ACPI: PCI: Rework acpi_get_pci_dev()
authorRafael J. Wysocki <rafael.j.wysocki@intel.com>
Sat, 10 Sep 2022 13:23:46 +0000 (15:23 +0200)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Thu, 22 Sep 2022 19:11:58 +0000 (21:11 +0200)
The PCI device returned by acpi_get_pci_dev() needs to be registered,
so if it corresponds to an ACPI device object, the struct acpi_device
representing that object must be registered too and, moreover, it
should be the ACPI companion of the given PCI device.  Thus it should
be sufficient to look for it in the ACPI device object's list of
physical nodes associated with it.

Modify the code accordingly.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/acpi/pci_root.c

index d57cf8454b93e3be6a67546a3a654458c8e71ddf..c8385ef54c370947686eaedb4a6cf24ca1d84d65 100644 (file)
@@ -312,76 +312,25 @@ struct acpi_handle_node {
  */
 struct pci_dev *acpi_get_pci_dev(acpi_handle handle)
 {
-       int dev, fn;
-       unsigned long long adr;
-       acpi_status status;
-       acpi_handle phandle;
-       struct pci_bus *pbus;
-       struct pci_dev *pdev = NULL;
-       struct acpi_handle_node *node, *tmp;
-       struct acpi_pci_root *root;
-       LIST_HEAD(device_list);
-
-       /*
-        * Walk up the ACPI CA namespace until we reach a PCI root bridge.
-        */
-       phandle = handle;
-       while (!acpi_is_root_bridge(phandle)) {
-               node = kzalloc(sizeof(struct acpi_handle_node), GFP_KERNEL);
-               if (!node)
-                       goto out;
-
-               INIT_LIST_HEAD(&node->node);
-               node->handle = phandle;
-               list_add(&node->node, &device_list);
-
-               status = acpi_get_parent(phandle, &phandle);
-               if (ACPI_FAILURE(status))
-                       goto out;
-       }
-
-       root = acpi_pci_find_root(phandle);
-       if (!root)
-               goto out;
+       struct acpi_device *adev = acpi_fetch_acpi_dev(handle);
+       struct acpi_device_physical_node *pn;
+       struct pci_dev *pci_dev = NULL;
 
-       pbus = root->bus;
-
-       /*
-        * Now, walk back down the PCI device tree until we return to our
-        * original handle. Assumes that everything between the PCI root
-        * bridge and the device we're looking for must be a P2P bridge.
-        */
-       list_for_each_entry(node, &device_list, node) {
-               acpi_handle hnd = node->handle;
-               status = acpi_evaluate_integer(hnd, "_ADR", NULL, &adr);
-               if (ACPI_FAILURE(status))
-                       goto out;
-               dev = (adr >> 16) & 0xffff;
-               fn  = adr & 0xffff;
-
-               pdev = pci_get_slot(pbus, PCI_DEVFN(dev, fn));
-               if (!pdev || hnd == handle)
-                       break;
+       if (!adev)
+               return NULL;
 
-               pbus = pdev->subordinate;
-               pci_dev_put(pdev);
+       mutex_lock(&adev->physical_node_lock);
 
-               /*
-                * This function may be called for a non-PCI device that has a
-                * PCI parent (eg. a disk under a PCI SATA controller).  In that
-                * case pdev->subordinate will be NULL for the parent.
-                */
-               if (!pbus) {
-                       dev_dbg(&pdev->dev, "Not a PCI-to-PCI bridge\n");
-                       pdev = NULL;
+       list_for_each_entry(pn, &adev->physical_node_list, node) {
+               if (dev_is_pci(pn->dev)) {
+                       pci_dev = to_pci_dev(pn->dev);
                        break;
                }
        }
-out:
-       list_for_each_entry_safe(node, tmp, &device_list, node)
-               kfree(node);
 
-       return pdev;
+       mutex_unlock(&adev->physical_node_lock);
+
+       return pci_dev;
 }
 EXPORT_SYMBOL_GPL(acpi_get_pci_dev);