Tests for parsing musl information

packaging/_musllinux.py

@@ -4,11 +4,11 @@
 linked against musl, and what musl version is used.
 """
 
+import contextlib
 import functools
 import operator
 import os
 import re
-import shutil
 import struct
 import subprocess
 import sys
@@ -19,10 +19,11 @@ def _read_unpacked(f: IO[bytes], fmt: str) -> Tuple[int, ...]:
     return struct.unpack(fmt, f.read(struct.calcsize(fmt)))
 
 
-def _get_ld_musl_elf(f: IO[bytes]) -> Optional[str]:
+def _parse_ld_musl_from_elf(f: IO[bytes]) -> Optional[str]:
     """Detect musl libc location by parsing the Python executable.
 
-    Based on https://gist.github.com/lyssdod/f51579ae8d93c8657a5564aefc2ffbca
+    Based on: https://gist.github.com/lyssdod/f51579ae8d93c8657a5564aefc2ffbca
+    ELF header: https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.eheader.html
     """
     f.seek(0)
     try:
@@ -67,39 +68,21 @@ def _get_ld_musl_elf(f: IO[bytes]) -> Optional[str]:
     return None
 
 
-def _get_ld_musl_ldd(executable: str) -> Optional[str]:
-    ldd = shutil.which("ldd")
-    if not ldd:  # No dynamic program loader.
-        return None
-    proc = subprocess.run(
-        [ldd, executable], stdout=subprocess.PIPE, universal_newlines=True
-    )
-    if proc.returncode != 0:  # Not a valid dynamic program.
-        return None
-    for line in proc.stdout.splitlines(keepends=False):
-        path = line.lstrip().rsplit(None, 1)[0]
-        if "musl" not in path:
-            continue
-        return path
-    return None
-
-
-def _get_ld_musl(executable: str) -> Optional[str]:
-    try:
-        with open(executable, "rb") as f:
-            return _get_ld_musl_elf(f)
-    except IOError:
-        return _get_ld_musl_ldd(executable)
-
-
-_version_pat = re.compile(r"Version (\d+)\.(\d+)")
-
-
 class _MuslVersion(NamedTuple):
     major: int
     minor: int
 
 
+def _parse_musl_version(output: str) -> Optional[_MuslVersion]:
+    lines = [n for n in (n.strip() for n in output.splitlines()) if n]
+    if len(lines) < 2 or lines[0][:4] != "musl":
+        return None
+    m = re.match(r"Version (\d+)\.(\d+)", lines[1])
+    if not m:
+        return None
+    return _MuslVersion(major=int(m.group(1)), minor=int(m.group(2)))
+
+
 @functools.lru_cache()
 def _get_musl_version(executable: str) -> Optional[_MuslVersion]:
     """Detect currently-running musl runtime version.
@@ -112,17 +95,16 @@ def _get_musl_version(executable: str) -> Optional[_MuslVersion]:
         Version 1.2.2
         Dynamic Program Loader
     """
-    ld = _get_ld_musl(executable)
+    with contextlib.ExitStack() as stack:
+        try:
+            f = stack.enter_context(open(executable, "rb"))
+        except IOError:
+            return None
+        ld = _parse_ld_musl_from_elf(f)
     if not ld:
         return None
     proc = subprocess.run([ld], stderr=subprocess.PIPE, universal_newlines=True)
-    lines = [n for n in (n.strip() for n in proc.stderr.splitlines()) if n]
-    if len(lines) < 2 or lines[0][:4] != "musl":
-        return None
-    m = re.match(r"Version (\d+)\.(\d+)", lines[1])
-    if not m:
-        return None
-    return _MuslVersion(major=int(m.group(1)), minor=int(m.group(2)))
+    return _parse_musl_version(proc.stderr)
 
 
 def platform_tags(arch: str) -> Iterator[str]:
@@ -141,7 +123,7 @@ def platform_tags(arch: str) -> Iterator[str]:
         yield f"musllinux_{sys_musl.major}_{minor}_{arch}"
 
 
-if __name__ == "__main__":
+if __name__ == "__main__":  # pragma: no cover
     import sysconfig
 
     plat = sysconfig.get_platform()

tests/musllinux/build.sh

@@ -0,0 +1,58 @@
+# Build helper binaries for musllinux tests.
+# Usages:
+#   build.sh              # Build everything.
+#   build.sh $DISTRO $ARCH  # Build one executable in $ARCH using $DISTRO.
+#
+# Either invocation ultimately runs this script in a Docker container with
+# `build.sh glibc|musl $ARCH` to actually build the executable.
+
+set -euo pipefail
+set -x
+
+build_one_in_ubuntu () {
+    $1 "multiarch/ubuntu-core:${2}-focal" \
+        bash "/home/hello-world/musllinux/build.sh" glibc "glibc-$2"
+}
+
+build_one_in_alpine () {
+    $1 "multiarch/alpine:${2}-edge" \
+        sh "/home/hello-world/musllinux/build.sh" musl "musl-$2"
+}
+
+build_in_container () {
+    local SOURCE="$(dirname $(dirname $(realpath ${BASH_SOURCE[0]})))"
+    local DOCKER="docker run --rm -v ${SOURCE}:/home/hello-world"
+
+    if [[ $# -ne 0 ]]; then
+        "build_one_in_${1}" "$DOCKER" "$2"
+        return
+    fi
+
+    build_one_in_alpine "$DOCKER" x86_64
+    build_one_in_alpine "$DOCKER" i386
+    build_one_in_alpine "$DOCKER" aarch64
+    build_one_in_ubuntu "$DOCKER" x86_64
+}
+
+if [[ $# -eq 0 ]]; then
+    build_in_container
+    exit 0
+elif [[ "$1" == "glibc" ]]; then
+    DEBIAN_FRONTEND=noninteractive apt-get update -qq \
+        && apt-get install -qqy --no-install-recommends gcc libc6-dev
+elif [[ "$1" == "musl" ]]; then
+    apk add -q build-base
+else
+    build_in_container "$@"
+    exit 0
+fi
+
+build () {
+    local CFLAGS=""
+    local OUT="/home/hello-world/musllinux/${2}"
+    gcc -Os ${CFLAGS} -o "$OUT-full" "/home/hello-world/hello-world.c"
+    head -c1024 "$OUT-full" > "$OUT"
+    rm -f "$OUT-full"
+}
+
+build "$@"

tests/test_musllinux.py

@@ -0,0 +1,124 @@
+import collections
+import io
+import pathlib
+import struct
+
+import pretend
+import pytest
+
+from packaging import _musllinux
+from packaging._musllinux import (
+    _MuslVersion,
+    _get_musl_version,
+    _parse_musl_version,
+    _parse_ld_musl_from_elf,
+)
+
+
+MUSL_AMD64_122 = "musl libc (x86_64)\nVersion 1.2.2\n"
+MUSL_I386_121 = "musl libc (i386)\nVersion 1.2.1\n"
+MUSL_AARCH64_1124 = "musl libc (aarch64)\nVersion 1.1.24\n"
+MUSL_INVALID = "musl libc (invalid)\n"
+MUSL_UNKNOWN = "musl libc (unknown)\nVersion unknown\n"
+
+MUSL_DIR = pathlib.Path(__file__).with_name("musllinux").resolve()
+
+BINARY_GLIBC_X86_64 = MUSL_DIR.joinpath("glibc-x86_64")
+BINARY_MUSL_X86_64 = MUSL_DIR.joinpath("musl-x86_64")
+BINARY_MUSL_I386 = MUSL_DIR.joinpath("musl-i386")
+BINARY_MUSL_AARCH64 = MUSL_DIR.joinpath("musl-aarch64")
+
+
+@pytest.mark.parametrize(
+    "output, version",
+    [
+        (MUSL_AMD64_122, _MuslVersion(1, 2)),
+        (MUSL_I386_121, _MuslVersion(1, 2)),
+        (MUSL_AARCH64_1124, _MuslVersion(1, 1)),
+        (MUSL_INVALID, None),
+        (MUSL_UNKNOWN, None),
+    ],
+    ids=["amd64-1.2.2", "i386-1.2.1", "aarch64-1.1.24", "invalid", "unknown"],
+)
+def test_parse_musl_version(output, version):
+    assert _parse_musl_version(output) == version
+
+
+@pytest.mark.parametrize(
+    "executable, location",
+    [
+        (BINARY_GLIBC_X86_64, None),
+        (BINARY_MUSL_X86_64, "/lib/ld-musl-x86_64.so.1"),
+        (BINARY_MUSL_I386, "/lib/ld-musl-i386.so.1"),
+        (BINARY_MUSL_AARCH64, "/lib/ld-musl-aarch64.so.1"),
+    ],
+    ids=["glibc", "x86_64", "i386", "aarch64"],
+)
+def test_parse_ld_musl_from_elf(executable, location):
+    with executable.open("rb") as f:
+        assert _parse_ld_musl_from_elf(f) == location
+
+
+@pytest.mark.parametrize(
+    "data",
+    [
+        # Too short for magic.
+        b"\0",
+        # Enough for magic, but not ELF.
+        b"#!/bin/bash" + b"\0" * 16,
+        # ELF, but unknown byte declaration.
+        b"\x7fELF\3" + b"\0" * 16,
+    ],
+    ids=["no-magic", "wrong-magic", "unknown-format"],
+)
+def test_parse_ld_musl_from_elf_invalid(data):
+    assert _parse_ld_musl_from_elf(io.BytesIO(data)) is None
+
+
+@pytest.mark.parametrize(
+    "head",
+    [
+        25,  # Enough for magic, but not the section definitions.
+        58,  # Enough for section definitions, but not the actual sections.
+    ],
+)
+def test_parse_ld_musl_from_elf_invalid_section(head):
+    data = BINARY_MUSL_X86_64.read_bytes()[:head]
+    assert _parse_ld_musl_from_elf(io.BytesIO(data)) is None
+
+
+def test_parse_ld_musl_from_elf_no_interpreter_section():
+    with BINARY_MUSL_X86_64.open("rb") as f:
+        data = f.read()
+
+    # Change all sections to *not* PT_INTERP.
+    unpacked = struct.unpack("16BHHIQQQIHHH", data[:58])
+    *_, e_phoff, _, _, _, e_phentsize, e_phnum = unpacked
+    for i in range(e_phnum + 1):
+        sb = e_phoff + e_phentsize * i
+        se = sb + 56
+        section = struct.unpack("IIQQQQQQ", data[sb:se])
+        data = data[:sb] + struct.pack("IIQQQQQQ", 0, *section[1:]) + data[se:]
+
+    assert _parse_ld_musl_from_elf(io.BytesIO(data)) is None
+
+
+@pytest.mark.parametrize(
+    "executable, output, version",
+    [
+        (MUSL_DIR.joinpath("does-not-exist"), "error", None),
+        (BINARY_GLIBC_X86_64, "error", None),
+        (BINARY_MUSL_X86_64, MUSL_AMD64_122, _MuslVersion(1, 2)),
+        (BINARY_MUSL_I386, MUSL_I386_121, _MuslVersion(1, 2)),
+        (BINARY_MUSL_AARCH64, MUSL_AARCH64_1124, _MuslVersion(1, 1)),
+    ],
+    ids=["does-not-exist", "glibc", "x86_64", "i386", "aarch64"],
+)
+def test_get_musl_version(monkeypatch, executable, output, version):
+    def mock_run(*args, **kwargs):
+        return collections.namedtuple("Proc", "stderr")(output)
+
+    run_recorder = pretend.call_recorder(mock_run)
+    monkeypatch.setattr(_musllinux.subprocess, "run", run_recorder)
+
+    assert _get_musl_version(str(executable)) == version