Initial PHP support using phpspy (#87)

Dockerfile

@@ -29,6 +29,12 @@ WORKDIR /bcc
 COPY ./scripts/pyperf_build.sh .
 RUN ./pyperf_build.sh
 
+# ubuntu:20.04
+FROM ubuntu@sha256:cf31af331f38d1d7158470e095b132acd126a7180a54f263d386da88eb681d93 as phpspy-builder
+RUN apt update && apt install -y git wget make gcc
+COPY scripts/phpspy_build.sh .
+RUN ./phpspy_build.sh
+
 
 # ubuntu 20.04
 FROM ubuntu@sha256:cf31af331f38d1d7158470e095b132acd126a7180a54f263d386da88eb681d93
@@ -47,6 +53,10 @@ COPY --from=bcc-builder /bcc/bcc/NOTICE gprofiler/resources/python/pyperf/
 COPY --from=pyspy-builder /py-spy/target/x86_64-unknown-linux-musl/release/py-spy gprofiler/resources/python/py-spy
 COPY --from=perf-builder /perf gprofiler/resources/perf
 
+RUN mkdir -p gprofiler/resources/python/phpspy
+COPY --from=phpspy-builder /phpspy/phpspy gprofiler/resources/php/phpspy
+COPY --from=phpspy-builder /binutils/binutils-2.25/bin/bin/objdump gprofiler/resources/php/objdump
+
 COPY scripts/build.sh scripts/build.sh
 RUN ./scripts/build.sh
 

README.md

@@ -114,6 +114,8 @@ Alongside `perf`, gProfiler invokes runtime-specific profilers for processes bas
 * The CPython interpreter, versions 2.7 and 3.5-3.9.
   * eBPF profiling (based on PyPerf) requires Linux 4.14 or higher. Profiling using eBPF incurs lower overhead. This requires kernel headers to be installed.
   * If eBPF is not available for whatever reason, py-spy is used.
+* PHP (Zend Engine), versions 7.0-8.0.
+  * Uses [Granulate's fork](https://github.com/Granulate/phpspy/) of the phpspy project.
 
 The runtime-specific profilers produce stack traces that include runtime information (i.e, stacks of Java/Python functions), unlike `perf` which produces native stacks of the JVM / CPython interpreter.
 The runtime stacks are then merged into the data collected by `perf`, substituting the *native* stacks `perf` has collected for those processes.
@@ -133,6 +135,7 @@ We recommend going through our [contribution guide](https://github.com/granulate
 * [async-profiler](https://github.com/jvm-profiling-tools/async-profiler) by [Andrei Pangin](https://github.com/apangin). See [our fork](https://github.com/Granulate/async-profiler).
 * [py-spy](https://github.com/benfred/py-spy) by [Ben Frederickson](https://github.com/benfred). See [our fork](https://github.com/Granulate/py-spy).
 * [bcc](https://github.com/iovisor/bcc) (for PyPerf) by the IO Visor project. See [our fork](https://github.com/Granulate/bcc).
+* [phpspy](https://github.com/adsr/phpspy) by [Adam Saponara](https://github.com/adsr). See [our fork](https://github.com/Granulate/phpspy).
 
 # Footnotes
 

gprofiler/main.py

@@ -15,7 +15,7 @@
 from pathlib import Path
 from socket import gethostname
 from threading import Event
-from typing import Dict, Optional
+from typing import Callable, Dict, Optional
 
 import configargparse
 from requests import RequestException, Timeout
@@ -24,6 +24,7 @@
 from .client import DEFAULT_UPLOAD_TIMEOUT, GRANULATE_SERVER_HOST, APIClient, APIError
 from .java import JavaProfiler
 from .perf import SystemProfiler
+from .php import PHPSpyProfiler
 from .profiler_base import NoopProfiler
 from .python import get_python_profiler
 from .utils import (
@@ -66,6 +67,14 @@ def sigint_handler(sig, frame):
         raise KeyboardInterrupt
 
 
+def create_profiler_or_noop(profiler_constructor_callback: Callable, runtime_name: str):
+    try:
+        return profiler_constructor_callback()
+    except Exception:
+        logger.exception(f"Couldn't create {runtime_name} profiler, continuing without this runtime profiler")
+        return NoopProfiler()
+
+
 class GProfiler:
     def __init__(
         self,
@@ -76,6 +85,7 @@ def __init__(
         rotating_output: bool,
         runtimes: Dict[str, bool],
         client: APIClient,
+        php_process_filter: str = PHPSpyProfiler.DEFAULT_PROCESS_FILTER,
     ):
         self._frequency = frequency
         self._duration = duration
@@ -93,14 +103,29 @@ def __init__(
         # files unnecessarily.
         self._temp_storage_dir = TemporaryDirectoryWithMode(dir=TEMPORARY_STORAGE_PATH, mode=0o755)
         self.java_profiler = (
-            JavaProfiler(self._frequency, self._duration, True, self._stop_event, self._temp_storage_dir.name)
+            create_profiler_or_noop(
+                lambda: JavaProfiler(
+                    self._frequency, self._duration, True, self._stop_event, self._temp_storage_dir.name
+                ),
+                "java",
+            )
             if self._runtimes["java"]
             else NoopProfiler()
         )
         self.system_profiler = SystemProfiler(
             self._frequency, self._duration, self._stop_event, self._temp_storage_dir.name
         )
         self.initialize_python_profiler()
+        self.php_profiler = (
+            create_profiler_or_noop(
+                lambda: PHPSpyProfiler(
+                    self._frequency, self._duration, self._stop_event, self._temp_storage_dir.name, php_process_filter
+                ),
+                "php",
+            )
+            if self._runtimes["php"]
+            else NoopProfiler()
+        )
 
     def __enter__(self):
         self.start()
@@ -111,12 +136,15 @@ def __exit__(self, exc_type, exc_val, exc_tb):
 
     def initialize_python_profiler(self) -> None:
         self.python_profiler = (
-            get_python_profiler(
-                self._frequency,
-                self._duration,
-                self._stop_event,
-                self._temp_storage_dir.name,
-                self.initialize_python_profiler,
+            create_profiler_or_noop(
+                lambda: get_python_profiler(
+                    self._frequency,
+                    self._duration,
+                    self._stop_event,
+                    self._temp_storage_dir.name,
+                    self.initialize_python_profiler,
+                ),
+                "python",
             )
             if self._runtimes["python"]
             else NoopProfiler()
@@ -179,6 +207,7 @@ def start(self):
             self.python_profiler,
             self.java_profiler,
             self.system_profiler,
+            self.php_profiler,
         ):
             prof.start()
 
@@ -190,6 +219,7 @@ def stop(self):
             self.python_profiler,
             self.java_profiler,
             self.system_profiler,
+            self.php_profiler,
         ):
             prof.stop()
 
@@ -201,11 +231,13 @@ def _snapshot(self):
         java_future.name = "java"
         python_future = self._executor.submit(self.python_profiler.snapshot)
         python_future.name = "python"
+        php_future = self._executor.submit(self.php_profiler.snapshot)
+        php_future.name = "php"
         system_future = self._executor.submit(self.system_profiler.snapshot)
         system_future.name = "system"
 
         process_perfs: Dict[int, Dict[str, int]] = {}
-        for future in concurrent.futures.as_completed([java_future, python_future]):
+        for future in concurrent.futures.as_completed([java_future, python_future, php_future]):
             # if either of these fail - log it, and continue.
             try:
                 process_perfs.update(future.result())
@@ -325,6 +357,19 @@ def parse_cmd_args():
         default=True,
         help="Do not invoke runtime-specific profilers for Python processes",
     )
+    parser.add_argument(
+        "--no-php",
+        dest="php",
+        action="store_false",
+        default=True,
+        help="Do not invoke runtime-specific profilers for PHP processes",
+    )
+    parser.add_argument(
+        "--php-proc-filter",
+        dest="php_process_filter",
+        default=PHPSpyProfiler.DEFAULT_PROCESS_FILTER,
+        help="Process filter for php processes (default: %(default)s)",
+    )
 
     parser.add_argument(
         "-u",
@@ -463,9 +508,16 @@ def main():
             logger.error(f"Failed to connect to server: {e}")
             return
 
-        runtimes = {"java": args.java, "python": args.python}
+        runtimes = {"java": args.java, "python": args.python, "php": args.php}
         gprofiler = GProfiler(
-            args.frequency, args.duration, args.output_dir, args.flamegraph, args.rotating_output, runtimes, client
+            args.frequency,
+            args.duration,
+            args.output_dir,
+            args.flamegraph,
+            args.rotating_output,
+            runtimes,
+            client,
+            args.php_process_filter,
         )
         logger.info("gProfiler initialized and ready to start profiling")