6.6-stable review patch. If anyone has any objections, please let me know.
------------------
From: Vitaly Lifshits vitaly.lifshits@intel.com
[ Upstream commit 0a6ad4d9e1690c7faa3a53f762c877e477093657 ]
Occasionally when the system goes into pm_suspend, the suspend might fail due to a PHY access error on the network adapter. Previously, this would have caused the whole system to fail to go to a low power state. An example of this was reported in the following Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=205015
[ 1663.694828] e1000e 0000:00:19.0 eth0: Failed to disable ULP [ 1664.731040] asix 2-3:1.0 eth1: link up, 100Mbps, full-duplex, lpa 0xC1E1 [ 1665.093513] e1000e 0000:00:19.0 eth0: Hardware Error [ 1665.596760] e1000e 0000:00:19.0: pci_pm_resume+0x0/0x80 returned 0 after 2975399 usecs
and then the system never recovers from it, and all the following suspend failed due to this [22909.393854] PM: pci_pm_suspend(): e1000e_pm_suspend+0x0/0x760 [e1000e] returns -2 [22909.393858] PM: dpm_run_callback(): pci_pm_suspend+0x0/0x160 returns -2 [22909.393861] PM: Device 0000:00:1f.6 failed to suspend async: error -2
This can be avoided by changing the return values of __e1000_shutdown and e1000e_pm_suspend functions so that they always return 0 (success). This is consistent with what other drivers do.
If the e1000e driver encounters a hardware error during suspend, potential side effects include slightly higher power draw or non-working wake on LAN. This is preferred to a system-level suspend failure, and a warning message is written to the system log, so that the user can be aware that the LAN controller experienced a problem during suspend.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=205015 Suggested-by: Dima Ruinskiy dima.ruinskiy@intel.com Signed-off-by: Vitaly Lifshits vitaly.lifshits@intel.com Tested-by: Mor Bar-Gabay morx.bar.gabay@intel.com Signed-off-by: Tony Nguyen anthony.l.nguyen@intel.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/net/ethernet/intel/e1000e/netdev.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c index 334f652c60601..d377a286c0e1b 100644 --- a/drivers/net/ethernet/intel/e1000e/netdev.c +++ b/drivers/net/ethernet/intel/e1000e/netdev.c @@ -6672,8 +6672,10 @@ static int __e1000_shutdown(struct pci_dev *pdev, bool runtime) if (adapter->flags2 & FLAG2_HAS_PHY_WAKEUP) { /* enable wakeup by the PHY */ retval = e1000_init_phy_wakeup(adapter, wufc); - if (retval) - return retval; + if (retval) { + e_err("Failed to enable wakeup\n"); + goto skip_phy_configurations; + } } else { /* enable wakeup by the MAC */ ew32(WUFC, wufc); @@ -6694,8 +6696,10 @@ static int __e1000_shutdown(struct pci_dev *pdev, bool runtime) * or broadcast. */ retval = e1000_enable_ulp_lpt_lp(hw, !runtime); - if (retval) - return retval; + if (retval) { + e_err("Failed to enable ULP\n"); + goto skip_phy_configurations; + } }
/* Force SMBUS to allow WOL */ @@ -6744,6 +6748,7 @@ static int __e1000_shutdown(struct pci_dev *pdev, bool runtime) hw->phy.ops.release(hw); }
+skip_phy_configurations: /* Release control of h/w to f/w. If f/w is AMT enabled, this * would have already happened in close and is redundant. */ @@ -6986,15 +6991,13 @@ static __maybe_unused int e1000e_pm_suspend(struct device *dev) e1000e_pm_freeze(dev);
rc = __e1000_shutdown(pdev, false); - if (rc) { - e1000e_pm_thaw(dev); - } else { + if (!rc) { /* Introduce S0ix implementation */ if (adapter->flags2 & FLAG2_ENABLE_S0IX_FLOWS) e1000e_s0ix_entry_flow(adapter); }
- return rc; + return 0; }
static __maybe_unused int e1000e_pm_resume(struct device *dev)