drm/amdgpu: add ACA bank dump debugfs support

add ACA bank dump debugfs support

Signed-off-by: Yang Wang <kevinyang.wang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
index e091e9ac5eddc84f039f476381ee94647af38806..6074a529caf7851cf8c27509b621235705b9f892 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
@@ -701,3 +701,122 @@ int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank
        return -EINVAL;
 }
 
+int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en)
+{
+       struct amdgpu_aca *aca = &adev->aca;
+       const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+
+       if (!smu_funcs || !smu_funcs->set_debug_mode)
+               return -EOPNOTSUPP;
+
+       return smu_funcs->set_debug_mode(adev, en);
+}
+
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_aca_smu_debug_mode_set(void *data, u64 val)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)data;
+       int ret;
+
+       ret = amdgpu_ras_set_aca_debug_mode(adev, val ? true : false);
+       if (ret)
+               return ret;
+
+       dev_info(adev->dev, "amdgpu set smu aca debug mode %s success\n", val ? "on" : "off");
+
+       return 0;
+}
+
+static void aca_dump_entry(struct seq_file *m, struct aca_bank *bank, enum aca_error_type type, int idx)
+{
+       struct aca_bank_info info;
+       int i, ret;
+
+       ret = aca_bank_info_decode(bank, &info);
+       if (ret)
+               return;
+
+       seq_printf(m, "aca entry[%d].type: %s\n", idx, type ==  ACA_ERROR_TYPE_UE ? "UE" : "CE");
+       seq_printf(m, "aca entry[%d].info: socketid:%d aid:%d hwid:0x%03x mcatype:0x%04x\n",
+                  idx, info.socket_id, info.die_id, info.hwid, info.mcatype);
+
+       for (i = 0; i < ARRAY_SIZE(aca_regs); i++)
+               seq_printf(m, "aca entry[%d].regs[%d]: 0x%016llx\n", idx, aca_regs[i].reg_idx, bank->regs[aca_regs[i].reg_idx]);
+}
+
+struct aca_dump_context {
+       struct seq_file *m;
+       int idx;
+};
+
+static int handler_aca_bank_dump(struct aca_handle *handle, struct aca_bank *bank,
+                                enum aca_error_type type, void *data)
+{
+       struct aca_dump_context *ctx = (struct aca_dump_context *)data;
+
+       aca_dump_entry(ctx->m, bank, type, ctx->idx++);
+
+       return handler_aca_log_bank_error(handle, bank, type, NULL);
+}
+
+static int aca_dump_show(struct seq_file *m, enum aca_error_type type)
+{
+       struct amdgpu_device *adev = (struct amdgpu_device *)m->private;
+       struct aca_dump_context context = {
+               .m = m,
+               .idx = 0,
+       };
+
+       return aca_banks_update(adev, type, handler_aca_bank_dump, (void *)&context);
+}
+
+static int aca_dump_ce_show(struct seq_file *m, void *unused)
+{
+       return aca_dump_show(m, ACA_ERROR_TYPE_CE);
+}
+
+static int aca_dump_ce_open(struct inode *inode, struct file *file)
+{
+       return single_open(file, aca_dump_ce_show, inode->i_private);
+}
+
+static const struct file_operations aca_ce_dump_debug_fops = {
+       .owner = THIS_MODULE,
+       .open = aca_dump_ce_open,
+       .read = seq_read,
+       .llseek = seq_lseek,
+       .release = single_release,
+};
+
+static int aca_dump_ue_show(struct seq_file *m, void *unused)
+{
+       return aca_dump_show(m, ACA_ERROR_TYPE_UE);
+}
+
+static int aca_dump_ue_open(struct inode *inode, struct file *file)
+{
+       return single_open(file, aca_dump_ue_show, inode->i_private);
+}
+
+static const struct file_operations aca_ue_dump_debug_fops = {
+       .owner = THIS_MODULE,
+       .open = aca_dump_ue_open,
+       .read = seq_read,
+       .llseek = seq_lseek,
+       .release = single_release,
+};
+
+DEFINE_DEBUGFS_ATTRIBUTE(aca_debug_mode_fops, NULL, amdgpu_aca_smu_debug_mode_set, "%llu\n");
+#endif
+
+void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root)
+{
+#if defined(CONFIG_DEBUG_FS)
+       if (!root || adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 6))
+               return;
+
+       debugfs_create_file("aca_debug_mode", 0200, root, adev, &aca_debug_mode_fops);
+       debugfs_create_file("aca_ue_dump", 0400, root, adev, &aca_ue_dump_debug_fops);
+       debugfs_create_file("aca_ce_dump", 0400, root, adev, &aca_ce_dump_debug_fops);
+#endif
+}

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
index 8aa3d8c4136c2eae9081522139e5e1303065d500..ac641952ab7126cf55946244ea573ba60fc55be9 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
@@ -192,4 +192,6 @@ int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle,
 void amdgpu_aca_remove_handle(struct aca_handle *handle);
 int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle,
                                     enum aca_error_type type, void *data);
+int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en);
+void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root);
 #endif

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index a86f91838301480a17a4891b532d5b03b5232751..ded19182dc792a4c9facaeeb0427c214c8664f0c 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -3431,6 +3431,20 @@ int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable)
        return ret;
 }
 
+int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable)
+{
+       struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+       int ret = 0;
+
+       if (con) {
+               ret = amdgpu_aca_smu_set_debug_mode(adev, enable);
+               if (!ret)
+                       con->is_mca_debug_mode = enable;
+       }
+
+       return ret;
+}
+
 bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev)
 {
        struct amdgpu_ras *con = amdgpu_ras_get_context(adev);

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index 80e574475aa363b4c6bdf2031bda76fb52eb99d5..9edc3909f17d7ab99cd81da5c9fcc372abca21ca 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -794,6 +794,7 @@ struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev);
 
 int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con);
 
+int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable);
 int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable);
 bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev);
 bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev,