test(profiling): add missing prof correctness tests for I/O profiling

.github/workflows/prof_correctness.yml

@@ -77,7 +77,7 @@ jobs:
           export DD_PROFILING_EXCEPTION_MESSAGE_ENABLED=1
           php -v
           php -d extension=target/profiler-release/libdatadog_php_profiling.so --ri datadog-profiling
-          for test_case in "allocations" "time" "strange_frames" "timeline" "exceptions"; do
+          for test_case in "allocations" "time" "strange_frames" "timeline" "exceptions" "io"; do
               mkdir -p profiling/tests/correctness/"$test_case"/
               export DD_PROFILING_OUTPUT_PPROF=$PWD/profiling/tests/correctness/"$test_case"/test.pprof
               php -d extension="${PWD}/target/profiler-release/libdatadog_php_profiling.so" "profiling/tests/correctness/${test_case}.php"
@@ -95,7 +95,7 @@ jobs:
           export DD_PROFILING_EXCEPTION_MESSAGE_ENABLED=1
           php -v
           php -d extension=target/profiler-release/libdatadog_php_profiling.so --ri datadog-profiling
-          for test_case in "allocations" "time" "strange_frames" "timeline" "exceptions"; do
+          for test_case in "allocations" "time" "strange_frames" "timeline" "exceptions" "io"; do
               mkdir -p profiling/tests/correctness/"$test_case"/
               export DD_PROFILING_OUTPUT_PPROF=$PWD/profiling/tests/correctness/"$test_case"/test.pprof
               php -d extension=$PWD/target/profiler-release/libdatadog_php_profiling.so profiling/tests/correctness/"$test_case".php
@@ -188,6 +188,12 @@ jobs:
           expected_json: profiling/tests/correctness/allocation_time_combined.json
           pprof_path: profiling/tests/correctness/allocation_time_combined/
 
+      - name: Check profiler correctness for IO
+        uses: Datadog/prof-correctness/analyze@main
+        with:
+          expected_json: profiling/tests/correctness/io.json
+          pprof_path: profiling/tests/correctness/io/
+
       - name: Check profiler correctness for exceptions ZTS
         if: matrix.phpts == 'zts'
         uses: Datadog/prof-correctness/analyze@main

profiling/src/profiling/sample_type_filter.rs

@@ -288,4 +288,46 @@ mod tests {
         );
         assert_eq!(values, vec![10, 20, 30, 70]);
     }
+
+    #[test]
+    fn filter_with_io_profiling() {
+        let mut settings = get_system_settings();
+        settings.profiling_enabled = true;
+        settings.profiling_io_enabled = true;
+
+        let sample_type_filter = SampleTypeFilter::new(&settings);
+        let values = sample_type_filter.filter(get_samples());
+        let types = sample_type_filter.sample_types();
+
+        assert_eq!(
+            types,
+            vec![
+                ValueType::new("sample", "count"),
+                ValueType::new("wall-time", "nanoseconds"),
+                ValueType::new("socket-read-time", "nanoseconds"),
+                ValueType::new("socket-read-time-samples", "count"),
+                ValueType::new("socket-write-time", "nanoseconds"),
+                ValueType::new("socket-write-time-samples", "count"),
+                ValueType::new("file-io-read-time", "nanoseconds"),
+                ValueType::new("file-io-read-time-samples", "count"),
+                ValueType::new("file-io-write-time", "nanoseconds"),
+                ValueType::new("file-io-write-time-samples", "count"),
+                ValueType::new("socket-read-size", "bytes"),
+                ValueType::new("socket-read-size-samples", "count"),
+                ValueType::new("socket-write-size", "bytes"),
+                ValueType::new("socket-write-size-samples", "count"),
+                ValueType::new("file-io-read-size", "bytes"),
+                ValueType::new("file-io-read-size-samples", "count"),
+                ValueType::new("file-io-write-size", "bytes"),
+                ValueType::new("file-io-write-size-samples", "count"),
+            ]
+        );
+        assert_eq!(
+            values,
+            vec![
+                10, 20, 80, 81, 90, 91, 100, 101, 110, 111, 120, 121, 130,
+                131, 140, 141, 150, 151
+            ]
+        );
+    }
 }

profiling/tests/correctness/io.json

@@ -0,0 +1,36 @@
+{
+  "scale_by_duration": true,
+  "test_name": "php_io",
+  "stacks": [
+    {
+      "profile-type": "file-io-write-size",
+      "stack-content": [
+        {
+          "regular_expression": "<?php;main;big_write;standard\\|file_put_contents$",
+          "percent": 66,
+          "error_margin": 2
+        },
+        {
+          "regular_expression": "<?php;main;small_write;standard\\|file_put_contents$",
+          "percent": 33,
+          "error_margin": 2
+        }
+      ]
+    },
+    {
+      "profile-type": "file-io-read-size",
+      "stack-content": [
+        {
+          "regular_expression": "<?php;main;big_read;standard\\|file_get_contents$",
+          "percent": 66,
+          "error_margin": 2
+        },
+        {
+          "regular_expression": "<?php;main;small_read;standard\\|file_get_contents$",
+          "percent": 33,
+          "error_margin": 2
+        }
+      ]
+    }
+  ]
+}

profiling/tests/correctness/io.php

@@ -0,0 +1,50 @@
+<?php
+// Test I/O profiling correctness.
+//
+// We write and read temporary files in known proportions to verify that
+// file-io-write-size and file-io-read-size samples are captured correctly.
+//
+// Function `big_write` writes 2x more data than `small_write`, so we expect
+// ~67% vs ~33% of file-io-write-size. Similarly for reads.
+//
+// The write functions return the file path so that read functions can reuse
+// the same files without creating their own (which would pollute the write
+// profile). Files are cleaned up in the main loop.
+
+function big_write() {
+    $tmp = tempnam(sys_get_temp_dir(), 'io_test');
+    // Write 2MB
+    file_put_contents($tmp, str_repeat('A', 1024 * 1024 * 2));
+    return $tmp;
+}
+
+function small_write() {
+    $tmp = tempnam(sys_get_temp_dir(), 'io_test');
+    // Write 1MB
+    file_put_contents($tmp, str_repeat('B', 1024 * 1024));
+    return $tmp;
+}
+
+function big_read($path) {
+    // Read 2MB
+    file_get_contents($path, false, null, 0, 1024 * 1024 * 2);
+}
+
+function small_read($path) {
+    // Read 1MB
+    file_get_contents($path, false, null, 0, 1024 * 1024);
+}
+
+function main() {
+    $duration = $_ENV["EXECUTION_TIME"] ?? 10;
+    $end = microtime(true) + $duration;
+    while (microtime(true) < $end) {
+        $big = big_write();
+        $small = small_write();
+        big_read($big);
+        small_read($small);
+        unlink($big);
+        unlink($small);
+    }
+}
+main();