The "Actual queue depth" statistics are what the driver has set internally
for the maximum number of commands it will allow the device to use, aka the
device's queue depth limit. The only reason it's even reported is that some
hard disks have hard queue depth limits. When the driver determines those
limits, it will change this number to avoid lots of useless QUEUE_FULL
status conditions from the drive. That shows up in these numbers. It also
provides a means by which to check and make sure the tag_info:{{x,x,x,x}}
line used when booting was properly given and parsed.
> Also I winder why 16 queues are displayed for a narrow controller
> like mine (AHA2940 plain).
The queue depth structures have to be allocated prior to the initialization
of the adapters (since they are passed in during the setup phase). As such,
each structure must be capable of supporting a wide controller, so each one
is 16 entries wide. Since they are all the same, there was little reason to
cut them down to 8 entries on narrow controllers merely for appearance
sake. This is especially true if you consider that someone may be trying to
set up a tag_info command line and they might need to see what is happening
to the values they pass in.
--Doug Ledford <dledford@dialnet.net> Opinions expressed are my own, but they should be everybody's.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu