test(mmds): add perf test for token generation and verification

Adds a perf test that generates and uses a MMDSv2 token 100 times in a
loop.

Signed-off-by: Riccardo Mancini <mancio@amazon.com>

Author: Manciukic

.buildkite/pipeline_perf.py

@@ -68,6 +68,10 @@
     "pmem": {
         "label": "💿 Pmem Performance",
         "tests": "integration_tests/performance/test_pmem.py",
+    },
+    "mmds": {
+        "label": ":information_source: MMDS",
+        "tests": "integration_tests/performance/test_mmds.py",
         "devtool_opts": "-c 1-10 -m 0",
     },
 }

tests/integration_tests/performance/test_mmds.py

@@ -0,0 +1,107 @@
+# Copyright 2025 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+# SPDX-License-Identifier: Apache-2.0
+"""Tests the performance of MMDS token generation and verification."""
+
+import re
+
+import pytest
+
+from framework.utils import configure_mmds, populate_data_store
+
+# Default IPv4 address for MMDS
+DEFAULT_IPV4 = "169.254.169.254"
+
+# Number of iterations for performance measurements
+ITERATIONS = 500
+
+
+def parse_curl_timing(timing_line):
+    """Parse curl timing output and extract timing information in milliseconds."""
+    # curl -w format outputs timing in seconds, convert to milliseconds
+    # Expected format: "time_total:0.123456"
+    match = re.search(r"time_total:([\d.]+)", timing_line)
+    if match:
+        return float(match.group(1)) * 1000  # Convert to milliseconds
+
+    raise ValueError(f"Could not parse timing from curl output: {timing_line}")
+
+
+@pytest.fixture
+def mmds_microvm(uvm_plain_any):
+    """Creates a microvm with MMDS configured for performance testing."""
+    uvm = uvm_plain_any
+    uvm.spawn(log_level="Info")
+    uvm.basic_config()
+    uvm.add_net_iface()
+
+    # Configure MMDS V2 (requires tokens)
+    configure_mmds(uvm, iface_ids=["eth0"], version="V2", ipv4_address=DEFAULT_IPV4)
+
+    # Populate with minimal test data
+    test_data = {"latest": {"meta-data": {"instance-id": "i-1234567890abcdef0"}}}
+    populate_data_store(uvm, test_data)
+
+    uvm.start()
+
+    uvm.ssh.check_output(f"ip route add {DEFAULT_IPV4} dev eth0")
+
+    return uvm
+
+
+@pytest.mark.nonci
+def test_mmds_token(mmds_microvm, metrics):
+    """
+    Test MMDS token generation performance using curl timing from within the guest.
+
+    This test measures the time it takes to generate MMDS session tokens
+    using curl's built-in timing capabilities.
+    """
+
+    metrics.set_dimensions(
+        {
+            "performance_test": "test_mmds_performance",
+            **mmds_microvm.dimensions,
+        }
+    )
+
+    # Measure token generation performance
+    for _ in range(ITERATIONS):
+        # Curl command to generate token with timing
+        token_cmd = (
+            f'curl -m 2 -s -w "\\ntime_total:%{{time_total}}" '
+            f'-X PUT -H "X-metadata-token-ttl-seconds: 60" '
+            f"http://{DEFAULT_IPV4}/latest/api/token"
+        )
+        _, stdout, stderr = mmds_microvm.ssh.check_output(token_cmd)
+        assert stderr == "", "Error generating token"
+
+        # Parse timing and token from output
+        lines = stdout.strip().split("\n")
+        token = lines[0].strip()  # First line is the token
+
+        # Verify token was generated successfully
+        assert len(token) > 0, f"Token generation failed. Output: {stdout}"
+
+        generation_time_ms = parse_curl_timing(lines[-1])
+        metrics.put_metric("token_generation_time", generation_time_ms, "Milliseconds")
+
+        # Curl command to verify token with timing
+        request_cmd = (
+            f'curl -m 2 -s -w "\\ntime_total:%{{time_total}}" '
+            f'-X GET -H "X-metadata-token: {token}" -H "Accept: application/json" '
+            f"http://{DEFAULT_IPV4}/latest/meta-data/instance-id"
+        )
+        _, stdout, stderr = mmds_microvm.ssh.check_output(request_cmd)
+        assert stderr == "", "MMDS request failed"
+
+        # Parse response and timing
+        lines = stdout.strip().split("\n")
+        response = lines[0].strip()  # First line is the response
+
+        # Verify request was successful
+        assert (
+            "i-1234567890abcdef0" in response
+        ), f"MMDS request failed. Response: {response}"
+
+        request_time_ms = parse_curl_timing(lines[-1])
+        metrics.put_metric("request_time", request_time_ms, "Milliseconds")