On Wed, Apr 17, 2024 at 05:38:05PM +0200, Niklas Cassel wrote:
> On Wed, Apr 17, 2024 at 03:47:48PM +0300, Dan Carpenter wrote:
> > Hello Manivannan Sadhasivam,
> >
> > Commit a01e7214bef9 ("PCI: endpoint: Remove "core_init_notifier"
> > flag") from Mar 27, 2024 (linux-next), leads to the following Smatch
> > static checker warning:
> >
> > drivers/pci/endpoint/functions/pci-epf-test.c:784 pci_epf_test_core_init()
> > error: we previously assumed 'epc_features' could be null (see line 747)
> >
> > drivers/pci/endpoint/functions/pci-epf-test.c
> > 734 static int pci_epf_test_core_init(struct pci_epf *epf)
> > 735 {
> > 736 struct pci_epf_test *epf_test = epf_get_drvdata(epf);
> > 737 struct pci_epf_header *header = epf->header;
> > 738 const struct pci_epc_features *epc_features;
> > 739 struct pci_epc *epc = epf->epc;
> > 740 struct device *dev = &epf->dev;
> > 741 bool linkup_notifier = false;
> > 742 bool msix_capable = false;
> > 743 bool msi_capable = true;
> > 744 int ret;
>
> We check pci_epc_get_features() in ->bind(), which comes before ->core_init()
> so in practice, this shouldn't be able to be NULL here.
>
Right.
>
> > 745
> > 746 epc_features = pci_epc_get_features(epc, epf->func_no, epf->vfunc_no);
> > 747 if (epc_features) {
> > ^^^^^^^^^^^^
> > epc_features can be NULL
>
> We should probably just chge this to:
>
> """
> if (!epc_features)
> return some_error;
>
> msix_capable = epc_features->msix_capable;
> msi_capable = epc_features->msi_capable;
> """
>
> Such that we don't need another check further down for linkup_notifier.
>
No. We should just remove this NULL check since the check is already present in
bind().
- Mani
--
மணிவண்ணன் சதாசிவம்
