Adapt GPU label support to debugfs DRM entry changes

GPU generation "gen" number is replaced in the capability files of
latest kernels with separate display, graphics, and media versions.

For compatibility with newer kernels, provide "gen" based on the new
labels (but without decimals), and for older kernel compatibility, new
labels based on the "gen".

Because different kernels match different items from the action map,
whole capability file will get parsed. Capability file parsing is
optimized by using prefix check instead of scanf.

"platform_gen" label is deprecated, and can be dropped whenever it
becomes inconvenient (lint complains about line count etc).

Author: eero-t

cmd/gpu_nfdhook/README.md

@@ -43,9 +43,11 @@ Depending on your kernel driver, running the NFD hook as root may introduce foll
 
 name | type | description|
 -----|------|------|
-|`gpu.intel.com/platform_gen`| string | GPU platform generation name, typically a number.
+|`gpu.intel.com/platform_gen`| string | GPU platform generation name, typically an integer. Deprecated.
+|`gpu.intel.com/media_version`| string | GPU platform Media pipeline generation name, typically a number.
+|`gpu.intel.com/graphics_version`| string | GPU platform graphics/compute pipeline generation name, typically a number.
 |`gpu.intel.com/platform_<PLATFORM_NAME>_.count`| number | GPU count for the named platform.
 |`gpu.intel.com/platform_<PLATFORM_NAME>_.tiles`| number | GPU tile count in the GPUs of the named platform.
 |`gpu.intel.com/platform_<PLATFORM_NAME>_.present`| string | "true" for indicating the presense of the GPU platform.
 
-For the above to work as intended, installed GPUs must be identical in their capabilities.
+For the above to work as intended, GPUs on the same node must be identical in their capabilities.

cmd/gpu_nfdhook/labeler.go

@@ -1,4 +1,4 @@
-// Copyright 2020 Intel Corporation. All Rights Reserved.
+// Copyright 2020-2021 Intel Corporation. All Rights Reserved.
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -171,33 +171,70 @@ func (l *labeler) createCapabilityLabels(cardNum string, numTiles uint64) {
 	}
 	defer file.Close()
 
-	// define strings to search from the file, and the actions to take in order to create labels from those strings (as funcs)
+	gen := ""
+	media := ""
+	graphics := ""
+	// define string prefixes to search from the file, and the actions to take in order to create labels from those strings (as funcs)
 	searchStringActionMap := map[string]func(string){
-		"platform: ": func(platformName string) {
+		"platform:": func(platformName string) {
 			l.labels.addNumericLabel(labelNamespace+"platform_"+platformName+".count", 1)
 			l.labels[labelNamespace+"platform_"+platformName+".tiles"] = strconv.FormatInt(int64(numTiles), 10)
 			l.labels[labelNamespace+"platform_"+platformName+".present"] = "true"
 		},
-		"gen: ": func(genName string) {
-			l.labels[labelNamespace+"platform_gen"] = genName
+		// there's also display block version, but that's not relevant
+		"media version:": func(version string) {
+			l.labels[labelNamespace+"media_version"] = version
+			media = version
+		},
+		"graphics version:": func(version string) {
+			l.labels[labelNamespace+"graphics_version"] = version
+			graphics = version
+		},
+		"gen:": func(version string) {
+			l.labels[labelNamespace+"platform_gen"] = version
+			gen = version
 		},
 	}
 
 	// Finally, read the file, and try to find the matches. Perform actions and reduce the search map size as we proceed. Return at 0 size.
 	scanner := bufio.NewScanner(file)
+scanning:
 	for scanner.Scan() {
-		for searchString, action := range searchStringActionMap {
-			var stringValue string
-			n, _ := fmt.Sscanf(scanner.Text(), searchString+"%s", &stringValue)
-			if n > 0 {
-				action(stringValue)
-				delete(searchStringActionMap, searchString)
-				if len(searchStringActionMap) == 0 {
-					return
-				}
-				break
+		line := scanner.Text()
+		for prefix, action := range searchStringActionMap {
+			if !strings.HasPrefix(line, prefix) {
+				continue
+			}
+			fields := strings.Split(line, ": ")
+			if len(fields) == 2 {
+				action(fields[1])
+			} else {
+				klog.Warningf("invalid '%s' line format: '%s'", file.Name(), line)
+			}
+			delete(searchStringActionMap, prefix)
+			if len(searchStringActionMap) == 0 {
+				break scanning
 			}
+			break
+		}
+	}
+	if gen == "" {
+		// TODO: drop gen label before engine types
+		// start to have diverging major gen values
+		if graphics != "" {
+			gen = graphics
+		} else if media != "" {
+			gen = media
+		}
+		if gen != "" {
+			// truncate to major value
+			gen = strings.SplitN(gen, ".", 2)[0]
+			l.labels[labelNamespace+"platform_gen"] = gen
 		}
+	} else if media == "" && graphics == "" {
+		// 5.14 or older kernels need this
+		l.labels[labelNamespace+"media_version"] = gen
+		l.labels[labelNamespace+"graphics_version"] = gen
 	}
 }
 

cmd/gpu_nfdhook/labeler_test.go

@@ -1,4 +1,4 @@
-// Copyright 2020 Intel Corporation. All Rights Reserved.
+// Copyright 2020-2021 Intel Corporation. All Rights Reserved.
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -58,6 +58,8 @@ func getTestCases() []testcase {
 				"gpu.intel.com/platform_new.count":   "1",
 				"gpu.intel.com/platform_new.present": "true",
 				"gpu.intel.com/platform_new.tiles":   "1",
+				"gpu.intel.com/graphics_version":     "9",
+				"gpu.intel.com/media_version":        "9",
 				"gpu.intel.com/platform_gen":         "9",
 				"gpu.intel.com/cards":                "card0",
 			},
@@ -104,6 +106,8 @@ func getTestCases() []testcase {
 				"gpu.intel.com/platform_new.count":   "1",
 				"gpu.intel.com/platform_new.present": "true",
 				"gpu.intel.com/platform_new.tiles":   "2",
+				"gpu.intel.com/graphics_version":     "9",
+				"gpu.intel.com/media_version":        "9",
 				"gpu.intel.com/platform_gen":         "9",
 				"gpu.intel.com/cards":                "card0",
 			},
@@ -132,6 +136,8 @@ func getTestCases() []testcase {
 				"gpu.intel.com/platform_new.count":   "1",
 				"gpu.intel.com/platform_new.present": "true",
 				"gpu.intel.com/platform_new.tiles":   "1",
+				"gpu.intel.com/graphics_version":     "9",
+				"gpu.intel.com/media_version":        "9",
 				"gpu.intel.com/platform_gen":         "9",
 				"gpu.intel.com/cards":                "card0",
 			},
@@ -157,6 +163,8 @@ func getTestCases() []testcase {
 				"gpu.intel.com/platform_new.count":   "1",
 				"gpu.intel.com/platform_new.present": "true",
 				"gpu.intel.com/platform_new.tiles":   "1",
+				"gpu.intel.com/graphics_version":     "9",
+				"gpu.intel.com/media_version":        "9",
 				"gpu.intel.com/platform_gen":         "9",
 				"gpu.intel.com/cards":                "card0",
 			},
@@ -184,6 +192,60 @@ func getTestCases() []testcase {
 				"gpu.intel.com/cards":                "card0",
 			},
 		},
+		{
+			sysfsdirs: []string{
+				"card0/device/drm/card0",
+			},
+			sysfsfiles: map[string][]byte{
+				"card0/device/vendor": []byte("0x8086"),
+			},
+			name:           "gen version missing, but media & graphics versions present",
+			memoryOverride: 16000000000,
+			capabilityFile: map[string][]byte{
+				"0/i915_capabilities": []byte(
+					"platform: new\n" +
+						"media version: 12.5\n" +
+						"graphics version: 12.2"),
+			},
+			expectedRetval: nil,
+			expectedLabels: labelMap{
+				"gpu.intel.com/millicores":           "1000",
+				"gpu.intel.com/memory.max":           "16000000000",
+				"gpu.intel.com/platform_new.count":   "1",
+				"gpu.intel.com/platform_new.present": "true",
+				"gpu.intel.com/platform_new.tiles":   "1",
+				"gpu.intel.com/graphics_version":     "12.2",
+				"gpu.intel.com/media_version":        "12.5",
+				"gpu.intel.com/platform_gen":         "12",
+				"gpu.intel.com/cards":                "card0",
+			},
+		},
+		{
+			sysfsdirs: []string{
+				"card0/device/drm/card0",
+			},
+			sysfsfiles: map[string][]byte{
+				"card0/device/vendor": []byte("0x8086"),
+			},
+			name:           "only media version present",
+			memoryOverride: 16000000000,
+			capabilityFile: map[string][]byte{
+				"0/i915_capabilities": []byte(
+					"platform: new\n" +
+						"media version: 12.5"),
+			},
+			expectedRetval: nil,
+			expectedLabels: labelMap{
+				"gpu.intel.com/millicores":           "1000",
+				"gpu.intel.com/memory.max":           "16000000000",
+				"gpu.intel.com/platform_new.count":   "1",
+				"gpu.intel.com/platform_new.present": "true",
+				"gpu.intel.com/platform_new.tiles":   "1",
+				"gpu.intel.com/media_version":        "12.5",
+				"gpu.intel.com/platform_gen":         "12",
+				"gpu.intel.com/cards":                "card0",
+			},
+		},
 		{
 			sysfsdirs: []string{
 				"card0/device/drm/card0",