[pdata/pprofile] MergeTo: reserve index 0 of empty destination dictionary tables

.chloggen/fix-pprofile-mergeto-dictionary-zero-value.yaml

@@ -0,0 +1,30 @@
+# Use this changelog template to create an entry for release notes.
+
+# One of 'breaking', 'deprecation', 'new_component', 'enhancement', 'bug_fix'
+change_type: bug_fix
+
+# The name of the component, or a single word describing the area of concern, (e.g. receiver/otlp)
+component: pkg/pprofile
+
+# A brief description of the change.  Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: "`Profiles.MergeTo` no longer overwrites index 0 of an empty destination dictionary table with real data"
+
+# One or more tracking issues or pull requests related to the change
+issues: [15661]
+
+# (Optional) One or more lines of additional information to render under the primary note.
+# These lines will be padded with 2 spaces and then inserted directly into the document.
+# Use pipe (|) for multiline entries.
+subtext: |
+  `MergeTo` did not reserve index 0 of the destination dictionary's tables for their zero value before
+  remapping entries into them. When the destination dictionary was empty, the first entry merged in would
+  land at index 0, the slot every unset reference (e.g. an unset Strindex) resolves to, producing a
+  dictionary that violates the `ProfilesDictionary` protocol. `MergeTo` now seeds index 0 of any table that
+  is still empty with that table's zero value before remapping.
+
+# Optional: The change log or logs in which this entry should be included.
+# e.g. '[user]' or '[user, api]'
+# Include 'user' if the change is relevant to end users.
+# Include 'api' if there is a change to a library API.
+# Default: '[user]'
+change_logs: [api]

pdata/pprofile/profiles_merge.go

@@ -13,6 +13,8 @@ func (ms Profiles) MergeTo(dest Profiles) error {
 		return nil
 	}
 
+	reserveDictionaryZeroValues(dest.Dictionary())
+
 	if err := ms.switchDictionary(ms.Dictionary(), dest.Dictionary()); err != nil {
 		return err
 	}
@@ -22,3 +24,34 @@ func (ms Profiles) MergeTo(dest Profiles) error {
 
 	return nil
 }
+
+// reserveDictionaryZeroValues seeds index 0 of each of dst's tables with that
+// table's zero value, but only for tables that are still empty. Per the
+// ProfilesDictionary contract, index 0 of every table MUST hold the zero
+// value for that table's element type, since unset references (e.g. an
+// unset Strindex) resolve to it. Tables that already hold entries are left
+// untouched, since inserting at index 0 would invalidate every existing
+// reference into them.
+func reserveDictionaryZeroValues(dst ProfilesDictionary) {
+	if dst.StringTable().Len() == 0 {
+		dst.StringTable().Append("")
+	}
+	if dst.AttributeTable().Len() == 0 {
+		dst.AttributeTable().AppendEmpty()
+	}
+	if dst.FunctionTable().Len() == 0 {
+		dst.FunctionTable().AppendEmpty()
+	}
+	if dst.LinkTable().Len() == 0 {
+		dst.LinkTable().AppendEmpty()
+	}
+	if dst.LocationTable().Len() == 0 {
+		dst.LocationTable().AppendEmpty()
+	}
+	if dst.MappingTable().Len() == 0 {
+		dst.MappingTable().AppendEmpty()
+	}
+	if dst.StackTable().Len() == 0 {
+		dst.StackTable().AppendEmpty()
+	}
+}

pdata/pprofile/profiles_merge_test.go

@@ -585,6 +585,59 @@ func TestProfilesMergeTo(t *testing.T) {
 	}
 }
 
+// TestProfilesMergeTo_ReservesDictionaryZeroValues verifies that MergeTo
+// reserves index 0 of every destination table for that table's zero value,
+// even when the destination dictionary starts out completely empty (not
+// pre-populated by the caller). See
+// https://github.com/open-telemetry/opentelemetry-collector/issues/15661.
+func TestProfilesMergeTo_ReservesDictionaryZeroValues(t *testing.T) {
+	src := NewProfiles()
+	dic := src.Dictionary()
+	dic.StringTable().Append("") // conformant source
+	dic.AttributeTable().AppendEmpty()
+	dic.StackTable().AppendEmpty()
+	dic.LocationTable().AppendEmpty()
+	dic.FunctionTable().AppendEmpty()
+	dic.MappingTable().AppendEmpty()
+	dic.LinkTable().AppendEmpty()
+	dic.StringTable().Append("inuse_space") // index 1
+
+	prof := src.ResourceProfiles().AppendEmpty().ScopeProfiles().AppendEmpty().Profiles().AppendEmpty()
+	prof.SampleType().SetTypeStrindex(1)
+
+	dest := NewProfiles()
+	require.NoError(t, src.MergeTo(dest))
+
+	assert.Equal(t, "", dest.Dictionary().StringTable().At(0))
+	assert.Equal(t, "inuse_space", dest.Dictionary().StringTable().At(1))
+
+	require.Equal(t, 1, dest.Dictionary().AttributeTable().Len())
+	attr := dest.Dictionary().AttributeTable().At(0)
+	assert.Zero(t, attr.KeyStrindex())
+	assert.Zero(t, attr.UnitStrindex())
+	assert.Equal(t, pcommon.ValueTypeEmpty, attr.Value().Type())
+
+	require.Equal(t, 1, dest.Dictionary().StackTable().Len())
+	assert.Zero(t, dest.Dictionary().StackTable().At(0).LocationIndices().Len())
+
+	require.Equal(t, 1, dest.Dictionary().LocationTable().Len())
+	loc := dest.Dictionary().LocationTable().At(0)
+	assert.Zero(t, loc.MappingIndex())
+	assert.Zero(t, loc.Address())
+	assert.Zero(t, loc.Lines().Len())
+
+	require.Equal(t, 1, dest.Dictionary().FunctionTable().Len())
+	fn := dest.Dictionary().FunctionTable().At(0)
+	assert.Zero(t, fn.NameStrindex())
+	assert.Zero(t, fn.SystemNameStrindex())
+	assert.Zero(t, fn.FilenameStrindex())
+
+	require.Equal(t, 1, dest.Dictionary().MappingTable().Len())
+	assert.Zero(t, dest.Dictionary().MappingTable().At(0).FilenameStrindex())
+
+	require.Equal(t, 1, dest.Dictionary().LinkTable().Len())
+}
+
 func TestProfilesMergeToSelf(t *testing.T) {
 	profiles := NewProfiles()
 	profiles.Dictionary().StringTable().Append("", "test")