> We did the hardware fix described in the BSP user's manual to > make PCIe work (according to the manual section 2.1, step 3). > I.e. > 1) removed R193 and R194 on the carrier card (rev 1.2) > 2) removed RN1 on the CPU daughter card > 2a) connected pad3 of RN1 to pin3 of U12 (IRQ0) > 2b) connected pad2 of RN1 to pin4 of U12 (IRQ1) > > Does this fix the interrupt polarity problem (as well)?
Yes, I think so. You can plug a PCIe ethernet card to test it. > > We applied the provided kernel patch as well. > IMHO that patch just moves the local PCIe root-complex > "out-of-space" so no detection of that one occurs any more. > This is what actually happens when "lspci" is run. > > However we still don't see any devices behind the PCIe switch > (e.g. a transparent PLX8516). It seems that the enumeration > process (traversing through the bus hierarchy) in the kernel > is somehow disabled. :-), Maybe it's need more study. Could you enable the DEBUG and post the kernel verbose message? > Bottom line: Only one device accessible at all on the PCIe port. > > Any further ideas? > > Thanks, > > Florian > > _______________________________________________ > Linuxppc-embedded mailing list > Linuxppc-embedded at ozlabs.org > https://ozlabs.org/mailman/listinfo/linuxppc-embedded >