9 transmit event response procedures, 10 receive event response procedures, Transmit event response procedures -70 – Freescale Semiconductor MPC8260 User Manual

Page 1172: Receive event response procedures -70

Advertising
background image

Inverse Multiplexing for ATM (IMA)

MPC8260 PowerQUICC II Family Reference Manual, Rev. 2

33-70

Freescale Semiconductor

33.5.4.9

Transmit Event Response Procedures

The following TX events may take place when operating in IMA mode. It is recommended that all events
be handled via an “exception/interrupt service routine” (ISR) as the response time inherent with interrupt
driven events should diminish the negative impact/propagation of such events:

1. TQU (Transmit Queue Underrun)—Indicates that a transmit queue was emptied. This implies that

the offending link (PHY) is requesting cells at a faster rate relative to the TRL. If the TRL is out of
spec, then you will see multiple links reporting TQUs (because all links in the group will be faster
relative to the TRL). The offending link should be removed (see

Section 33.5.4.5, “Link Removal

Procedure

), check the following: PHY, TX Queue depth (start and end pointers should not be the

same), TNUMLINKS is equal (not less or greater) than the number of active links.

2. TQO (Transmit Queue Overflow)—Indicates that a transmit queue was not ready to receive a cell

(full). This implies that the offending link (PHY) is requesting cells at a slower rate relative to the
TRL. If the TRL is out of spec, then you will see multiple links reporting TQOs (because all links
in the group will be slower relative to the TRL). The offending link should be removed (see

Section 33.5.4.5, “Link Removal Procedure”

), check the following: PHY, TX Queue depth (depth

should be the same as other queues), TNUMLINKS is equal (not less or greater) than the number
of active links.

33.5.4.10 Receive Event Response Procedures

The following RX events may take place when operating in IMA mode. It is recommended that all events
be handled via an “exception/interrupt service routine” (ISR) as the response time inherent with interrupt
driven events should diminish the negative impact/propagation of such events:

1. LS (Link Stalled)—The link’s DCB has emptied, either because the PHY is no longer functioning

or it is relatively slower than the other links. The offending link should be removed (see

Section 33.5.4.5, “Link Removal Procedure”

). Make sure the DCB start and end pointers are not

the same. If only one link is used in a group the software must monitor the TC layer to detect that
this link has stalled. No LS event is reported in the IMA interrupt queue. To re-start an IMA group
properly after link(s) have recovered, the user must reset all the microcode-managed parameters in
the IMA receive group to zero.

2. DCBO (DCB Overflow)—Indicates that the Delay Compensation Buffer has no more space. The

source of the problem can be varied: 1) the offending PHY is sending at a faster rate (out of spec.),
2) the propagation delay of one of the other links in the group is greater than anticipated (need to
make DCB larger) or, 3) the size of the offending link was programmed incorrectly, i.e., smaller
(the size of all DCBs in the group should be the same). The offending link should be removed (see

Section 33.5.4.5, “Link Removal Procedure”

). This exception corresponds to the LODS (Link Out

of Delay Synchronization Defect) and should be reported to the FE (via the ICP cell, RxDefect =
LODS) of the problem. The PowerQUICC II will continue to report this exception if the condition
persists (unless the event is masked).

3. LDS (Link Delay Synchronized)—Indicates that the new/added link (to an existing group) has

achieved synchronization (link delay). The link is capable of receiving data at this point, (see

Section 33.5.4.4, “Link Addition Procedure”

).

Advertising