Thread (12 messages) 12 messages, 4 authors, 2025-11-17

Re: [PATCH] PCI: cadence: Kconfig: change PCIE_CADENCE configs from tristate to bool

From: Siddharth Vadapalli <s-vadapalli@ti.com>
Date: 2025-11-17 09:22:55
Also in: linux-pci, lkml, stable

On Mon, 2025-11-17 at 10:06 +0100, Arnd Bergmann wrote:
quoted hunk ↗ jump to hunk
On Mon, Nov 17, 2025, at 07:05, Siddharth Vadapalli wrote:
quoted
On Fri, 2025-11-14 at 08:03 +0100, Arnd Bergmann wrote:
quoted
On Fri, Nov 14, 2025, at 06:47, Siddharth Vadapalli wrote:
quoted
I understand that the solution should be fixing the pci-j721e.c driver
rather than updating Kconfig or Makefile. Thank you for the feedback. I
will update the pci-j721e.c driver to handle the case that is triggering
the build error.
Ok, thanks!

I think a single if(IS_ENABLED(CONFIG_PCI_J721E_HOST)) check
is probably enough to avoid the link failure
--- a/drivers/pci/controller/cadence/pci-j721e.c
+++ b/drivers/pci/controller/cadence/pci-j721e.c
@@ -669,7 +669,8 @@ static void j721e_pcie_remove(struct platform_device *pdev)
        struct cdns_pcie_ep *ep;
        struct cdns_pcie_rc *rc;
 
-       if (pcie->mode == PCI_MODE_RC) {
+       if (IS_ENABLED(CONFIG_PCI_J721E_HOST) &&
+           pcie->mode == PCI_MODE_RC) {
                rc = container_of(cdns_pcie, struct cdns_pcie_rc, pcie);
                cdns_pcie_host_disable(rc);

but you may want to split it up further to get better dead
code elimination and prevent similar bugs from reappearing when
another call gets added without this type of check.

If you split j721e_pcie_driver into a host and an ep driver
structure with their own probe/remove callbacks, you can
move the IS_ENABLED() check all the way into module_init()
function.
Thank you for the suggestion :)

Would it work if I send a quick fix for `cdns_pcie_host_disable` using
IS_ENABLED in the existing driver implementation and then send the
refactoring series later? This is to resolve the build error quickly until
the refactoring series is ready.

On the other hand, if it should be fixed the right way by refactoring, I
will not post the temporary fix. Please let me know.

Regards,
Siddharth.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help