Group by JAR file in code area breakdown.

fniephaus · fniephaus · commit 188a706eba09 · 2023-01-05T09:59:59.000+01:00
And extend documentation on code area.
diff --git a/docs/reference-manual/native-image/BuildOutput.md b/docs/reference-manual/native-image/BuildOutput.md
@@ -20,51 +20,50 @@ Below is the example output when building a native executable of the `HelloWorld
 ================================================================================
 GraalVM Native Image: Generating 'helloworld' (executable)...
 ================================================================================
-[1/8] Initializing...                                            (2.5s @ 0.21GB)
- Version info: 'GraalVM dev Java 11 CE'
- C compiler: gcc (linux, x86_64, 9.3.0)
+[1/8] Initializing...                                            (3.7s @ 0.25GB)
+ Version info: 'GraalVM dev Java 19+36-jvmci-23.0-b01 CE'
+ Java version info: '19+36-jvmci-23.0-b01'
+ C compiler: gcc (linux, x86_64, 11.3.0)
  Garbage collector: Serial GC
-[2/8] Performing analysis...  [*******]                          (5.6s @ 0.46GB)
-   2,718 (72.93%) of  3,727 types reachable
-   3,442 (53.43%) of  6,442 fields reachable
-  12,128 (44.82%) of 27,058 methods reachable
-      27 types,     0 fields, and   271 methods registered for reflection
-      58 types,    59 fields, and    52 methods registered for JNI access
+[2/8] Performing analysis...  [****]                             (8.4s @ 2.56GB)
+   2,880 (71.93%) of  4,004 types reachable
+   3,519 (51.06%) of  6,892 fields reachable
+  13,339 (45.71%) of 29,181 methods reachable
+     879 types,     0 fields, and   356 methods registered for reflection
+      57 types,    56 fields, and    52 methods registered for JNI access
        4 native libraries: dl, pthread, rt, z
-[3/8] Building universe...                                       (0.5s @ 0.61GB)
-[4/8] Parsing methods...      [*]                                (0.5s @ 0.86GB)
-[5/8] Inlining methods...     [****]                             (0.5s @ 0.73GB)
-[6/8] Compiling methods...    [**]                               (3.7s @ 2.38GB)
-[7/8] Layouting methods...    [*]                                (0.5s @ 0.71GB)
-[8/8] Creating image...                                          (2.1s @ 1.04GB)
-   4.00MB (28.31%) for code area:     7,073 compilation units
-   5.90MB (41.70%) for image heap:   83,319 objects and 5 resources
-   3.24MB (22.91%) for debug info generated in 1.0s
-   1.00MB ( 7.08%) for other data
-  14.15MB in total
+[3/8] Building universe...                                       (1.4s @ 1.85GB)
+[4/8] Parsing methods...      [*]                                (1.3s @ 2.17GB)
+[5/8] Inlining methods...     [***]                              (0.9s @ 0.37GB)
+[6/8] Compiling methods...    [***]                              (8.4s @ 3.28GB)
+[7/8] Layouting methods...    [**]                               (4.5s @ 2.41GB)
+[8/8] Creating image...                                          (4.5s @ 4.90GB)
+   4.52MB (22.97%) for code area:     7,470 compilation units
+   7.06MB (35.89%) for image heap:  101,746 objects and 5 resources
+   7.52MB (38.20%) for debug info generated in 1.9s
+ 593.66KB ( 2.95%) for other data
+  19.68MB in total
 --------------------------------------------------------------------------------
-Top 10 packages in code area:           Top 10 object types in image heap:
- 632.68KB java.util                      871.62KB byte[] for code metadata
- 324.42KB java.lang                      798.53KB java.lang.String
- 223.90KB java.util.regex                774.91KB byte[] for general heap data
- 221.62KB java.text                      614.06KB java.lang.Class
- 198.30KB com.oracle.svm.jni             492.51KB byte[] for java.lang.String
- 166.02KB java.util.concurrent           314.81KB java.util.HashMap$Node
- 115.44KB java.math                      233.58KB c.o.s.c.h.DynamicHubCompanion
-  98.48KB sun.text.normalizer            154.84KB java.lang.String[]
-  97.42KB java.util.logging              139.54KB byte[] for embedded resources
-  95.18KB c.oracle.svm.core.genscavenge  139.04KB char[]
-   1.83MB for 118 more packages            1.29MB for 753 more object types
+Top 10 origins of code area:            Top 10 object types in image heap:
+ 761.01KB svm.jar                          1.01MB byte[] for code metadata
+ 700.69KB java.util                     1000.69KB java.lang.String
+ 361.45KB java.lang                      884.17KB byte[] for general heap data
+ 289.08KB java.text                      686.76KB byte[] for java.lang.String
+ 239.66KB java.util.regex                659.87KB java.lang.Class
+ 206.52KB java.util.concurrent           247.50KB c.o.s.c.h.DynamicHubCompanion
+ 156.60KB java.math                      239.21KB java.lang.Object[]
+ 127.50KB java.util.stream               225.42KB java.util.HashMap$Node
+ 122.44KB java.lang.invoke               173.15KB java.lang.String[]
+ 100.94KB java.util.logging              163.13KB j.u.c.ConcurrentHashMap$Node
+   1.47MB for 76 more origins              1.70MB for 808 more object types
 --------------------------------------------------------------------------------
-    0.9s (5.6% of total time) in 17 GCs | Peak RSS: 3.22GB | CPU load: 10.87
+    0.6s (1.7% of total time) in 23 GCs | Peak RSS: 5.60GB | CPU load: 9.39
 --------------------------------------------------------------------------------
 Produced artifacts:
- /home/janedoe/helloworld/helloworld (executable)
+ /home/janedoe/helloworld/helloworld (executable, debug_info)
  /home/janedoe/helloworld/sources (debug_info)
- /home/janedoe/helloworld/helloworld (debug_info)
- /home/janedoe/helloworld/helloworld.build_artifacts.txt
 ================================================================================
-Finished generating 'helloworld' in 16.2s.
+Finished generating 'helloworld' in 33.6s.
 ```
 
 ## Build Stages
@@ -148,6 +147,14 @@ Debug info is also generated as part of this stage (if requested).
 The code area contains machine code produced by the Graal compiler for all reachable methods.
 Therefore, reducing the number of [reachable methods](#glossary-reachability) also reduces the size of the code area.
 
+##### <a name="glossary-code-area-origins"></a>Origins of Code Area
+To help users understand where the machine code of the code area comes from, the build output shows a breakdown of the top origins.
+An origin is a group of Java sources and can be a JAR file, a package name, or a class name, depending on the information available.
+The package `java.util`, for example, contains `HashMap` and other classes from the JDK.
+The `svm.jar` file contains internal sources for the Native Image runtime called SubstrateVM.
+To reduce the size of the code area and with that, the total size of the native executable, re-evaluate the dependencies of your application based on the code area breakdown.
+Some libraries and frameworks are better prepared for Native Image than others, and newer versions of a library or framework may improve (or worsen) their code footprint.
+
 #### <a name="glossary-image-heap"></a>Image Heap
 The heap contains reachable objects such as static application data, metadata, and `byte[]` for different purposes (see below).
 
diff --git a/substratevm/src/com.oracle.svm.hosted/src/com/oracle/svm/hosted/ProgressReporter.java b/substratevm/src/com.oracle.svm.hosted/src/com/oracle/svm/hosted/ProgressReporter.java
@@ -33,6 +33,7 @@
 import java.lang.management.OperatingSystemMXBean;
 import java.lang.reflect.Field;
 import java.nio.file.Path;
+import java.security.CodeSource;
 import java.util.ArrayList;
 import java.util.Collection;
 import java.util.Comparator;
@@ -108,8 +109,6 @@ public class ProgressReporter {
     private static final boolean IS_CI = System.console() == null || System.getenv("CI") != null;
     private static final boolean IS_DUMB_TERM = isDumbTerm();
     private static final int MAX_NUM_BREAKDOWN = 10;
-    private static final String CODE_BREAKDOWN_TITLE = String.format("Top %d packages in code area:", MAX_NUM_BREAKDOWN);
-    private static final String HEAP_BREAKDOWN_TITLE = String.format("Top %d object types in image heap:", MAX_NUM_BREAKDOWN);
     private static final String STAGE_DOCS_URL = "https://github.com/oracle/graal/blob/master/docs/reference-manual/native-image/BuildOutput.md";
     private static final double EXCESSIVE_GC_MIN_THRESHOLD_MILLIS = 15_000;
     private static final double EXCESSIVE_GC_RATIO = 0.5;
@@ -487,12 +486,23 @@ public void createBreakdowns(HostedMetaAccess metaAccess, Collection<CompileTask
 
     private void calculateCodeBreakdown(Collection<CompileTask> compilationTasks) {
         for (CompileTask task : compilationTasks) {
-            String classOrPackageName = task.method.format("%H");
-            int lastDotIndex = classOrPackageName.lastIndexOf('.');
-            if (lastDotIndex > 0) {
-                classOrPackageName = classOrPackageName.substring(0, lastDotIndex);
+            CodeSource codeSource = task.method.getDeclaringClass().getJavaClass().getProtectionDomain().getCodeSource();
+            String key = null;
+            if (codeSource != null && codeSource.getLocation() != null) {
+                String path = codeSource.getLocation().getPath();
+                if (path.endsWith(".jar")) {
+                    // Use String API to determine basename of path to handle both / and \.
+                    key = path.substring(Math.max(path.lastIndexOf('/') + 1, path.lastIndexOf('\\') + 1));
+                }
+            }
+            if (key == null) {
+                key = task.method.format("%H");
+                int lastDotIndex = key.lastIndexOf('.');
+                if (lastDotIndex > 0) {
+                    key = key.substring(0, lastDotIndex);
+                }
             }
-            codeBreakdown.merge(classOrPackageName, (long) task.result.getTargetCodeSize(), Long::sum);
+            codeBreakdown.merge(key, (long) task.result.getTargetCodeSize(), Long::sum);
         }
     }
 
@@ -555,7 +565,9 @@ private void printBreakdowns() {
                         .sorted(Entry.comparingByValue(Comparator.reverseOrder())).iterator();
 
         final TwoColumnPrinter p = new TwoColumnPrinter();
-        p.l().yellowBold().a(CODE_BREAKDOWN_TITLE).jumpToMiddle().a(HEAP_BREAKDOWN_TITLE).reset().flushln();
+        p.l().yellowBold().a(String.format("Top %d ", MAX_NUM_BREAKDOWN)).doclink("origins", "#glossary-code-area-origins").a(" of code area:")
+                        .jumpToMiddle()
+                        .a(String.format("Top %d object types in image heap:", MAX_NUM_BREAKDOWN)).reset().flushln();
 
         long printedCodeBytes = 0;
         long printedHeapBytes = 0;
@@ -593,7 +605,7 @@ private void printBreakdowns() {
         int numHeapItems = heapBreakdown.size();
         long totalCodeBytes = codeBreakdown.values().stream().collect(Collectors.summingLong(Long::longValue));
         long totalHeapBytes = heapBreakdown.values().stream().collect(Collectors.summingLong(Long::longValue));
-        p.l().a(String.format("%9s for %s more packages", Utils.bytesToHuman(totalCodeBytes - printedCodeBytes), numCodeItems - printedCodeItems))
+        p.l().a(String.format("%9s for %s more ", Utils.bytesToHuman(totalCodeBytes - printedCodeBytes), numCodeItems - printedCodeItems)).doclink("origins", "#glossary-code-area-origins")
                         .jumpToMiddle()
                         .a(String.format("%9s for %s more object types", Utils.bytesToHuman(totalHeapBytes - printedHeapBytes), numHeapItems - printedHeapItems)).flushln();
     }
