From b4c7d2076b4e767dd2e075a2b3a9e57753fc67f5 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Tue, 2 Feb 2021 14:17:54 -0600 Subject: PCI/LINK: Remove bandwidth notification The PCIe Bandwidth Change Notification feature logs messages when the link bandwidth changes. Some users have reported that these messages occur often enough to significantly reduce NVMe performance. GPUs also seem to generate these messages. We don't know why the link bandwidth changes, but in the reported cases there's no indication that it's caused by hardware failures. Remove the bandwidth change notifications for now. Hopefully we can add this back when we have a better understanding of why this happens and how we can make the messages useful instead of overwhelming. Link: https://lore.kernel.org/r/20200115221008.GA191037@google.com/ Link: https://lore.kernel.org/r/155605909349.3575.13433421148215616375.stgit@gimli.home/ Link: https://bugzilla.kernel.org/show_bug.cgi?id=206197 Signed-off-by: Bjorn Helgaas --- drivers/pci/pcie/Kconfig | 8 --- drivers/pci/pcie/Makefile | 1 - drivers/pci/pcie/bw_notification.c | 138 ------------------------------------- drivers/pci/pcie/portdrv.h | 6 -- drivers/pci/pcie/portdrv_pci.c | 1 - 5 files changed, 154 deletions(-) delete mode 100644 drivers/pci/pcie/bw_notification.c (limited to 'drivers/pci/pcie') diff --git a/drivers/pci/pcie/Kconfig b/drivers/pci/pcie/Kconfig index 3946555a6042..45a2ef702b45 100644 --- a/drivers/pci/pcie/Kconfig +++ b/drivers/pci/pcie/Kconfig @@ -133,14 +133,6 @@ config PCIE_PTM This is only useful if you have devices that support PTM, but it is safe to enable even if you don't. -config PCIE_BW - bool "PCI Express Bandwidth Change Notification" - depends on PCIEPORTBUS - help - This enables PCI Express Bandwidth Change Notification. If - you know link width or rate changes occur only to correct - unreliable links, you may answer Y. - config PCIE_EDR bool "PCI Express Error Disconnect Recover support" depends on PCIE_DPC && ACPI diff --git a/drivers/pci/pcie/Makefile b/drivers/pci/pcie/Makefile index d9697892fa3e..b2980db88cc0 100644 --- a/drivers/pci/pcie/Makefile +++ b/drivers/pci/pcie/Makefile @@ -12,5 +12,4 @@ obj-$(CONFIG_PCIEAER_INJECT) += aer_inject.o obj-$(CONFIG_PCIE_PME) += pme.o obj-$(CONFIG_PCIE_DPC) += dpc.o obj-$(CONFIG_PCIE_PTM) += ptm.o -obj-$(CONFIG_PCIE_BW) += bw_notification.o obj-$(CONFIG_PCIE_EDR) += edr.o diff --git a/drivers/pci/pcie/bw_notification.c b/drivers/pci/pcie/bw_notification.c deleted file mode 100644 index 565d23cccb8b..000000000000 --- a/drivers/pci/pcie/bw_notification.c +++ /dev/null @@ -1,138 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0+ -/* - * PCI Express Link Bandwidth Notification services driver - * Author: Alexandru Gagniuc - * - * Copyright (C) 2019, Dell Inc - * - * The PCIe Link Bandwidth Notification provides a way to notify the - * operating system when the link width or data rate changes. This - * capability is required for all root ports and downstream ports - * supporting links wider than x1 and/or multiple link speeds. - * - * This service port driver hooks into the bandwidth notification interrupt - * and warns when links become degraded in operation. - */ - -#define dev_fmt(fmt) "bw_notification: " fmt - -#include "../pci.h" -#include "portdrv.h" - -static bool pcie_link_bandwidth_notification_supported(struct pci_dev *dev) -{ - int ret; - u32 lnk_cap; - - ret = pcie_capability_read_dword(dev, PCI_EXP_LNKCAP, &lnk_cap); - return (ret == PCIBIOS_SUCCESSFUL) && (lnk_cap & PCI_EXP_LNKCAP_LBNC); -} - -static void pcie_enable_link_bandwidth_notification(struct pci_dev *dev) -{ - u16 lnk_ctl; - - pcie_capability_write_word(dev, PCI_EXP_LNKSTA, PCI_EXP_LNKSTA_LBMS); - - pcie_capability_read_word(dev, PCI_EXP_LNKCTL, &lnk_ctl); - lnk_ctl |= PCI_EXP_LNKCTL_LBMIE; - pcie_capability_write_word(dev, PCI_EXP_LNKCTL, lnk_ctl); -} - -static void pcie_disable_link_bandwidth_notification(struct pci_dev *dev) -{ - u16 lnk_ctl; - - pcie_capability_read_word(dev, PCI_EXP_LNKCTL, &lnk_ctl); - lnk_ctl &= ~PCI_EXP_LNKCTL_LBMIE; - pcie_capability_write_word(dev, PCI_EXP_LNKCTL, lnk_ctl); -} - -static irqreturn_t pcie_bw_notification_irq(int irq, void *context) -{ - struct pcie_device *srv = context; - struct pci_dev *port = srv->port; - u16 link_status, events; - int ret; - - ret = pcie_capability_read_word(port, PCI_EXP_LNKSTA, &link_status); - events = link_status & PCI_EXP_LNKSTA_LBMS; - - if (ret != PCIBIOS_SUCCESSFUL || !events) - return IRQ_NONE; - - pcie_capability_write_word(port, PCI_EXP_LNKSTA, events); - pcie_update_link_speed(port->subordinate, link_status); - return IRQ_WAKE_THREAD; -} - -static irqreturn_t pcie_bw_notification_handler(int irq, void *context) -{ - struct pcie_device *srv = context; - struct pci_dev *port = srv->port; - struct pci_dev *dev; - - /* - * Print status from downstream devices, not this root port or - * downstream switch port. - */ - down_read(&pci_bus_sem); - list_for_each_entry(dev, &port->subordinate->devices, bus_list) - pcie_report_downtraining(dev); - up_read(&pci_bus_sem); - - return IRQ_HANDLED; -} - -static int pcie_bandwidth_notification_probe(struct pcie_device *srv) -{ - int ret; - - /* Single-width or single-speed ports do not have to support this. */ - if (!pcie_link_bandwidth_notification_supported(srv->port)) - return -ENODEV; - - ret = request_threaded_irq(srv->irq, pcie_bw_notification_irq, - pcie_bw_notification_handler, - IRQF_SHARED, "PCIe BW notif", srv); - if (ret) - return ret; - - pcie_enable_link_bandwidth_notification(srv->port); - pci_info(srv->port, "enabled with IRQ %d\n", srv->irq); - - return 0; -} - -static void pcie_bandwidth_notification_remove(struct pcie_device *srv) -{ - pcie_disable_link_bandwidth_notification(srv->port); - free_irq(srv->irq, srv); -} - -static int pcie_bandwidth_notification_suspend(struct pcie_device *srv) -{ - pcie_disable_link_bandwidth_notification(srv->port); - return 0; -} - -static int pcie_bandwidth_notification_resume(struct pcie_device *srv) -{ - pcie_enable_link_bandwidth_notification(srv->port); - return 0; -} - -static struct pcie_port_service_driver pcie_bandwidth_notification_driver = { - .name = "pcie_bw_notification", - .port_type = PCIE_ANY_PORT, - .service = PCIE_PORT_SERVICE_BWNOTIF, - .probe = pcie_bandwidth_notification_probe, - .suspend = pcie_bandwidth_notification_suspend, - .resume = pcie_bandwidth_notification_resume, - .remove = pcie_bandwidth_notification_remove, -}; - -int __init pcie_bandwidth_notification_init(void) -{ - return pcie_port_service_register(&pcie_bandwidth_notification_driver); -} diff --git a/drivers/pci/pcie/portdrv.h b/drivers/pci/pcie/portdrv.h index af7cf237432a..2ff5724b8f13 100644 --- a/drivers/pci/pcie/portdrv.h +++ b/drivers/pci/pcie/portdrv.h @@ -53,12 +53,6 @@ int pcie_dpc_init(void); static inline int pcie_dpc_init(void) { return 0; } #endif -#ifdef CONFIG_PCIE_BW -int pcie_bandwidth_notification_init(void); -#else -static inline int pcie_bandwidth_notification_init(void) { return 0; } -#endif - /* Port Type */ #define PCIE_ANY_PORT (~0) diff --git a/drivers/pci/pcie/portdrv_pci.c b/drivers/pci/pcie/portdrv_pci.c index 0b250bc5f405..8bd4992a4f32 100644 --- a/drivers/pci/pcie/portdrv_pci.c +++ b/drivers/pci/pcie/portdrv_pci.c @@ -255,7 +255,6 @@ static void __init pcie_init_services(void) pcie_pme_init(); pcie_dpc_init(); pcie_hp_init(); - pcie_bandwidth_notification_init(); } static int __init pcie_portdrv_init(void) -- cgit v1.2.3 From 7d7cbeaba5b7aea8e1e4eb988d6b5e7cb3c34490 Mon Sep 17 00:00:00 2001 From: Keith Busch Date: Mon, 4 Jan 2021 15:02:56 -0800 Subject: PCI/ERR: Clear status of the reporting device Error handling operates on the first Downstream Port above the detected error, but the error may have been reported by a downstream device. Clear the AER status of the device that reported the error rather than the first Downstream Port. Link: https://lore.kernel.org/r/20210104230300.1277180-2-kbusch@kernel.org Tested-by: Hedi Berriche Signed-off-by: Keith Busch Signed-off-by: Bjorn Helgaas Acked-by: Sean V Kelley Acked-by: Hedi Berriche --- drivers/pci/pcie/err.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) (limited to 'drivers/pci/pcie') diff --git a/drivers/pci/pcie/err.c b/drivers/pci/pcie/err.c index 510f31f0ef6d..a84f0bf4c1e2 100644 --- a/drivers/pci/pcie/err.c +++ b/drivers/pci/pcie/err.c @@ -231,15 +231,14 @@ pci_ers_result_t pcie_do_recovery(struct pci_dev *dev, pci_walk_bridge(bridge, report_resume, &status); /* - * If we have native control of AER, clear error status in the Root - * Port or Downstream Port that signaled the error. If the - * platform retained control of AER, it is responsible for clearing - * this status. In that case, the signaling device may not even be - * visible to the OS. + * If we have native control of AER, clear error status in the device + * that detected the error. If the platform retained control of AER, + * it is responsible for clearing this status. In that case, the + * signaling device may not even be visible to the OS. */ if (host->native_aer || pcie_ports_native) { - pcie_clear_device_status(bridge); - pci_aer_clear_nonfatal_status(bridge); + pcie_clear_device_status(dev); + pci_aer_clear_nonfatal_status(dev); } pci_info(bridge, "device recovery successful\n"); return status; -- cgit v1.2.3 From 7a8a22be35a5058366429e311017e05206c43137 Mon Sep 17 00:00:00 2001 From: Keith Busch Date: Mon, 4 Jan 2021 15:02:57 -0800 Subject: PCI/AER: Clear AER status from Root Port when resetting Downstream Port The pci_dev parameter given to aer_root_reset() may be a Downstream Port rather than the Root Port. Get the Root Port from the provided device in order to clear the root's AER status. Link: https://lore.kernel.org/r/20210104230300.1277180-3-kbusch@kernel.org Tested-by: Hedi Berriche Signed-off-by: Keith Busch Signed-off-by: Bjorn Helgaas Acked-by: Sean V Kelley Acked-by: Hedi Berriche --- drivers/pci/pcie/aer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/pci/pcie') diff --git a/drivers/pci/pcie/aer.c b/drivers/pci/pcie/aer.c index 77b0f2c45bc0..3fd4aaaa627e 100644 --- a/drivers/pci/pcie/aer.c +++ b/drivers/pci/pcie/aer.c @@ -1388,7 +1388,7 @@ static pci_ers_result_t aer_root_reset(struct pci_dev *dev) if (type == PCI_EXP_TYPE_RC_END) root = dev->rcec; else - root = dev; + root = pcie_find_root_port(dev); /* * If the platform retained control of AER, an RCiEP may not have -- cgit v1.2.3 From 387c72cdd7fb6bef650fb078d0f6ae9682abf631 Mon Sep 17 00:00:00 2001 From: Keith Busch Date: Mon, 4 Jan 2021 15:02:58 -0800 Subject: PCI/ERR: Retain status from error notification Overwriting the frozen detected status with the result of the link reset loses the NEED_RESET result that drivers are depending on for error handling to report the .slot_reset() callback. Retain this status so that subsequent error handling has the correct flow. Link: https://lore.kernel.org/r/20210104230300.1277180-4-kbusch@kernel.org Reported-by: Hinko Kocevar Tested-by: Hedi Berriche Signed-off-by: Keith Busch Signed-off-by: Bjorn Helgaas Acked-by: Sean V Kelley Acked-by: Hedi Berriche --- drivers/pci/pcie/err.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'drivers/pci/pcie') diff --git a/drivers/pci/pcie/err.c b/drivers/pci/pcie/err.c index a84f0bf4c1e2..b576aa890c76 100644 --- a/drivers/pci/pcie/err.c +++ b/drivers/pci/pcie/err.c @@ -198,8 +198,7 @@ pci_ers_result_t pcie_do_recovery(struct pci_dev *dev, pci_dbg(bridge, "broadcast error_detected message\n"); if (state == pci_channel_io_frozen) { pci_walk_bridge(bridge, report_frozen_detected, &status); - status = reset_subordinates(bridge); - if (status != PCI_ERS_RESULT_RECOVERED) { + if (reset_subordinates(bridge) != PCI_ERS_RESULT_RECOVERED) { pci_warn(bridge, "subordinate device reset failed\n"); goto failed; } -- cgit v1.2.3 From 33ac78bd3b509d36e7f109a447e28af42e637cb2 Mon Sep 17 00:00:00 2001 From: Keith Busch Date: Mon, 4 Jan 2021 15:02:59 -0800 Subject: PCI/AER: Specify the type of Port that was reset The AER driver may be called upon to reset either a Downstream or a Root Port. Check which type it is to properly identify it when logging that the reset occurred. Link: https://lore.kernel.org/r/20210104230300.1277180-5-kbusch@kernel.org Tested-by: Hedi Berriche Signed-off-by: Keith Busch Signed-off-by: Bjorn Helgaas Acked-by: Hedi Berriche --- drivers/pci/pcie/aer.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/pci/pcie') diff --git a/drivers/pci/pcie/aer.c b/drivers/pci/pcie/aer.c index 3fd4aaaa627e..ba22388342d1 100644 --- a/drivers/pci/pcie/aer.c +++ b/drivers/pci/pcie/aer.c @@ -1414,7 +1414,8 @@ static pci_ers_result_t aer_root_reset(struct pci_dev *dev) } } else { rc = pci_bus_error_reset(dev); - pci_info(dev, "Root Port link has been reset (%d)\n", rc); + pci_info(dev, "%s Port link has been reset (%d)\n", + pci_is_root_bus(dev->bus) ? "Root" : "Downstream", rc); } if ((host->native_aer || pcie_ports_native) && aer) { -- cgit v1.2.3 From ba952824e6c106f979c07814c8e3ef7405dd7b29 Mon Sep 17 00:00:00 2001 From: Keith Busch Date: Mon, 4 Jan 2021 15:03:00 -0800 Subject: PCI/portdrv: Report reset for frozen channel The PCI error recovery always resets the link for a frozen state, so the port driver should return that a reset is required for its result. This will get the .slot_reset() callback invoked, which is necessary to restore the port's config space. Without this, the driver had been relying on downstream drivers to return this status. Link: https://lore.kernel.org/r/20210104230300.1277180-6-kbusch@kernel.org Tested-by: Hedi Berriche Signed-off-by: Keith Busch Signed-off-by: Bjorn Helgaas Acked-by: Hedi Berriche --- drivers/pci/pcie/portdrv_pci.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/pci/pcie') diff --git a/drivers/pci/pcie/portdrv_pci.c b/drivers/pci/pcie/portdrv_pci.c index 0b250bc5f405..de141bfb0bc2 100644 --- a/drivers/pci/pcie/portdrv_pci.c +++ b/drivers/pci/pcie/portdrv_pci.c @@ -153,7 +153,8 @@ static void pcie_portdrv_remove(struct pci_dev *dev) static pci_ers_result_t pcie_portdrv_error_detected(struct pci_dev *dev, pci_channel_state_t error) { - /* Root Port has no impact. Always recovers. */ + if (error == pci_channel_io_frozen) + return PCI_ERS_RESULT_NEED_RESET; return PCI_ERS_RESULT_CAN_RECOVER; } -- cgit v1.2.3