Martin et al.
This is the next step in stablizing the transport layer.
I have not booted but it will compile, and it is nearly identical to what
I generated for 2.4 to be released soon.
The comments are harsh on the interface but it functionally correct.
If you get an device error in PIO, bad things can happen to the data.
This is no different in the stock 2.4.0->2.4.18->19x.
Of of all the transport data handlers.
CLEAN and SAFE:
DMA read/write is safe and has always been.
Single sector PIO WRITING is clean and safe.
DIRTY but operational (error events in the hardware will cause data problems)
Single sector PIO READING can corrupt a single sector if there
is a device error.
Multi-Read/Write will corrupt and misreport data only on an error.
What is still lacking in block is the much needed in proccess bio walker.
Once I can finish coding this fix into BLOCK, then I can complete the
transport layer and slap it on a bus analyzer and force articial errors on
the buss to see if the driver behaves correctly. If this passes, we are
good to run like the wind.
Regards,
Andre Hedrick
LAD Storage Consulting Group
This archive was generated by hypermail 2b29 : Sat Mar 23 2002 - 22:00:30 EST