PCI bus isolation using AIX, Linux, or the HMC

If you have an HMC, then this procedure should be performed from the HMC as part of the HMC directed service.

If you do not have an HMC, then you should perform this procedure when directed by the maintenance package.

Isolating a PCI bus problem while running AIX or Linux

Can an IPL be run on the operating system?

• No: Perform MABIP52. This ends the procedure.
• Yes: Choose from the following:
  • If you are running AIX, go to Running the online and eServer stand-alone diagnostics to isolate the PCI bus failure with online diagnostics in concurrent mode.
  • If you are running Linux, go to Running the online and eServer stand-alone diagnostics to isolate the PCI bus failure with stand-alone diagnostics.
  This ends the procedure.

Isolating a PCI bus problem from the HMC

To isolate a PCI bus problem from the HMC, check the serviceable event view for the server for FRU part locations associated with the serviceable event, then continue with this procedure:

1. Did the serviceable event view provide the location(s) for the failing FRU(s)?
   • Yes: Use those locations to exchange the given FRUs one at a time until the problem is resolved. This ends the procedure.
   • No: Continue with the next step.
2. Go to DSA translation to determine the Direct Select Address (DSA).
3. Perform the following:
   1. Record the bus number value (BBBB) from the DSA and convert it to decimal format.
   2. Search for the decimal system bus number in the partition resources screens on the HMC.
   3. Record the frame or unit type and continue with the next step.
4. Record the Cc value from the DSA. Is the Cc value greater than 00?
   • Yes: Continue with the next step.
   • No: The multi-adapter bridge number and the multi-adapter function number have not been identified, and so the card slot cannot be identified using the DSA. Look in the HMC partition resources for non-reporting or non-operational hardware. That will indicate which cards in which positions need to be replaced. Refer to Finding part locations for the model you are working on for information about the multi-adapter bridge that controls those card slots. That multi-adapter bridge is also a FRU. This ends the procedure.
5. Is the right-most character (c) F?
   • No: Continue with the next step.
   • Yes: Only the multi-adapter bridge number has been identified. Record the multi-adapter bridge number (left-most character of Cc) for later use. Since the card slot cannot be identified with the DSA, refer to Finding part locations for the model you are working on for information about the multi-adapter bridge that controls the card slots. Consider all card slots controlled by the multi-adapter bridge to be FRUs. This ends the procedure.
6. Refer to Card positions and use the BBBB and Cc values that you recorded to identify the card position. Then return to the procedure that sent you here. This ends the procedure.