[PATCH v2 02/12] drm/amdgpu: Introduce funcs for populating CPER
Xiang Liu
xiang.liu at amd.com
Fri Feb 14 08:07:09 UTC 2025
From: Hawking Zhang <Hawking.Zhang at amd.com>
Introduce utility functions designed to assist
in populating CPER records.
v2: call cper_init/fini in device_ip_init/fini.
Signed-off-by: Hawking Zhang <Hawking.Zhang at amd.com>
Reviewed-by: Tao Zhou <tao.zhou1 at amd.com>
---
drivers/gpu/drm/amd/amdgpu/Makefile | 2 +-
drivers/gpu/drm/amd/amdgpu/amdgpu.h | 4 +
drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c | 281 +++++++++++++++++++++
drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h | 91 +++++++
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 4 +
5 files changed, 381 insertions(+), 1 deletion(-)
create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c
create mode 100644 drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index 04247303b3cf..84bb3dfa39a9 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -66,7 +66,7 @@ amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o amdgpu_kms.o \
amdgpu_fw_attestation.o amdgpu_securedisplay.o \
amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \
amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o amdgpu_dev_coredump.o \
- amdgpu_userq_fence.o amdgpu_eviction_fence.o
+ amdgpu_userq_fence.o amdgpu_eviction_fence.o amdgpu_cper.o
amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index dc1f8d6fd0c4..db0a26800927 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -110,6 +110,7 @@
#include "amdgpu_mca.h"
#include "amdgpu_aca.h"
#include "amdgpu_ras.h"
+#include "amdgpu_cper.h"
#include "amdgpu_xcp.h"
#include "amdgpu_seq64.h"
#include "amdgpu_reg_state.h"
@@ -1128,6 +1129,9 @@ struct amdgpu_device {
/* ACA */
struct amdgpu_aca aca;
+ /* CPER */
+ struct amdgpu_cper cper;
+
struct amdgpu_ip_block ip_blocks[AMDGPU_MAX_IP_NUM];
uint32_t harvest_ip_mask;
int num_ip_blocks;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c
new file mode 100644
index 000000000000..8ce5dc6efcf9
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c
@@ -0,0 +1,281 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+
+static const guid_t MCE = CPER_NOTIFY_MCE;
+static const guid_t CMC = CPER_NOTIFY_CMC;
+static const guid_t BOOT = BOOT_TYPE;
+
+static const guid_t CRASHDUMP = AMD_CRASHDUMP;
+static const guid_t RUNTIME = AMD_GPU_NONSTANDARD_ERROR;
+
+static void __inc_entry_length(struct cper_hdr *hdr, uint32_t size)
+{
+ hdr->record_length += size;
+}
+
+void amdgpu_cper_entry_fill_hdr(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ enum amdgpu_cper_type type,
+ enum cper_error_severity sev)
+{
+ hdr->signature[0] = 'C';
+ hdr->signature[1] = 'P';
+ hdr->signature[2] = 'E';
+ hdr->signature[3] = 'R';
+ hdr->revision = CPER_HDR_REV_1;
+ hdr->signature_end = 0xFFFFFFFF;
+ hdr->error_severity = sev;
+
+ hdr->valid_bits.platform_id = 1;
+ hdr->valid_bits.partition_id = 1;
+ hdr->valid_bits.timestamp = 1;
+ /*TODO need to initialize hdr->timestamp */
+
+ snprintf(hdr->record_id, 8, "%d", atomic_inc_return(&adev->cper.unique_id));
+ snprintf(hdr->platform_id, 16, "0x%04X:0x%04X",
+ adev->pdev->vendor, adev->pdev->device);
+ /* pmfw version should be part of creator_id according to CPER spec */
+ snprintf(hdr->creator_id, 16, "%s", CPER_CREATOR_ID_AMDGPU);
+
+ switch (type) {
+ case AMDGPU_CPER_TYPE_BOOT:
+ hdr->notify_type = BOOT;
+ break;
+ case AMDGPU_CPER_TYPE_FATAL:
+ case AMDGPU_CPER_TYPE_BP_THRESHOLD:
+ hdr->notify_type = MCE;
+ break;
+ case AMDGPU_CPER_TYPE_RUNTIME:
+ if (sev == CPER_SEV_NON_FATAL_CORRECTED)
+ hdr->notify_type = CMC;
+ else
+ hdr->notify_type = MCE;
+ break;
+ default:
+ dev_err(adev->dev, "Unknown CPER Type\n");
+ break;
+ }
+
+ __inc_entry_length(hdr, HDR_LEN);
+}
+
+static int amdgpu_cper_entry_fill_section_desc(struct amdgpu_device *adev,
+ struct cper_sec_desc *section_desc,
+ bool bp_threshold,
+ bool poison,
+ enum cper_error_severity sev,
+ guid_t sec_type,
+ uint32_t section_length,
+ uint32_t section_offset)
+{
+ section_desc->revision_minor = CPER_SEC_MINOR_REV_1;
+ section_desc->revision_major = CPER_SEC_MAJOR_REV_22;
+ section_desc->sec_offset = section_offset;
+ section_desc->sec_length = section_length;
+ section_desc->valid_bits.fru_id = 1;
+ section_desc->valid_bits.fru_text = 1;
+ section_desc->flag_bits.primary = 1;
+ section_desc->severity = sev;
+ section_desc->sec_type = sec_type;
+
+ if (adev->smuio.funcs &&
+ adev->smuio.funcs->get_socket_id)
+ snprintf(section_desc->fru_text, 20, "OAM%d",
+ adev->smuio.funcs->get_socket_id(adev));
+ /* TODO: fru_id is 16 bytes in CPER spec, but driver defines it as 20 bytes */
+ snprintf(section_desc->fru_id, 16, "%llx", adev->unique_id);
+
+ if (bp_threshold)
+ section_desc->flag_bits.exceed_err_threshold = 1;
+ if (poison)
+ section_desc->flag_bits.latent_err = 1;
+
+ return 0;
+}
+
+int amdgpu_cper_entry_fill_fatal_section(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ uint32_t idx,
+ struct cper_sec_crashdump_reg_data reg_data)
+{
+ struct cper_sec_desc *section_desc;
+ struct cper_sec_crashdump_fatal *section;
+
+ section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+ section = (struct cper_sec_crashdump_fatal *)((uint8_t *)hdr +
+ FATAL_SEC_OFFSET(hdr->sec_cnt, idx));
+
+ amdgpu_cper_entry_fill_section_desc(adev, section_desc, false, false,
+ CPER_SEV_FATAL, CRASHDUMP, FATAL_SEC_LEN,
+ FATAL_SEC_OFFSET(hdr->sec_cnt, idx));
+
+ section->body.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+ section->body.reg_arr_size = sizeof(reg_data);
+ section->body.data = reg_data;
+
+ __inc_entry_length(hdr, SEC_DESC_LEN + FATAL_SEC_LEN);
+
+ return 0;
+}
+
+int amdgpu_cper_entry_fill_runtime_section(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ uint32_t idx,
+ enum cper_error_severity sev,
+ uint32_t *reg_dump,
+ uint32_t reg_count)
+{
+ struct cper_sec_desc *section_desc;
+ struct cper_sec_nonstd_err *section;
+ bool poison;
+
+ poison = (sev == CPER_SEV_NON_FATAL_CORRECTED) ? false : true;
+ section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+ section = (struct cper_sec_nonstd_err *)((uint8_t *)hdr +
+ NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+ amdgpu_cper_entry_fill_section_desc(adev, section_desc, false, poison,
+ sev, RUNTIME, NONSTD_SEC_LEN,
+ NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+ reg_count = min(reg_count, CPER_ACA_REG_COUNT);
+
+ section->hdr.valid_bits.err_info_cnt = 1;
+ section->hdr.valid_bits.err_context_cnt = 1;
+
+ section->info.error_type = RUNTIME;
+ section->info.ms_chk_bits.err_type_valid = 1;
+ section->ctx.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+ section->ctx.reg_arr_size = sizeof(section->ctx.reg_dump);
+
+ memcpy(section->ctx.reg_dump, reg_dump, reg_count * sizeof(uint32_t));
+
+ __inc_entry_length(hdr, SEC_DESC_LEN + NONSTD_SEC_LEN);
+
+ return 0;
+}
+
+int amdgpu_cper_entry_fill_bad_page_threshold_section(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ uint32_t idx)
+{
+ struct cper_sec_desc *section_desc;
+ struct cper_sec_nonstd_err *section;
+
+ section_desc = (struct cper_sec_desc *)((uint8_t *)hdr + SEC_DESC_OFFSET(idx));
+ section = (struct cper_sec_nonstd_err *)((uint8_t *)hdr +
+ NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+ amdgpu_cper_entry_fill_section_desc(adev, section_desc, true, false,
+ CPER_SEV_FATAL, RUNTIME, NONSTD_SEC_LEN,
+ NONSTD_SEC_OFFSET(hdr->sec_cnt, idx));
+
+ section->hdr.valid_bits.err_info_cnt = 1;
+ section->hdr.valid_bits.err_context_cnt = 1;
+
+ section->info.error_type = RUNTIME;
+ section->info.ms_chk_bits.err_type_valid = 1;
+ section->ctx.reg_ctx_type = CPER_CTX_TYPE_CRASH;
+ section->ctx.reg_arr_size = sizeof(section->ctx.reg_dump);
+
+ /* Hardcoded Reg dump for bad page threshold CPER */
+ section->ctx.reg_dump[CPER_ACA_REG_CTL_LO] = 0x1;
+ section->ctx.reg_dump[CPER_ACA_REG_CTL_HI] = 0x0;
+ section->ctx.reg_dump[CPER_ACA_REG_STATUS_LO] = 0x137;
+ section->ctx.reg_dump[CPER_ACA_REG_STATUS_HI] = 0xB0000000;
+ section->ctx.reg_dump[CPER_ACA_REG_ADDR_LO] = 0x0;
+ section->ctx.reg_dump[CPER_ACA_REG_ADDR_HI] = 0x0;
+ section->ctx.reg_dump[CPER_ACA_REG_MISC0_LO] = 0x0;
+ section->ctx.reg_dump[CPER_ACA_REG_MISC0_HI] = 0x0;
+ section->ctx.reg_dump[CPER_ACA_REG_CONFIG_LO] = 0x2;
+ section->ctx.reg_dump[CPER_ACA_REG_CONFIG_HI] = 0x1ff;
+ section->ctx.reg_dump[CPER_ACA_REG_IPID_LO] = 0x0;
+ section->ctx.reg_dump[CPER_ACA_REG_IPID_HI] = 0x96;
+ section->ctx.reg_dump[CPER_ACA_REG_SYND_LO] = 0x0;
+ section->ctx.reg_dump[CPER_ACA_REG_SYND_HI] = 0x0;
+
+ __inc_entry_length(hdr, SEC_DESC_LEN + NONSTD_SEC_LEN);
+
+ return 0;
+}
+
+struct cper_hdr *amdgpu_cper_alloc_entry(struct amdgpu_device *adev,
+ enum amdgpu_cper_type type,
+ uint16_t section_count)
+{
+ struct cper_hdr *hdr;
+ uint32_t size = 0;
+
+ size += HDR_LEN;
+ size += (SEC_DESC_LEN * section_count);
+
+ switch (type) {
+ case AMDGPU_CPER_TYPE_RUNTIME:
+ case AMDGPU_CPER_TYPE_BP_THRESHOLD:
+ size += (NONSTD_SEC_LEN * section_count);
+ break;
+ case AMDGPU_CPER_TYPE_FATAL:
+ size += (FATAL_SEC_LEN * section_count);
+ break;
+ case AMDGPU_CPER_TYPE_BOOT:
+ size += (BOOT_SEC_LEN * section_count);
+ break;
+ default:
+ dev_err(adev->dev, "Unknown CPER Type!\n");
+ return NULL;
+ }
+
+ hdr = kzalloc(size, GFP_KERNEL);
+ if (!hdr)
+ return NULL;
+
+ /* Save this early */
+ hdr->sec_cnt = section_count;
+
+ return hdr;
+}
+
+int amdgpu_cper_init(struct amdgpu_device *adev)
+{
+ mutex_init(&adev->cper.cper_lock);
+
+ adev->cper.enabled = true;
+ adev->cper.max_count = CPER_MAX_ALLOWED_COUNT;
+
+ /*TODO: initialize cper ring*/
+
+ return 0;
+}
+
+int amdgpu_cper_fini(struct amdgpu_device *adev)
+{
+ adev->cper.enabled = false;
+
+ /*TODO: free cper ring */
+ adev->cper.count = 0;
+ adev->cper.wptr = 0;
+
+ return 0;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h
new file mode 100644
index 000000000000..0ae845420983
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h
@@ -0,0 +1,91 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright 2025 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_CPER_H__
+#define __AMDGPU_CPER_H__
+
+#include "amd_cper.h"
+
+#define CPER_MAX_ALLOWED_COUNT 0x1000
+#define HDR_LEN (sizeof(struct cper_hdr))
+#define SEC_DESC_LEN (sizeof(struct cper_sec_desc))
+
+#define BOOT_SEC_LEN (sizeof(struct cper_sec_crashdump_boot))
+#define FATAL_SEC_LEN (sizeof(struct cper_sec_crashdump_fatal))
+#define NONSTD_SEC_LEN (sizeof(struct cper_sec_nonstd_err))
+
+#define SEC_DESC_OFFSET(idx) (HDR_LEN + (SEC_DESC_LEN * idx))
+
+#define BOOT_SEC_OFFSET(count, idx) (HDR_LEN + (SEC_DESC_LEN * count) + (BOOT_SEC_LEN * idx))
+#define FATAL_SEC_OFFSET(count, idx) (HDR_LEN + (SEC_DESC_LEN * count) + (FATAL_SEC_LEN * idx))
+#define NONSTD_SEC_OFFSET(count, idx) (HDR_LEN + (SEC_DESC_LEN * count) + (NONSTD_SEC_LEN * idx))
+
+enum amdgpu_cper_type {
+ AMDGPU_CPER_TYPE_RUNTIME,
+ AMDGPU_CPER_TYPE_FATAL,
+ AMDGPU_CPER_TYPE_BOOT,
+ AMDGPU_CPER_TYPE_BP_THRESHOLD,
+};
+
+struct amdgpu_cper {
+ bool enabled;
+
+ atomic_t unique_id;
+ struct mutex cper_lock;
+
+ /* Lifetime CPERs generated */
+ uint32_t count;
+ uint32_t max_count;
+
+ uint32_t wptr;
+
+ void *ring[CPER_MAX_ALLOWED_COUNT];
+};
+
+void amdgpu_cper_entry_fill_hdr(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ enum amdgpu_cper_type type,
+ enum cper_error_severity sev);
+int amdgpu_cper_entry_fill_fatal_section(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ uint32_t idx,
+ struct cper_sec_crashdump_reg_data reg_data);
+int amdgpu_cper_entry_fill_runtime_section(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ uint32_t idx,
+ enum cper_error_severity sev,
+ uint32_t *reg_dump,
+ uint32_t reg_count);
+int amdgpu_cper_entry_fill_bad_page_threshold_section(struct amdgpu_device *adev,
+ struct cper_hdr *hdr,
+ uint32_t section_idx);
+
+struct cper_hdr *amdgpu_cper_alloc_entry(struct amdgpu_device *adev,
+ enum amdgpu_cper_type type,
+ uint16_t section_count);
+
+int amdgpu_cper_init(struct amdgpu_device *adev);
+int amdgpu_cper_fini(struct amdgpu_device *adev);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index f9a7bd741a41..f5c65e89b23b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -3075,6 +3075,8 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
amdgpu_fru_get_product_info(adev);
+ r = amdgpu_cper_init(adev);
+
init_failed:
return r;
@@ -3435,6 +3437,8 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
{
int i, r;
+ amdgpu_cper_fini(adev);
+
if (amdgpu_sriov_vf(adev) && adev->virt.ras_init_done)
amdgpu_virt_release_ras_err_handler_data(adev);
--
2.34.1
More information about the amd-gfx
mailing list