On 09/14/2020 12:31 PM, Huacai Chen wrote:
Hi, Tiezhu------------------ Original ------------------ From: "Tiezhu Yang"<yangtie...@loongson.cn>; Date: Mon, Sep 14, 2020 04:29 AM To: "Bjorn Helgaas"<bhelg...@google.com>; Cc: "Konstantin Khlebnikov"<khlebni...@openvz.org>; "Khalid Aziz"<khalid.a...@oracle.com>; "Vivek Goyal"<vgo...@redhat.com>; "Lukas Wunner"<lu...@wunner.de>; "oohall"<ooh...@gmail.com>; "rafael.j.wysocki"<rafael.j.wyso...@intel.com>; "Xuefeng Li"<lixuef...@loongson.cn>; "Huacai Chen"<che...@lemote.com>; "Jiaxun Yang"<jiaxun.y...@flygoat.com>; "linux-pci"<linux-...@vger.kernel.org>; "linux-kernel"<linux-kernel@vger.kernel.org>; Subject: [RFC PATCH v2] PCI/portdrv: Only disable Bus Master on kexec reboot and connected PCI devices After commit 745be2e700cd ("PCIe: portdrv: call pci_disable_device during remove") and commit cc27b735ad3a ("PCI/portdrv: Turn off PCIe services during shutdown"), it also calls pci_disable_device() during shutdown, this leads to shutdown or reboot failure occasionally due to clear PCI_COMMAND_MASTER on the device in do_pci_disable_device(). drivers/pci/pci.c static void do_pci_disable_device(struct pci_dev *dev) { u16 pci_command; pci_read_config_word(dev, PCI_COMMAND, &pci_command); if (pci_command & PCI_COMMAND_MASTER) { pci_command &= ~PCI_COMMAND_MASTER; pci_write_config_word(dev, PCI_COMMAND, pci_command); } pcibios_disable_device(dev); } When remove "pci_command &= ~PCI_COMMAND_MASTER;", it can work well. As Oliver O'Halloran said, no need to call pci_disable_device() when actually shutting down, but we should call pci_disable_device() before handing over to the new kernel on kexec reboot, so we can do some condition checks which are similar with pci_device_shutdown(), this is done by commit 4fc9bbf98fd6 ("PCI: Disable Bus Master only on kexec reboot") and commit 6e0eda3c3898 ("PCI: Don't try to disable Bus Master on disconnected PCI devices"). drivers/pci/pci-driver.c static void pci_device_shutdown(struct device *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 * devices in D3cold or unknown states. * If it is not a kexec reboot, firmware will hit the PCI * devices with big hammer and stop their DMA any way. */ if (kexec_in_progress && (pci_dev->current_state <= PCI_D3hot)) pci_clear_master(pci_dev); }Have you really tried kexec? Why do you think kexec can disable pci device successfully while normal reboot/poweroff cannot?
Yes, I test it on kexec reboot, it works well.
HuacaiSigned-off-by: Tiezhu Yang <yangtie...@loongson.cn> --- drivers/pci/pcie/portdrv_core.c | 1 - drivers/pci/pcie/portdrv_pci.c | 25 ++++++++++++++++++++++++- 2 files changed, 24 insertions(+), 2 deletions(-) diff --git a/drivers/pci/pcie/portdrv_core.c b/drivers/pci/pcie/portdrv_core.c index 50a9522..1991aca 100644 --- a/drivers/pci/pcie/portdrv_core.c +++ b/drivers/pci/pcie/portdrv_core.c @@ -491,7 +491,6 @@ void pcie_port_device_remove(struct pci_dev *dev) { device_for_each_child(&dev->dev, NULL, remove_iter); pci_free_irq_vectors(dev); - pci_disable_device(dev); } /** diff --git a/drivers/pci/pcie/portdrv_pci.c b/drivers/pci/pcie/portdrv_pci.c index 3a3ce40..ce89a9e8 100644 --- a/drivers/pci/pcie/portdrv_pci.c +++ b/drivers/pci/pcie/portdrv_pci.c @@ -15,6 +15,7 @@ #include <linux/init.h> #include <linux/aer.h> #include <linux/dmi.h> +#include <linux/kexec.h> #include "../pci.h" #include "portdrv.h" @@ -143,6 +144,28 @@ static void pcie_portdrv_remove(struct pci_dev *dev) } pcie_port_device_remove(dev); + pci_disable_device(dev); +} + +static void pcie_portdrv_shutdown(struct pci_dev *dev) +{ + if (pci_bridge_d3_possible(dev)) { + pm_runtime_forbid(&dev->dev); + pm_runtime_get_noresume(&dev->dev); + pm_runtime_dont_use_autosuspend(&dev->dev); + } + + pcie_port_device_remove(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 + * devices in D3cold or unknown states. + * If it is not a kexec reboot, firmware will hit the PCI + * devices with big hammer and stop their DMA any way. + */ + if (kexec_in_progress && (dev->current_state <= PCI_D3hot)) + pci_disable_device(dev); } static pci_ers_result_t pcie_portdrv_error_detected(struct pci_dev *dev, @@ -211,7 +234,7 @@ static void pcie_portdrv_err_resume(struct pci_dev *dev) .probe = pcie_portdrv_probe, .remove = pcie_portdrv_remove, - .shutdown = pcie_portdrv_remove, + .shutdown = pcie_portdrv_shutdown, .err_handler = &pcie_portdrv_err_handler, -- 1.8.3.1
