|
| 1 | +From 4fb328cb883504dde04dfdd0b4d182a0130a0909 Mon Sep 17 00:00:00 2001 |
| 2 | +From: Geoffrey Thomas <geofft@ldpreload.com> |
| 3 | +Date: Mon, 1 Dec 2025 14:11:43 -0500 |
| 4 | +Subject: [PATCH 1/1] getpath: Fix library detection and canonicalize paths on |
| 5 | + Linux |
| 6 | +Forwarded: no |
| 7 | + |
| 8 | +The code in getpath.py to look for the stdlib relative to the Python |
| 9 | +library did not work in the common layout where libpython itself is in |
| 10 | +the lib/ directory; it added an extra lib/ segment. It is also equally |
| 11 | +applicable and useful when statically linking libpython into bin/python; |
| 12 | +in both cases, we want to go up a directory and then look into |
| 13 | +lib/python3.x/. Add an extra dirname() call in getpath.py, and |
| 14 | +unconditionally attempt to fill in the "library" variable in getpath.c, |
| 15 | +even on builds that are statically linking libpython. |
| 16 | + |
| 17 | +Also, we want to use the realpath'd version of the library's path to |
| 18 | +locate the standard library, particularly in the case where the library |
| 19 | +is a symlink to an executable statically linking libpython. On macOS |
| 20 | +dyld, this is done automatically. On glibc and musl, we often get |
| 21 | +relative paths and they are not canonicalized, so instead, use |
| 22 | +/proc/self/maps to find the file where libpython is coming from. |
| 23 | + |
| 24 | +(We could instead use the origin, which is canonicalized, but there is |
| 25 | +no safe API on glibc to read it and no API at all on musl. Note that and |
| 26 | +glibc also uses procfs to do so; see discussion at |
| 27 | +https://sourceware.org/bugzilla/show_bug.cgi?id=25263) |
| 28 | +--- |
| 29 | + Modules/getpath.c | 52 ++++++++++++++++++++++++++++++++++++++++------ |
| 30 | + Modules/getpath.py | 4 ++-- |
| 31 | + 2 files changed, 48 insertions(+), 8 deletions(-) |
| 32 | + |
| 33 | +diff --git a/Modules/getpath.c b/Modules/getpath.c |
| 34 | +index 1e75993480a..72860807133 100644 |
| 35 | +--- a/Modules/getpath.c |
| 36 | ++++ b/Modules/getpath.c |
| 37 | +@@ -802,14 +802,19 @@ progname_to_dict(PyObject *dict, const char *key) |
| 38 | + } |
| 39 | + |
| 40 | + |
| 41 | ++static void |
| 42 | ++fclose_cleanup(FILE **pf) { |
| 43 | ++ if (*pf) { |
| 44 | ++ fclose(*pf); |
| 45 | ++ *pf = NULL; |
| 46 | ++ } |
| 47 | ++} |
| 48 | ++ |
| 49 | ++ |
| 50 | + /* Add the runtime library's path to the dict */ |
| 51 | + static int |
| 52 | + library_to_dict(PyObject *dict, const char *key) |
| 53 | + { |
| 54 | +-/* macOS framework builds do not link against a libpython dynamic library, but |
| 55 | +- instead link against a macOS Framework. */ |
| 56 | +-#if defined(Py_ENABLE_SHARED) || defined(WITH_NEXT_FRAMEWORK) |
| 57 | +- |
| 58 | + #ifdef MS_WINDOWS |
| 59 | + extern HMODULE PyWin_DLLhModule; |
| 60 | + if (PyWin_DLLhModule) { |
| 61 | +@@ -817,12 +822,47 @@ library_to_dict(PyObject *dict, const char *key) |
| 62 | + } |
| 63 | + #endif |
| 64 | + |
| 65 | ++ const void *target = (void *)Py_Initialize; |
| 66 | ++ |
| 67 | ++#ifdef __linux__ |
| 68 | ++ /* Linux libcs do not reliably report the realpath in dladdr dli_fname and |
| 69 | ++ * sometimes return relative paths, especially if the returned object is |
| 70 | ++ * the main program itself. However, /proc/self/maps will give absolute |
| 71 | ++ * realpaths (from the kernel, for the same reason that /proc/self/exe is |
| 72 | ++ * canonical), so try to parse and look it up there. (dyld seems to |
| 73 | ++ * reliably report the canonical path, so doing this matches the behavior |
| 74 | ++ * on macOS.) */ |
| 75 | ++ |
| 76 | ++ __attribute__((cleanup(fclose_cleanup))) |
| 77 | ++ FILE *maps = fopen("/proc/self/maps", "r"); |
| 78 | ++ if (maps != NULL) { |
| 79 | ++ /* See implementation in fs/proc/task_mmu.c for spacing. The pathname |
| 80 | ++ * is the last field and has any \n characters escaped, so we can read |
| 81 | ++ * until \n. Note that the filename may have " (deleted)" appended; |
| 82 | ++ * we don't bother to handle that specially as the only user of this |
| 83 | ++ * value calls dirname() anyway. |
| 84 | ++ * TODO(geofft): Consider using PROCMAP_QUERY if supported. |
| 85 | ++ */ |
| 86 | ++ uintptr_t low, high; |
| 87 | ++ char filename[PATH_MAX]; |
| 88 | ++ while (fscanf(maps, |
| 89 | ++ "%lx-%lx %*s %*s %*s %*s %[^\n]", |
| 90 | ++ &low, &high, filename) == 3) { |
| 91 | ++ if (low <= (uintptr_t)target && (uintptr_t)target < high) { |
| 92 | ++ if (filename[0] == '/') { |
| 93 | ++ return decode_to_dict(dict, key, filename); |
| 94 | ++ } |
| 95 | ++ break; |
| 96 | ++ } |
| 97 | ++ } |
| 98 | ++ } |
| 99 | ++#endif |
| 100 | ++ |
| 101 | + #if HAVE_DLADDR |
| 102 | + Dl_info libpython_info; |
| 103 | +- if (dladdr(&Py_Initialize, &libpython_info) && libpython_info.dli_fname) { |
| 104 | ++ if (dladdr(target, &libpython_info) && libpython_info.dli_fname) { |
| 105 | + return decode_to_dict(dict, key, libpython_info.dli_fname); |
| 106 | + } |
| 107 | +-#endif |
| 108 | + #endif |
| 109 | + |
| 110 | + return PyDict_SetItemString(dict, key, Py_None) == 0; |
| 111 | +diff --git a/Modules/getpath.py b/Modules/getpath.py |
| 112 | +index b89d7427e3f..8c431e53be2 100644 |
| 113 | +--- a/Modules/getpath.py |
| 114 | ++++ b/Modules/getpath.py |
| 115 | +@@ -436,7 +436,7 @@ def search_up(prefix, *landmarks, test=isfile): |
| 116 | + |
| 117 | + if not executable_dir and os_name == 'darwin' and library: |
| 118 | + # QUIRK: macOS checks adjacent to its library early |
| 119 | +- library_dir = dirname(library) |
| 120 | ++ library_dir = dirname(dirname(library)) |
| 121 | + if any(isfile(joinpath(library_dir, p)) for p in STDLIB_LANDMARKS): |
| 122 | + # Exceptions here should abort the whole process (to match |
| 123 | + # previous behavior) |
| 124 | +@@ -570,7 +570,7 @@ def search_up(prefix, *landmarks, test=isfile): |
| 125 | + |
| 126 | + # First try to detect prefix by looking alongside our runtime library, if known |
| 127 | + if library and not prefix: |
| 128 | +- library_dir = dirname(library) |
| 129 | ++ library_dir = dirname(dirname(library)) |
| 130 | + if ZIP_LANDMARK: |
| 131 | + if os_name == 'nt': |
| 132 | + # QUIRK: Windows does not search up for ZIP file |
| 133 | +-- |
| 134 | +2.50.1 (Apple Git-155) |
| 135 | + |
0 commit comments