python / python/cpython

posix_spawn crashes when python is run with gprofng

Open
#149,509 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

extension-modules type-bug
Dominant language
Python
Stars
77.2k
Forks
35.9k
PR merge metrics
PR metrics pending

Description

Bug report

Bug description:

I've been observing crashes when python code calls posix_spawn when being
run under gprofng.

As an example, I tested on a Fedora 42 x86_64 system with python3-3.13.13-1.fc42.x86_64
[Python 3.13.13 (main, Apr 8 2026, 00:00:00) [GCC 15.2.1 20260123 (Red Hat 15.2.1-7)] on linux]
and binutils-2.44-12.fc42.x86_64
[GNU gprofng binutils version 2.44]

Consider this script:

import os
os.posix_spawn("/usr/bin/echo", ["/usr/bin/echo", "world"], None)
print ("hello")

If I run this on its own, it works as expected:

$ python spawn.py
hello
world

But if I run it with gprofng, it crashes:

$ gprofng collect app python spawn.py
Creating experiment directory test.1.er (Process ID: 591620) ...
free(): invalid pointer
Aborted (core dumped)
$ world

The problem goes away if I change the third argument of the posix_spawn
command from None to os.environ. I also reproduce the crash with
cpython git main as of May 6 (65ed109b5de7bab28f1051336f0ae312205c4233).

The issue is that py_posix_spawn assumes that environ does not change
over the call to posix_spawn. However, when gprofng is used, it
interposes code wrapping the posix_spawn call that can modify the
environment (apparently to sanitize the environment from things like
LD_PRELOAD entries used by gprofng) and thus change environ.

In more detail, in py_posix_spawn, we have this before the spawn call:

    EXECV_CHAR **envlist = NULL;
    ...
    Py_ssize_t argc, envc;
    ...
    if (env == Py_None) {
#ifdef USE_DARWIN_NS_GET_ENVIRON
        environ = *_NSGetEnviron();
#endif
        envlist = environ;
    } else {
        envlist = parse_envlist(env, &envc);
        if (envlist == NULL) {
            goto exit;
        }
    }

and after the call:

    if (envlist && envlist != environ) {
        free_string_array(envlist, envc);
    }

So if the env argument is null and environ is changed by the call,
then we end up trying to free environ,
using a length taken from an uninitialized variable.

Here's one possible fix:

diff --git a/Modules/posixmodule.c b/Modules/posixmodule.c
index 5bd53c2146a..4fef8a60647 100644
--- a/Modules/posixmodule.c
+++ b/Modules/posixmodule.c
@@ -7961,6 +7961,7 @@ py_posix_spawn(int use_posix_spawnp, PyObject *module, path_t *path, PyObject *a
         environ = *_NSGetEnviron();
 #endif
         envlist = environ;
+        envc = (Py_ssize_t)-1;
     } else {
         envlist = parse_envlist(env, &envc);
         if (envlist == NULL) {
@@ -8028,7 +8029,10 @@ py_posix_spawn(int use_posix_spawnp, PyObject *module, path_t *path, PyObject *a
     if (attrp) {
         (void)posix_spawnattr_destroy(attrp);
     }
-    if (envlist && envlist != environ) {
+    /* Can't just test for envlist != environ because some tools, such as
+       gprofng, interpose code around the posix_spawn call that can change
+       environ. */
+    if (envlist && envc != (Py_ssize_t)-1) {
         free_string_array(envlist, envc);
     }
     if (argvlist) {
CPython versions tested on:

3.13, CPython main branch

Operating systems tested on:

Linux

Linked PRs
  • gh-149724

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start in Modules/posixmodule.c at py_posix_spawn and compare the environ handling before and after the posix_spawn call with the provided reproducer under gprofng. Done means the reproducer no longer crashes when the environment changes during the call, while ordinary posix_spawn behavior remains intact.

Written by the indexing model from the issue text.

Assessment

Tech stack
c, python
Domain
operating-systems
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.