therefore, we want a kernel thread. The _request function should do nothing 
but take requests off the kernel's queue and stick them on a private queue, 
to be served by the kernel thread.

Once we do proper concurrent access to individual flash chips within the 
DiskOnChip, that'll become _very_ important.

--
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo at infradead.org