Re: [patch v2] tg3: Disable tg3 PCIe AER on system reboot
From: Lenny Szubowicz
Date: Thu Jan 30 2025 - 14:40:46 EST
On 12/2/24 2:00 AM, Pavan Chebbi wrote:
On Sat, Nov 30, 2024 at 2:06 AM Lenny Szubowicz <lszubowi@xxxxxxxxxx> wrote:
Disable PCIe AER on the tg3 device on system reboot on a limited
list of Dell PowerEdge systems. This prevents a fatal PCIe AER event
on the tg3 device during the ACPI _PTS (prepare to sleep) method for
S5 on those systems. The _PTS is invoked by acpi_enter_sleep_state_prep()
as part of the kernel's reboot sequence as a result of commit
38f34dba806a ("PM: ACPI: reboot: Reinstate S5 for reboot").
There was an earlier fix for this problem by commit 2ca1c94ce0b6
("tg3: Disable tg3 device on system reboot to avoid triggering AER").
Are you saying that if we have tg3_power_down() done then the current
new issue won't be seen?
First, thank you for your review and sorry for not responding sooner.
I did not see your comments until now because I had some unhelpful
email filters in place.
Yes, if tg3_power_down() is called on a restart, then the PCIe AER does
not occur. The other calls added by upstream commit 2ca1c94ce0b6 ("tg3:
Disable tg3 device on system reboot to avoid triggering AER"), i.e.
tg3_reset_task_cancel() and pci_disable_device(), are not
sufficient to prevent the AER.
But it was discovered that this earlier fix caused a reboot hang
when some Dell PowerEdge servers were booted via ipxe. To address
this reboot hang, the earlier fix was essentially reverted by commit
9fc3bc764334 ("tg3: power down device only on SYSTEM_POWER_OFF").
This re-exposed the tg3 PCIe AER on reboot problem.
This fix is not an ideal solution because the root cause of the AER
is in system firmware. Instead, it's a targeted work-around in the
tg3 driver.
Note also that the PCIe AER must be disabled on the tg3 device even
if the system is configured to use "firmware first" error handling.
Not too sure about this. The list has some widely used latest Dell
servers. The first fix only did pci_disable_device()
But looks like this fix should be the right one for the first ever
reported issue in commit 2ca1c94ce0b6 ?
I have reproduced this problem on an example system for every entry in
the DMI match table tg3_restart_aer_quirk_table that is in the patch.
I don't have access to all models of Dell PowerEdge servers. So that
list might not be comprehensive. I was not able to reproduce the
problem on newer PowerEdge servers that I tested.
I reproduced the problem this morning with the latest upstream
kernel on a Dell PowerEdge R640 with the latest BIOS system firmware.
root@dell-per640-02:~# uname -a
Linux dell-per640-02.khw.eng.rdu2.dc.redhat.com 6.13.0.lss001+ #23 SMP PREEMPT_DYNAMIC Thu Jan 30 08:56:44 EST 2025 x86_64 GNU/Linux
root@dell-per640-02:~# lspci -s 01:00.1
01:00.1 Ethernet controller: Broadcom Inc. and subsidiaries NetXtreme BCM5720 Gigabit Ethernet PCIe
root@dell-per640-02:~# dmidecode
...
System Information
Manufacturer: Dell Inc.
Product Name: PowerEdge R640
...
BIOS Information
Vendor: Dell Inc.
Version: 2.22.2
Release Date: 09/12/2024
root@dell-per640-02:~# reboot
...
[ 68.497693] ACPI: PM: Preparing to enter system sleep state S5
[ 71.100607] {1}[Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 5
[ 71.100610] {1}[Hardware Error]: event severity: fatal
[ 71.100611] {1}[Hardware Error]: Error 0, type: fatal
[ 71.100613] {1}[Hardware Error]: section_type: PCIe error
[ 71.100614] {1}[Hardware Error]: port_type: 0, PCIe end point
[ 71.100616] {1}[Hardware Error]: version: 3.0
[ 71.100618] {1}[Hardware Error]: command: 0x0002, status: 0x0010
[ 71.100619] {1}[Hardware Error]: device_id: 0000:01:00.1
[ 71.100621] {1}[Hardware Error]: slot: 0
[ 71.100622] {1}[Hardware Error]: secondary_bus: 0x00
[ 71.100623] {1}[Hardware Error]: vendor_id: 0x14e4, device_id: 0x165f
[ 71.100625] {1}[Hardware Error]: class_code: 020000
[ 71.100626] {1}[Hardware Error]: aer_cor_status: 0x00002000, aer_cor_mask: 0x000031c0
[ 71.100627] {1}[Hardware Error]: aer_uncor_status: 0x00100000, aer_uncor_mask: 0x00010000
[ 71.100629] {1}[Hardware Error]: aer_uncor_severity: 0x000ef030
[ 71.100630] {1}[Hardware Error]: TLP Header: 40000001 0000030f 90028090 00000000
[ 71.100632] GHES: Fatal hardware error but panic disabled
[ 71.100633] Kernel panic - not syncing: GHES: Fatal hardware error
[ 71.100635] CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Not tainted 6.13.0.lss001+ #23
[ 71.100638] Hardware name: Dell Inc. PowerEdge R640/0W23H8, BIOS 2.22.2 09/12/2024
[ 71.100640] Call Trace:
[ 71.100642] <NMI>
[ 71.100643] dump_stack_lvl+0x4e/0x70
[ 71.100649] panic+0x113/0x2dd
[ 71.100654] __ghes_panic.cold+0x28/0x28
[ 71.100657] ghes_in_nmi_queue_one_entry.constprop.0+0x23f/0x2c0
[ 71.100662] ghes_notify_nmi+0x5d/0xd0
[ 71.100665] nmi_handle+0x5e/0x120
[ 71.100668] default_do_nmi+0x40/0x130
[ 71.100673] exc_nmi+0x103/0x180
[ 71.100676] end_repeat_nmi+0xf/0x53
[ 71.100681] RIP: 0010:intel_idle+0x59/0xa0
...
Also you may want to address the warnings generated. Also note that
netdev requires you to wait 24hours before posting a new revision of
the patch.
Yes, I'll have a fix for the sparse warnings in V3 of the patch.
That was a sloppy error on my part.
Fixes: 9fc3bc764334 ("tg3: power down device only on SYSTEM_POWER_OFF")
Signed-off-by: Lenny Szubowicz <lszubowi@xxxxxxxxxx>
---
drivers/net/ethernet/broadcom/tg3.c | 59 +++++++++++++++++++++++++++++
1 file changed, 59 insertions(+)
diff --git a/drivers/net/ethernet/broadcom/tg3.c b/drivers/net/ethernet/broadcom/tg3.c
index 9cc8db10a8d6..12ae5a976ca7 100644
--- a/drivers/net/ethernet/broadcom/tg3.c
+++ b/drivers/net/ethernet/broadcom/tg3.c
@@ -55,6 +55,7 @@
#include <linux/hwmon.h>
#include <linux/hwmon-sysfs.h>
#include <linux/crc32poly.h>
+#include <linux/dmi.h>
#include <net/checksum.h>
#include <net/gso.h>
@@ -18192,6 +18193,51 @@ static int tg3_resume(struct device *device)
static SIMPLE_DEV_PM_OPS(tg3_pm_ops, tg3_suspend, tg3_resume);
+/*
+ * Systems where ACPI _PTS (Prepare To Sleep) S5 will result in a fatal
+ * PCIe AER event on the tg3 device if the tg3 device is not, or cannot
+ * be, powered down.
+ */
+static const struct dmi_system_id tg3_restart_aer_quirk_table[] = {
+ {
+ .matches = {
+ DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+ DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R440"),
+ },
+ },
+ {
+ .matches = {
+ DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+ DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R540"),
+ },
+ },
+ {
+ .matches = {
+ DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+ DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R640"),
+ },
+ },
+ {
+ .matches = {
+ DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+ DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R650"),
+ },
+ },
+ {
+ .matches = {
+ DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+ DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R740"),
+ },
+ },
+ {
+ .matches = {
+ DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
+ DMI_MATCH(DMI_PRODUCT_NAME, "PowerEdge R750"),
+ },
+ },
+ {}
+};
+
static void tg3_shutdown(struct pci_dev *pdev)
{
struct net_device *dev = pci_get_drvdata(pdev);
@@ -18208,6 +18254,19 @@ static void tg3_shutdown(struct pci_dev *pdev)
if (system_state == SYSTEM_POWER_OFF)
tg3_power_down(tp);
+ else if (system_state == SYSTEM_RESTART &&
+ dmi_first_match(tg3_restart_aer_quirk_table) &&
+ pdev->current_state <= PCI_D3hot) {
+ /*
+ * Disable PCIe AER on the tg3 to avoid a fatal
+ * error during this system restart.
+ */
+ pcie_capability_clear_word(pdev, PCI_EXP_DEVCTL,
+ PCI_EXP_DEVCTL_CERE |
+ PCI_EXP_DEVCTL_NFERE |
+ PCI_EXP_DEVCTL_FERE |
+ PCI_EXP_DEVCTL_URRE);
+ }
rtnl_unlock();
--
2.45.2