Resending the email to kernelnewbies list as well (have sent the same to edac mailing list). Appreciate any pointers on below query, if any one has worked on edac. On Wednesday 06 July 2011 03:56 PM, nilesh wrote:
Hi,
I am fairly new to the edac driver and have some basic query regarding the error detection in case of PCI devices. I could see the module for memory controller's error detection is done s.t. UE/CE/controller specific errors are scanned and detected, also the PCI errors are also scanned.
E.g. in case of i5400 : i5400_edac.c : i5400_probe1() {
1285 /* Set the function pointer to an actual operation function */ 1286 mci->edac_check = i5400_check_error; ... 1312 /* allocating generic PCI control info */ 1313 i5400_pci = edac_pci_create_generic_ctl(&pdev->dev, EDAC_MOD_STR); }
And we check the PCI status registers for the errors. edac_pci_generic_check() --> edac_pci_do_parity_check() --> edac_pci_dev_parity_test()
So, i5400_check_error - detect UE/CE. edac_pci_generic_check - detect PCI errors.
Query: ------ * If one wants to get the errors harvested for non-MC device (e.g. ethernet controller etc.) is it going to be done through edac_pci_generic_check() automatically - as we already scan through all PCI devices in a loop? Is it required to add the separate module for that device?
-- Thanks, Nilesh