Intel® Software Network

Hello,

I'm using IXP435 network processor with Intel Access Library version 2.4 . I'm running ATM MPHY 4-Port and HSS NPE image. I'm having a problem with it but cannot find support on Intel's website (I get this from Intel website: Intel does not provide email support for this product.). I've also reported this problem to my local Intel Authorized Distributor, but they appear to be less than helpful, so I'm posting my questions here.

The problem I have is the NPE that is running ATM and HSS crashes everytime when there's an irregular frame pulse (i.e. more clock pulses in a frame than usual). I don't know what other conditions can cause the NPE to crash. When the NPE crash, it brings the ATM  and HSS down with it. Does anyone know if this bug has been fixed in Intel Access Library version 3.0?

I got this message from log. "Failed to send message" and "ixHssAccComNpeCmdMsgSend - NpeMh failed to send". I got a lot this same message appearing on my log. The ATM stops working when I get this message. My question is, what may have caused the NPE MH to fail?

Any help appreciated.

-JA-

Hi,

Unfortunately, we don't provide product support in this forum but you might want to check out the release notes of version 3.0:

http://softwarecommunity.intel.com/isn/Community/en-US/forums/thread/30245204.aspx

- Thai

Hi,

Our engineer needs some additional information:

-          We are  assuming that when there is no over clock pulses, the whole system behave normally, right?

-          When you say more clock pulses, is it from HSS or ATM?

-Thai

Hello Thai,

Thank you for replying.

- Yes, the system behaves normally when there is no over clock pulses.
- The clock pulses are for the HSS port. The HSS port is setup as a slave, i.e. it relies on external clock and frame pulse (or some refer it as frame sync).

I have a master port generating clock and frame pulse down to the HSS slave port. The master port can turn off at anytime, at any point in a data frame, like when rebooting or power lost. When the master port is re-enabled, it may send out a few clocks before the first frame pulse is sent out. From the HSS perspective, now the data frame has irregular number of clocks cycles. The NPE crashes when the data frame has more clock cycles than normal. The NPE works normally if the data frame has less clock cycles than normal, I only get corrupted data which is expected.


-JA-