kernel: profiling infrastructure

prof/prof.py

@@ -0,0 +1,102 @@
+#!/usr/bin/env python3
+
+# This script takes data generated by the profiler in Astral and creates a flamegraph out of it.
+# Format of the profiler file:
+# 1 byte indicating the number of calls in the backtrace
+# 8-byte Pointers starting by deepest call in the stack
+# Repeat until EOF
+
+import sys
+import os
+import subprocess
+
+if len(sys.argv) != 3:
+    print(sys.argv[0] + ": usage: " + sys.argv[0] + " INPUT_FILE OUTPUT_FILE")
+    exit(1)
+
+# Open needed files
+in_file = open(sys.argv[1], "rb")
+out_file = open(sys.argv[2], "w")
+
+# Download the flamegraph script if nescessary
+os.chdir(os.path.dirname(__file__))
+
+if not os.access("flamegraph.pl", os.R_OK):
+    os.system("wget https://raw.githubusercontent.com/brendangregg/FlameGraph/refs/heads/master/flamegraph.pl")
+    os.system("chmod +x flamegraph.pl")
+
+print("Starting subprocesses")
+# Start addr2line TODO: other architectures
+proc_a2l = subprocess.Popen(
+        ["addr2line", "-s", "-f", "-e", "../build-x86_64/iso/astral"],
+        stdin=subprocess.PIPE,
+        stdout=subprocess.PIPE,
+        text=True
+)
+
+# Start the flamegraph renderer
+proc_fg = subprocess.Popen(
+    ["./flamegraph.pl"],
+    stdin=subprocess.PIPE,
+    stdout=out_file,
+    text=True
+)
+
+traces = {}
+
+print("Getting function information")
+# get info out of addr2line
+while True:
+    backtrace_size = in_file.read(1)
+    if not backtrace_size:
+        # EOF
+        break
+
+    backtrace_size = int.from_bytes(backtrace_size, byteorder='little', signed=False)
+
+    # Send the addresses to addr2line
+    i = backtrace_size
+    while i > 0:
+        i -= 1
+
+        addr = in_file.read(8)
+        if not addr:
+            print(("Corrupted file (unexpected EOF, reading %d entries)" % (backtrace_size)), file=sys.stderr)
+            exit(1)
+
+        addr = hex(int.from_bytes(addr, byteorder='little', signed=False))
+
+        proc_a2l.stdin.write(addr + "\n")
+        proc_a2l.stdin.flush()
+
+    # Read back the lines from addr2line and format them in a way flamegraph expects
+    i = backtrace_size
+    formatted_str = ""
+    while i > 0:
+        i -= 1
+
+        function = proc_a2l.stdout.readline().replace('\n', '')
+        file, line = proc_a2l.stdout.readline().replace('\n', '').split(":")
+
+        formatted_str = file + "`" + function + ";" + formatted_str
+
+    formatted_str = formatted_str[:-1] # remove ;
+
+    if not formatted_str in traces:
+        traces[formatted_str] = 0
+
+    traces[formatted_str] += 1
+        
+proc_a2l.stdin.close()
+
+print("Generating flamegraph")
+# send the trace over to flamegraph
+for trace in traces:
+    final_string = trace + " " + str(traces[trace])
+    proc_fg.stdin.write(final_string + "\n")
+    proc_fg.stdin.flush()
+
+print("Waiting...");
+# wait until it is done with its thing
+proc_fg.stdin.close()
+proc_fg.wait()