[CRITICAL][v2.8.0] Data resources cause provider segmentation fault on new state

[nfx]

SDK version

SDK v2.8.0 crashes provider data resources: databricks/terraform-provider-databricks#854

Logs say: Instance Diff is nil in Diff(), which is a valid case for starting new terraform state:

        if instanceDiff != nil {
		if err := t.DiffEncode(instanceDiff); err != nil {
			log.Printf("[ERR] Error encoding timeout to instance diff: %s", err)
		}
	} else {
		log.Printf("[DEBUG] Instance Diff is nil in Diff()")
	}

	return instanceDiff, err

Closer inspection of SDK updates - v2.7.0...v2.8.0 - show, that diff.RawConfig = configVal was added without a wrapping nil check.

Terraform Configuration Files

terraform {
  required_providers {
    databricks = {
      source  = "databrickslabs/databricks"
      version = "0.3.8"
    }
  }
}

provider "databricks" {
  host  = "https://XXX.cloud.databricks.com/"
  token = "YYY"
}

data "databricks_node_type" "smallest" {
}

Debug Output

[WARN]  ValidateProviderConfig from "provider[\"registry.terraform.io/databrickslabs/databricks\"]" changed the config value, but that value is unused
[INFO]  provider.terraform-provider-databricks: Explicit and implicit attributes: host, token: timestamp=2021-10-11T11:53:57.183+0200
[DEBUG] Resource instance state not found for node "data.databricks_node_type.smallest", instance data.databricks_node_type.smallest
[DEBUG] ReferenceTransformer: "data.databricks_node_type.smallest" references: []
[DEBUG] provider.terraform-provider-databricks: Instance Diff is nil in Diff(): timestamp=2021-10-11T11:53:57.185+0200
[DEBUG] provider.terraform-provider-databricks: PANIC: runtime error: invalid memory address or nil pointer dereference
[DEBUG] provider.terraform-provider-databricks: [signal SIGSEGV: segmentation violation code=0x1 addr=0x18 pc=0x1625584]
2021-10-11T11:53:57.187+0200 [DEBUG] provider.terraform-provider-databricks:
2021-10-11T11:53:57.187+0200 [DEBUG] provider.terraform-provider-databricks: goroutine 71 [running]:
2021-10-11T11:53:57.187+0200 [DEBUG] provider.terraform-provider-databricks: github.com/hashicorp/terraform-plugin-sdk/v2/helper/schema.(*GRPCProviderServer).ReadDataSource(0xc00054c930, {0x1d87d48, 0xc00053dac0}, 0xc0002d3a00)
2021-10-11T11:53:57.187+0200 [DEBUG] provider.terraform-provider-databricks: 	.../vendor/github.com/hashicorp/terraform-plugin-sdk/v2/helper/schema/grpc_provider.go:1126 +0x304

Expected Behavior

• Any version of SDK must not get released with segfaults in happy flow.
• github.com/hashicorp/terraform-plugin-sdk/v2/helper/resource/testing.go should catch errors on acceptance test level. SDK v2.8.0 does not give the ability to detect such errors.

Actual Behavior

 Error: Plugin did not respond
│
│   with data.databricks_node_type.smallest,
│   on main.tf line 15, in data "databricks_node_type" "smallest":
│   15: data "databricks_node_type" "smallest" {
│
│ The plugin encountered an error, and failed to respond to the plugin.(*GRPCProvider).ReadDataSource call. The plugin logs may contain more details.

Steps to Reproduce

terraform init && terraform plan

References

• Prototype exposing cty.Values from ResourceData. #802 - @paddycarver "This still needs unit tests" - i think it really does...

[bflad]

Hi @nfx 👋 Thank you for raising this and sorry you ran into trouble. I was able to create covering testing for this situation and will submit the fix shortly.

[github-actions]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.
If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.