On 5/6/2025 2:30 PM, Rafael J. Wysocki wrote:
On Tue, May 6, 2025 at 7:32 PM Mario Limonciello <superm1@xxxxxxxxxx> wrote:
On 5/6/2025 10:50 AM, Rafael J. Wysocki wrote:
On Tue, May 6, 2025 at 6:19 AM Mario Limonciello <superm1@xxxxxxxxxx> wrote:
From: Kai-Heng Feng <kaihengf@xxxxxxxxxx>
Some laptops wake up after poweroff when HP Thunderbolt Dock G4 is
connected.
The following error message can be found during shutdown:
pcieport 0000:00:1d.0: AER: Correctable error message received from 0000:09:04.0
pcieport 0000:09:04.0: PCIe Bus Error: severity=Correctable, type=Data Link Layer, (Receiver ID)
pcieport 0000:09:04.0: device [8086:0b26] error status/mask=00000080/00002000
pcieport 0000:09:04.0: [ 7] BadDLLP
Calling aer_remove() during shutdown can quiesce the error message,
however the spurious wakeup still happens.
The issue won't happen if the device is in D3 before system shutdown, so
putting device to low power state before shutdown to solve the issue.
ACPI Spec 6.5, "7.4.2.5 System \_S4 State" says "Devices states are
compatible with the current Power Resource states. In other words, all
devices are in the D3 state when the system state is S4."
The following "7.4.2.6 System \_S5 State (Soft Off)" states "The S5
state is similar to the S4 state except that OSPM does not save any
context." so it's safe to assume devices should be at D3 for S5.
That's fine as long as you assume that ->shutdown() is only used for
implementing ACPI S5, but it is not.
I suppose you're meaning things like:
kernel_restart_prepare()
->device_shutdown()
->->each device's shutdown() CB
->->each driver's shutdown() CB
Is there somewhere "better" to do this so it's truly only tied to S5?
On systems that use ACPI reboot is also a flavor of S5, but what about
systems that have PCI, but don't use ACPI?
The change you are proposing is going to affect more systems than just
the ones having the problem described in the patch changelog, so it is
kind of risky and breaking reboot is a big deal.
How would you feel about if instead this was using one of the power off
notifiers? That could narrow this down to only being applied on ACPI
systems.
Something like this (not yet tested):
diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
index af370628e5839..df83aed7d474c 100644
--- a/drivers/pci/pci-acpi.c
+++ b/drivers/pci/pci-acpi.c
@@ -10,6 +10,7 @@
#include <linux/delay.h>
#include <linux/init.h>
#include <linux/irqdomain.h>
+#include <linux/kexec.h>
#include <linux/pci.h>
#include <linux/msi.h>
#include <linux/pci_hotplug.h>
@@ -18,6 +19,7 @@
#include <linux/pci-ecam.h>
#include <linux/pm_runtime.h>
#include <linux/pm_qos.h>
+#include <linux/reboot.h>
#include <linux/rwsem.h>
#include "pci.h"
@@ -1438,6 +1440,16 @@ static void pci_acpi_set_external_facing(struct
pci_dev *dev)
dev->external_facing = 1;
}
+static int pci_acpi_power_off(struct sys_off_data *data)
+{
+ struct pci_dev *pci_dev = to_pci_dev(data->dev);
+
+ if (!kexec_in_progress && pci_dev->current_state == PCI_D0)
+ pci_prepare_to_sleep(pci_dev);
+
+ return NOTIFY_DONE;
+}
+
void pci_acpi_setup(struct device *dev, struct acpi_device *adev)
{
struct pci_dev *pci_dev = to_pci_dev(dev);
@@ -1465,6 +1477,11 @@ void pci_acpi_setup(struct device *dev, struct
acpi_device *adev)
if (pci_is_bridge(pci_dev))
acpi_dev_power_up_children_with_adr(adev);
+
+ devm_register_sys_off_handler(dev, SYS_OFF_MODE_POWER_OFF_PREPARE,
+ SYS_OFF_PRIO_DEFAULT,
+ &pci_acpi_power_off, NULL);
+
}
void pci_acpi_cleanup(struct device *dev, struct acpi_device *adev)
Link: https://bugzilla.kernel.org/show_bug.cgi?id=219036
Cc: AceLan Kao <acelan.kao@xxxxxxxxxxxxx>
Reviewed-by: Mario Limonciello <mario.limonciello@xxxxxxx>
Tested-by: Mario Limonciello <mario.limonciello@xxxxxxx>
Signed-off-by: Kai-Heng Feng <kaihengf@xxxxxxxxxx>
Tested-by: Mark Pearson <mpearson-lenovo@xxxxxxxxx>
Tested-by: Denis Benato <benato.denis96@xxxxxxxxx>
Tested-by: Merthan Karakaş <m3rthn.k@xxxxxxxxx>
Link: https://lore.kernel.org/r/20241208074147.22945-1-kaihengf@xxxxxxxxxx
Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxx>
---
v3:
* Pick up tags
* V2 was waiting for Rafael to review, rebase on pci/next and resend.
Is this change going to break kexec?
There is an explicit check in the below code for kexec_in_progress, so
my expectation was that kexec kept working. I didn't explicitly test
this myself when I tested KH's change before sending it again.
KH, Did you double check that on your side?
---
drivers/pci/pci-driver.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index 0c5bdb8c2c07b..5bbe8af996390 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -510,6 +510,14 @@ static void pci_device_shutdown(struct device *dev)
if (drv && drv->shutdown)
drv->shutdown(pci_dev);
+ /*
+ * If driver already changed device's power state, it can mean the
+ * wakeup setting is in place, or a workaround is used. Hence keep it
+ * as is.
+ */
+ if (!kexec_in_progress && pci_dev->current_state == PCI_D0)
+ pci_prepare_to_sleep(pci_dev);
+
/*
* If this is a kexec reboot, turn off Bus Master bit on the
* device to tell it to not continue to do DMA. Don't touch
--
