AER error recovery is part of the AER error handling subsystem in the Linux kernel. On large modular systems with a complex PCIe tree, AER recovery could cause unexpected behavior and side effects. Sometimes it would be nice to have the option to keep the system in an unmodified state and be able to handle PCIe errors from userspace.
Add pci=noaer_recovery kernel boot option to disable AER error recovery when an uncorrectable error is reported. When this option is set, the error is still logged but no recovery actions are taken. AER error status bits are preserved so userspace can inspect the error state. Signed-off-by: Yury Murashka <[email protected]> --- .../admin-guide/kernel-parameters.txt | 6 ++++ drivers/pci/pci.c | 2 ++ drivers/pci/pci.h | 2 ++ drivers/pci/pcie/aer.c | 30 ++++++++++++------- 4 files changed, 30 insertions(+), 10 deletions(-) diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index b5493a7f8f22..13c6b53bb9ee 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -5053,6 +5053,12 @@ Kernel parameters noaer [PCIE] If the PCIEAER kernel config parameter is enabled, this kernel boot option can be used to disable the use of PCIE advanced error reporting. + noaer_recovery [PCIE] If the PCIEAER kernel config parameter is + enabled, this kernel boot option can be used to + disable AER error recovery when an uncorrectable + error is reported. AER error status bits are + preserved so userspace can inspect the error + state. nodomains [PCI] Disable support for multiple PCI root domains (aka PCI segments, in ACPI-speak). nommconf [X86] Disable use of MMCONFIG for PCI diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index 77b17b13ee61..a215dd567d5d 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -6756,6 +6756,8 @@ static int __init pci_setup(char *str) pcie_ats_disabled = true; } else if (!strcmp(str, "noaer")) { pci_no_aer(); + } else if (!strcmp(str, "noaer_recovery")) { + pci_no_aer_recovery(); } else if (!strcmp(str, "earlydump")) { pci_early_dump = true; } else if (!strncmp(str, "realloc=", 8)) { diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h index 4469e1a77f3c..c4a42bbc277b 100644 --- a/drivers/pci/pci.h +++ b/drivers/pci/pci.h @@ -1283,6 +1283,7 @@ static inline void of_pci_remove_host_bridge_node(struct pci_host_bridge *bridge #ifdef CONFIG_PCIEAER void pci_no_aer(void); +void pci_no_aer_recovery(void); void pci_aer_init(struct pci_dev *dev); void pci_aer_exit(struct pci_dev *dev); extern const struct attribute_group aer_stats_attr_group; @@ -1294,6 +1295,7 @@ void pci_save_aer_state(struct pci_dev *dev); void pci_restore_aer_state(struct pci_dev *dev); #else static inline void pci_no_aer(void) { } +static inline void pci_no_aer_recovery(void) { } static inline void pci_aer_init(struct pci_dev *d) { } static inline void pci_aer_exit(struct pci_dev *d) { } static inline void pci_aer_clear_fatal_status(struct pci_dev *dev) { } diff --git a/drivers/pci/pcie/aer.c b/drivers/pci/pcie/aer.c index c4fd9c0b2a54..fb79990ce8ff 100644 --- a/drivers/pci/pcie/aer.c +++ b/drivers/pci/pcie/aer.c @@ -128,6 +128,7 @@ struct aer_info { PCI_ERR_ROOT_MULTI_UNCOR_RCV) static bool pcie_aer_disable; +static bool pcie_aer_recovery_disable; static pci_ers_result_t aer_root_reset(struct pci_dev *dev); void pci_no_aer(void) @@ -135,6 +136,11 @@ void pci_no_aer(void) pcie_aer_disable = true; } +void pci_no_aer_recovery(void) +{ + pcie_aer_recovery_disable = true; +} + bool pci_aer_available(void) { return !pcie_aer_disable && pci_msi_enabled(); @@ -1187,10 +1193,12 @@ static void pci_aer_handle_error(struct pci_dev *dev, struct aer_err_info *info) pdrv->err_handler->cor_error_detected(dev); pcie_clear_device_status(dev); } - } else if (info->severity == AER_NONFATAL) - pcie_do_recovery(dev, pci_channel_io_normal, aer_root_reset); - else if (info->severity == AER_FATAL) - pcie_do_recovery(dev, pci_channel_io_frozen, aer_root_reset); + } else if (!pcie_aer_recovery_disable) { + if (info->severity == AER_NONFATAL) + pcie_do_recovery(dev, pci_channel_io_normal, aer_root_reset); + else if (info->severity == AER_FATAL) + pcie_do_recovery(dev, pci_channel_io_frozen, aer_root_reset); + } } static void handle_error_source(struct pci_dev *dev, struct aer_err_info *info) @@ -1242,12 +1250,14 @@ static void aer_recover_work_func(struct work_struct *work) ghes_estatus_pool_region_free((unsigned long)entry.regs, sizeof(struct aer_capability_regs)); - if (entry.severity == AER_NONFATAL) - pcie_do_recovery(pdev, pci_channel_io_normal, - aer_root_reset); - else if (entry.severity == AER_FATAL) - pcie_do_recovery(pdev, pci_channel_io_frozen, - aer_root_reset); + if (!pcie_aer_recovery_disable) { + if (entry.severity == AER_NONFATAL) + pcie_do_recovery(pdev, pci_channel_io_normal, + aer_root_reset); + else if (entry.severity == AER_FATAL) + pcie_do_recovery(pdev, pci_channel_io_frozen, + aer_root_reset); + } pci_dev_put(pdev); } } -- 2.51.0
