From 5b412e1f7bdb3e0667b2bc8b216ad216d59d8373 Mon Sep 17 00:00:00 2001
From: Stan Ulbrych <stan@python.org>
Date: Mon, 8 Jun 2026 11:55:32 +0200
Subject: [PATCH] gh-150599: Prevent bz2 decompressor reuse after errors
 (GH-150600)

CVE: CVE-2026-9669
Upstream-Status: Backport [https://github.com/python/cpython/commit/5755d0f083949ff3c5bf3a37e673e24e306b036e]

Signed-off-by: Benjamin Robin <benjamin.robin@bootlin.com>
---
 Lib/test/test_bz2.py | 15 +++++++++++++++
 Modules/_bz2module.c | 18 +++++++++++++++---
 2 files changed, 30 insertions(+), 3 deletions(-)

diff --git a/Lib/test/test_bz2.py b/Lib/test/test_bz2.py
index cb730a1a46e2..dcbf6a298264 100644
--- a/Lib/test/test_bz2.py
+++ b/Lib/test/test_bz2.py
@@ -958,6 +958,21 @@ def test_failure(self):
         # Previously, a second call could crash due to internal inconsistency
         self.assertRaises(Exception, bzd.decompress, self.BAD_DATA * 30)

+    def test_decompress_after_data_error(self):
+        data = bytes.fromhex(
+            "425a6839314159265359000000000000007fffff000000000000000000000000"
+            "00000000000000000000000000000000000000e0370000000000000000000000"
+            "000000000000000000000000000000000000000000000000000083f3"
+        )
+        bzd = BZ2Decompressor()
+        with self.assertRaisesRegex(OSError, "Invalid data stream"):
+            bzd.decompress(data)
+        # Previously, a second call could crash due to internal inconsistency
+        self.assertFalse(bzd.needs_input)
+        self.assertFalse(bzd.eof)
+        with self.assertRaisesRegex(ValueError, "previous error"):
+            bzd.decompress(b'\x00' * 18)
+
     @support.refcount_test
     def test_refleaks_in___init__(self):
         gettotalrefcount = support.get_attribute(sys, 'gettotalrefcount')
diff --git a/Modules/_bz2module.c b/Modules/_bz2module.c
index 97bd44b4ac96..0b0916142f57 100644
--- a/Modules/_bz2module.c
+++ b/Modules/_bz2module.c
@@ -114,6 +114,7 @@ typedef struct {
 typedef struct {
     PyObject_HEAD
     bz_stream bzs;
+    int bzerror;
     char eof;           /* T_BOOL expects a char */
     PyObject *unused_data;
     char needs_input;
@@ -453,8 +454,11 @@ decompress_buf(BZ2Decompressor *d, Py_ssize_t max_length)

         d->bzs_avail_in_real += bzs->avail_in;

-        if (catch_bz2_error(bzret))
+        if (catch_bz2_error(bzret)) {
+            d->bzerror = bzret;
+            d->needs_input = 0;
             goto error;
+        }
         if (bzret == BZ_STREAM_END) {
             d->eof = 1;
             break;
@@ -621,10 +625,17 @@ _bz2_BZ2Decompressor_decompress_impl(BZ2Decompressor *self, Py_buffer *data,
     PyObject *result = NULL;

     ACQUIRE_LOCK(self);
-    if (self->eof)
+    if (self->eof) {
         PyErr_SetString(PyExc_EOFError, "End of stream already reached");
-    else
+    }
+    else if (self->bzerror) {
+        // Re-entering BZ2_bzDecompress() after an error can write out of bounds.
+        PyErr_SetString(PyExc_ValueError,
+                        "Decompressor is unusable after a previous error");
+    }
+    else {
         result = decompress(self, data->buf, data->len, max_length);
+    }
     RELEASE_LOCK(self);
     return result;
 }
@@ -658,6 +669,7 @@ _bz2_BZ2Decompressor_impl(PyTypeObject *type)
         return NULL;
     }

+    self->bzerror = 0;
     self->needs_input = 1;
     self->bzs_avail_in_real = 0;
     self->input_buffer = NULL;
--
2.54.0