pci: Fix hotplug of Express Module with pci bridges
authorYinghai Lu <yinghai@kernel.org>
Sun, 18 Dec 2011 02:33:37 +0000 (18:33 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sun, 18 Dec 2011 22:10:16 +0000 (14:10 -0800)
I noticed that hotplug of one setup does not work with recent change in
pci tree.

After checking the bridge conf setup, I noticed that the bridges get
assigned but do not get enabled.

The reason is the following commit, while simply ignores bridge
resources when enabling a pci device:

| commit bbef98ab0f019f1b0c25c1acdf1683c68933d41b
| Author: Ram Pai <linuxram@us.ibm.com>
| Date:   Sun Nov 6 10:33:10 2011 +0800
|
|    PCI: defer enablement of SRIOV BARS
|...
|    NOTE: Note, there is subtle change in the pci_enable_device() API.  Any
|    driver that depends on SRIOV BARS to be enabled in pci_enable_device()
|    can fail.

Put back bridge resource and ROM resource checking to fix the problem.

That should fix regression like BIOS does not assign correct resource to
bridge.

Discussion can be found at:
http://www.spinics.net/lists/linux-pci/msg12874.html

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
drivers/pci/pci.c

index faccb89377068e65d76b251be556fe86faff09ed..6d4a5319148d7eb293eb499cb98cfc56e496f078 100644 (file)
@@ -1129,7 +1129,11 @@ static int __pci_enable_device_flags(struct pci_dev *dev,
        if (atomic_add_return(1, &dev->enable_cnt) > 1)
                return 0;               /* already enabled */
 
-       for (i = 0; i < PCI_ROM_RESOURCE; i++)
+       /* only skip sriov related */
+       for (i = 0; i <= PCI_ROM_RESOURCE; i++)
+               if (dev->resource[i].flags & flags)
+                       bars |= (1 << i);
+       for (i = PCI_BRIDGE_RESOURCES; i < DEVICE_COUNT_RESOURCE; i++)
                if (dev->resource[i].flags & flags)
                        bars |= (1 << i);