[PATCH v2] PCI/PM: Restore BARs after powering up during enumeration

[Keng-Yu Lin]

Commit 4d4c10f763d7 ("PCI: Explicitly put devices into D0 when
initializing") added pci_pm_power_up_and_verify_state() to pci_pm_init().
The helper ignores pci_power_up()'s return value indicating that BARs
need restoring.  Its resume callers restore saved configuration
afterwards; pci_pm_init() does not.

D3hot-to-D0 may reset a device and lose its BARs (PCI PM 1.2, sec 5.4.1).
After enumeration powers it up, pci_enable_device() reads D0 and skips
the full power-up path.  Previously, a device still in D3hot at enable
time took that path, including BAR restoration.

On AMD Venice SP7, two AMD xHCI controllers [1022:17d2] are in runtime
D3hot with No_Soft_Reset clear before kdump.  Diagnostic reads on
v7.3-rc6-37 show the following for 0000:01:00.4 in the crash kernel:

                                   PMCSR      BAR0
  before the platform power call   0003       f4100004
  after the platform power call    0003       f4100004
  after native D0 power-up         0000       00000004

The upper BAR DWORD stays zero.  pci_power_up() returns 1, but the BAR
address remains zero while dev->resource[0] still contains 0xf4100000.
Without BAR restoration, the last console messages are around:

  pci 0000:01:00.4: enabling device (0000 -> 0002)

Restoring both controllers' BARs gives matching readbacks and allows
kdump to save a vmcore and reboot.  An AMD Turin xHCI [1022:1557] shows
the same BAR loss; restoring its BAR also allows kdump to complete.

Use pci_set_full_power_state() in pci_pm_init() to power up the device
and restore its BARs when needed.  It already checks the native power
state before restoring and updates ASPM when link state is present.
Leave pci_pm_power_up_and_verify_state() unchanged for the resume
callers, which restore saved configuration afterwards.

Read the initial power state first, as pci_enable_device_flags() does.
Otherwise, PCI_UNKNOWN can request a BAR restore even for devices
already in D0.  Keep the final pci_update_current_state() call so the
final cached state still accounts for platform power and failed config
reads.

Suggested-by: Mario Limonciello <mario.limonciello@xxxxxxx>
Fixes: 4d4c10f763d7 ("PCI: Explicitly put devices into D0 when initializing")
Cc: stable@xxxxxxxxxxxxxxx
Signed-off-by: Keng-Yu Lin <keng-yu.lin@xxxxxxx>
---

Changes in v2:
- Use pci_set_full_power_state() in pci_pm_init(), following Mario's
  suggestion to reuse the existing BAR restore code.
- Leave the shared resume helper and its interface unchanged.
- Keep the initial state readback and the final platform-aware refresh.

Unlike v1, BAR restoration uses the full helper's native PMCSR check.
It also updates ASPM when link state exists.

Tested v2 with kdump on both systems.  Normal boots made no BAR writes
from this path.  The crash kernels wrote only the cleared xHCI BARs:
two on Venice SP7 and one on Turin.  All readbacks matched.  kdump saved
2.53 GB and 1.10 GB vmcores respectively before rebooting.

v1: https://lore.kernel.org/r/20261008133708.947908-1-keng-yu.lin@xxxxxxx/
Review: https://lore.kernel.org/r/04f5463b-f123-4079-9286-06a241976d3a@xxxxxxx/

 drivers/pci/pci.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 8cc6a89130b5..d3a5871e1b36 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -3255,7 +3255,9 @@ void pci_pm_init(struct pci_dev *dev)
 	}
 
 poweron:
-	pci_pm_power_up_and_verify_state(dev);
+	pci_update_current_state(dev, dev->current_state);
+	pci_set_full_power_state(dev, false);
+	pci_update_current_state(dev, PCI_D0);
 	pm_runtime_forbid(&dev->dev);
 
 	/*

base-commit: 7b63ef2d55f24519e7e9e5f4d15dbea03f126e40
-- 
2.43.0