> From: Intel-wired-lan <intel-wired-lan-boun...@osuosl.org> On Behalf Of > Vaibhav Gupta > Sent: Monday, June 29, 2020 2:30 AM > To: Bjorn Helgaas <helg...@kernel.org>; Bjorn Helgaas > <bhelg...@google.com>; bj...@helgaas.com; Vaibhav Gupta > <vaibhav.varo...@gmail.com>; David S. Miller <da...@davemloft.net>; > Jakub Kicinski <k...@kernel.org>; Kirsher, Jeffrey T > <jeffrey.t.kirs...@intel.com> > Cc: Vaibhav Gupta <vaibhavgupt...@gmail.com>; net...@vger.kernel.org; > linux-kernel@vger.kernel.org; intel-wired-...@lists.osuosl.org; > sk...@linuxfoundation.org; linux-kernel-ment...@lists.linuxfoundation.org > Subject: [Intel-wired-lan] [PATCH v1 5/5] e100: use generic power management > > With legacy PM hooks, it was the responsibility of a driver to manage PCI > states and also the device's power state. The generic approach is to let > PCI core handle the work. > > e100_suspend() calls __e100_shutdown() to perform intermediate tasks. > __e100_shutdown() calls pci_save_state() which is not recommended. > > e100_suspend() also calls __e100_power_off() which is calling PCI helper > functions, pci_prepare_to_sleep(), pci_set_power_state(), along with > pci_wake_from_d3(...,false). Hence, the functin call is removed and wol is > disabled as earlier using device_wakeup_disable(). > > Compile-tested only. > > Signed-off-by: Vaibhav Gupta <vaibhavgupt...@gmail.com> > --- > drivers/net/ethernet/intel/e100.c | 31 +++++++++++++------------------ > 1 file changed, 13 insertions(+), 18 deletions(-)
I do have several e100 based adapters still working and a few old systems with plain old PCI that still function, however all of these older systems have broken power management. Regardless of if I use the kernel before or after this patch is applied, or even if the e100 driver is loaded or not I can't get a reliable suspend / resume cycle to work on them. I did run some basic regression with this patch against the remaining pro100 cards I could scrounge up and aside from broken power management (again with or without patch) the system seems good, so (hesitantly) from a regression perspective I will go ahead and say... Tested-by: Aaron Brown <aaron.f.br...@intel.com>
