Add support for decoding NVIDIA-specific CPER sections delivered via the APEI GHES vendor record notifier chain. NVIDIA hardware generates vendor-specific CPER sections containing error signatures and diagnostic register dumps. This implementation registers a notifier_block with the GHES vendor record notifier and decodes these sections, printing error details via dev_info().
The driver binds to ACPI device NVDA2012, present on NVIDIA server platforms. The NVIDIA CPER section contains a fixed header with error metadata (signature, error type, severity, socket) followed by variable-length register address-value pairs for hardware diagnostics. This work is based on libcper [0]. Example output: nvidia-ghes NVDA2012:00: NVIDIA CPER section, error_data_length: 544 nvidia-ghes NVDA2012:00: signature: CMET-INFO nvidia-ghes NVDA2012:00: error_type: 0 nvidia-ghes NVDA2012:00: error_instance: 0 nvidia-ghes NVDA2012:00: severity: 3 nvidia-ghes NVDA2012:00: socket: 0 nvidia-ghes NVDA2012:00: number_regs: 32 nvidia-ghes NVDA2012:00: instance_base: 0x0000000000000000 nvidia-ghes NVDA2012:00: register[0]: address=0x8000000100000000 value=0x0000000100000000 [0] https://github.com/openbmc/libcper/commit/683e055061ce Cc: Jonathan Cameron <[email protected]> Cc: Shiju Jose <[email protected]> Signed-off-by: Kai-Heng Feng <[email protected]> --- v2: - Use right headers. - Use embedded struct and __counted_by. - Drop __packed. - Remove unecessary casts. - Use * in sizeof(). - Use devm_kmalloc() and struct assignment. - Use dev_err_probe and new devm helper. MAINTAINERS | 6 ++ drivers/acpi/apei/Kconfig | 14 +++ drivers/acpi/apei/Makefile | 1 + drivers/acpi/apei/nvidia-ghes.c | 146 ++++++++++++++++++++++++++++++++ 4 files changed, 167 insertions(+) create mode 100644 drivers/acpi/apei/nvidia-ghes.c diff --git a/MAINTAINERS b/MAINTAINERS index d7241695df96..a9be03fbf1a9 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -18908,6 +18908,12 @@ S: Maintained F: drivers/video/fbdev/nvidia/ F: drivers/video/fbdev/riva/ +NVIDIA GHES VENDOR CPER RECORD HANDLER +M: Kai-Heng Feng <[email protected]> +L: [email protected] +S: Maintained +F: drivers/acpi/apei/nvidia-ghes.c + NVIDIA VRS RTC DRIVER M: Shubhi Garg <[email protected]> L: [email protected] diff --git a/drivers/acpi/apei/Kconfig b/drivers/acpi/apei/Kconfig index 070c07d68dfb..7dc49f14f223 100644 --- a/drivers/acpi/apei/Kconfig +++ b/drivers/acpi/apei/Kconfig @@ -74,6 +74,20 @@ config ACPI_APEI_EINJ_CXL If unsure say 'n' +config ACPI_APEI_NVIDIA_GHES + tristate "NVIDIA GHES vendor record handler" + depends on ACPI_APEI_GHES + help + Support for decoding NVIDIA-specific CPER sections delivered via + the APEI GHES vendor record notifier chain. Registers a handler + for the NVIDIA section GUID and logs error signatures, severity, + socket, and diagnostic register address-value pairs. + + Enable on NVIDIA server platforms (e.g. DGX, HGX) that expose + ACPI device NVDA2012 in their firmware tables. + + If unsure, say N. + config ACPI_APEI_ERST_DEBUG tristate "APEI Error Record Serialization Table (ERST) Debug Support" depends on ACPI_APEI diff --git a/drivers/acpi/apei/Makefile b/drivers/acpi/apei/Makefile index 1a0b85923cd4..4a883f67d698 100644 --- a/drivers/acpi/apei/Makefile +++ b/drivers/acpi/apei/Makefile @@ -10,5 +10,6 @@ obj-$(CONFIG_ACPI_APEI_EINJ) += einj.o einj-y := einj-core.o einj-$(CONFIG_ACPI_APEI_EINJ_CXL) += einj-cxl.o obj-$(CONFIG_ACPI_APEI_ERST_DEBUG) += erst-dbg.o +obj-$(CONFIG_ACPI_APEI_NVIDIA_GHES) += nvidia-ghes.o apei-y := apei-base.o hest.o erst.o bert.o diff --git a/drivers/acpi/apei/nvidia-ghes.c b/drivers/acpi/apei/nvidia-ghes.c new file mode 100644 index 000000000000..aa2e3a387b49 --- /dev/null +++ b/drivers/acpi/apei/nvidia-ghes.c @@ -0,0 +1,146 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * NVIDIA GHES vendor record handler + * + * Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved. + */ + +#include <linux/acpi.h> +#include <linux/module.h> +#include <linux/platform_device.h> +#include <linux/types.h> +#include <linux/uuid.h> +#include <acpi/ghes.h> + +static const guid_t nvidia_sec_guid = + GUID_INIT(0x6d5244f2, 0x2712, 0x11ec, + 0xbe, 0xa7, 0xcb, 0x3f, 0xdb, 0x95, 0xc7, 0x86); + +struct cper_sec_nvidia { + char signature[16]; + __le16 error_type; + __le16 error_instance; + u8 severity; + u8 socket; + u8 number_regs; + u8 reserved; + __le64 instance_base; + struct { + __le64 addr; + __le64 val; + } regs[] __counted_by(number_regs); +}; + +struct nvidia_ghes_private { + struct notifier_block nb; + struct device *dev; +}; + +static void nvidia_ghes_print_error(struct device *dev, + const struct cper_sec_nvidia *nvidia_err, + size_t error_data_length, bool fatal) +{ + const char *level = fatal ? KERN_ERR : KERN_INFO; + size_t min_size; + int i; + + dev_printk(level, dev, "signature: %.16s\n", nvidia_err->signature); + dev_printk(level, dev, "error_type: %u\n", le16_to_cpu(nvidia_err->error_type)); + dev_printk(level, dev, "error_instance: %u\n", le16_to_cpu(nvidia_err->error_instance)); + dev_printk(level, dev, "severity: %u\n", nvidia_err->severity); + dev_printk(level, dev, "socket: %u\n", nvidia_err->socket); + dev_printk(level, dev, "number_regs: %u\n", nvidia_err->number_regs); + dev_printk(level, dev, "instance_base: 0x%016llx\n", + le64_to_cpu(nvidia_err->instance_base)); + + if (nvidia_err->number_regs == 0) + return; + + /* + * Validate that all registers fit within error_data_length. + * Each register pair is two little-endian u64s. + */ + min_size = struct_size(nvidia_err, regs, nvidia_err->number_regs); + if (error_data_length < min_size) { + dev_err(dev, "Invalid number_regs %u (section size %zu, need %zu)\n", + nvidia_err->number_regs, error_data_length, min_size); + return; + } + + for (i = 0; i < nvidia_err->number_regs; i++) + dev_printk(level, dev, "register[%d]: address=0x%016llx value=0x%016llx\n", + i, le64_to_cpu(nvidia_err->regs[i].addr), + le64_to_cpu(nvidia_err->regs[i].val)); +} + +static int nvidia_ghes_notify(struct notifier_block *nb, + unsigned long event, void *data) +{ + struct acpi_hest_generic_data *gdata = data; + struct nvidia_ghes_private *priv; + const struct cper_sec_nvidia *nvidia_err; + guid_t sec_guid; + + import_guid(&sec_guid, gdata->section_type); + if (!guid_equal(&sec_guid, &nvidia_sec_guid)) + return NOTIFY_DONE; + + priv = container_of(nb, struct nvidia_ghes_private, nb); + + if (acpi_hest_get_error_length(gdata) < sizeof(*nvidia_err)) { + dev_err(priv->dev, "Section too small (%d < %zu)\n", + acpi_hest_get_error_length(gdata), sizeof(*nvidia_err)); + return NOTIFY_OK; + } + + nvidia_err = acpi_hest_get_payload(gdata); + + if (event >= GHES_SEV_RECOVERABLE) + dev_err(priv->dev, "NVIDIA CPER section, error_data_length: %u\n", + acpi_hest_get_error_length(gdata)); + else + dev_info(priv->dev, "NVIDIA CPER section, error_data_length: %u\n", + acpi_hest_get_error_length(gdata)); + + nvidia_ghes_print_error(priv->dev, nvidia_err, acpi_hest_get_error_length(gdata), + event >= GHES_SEV_RECOVERABLE); + + return NOTIFY_OK; +} + +static int nvidia_ghes_probe(struct platform_device *pdev) +{ + struct nvidia_ghes_private *priv; + + priv = devm_kmalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + *priv = (struct nvidia_ghes_private) { + .nb.notifier_call = nvidia_ghes_notify, + .dev = &pdev->dev, + }; + + return dev_err_probe(&pdev->dev, + devm_ghes_register_vendor_record_notifier(&pdev->dev, &priv->nb), + "Failed to register NVIDIA GHES vendor record notifier\n"); +} + +static const struct acpi_device_id nvidia_ghes_acpi_match[] = { + { "NVDA2012" }, + { } +}; +MODULE_DEVICE_TABLE(acpi, nvidia_ghes_acpi_match); + +static struct platform_driver nvidia_ghes_driver = { + .driver = { + .name = "nvidia-ghes", + .acpi_match_table = nvidia_ghes_acpi_match, + }, + .probe = nvidia_ghes_probe, +}; +module_platform_driver(nvidia_ghes_driver); + +MODULE_AUTHOR("Kai-Heng Feng <[email protected]>"); +MODULE_DESCRIPTION("NVIDIA GHES vendor CPER record handler"); +MODULE_LICENSE("GPL"); -- 2.50.1 (Apple Git-155)
