Mark Lord wrote:

quoted

quoted

Dmesg: hotplug-in 3726 into 6145 , and the unplug, finally plug again
ata9: exception Emask 0x10 SAct 0x0 SErr 0x0 action 0xe frozen
ata9: irq_stat 0x02400000, PHY RDY changed
ata9: hard resetting link
ata9: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
ata9.15: qc timeout (cmd 0xe4)
ata9.15: failed to read PMP GSCR[0] (Emask=0x4)

Heh.. failing PMP READs.  Doesn't look like there's anything else to do
than disabling it.  Will post a patch.

..

No kernel timestamps there, but I do wonder if somehow
the driver is polling too soon for the PMP?
I wonder if a short delay in there, before issuing the PMP read,
might help (or not)?

The timeout is currently 3 secs and the timeout is mostly dependent on
the PMP device and 3726 fares off fine with < 1sec timeouts, so I don't
think it's because of short timeout.

Thanks.

-- 
tejun