Spec currently states:

The INVALID DWORD COUNT field indicates the number of invalid dwords (see 3.1.114) that have been received
outside of phy reset sequences (i.e., between when the SP state machine (see 6.8) sends a Phy Layer Ready
(SAS) confirmation or Phy Layer Ready (SATA) confirmation and when it sends a Phy Layer Not Ready
confirmation to the link layer).

As written, this would then count the potentially large amount of invalid dwords that occur after DwsLost happens until DwsReset, if the SP state machine attempts to reestablish sync in the SP15:SAS_PHY_Ready state (and likewise for the equivalent SATA state).

Is this correct intention for this counter?  Since the number of invalid dwords during that time period could be very undeterministic, this seems to really limit the usefulness of the actual count value.

Thoughts?

__________
Brian Day
LSI Corp.
(719) 533-7468