| Message ID | CAOMZO5Co_kXpMNX5O66s49HBv3TMnuOWWDekGoecyR+ZozsDyw@mail.gmail.com |
|---|---|
| State | RFC |
| Delegated to: | Simon Glass |
| Headers | show |
On Wed, Oct 7, 2015 at 8:24 PM, Fabio Estevam <festevam@gmail.com> wrote: > On Wed, Oct 7, 2015 at 8:07 PM, Fabio Estevam <festevam@gmail.com> wrote: >> Hi Simon, >> >> On Wed, Oct 7, 2015 at 7:30 PM, Simon Glass <sjg@chromium.org> wrote: >> >>> This added error checking to pci_read_config_word(). It might be worth >>> looking at why the access is failing. >> >> I noticed that when VendorID !=0xffff then pci_read_config_word() returns 0. >> >> In the case when VendorID ==0xffff then it returns -1. >> >> If I do like this: >> >> --- a/common/cmd_pci.c >> +++ b/common/cmd_pci.c >> @@ -77,7 +77,7 @@ void pciinfo(int BusNum, int ShortPCIListing) >> >> ret = pci_read_config_word(dev, PCI_VENDOR_ID, >> &VendorID); >> - if (ret) >> + if (ret && (VendorID != 0xFFFF)) >> goto error; >> if ((VendorID == 0xFFFF) || (VendorID == 0x0000)) >> continue; >> >> Then the error message is not printed. >> >> Not familiar with the PCI code, so any suggestions are welcome. > > If I read the code right it seems that in this loop all the PCI > devices will be scanned. > > When no more devices are found then pci_read_config_word() returns -1 > and VendoID == 0xffff. > > Maybe all we need to do is to filter the error message like this: > > --- a/common/cmd_pci.c > +++ b/common/cmd_pci.c > @@ -100,7 +100,8 @@ void pciinfo(int BusNum, int ShortPCIListing) > > return; > error: > - printf("Cannot read bus configuration: %d\n", ret); > + if (VendorID != 0xFFFF) > + printf("Cannot read bus configuration: %d\n", ret); > } > > Makes sense? Ok, sent a patch that restored the original behaviour.
--- a/common/cmd_pci.c +++ b/common/cmd_pci.c @@ -100,7 +100,8 @@ void pciinfo(int BusNum, int ShortPCIListing) return; error: - printf("Cannot read bus configuration: %d\n", ret); + if (VendorID != 0xFFFF) + printf("Cannot read bus configuration: %d\n", ret); } Makes sense?