From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([209.132.183.28]:21087 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932970Ab2AIVKJ (ORCPT ); Mon, 9 Jan 2012 16:10:09 -0500 Date: Mon, 9 Jan 2012 16:10:06 -0500 From: MUNEDA Takahiro To: linux-pci@vger.kernel.org Cc: MUNEDA Takahiro Message-Id: <20120109211006.2135.13551.sendpatchset@dhcp-189-101.bos.redhat.com> Subject: [PATCH v3] Add pcie_hp=nomsi to disable MSI/MSI-X for pciehp driver Sender: linux-pci-owner@vger.kernel.org List-ID: Not to use MSI/MSI-X for pciehp, add a kernel parameter, pcie_hp=nomsi. In my environment, while shutting down, following stack trace is shown sometimes. irq 16: nobody cared (try booting with the "irqpoll" option) Pid: 1081, comm: reboot Not tainted 3.2.0 #1 Call Trace: [] __report_bad_irq+0x3d/0xe0 [] note_interrupt+0x15c/0x210 [] handle_irq_event_percpu+0xb5/0x210 [] handle_irq_event+0x41/0x70 [] handle_fasteoi_irq+0x55/0xc0 [] handle_irq+0x46/0xb0 [] do_IRQ+0x5d/0xe0 [] common_interrupt+0x6e/0x6e [] ? __do_softirq+0x60/0x210 [] ? hrtimer_interrupt+0x151/0x240 [] call_softirq+0x1c/0x30 [] do_softirq+0x65/0xa0 [] irq_exit+0xbd/0xe0 [] smp_apic_timer_interrupt+0x6e/0x99 [] apic_timer_interrupt+0x6e/0x80 [] ? _raw_spin_unlock_irqrestore+0x11/0x20 [] pci_bus_write_config_word+0x6c/0x80 [] pci_intx+0x52/0xa0 [] pci_intx_for_msi+0x1d/0x30 [] pci_msi_shutdown+0x7b/0x110 [] pci_device_shutdown+0x34/0x50 [] device_shutdown+0x2f/0x140 [] kernel_restart_prepare+0x31/0x40 [] kernel_restart+0x16/0x60 [] sys_reboot+0x1ad/0x220 [] ? do_page_fault+0x1e0/0x460 [] ? __sync_filesystem+0x90/0x90 [] ? __cond_resched+0x2a/0x40 [] ? _cond_resched+0x30/0x40 [] ? iterate_supers+0xb7/0xd0 [] system_call_fastpath+0x16/0x1b handlers: [] usb_hcd_irq [] usb_hcd_irq [] usb_hcd_irq Disabling IRQ #16 An un-wanted interrupt is generated when PCI driver switches from MSI/MSI-X to INTx while shutting down the device. The interrupt does not happen if MSI/MSI-X is not used on the device. I confirmed that this problem does not happen if pcie_hp=nomsi was specified and hotplug operation worked fine as usual. (From v2, the pcie_hp=nomsi does not need any more against the device 111d:807f, since MSI/MSI-X is disabled automatically) v2: Automatically disable MSI/MSI-X against following device: PCI bridge: Integrated Device Technology, Inc. Device 807f (rev 02) v3: Based on the review comment, combile the if statements. Signed-off-by: MUNEDA Takahiro --- Documentation/kernel-parameters.txt | 4 ++++ drivers/pci/hotplug/pciehp_core.c | 12 ++++++++++++ drivers/pci/pcie/portdrv.h | 12 ++++++++++++ drivers/pci/pcie/portdrv_core.c | 7 +++++-- 4 files changed, 33 insertions(+), 2 deletions(-) Index: linux-3.2/drivers/pci/hotplug/pciehp_core.c =================================================================== --- linux-3.2.orig/drivers/pci/hotplug/pciehp_core.c +++ linux-3.2/drivers/pci/hotplug/pciehp_core.c @@ -42,6 +42,7 @@ int pciehp_debug; int pciehp_poll_mode; int pciehp_poll_time; int pciehp_force; +int pciehp_msi_disabled; struct workqueue_struct *pciehp_wq; struct workqueue_struct *pciehp_ordered_wq; @@ -57,10 +58,12 @@ module_param(pciehp_debug, bool, 0644); module_param(pciehp_poll_mode, bool, 0644); module_param(pciehp_poll_time, int, 0644); module_param(pciehp_force, bool, 0644); +module_param(pciehp_msi_disabled, bool, 0644); MODULE_PARM_DESC(pciehp_debug, "Debugging mode enabled or not"); MODULE_PARM_DESC(pciehp_poll_mode, "Using polling mechanism for hot-plug events or not"); MODULE_PARM_DESC(pciehp_poll_time, "Polling mechanism frequency, in seconds"); MODULE_PARM_DESC(pciehp_force, "Force pciehp, even if OSHP is missing"); +MODULE_PARM_DESC(pciehp_msi, "MSI/MSI-X enabled or not"); #define PCIE_MODULE_NAME "pciehp" @@ -72,6 +75,15 @@ static int get_attention_status (struct static int get_latch_status (struct hotplug_slot *slot, u8 *value); static int get_adapter_status (struct hotplug_slot *slot, u8 *value); +static int __init pciehp_setup(char *str) +{ + if (!strncmp(str, "nomsi", 5)) + pciehp_msi_disabled = true; + + return 1; +} +__setup("pcie_hp=", pciehp_setup); + /** * release_slot - free up the memory used by a slot * @hotplug_slot: slot to free Index: linux-3.2/drivers/pci/pcie/portdrv.h =================================================================== --- linux-3.2.orig/drivers/pci/pcie/portdrv.h +++ linux-3.2/drivers/pci/pcie/portdrv.h @@ -34,6 +34,18 @@ struct pci_dev; extern void pcie_clear_root_pme_status(struct pci_dev *dev); +#ifdef CONFIG_HOTPLUG_PCI_PCIE +extern int pciehp_msi_disabled; + +static inline bool pciehp_no_msi(void) +{ + return pciehp_msi_disabled; +} + +#else /* !CONFIG_HOTPLUG_PCI_PCIE */ +static inline bool pciehp_no_msi(void) { return false; } +#endif /* !CONFIG_HOTPLUG_PCI_PCIE */ + #ifdef CONFIG_PCIE_PME extern bool pcie_pme_msi_disabled; Index: linux-3.2/drivers/pci/pcie/portdrv_core.c =================================================================== --- linux-3.2.orig/drivers/pci/pcie/portdrv_core.c +++ linux-3.2/drivers/pci/pcie/portdrv_core.c @@ -189,8 +189,11 @@ static int init_service_irqs(struct pci_ { int i, irq = -1; - /* We have to use INTx if MSI cannot be used for PCIe PME. */ - if ((mask & PCIE_PORT_SERVICE_PME) && pcie_pme_no_msi()) { + /* We have to use INTx if MSI cannot be used for PCIe PME or pciehp. */ + if (((mask & PCIE_PORT_SERVICE_PME) && pcie_pme_no_msi()) || + ((mask & PCIE_PORT_SERVICE_HP) && pciehp_no_msi()) || + ((mask & PCIE_PORT_SERVICE_HP) && + (dev->vendor == PCI_VENDOR_ID_IDT && dev->device == 0x807f))) { if (dev->pin) irq = dev->irq; goto no_msi; Index: linux-3.2/Documentation/kernel-parameters.txt =================================================================== --- linux-3.2.orig/Documentation/kernel-parameters.txt +++ linux-3.2/Documentation/kernel-parameters.txt @@ -2076,6 +2076,10 @@ bytes respectively. Such letter suffixes force Enable ASPM even on devices that claim not to support it. WARNING: Forcing ASPM on may cause system lockups. + pcie_hp= [PCIE] PCI Express Hotplug driver options: + nomsi Do not use MSI for PCI Express Native Hotplug (this + makes all PCIe ports use INTx for hotplug services). + pcie_ports= [PCIE] PCIe ports handling: auto Ask the BIOS whether or not to use native PCIe services associated with PCIe ports (PME, hot-plug, AER). Use