iHore/vLLM-4x-RTX-3090
8
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
main
vLLM-4x-RTX-3090/patches/0054-ple_offload-the-dying-path-traceback-runs-no-excepti.patch
T
MOVZX 777e910dd0 Initial commit 
Signed-off-by: MOVZX <movzx@yahoo.com>
2026-10-15 01:01:52 +07:00

287 lines
11 KiB
Diff
Raw Permalink Blame History

From 0ed590631391b088b17c35ffb5c54fca34dfc0b8 Mon Sep 17 00:00:00 2001
From: halt95 <halt95@users.noreply.github.com>
Date: Wed, 23 Sep 2026 12:30:20 -0400
Subject: [PATCH 054/103] ple_offload: the dying-path traceback runs no
exception code
ple_fail_closed formatted its traceback with traceback.format_exception, which
runs exception-defined __str__/__getattribute__; if one blocked, the rank never
reached os._exit(1) and kept its GPUs. A safe formatter reads frames and plain
string arguments only.
---
tests/v1/ple_offload/test_test_fault.py | 11 +-
.../test_transport_fatal_helper.py | 69 ++++++++++
vllm/v1/ple_offload/fatal.py | 125 +++++++++++++++++-
3 files changed, 196 insertions(+), 9 deletions(-)
diff --git a/tests/v1/ple_offload/test_test_fault.py b/tests/v1/ple_offload/test_test_fault.py
index a199207..ace9fab 100644
--- a/tests/v1/ple_offload/test_test_fault.py
+++ b/tests/v1/ple_offload/test_test_fault.py
@@ -270,11 +270,12 @@ def test_partial_ready_times_out_and_terminates_production_child(
f"cut={cut},ordinal=1,rank=any",
)
monkeypatch.setenv("VLLM_PLE_TEST_FAULT_LOG", str(tmp_path / "fault.log"))
- monkeypatch.setattr(
- worker_module.envs,
- "VLLM_PLE_OFFLOAD_READY_TIMEOUT",
- 1.0,
- )
+ # setenv, not setattr(envs, ...): envs serves this name from a module
+ # __getattr__, so monkeypatch's undo would setattr the pre-test default
+ # (600.0) back as a REAL module attribute that shadows __getattr__ for the
+ # rest of the session -- tests/test_envs.py then reads 600.0 whatever the
+ # environment says (a test-ordering failure seen on the reference host).
+ monkeypatch.setenv("VLLM_PLE_OFFLOAD_READY_TIMEOUT", "1")
handle = worker_module.PleOffloadWorker.make_process(
object(), # The bootstrap fault fires before config reconstruction.
diff --git a/tests/v1/ple_offload/test_transport_fatal_helper.py b/tests/v1/ple_offload/test_transport_fatal_helper.py
index f1abd27..16aaaf9 100644
--- a/tests/v1/ple_offload/test_transport_fatal_helper.py
+++ b/tests/v1/ple_offload/test_transport_fatal_helper.py
@@ -107,6 +107,18 @@ elif scenario == "str-subclass-argument":
threading.Event().wait()
error = RuntimeError(BlockingString("detail"))
+elif scenario == "blocking-str-cause":
+ class Cause(Exception):
+ def __str__(self):
+ threading.Event().wait()
+
+ try:
+ try:
+ raise Cause()
+ except Cause as cause:
+ raise RuntimeError("outer") from cause
+ except RuntimeError as outer:
+ error = outer
elif scenario == "huge-string":
error = RuntimeError("x" * (32 << 20))
else:
@@ -149,6 +161,7 @@ def test_fatal_helper_avoids_exception_controlled_descriptors(
"blocking-str-method",
"overridden-getattribute",
"str-subclass-argument",
+ "blocking-str-cause",
"huge-string",
],
)
@@ -164,6 +177,62 @@ def test_fatal_helper_hard_exit_controls(
)
+_FATAL_TRACEBACK_SOURCE = r"""
+from vllm.v1.ple_offload.fatal import ple_fail_closed
+
+
+class AcceleratorLikeError(RuntimeError):
+ pass
+
+
+def launch_kernel():
+ raise ValueError("root cause detail")
+
+
+def step():
+ try:
+ launch_kernel()
+ except ValueError as cause:
+ raise AcceleratorLikeError("CUDA error: an illegal memory access") from cause
+
+
+try:
+ step()
+except BaseException as error:
+ ple_fail_closed(
+ where="transport-traceback",
+ rank=0,
+ layer="ple",
+ sequence=7,
+ error=error,
+ )
+"""
+
+
+def test_fatal_helper_traceback_names_subclass_frames_and_cause() -> None:
+ """The dying-path traceback still carries what the early boot lacked: the
+ real (subclass) type, its message, the raising frames and the chain."""
+ completed = subprocess.run(
+ [sys.executable, "-c", _FATAL_TRACEBACK_SOURCE],
+ stdout=subprocess.DEVNULL,
+ stderr=subprocess.PIPE,
+ timeout=120,
+ )
+ stderr = completed.stderr.decode("utf-8", "replace")
+ assert completed.returncode == 1, stderr
+ assert "PLE fatal: where=transport-traceback" in stderr
+ assert "PLE fatal traceback follows:" in stderr
+ assert "AcceleratorLikeError: CUDA error: an illegal memory access" in stderr
+ assert "ValueError: root cause detail" in stderr
+ assert "direct cause of the following exception" in stderr
+ assert "in launch_kernel" in stderr
+ assert "in step" in stderr
+ # Interpreter order: the cause's block precedes the outer exception's.
+ assert stderr.index("ValueError: root cause detail") < stderr.index(
+ "AcceleratorLikeError: CUDA error"
+ )
+
+
@pytest.mark.skipif(sys.platform == "win32", reason="process groups are POSIX-only")
def test_wait_for_ready_rejects_non_string_status_without_comparison(
tmp_path: Path,
diff --git a/vllm/v1/ple_offload/fatal.py b/vllm/v1/ple_offload/fatal.py
index 90ab4a0..a96515f 100644
--- a/vllm/v1/ple_offload/fatal.py
+++ b/vllm/v1/ple_offload/fatal.py
@@ -5,8 +5,8 @@
import functools
import os
import time
-import traceback
from collections.abc import Callable
+from types import TracebackType
from typing import Any, NoReturn, TypeVar
_F = TypeVar("_F", bound=Callable[..., Any])
@@ -26,6 +26,122 @@ _TRUSTED_ERROR_LABELS = (
(KeyboardInterrupt, "KeyboardInterrupt"),
)
+# Slot readers for the dying path. Each is the C getter of the base type,
+# called directly, so no __getattribute__, property, metaclass attribute or
+# descriptor defined by the exception (or its class) can run. The traceback
+# module is NOT safe here: traceback.format_exception calls str(error) (and
+# str() on every chained exception) and reads error.__traceback__ through the
+# instance's own __getattribute__, so an exception whose __str__ or
+# __getattribute__ blocks kept the process alive forever -- the fail-closed
+# path hung instead of dying (found by the blocking-exception transport tests).
+_EXC_ARGS = BaseException.args.__get__
+_EXC_TRACEBACK = BaseException.__traceback__.__get__
+_EXC_CAUSE = BaseException.__cause__.__get__
+_EXC_CONTEXT = BaseException.__context__.__get__
+_EXC_SUPPRESS_CONTEXT = BaseException.__suppress_context__.__get__
+_TYPE_QUALNAME = type.__dict__["__qualname__"].__get__
+_TYPE_MODULE = type.__dict__["__module__"].__get__
+_MAX_CHAIN = 16
+_MAX_ARG_CHARS = 1024
+
+
+def _safe_type_label(cls: type) -> str:
+ """``module.qualname`` of ``cls`` without running class-defined code."""
+ try:
+ qualname = _TYPE_QUALNAME(cls)
+ module = _TYPE_MODULE(cls)
+ except BaseException:
+ return "<unknown type>"
+ if type(qualname) is not str:
+ qualname = "<unknown type>"
+ if type(module) is str and module != "builtins":
+ return f"{module}.{qualname}"
+ return qualname
+
+
+def _safe_exception_line(error: BaseException) -> str:
+ """Type label plus the exact-``str``/``int`` args; nothing else is shown."""
+ label = _safe_type_label(type(error))
+ args = _EXC_ARGS(error)
+ if type(args) is not tuple or len(args) == 0:
+ return label
+ parts = []
+ for arg in args[:4]:
+ if type(arg) is str:
+ parts.append(arg[:_MAX_ARG_CHARS])
+ elif type(arg) is int and -(10**18) < arg < 10**18:
+ parts.append(str(arg))
+ else:
+ parts.append(f"<{_safe_type_label(type(arg))}>")
+ if len(args) > 4:
+ parts.append("...")
+ if len(args) == 1:
+ return f"{label}: {parts[0]}"
+ return f"{label}: ({', '.join(parts)})"
+
+
+def _safe_frames(error: BaseException) -> list[str]:
+ lines = []
+ tb = _EXC_TRACEBACK(error)
+ while type(tb) is TracebackType:
+ code = tb.tb_frame.f_code
+ filename = code.co_filename
+ name = code.co_qualname
+ lineno = tb.tb_lineno
+ lines.append(
+ f' File "{filename if type(filename) is str else "?"}", '
+ f"line {lineno if type(lineno) is int else '?'}, "
+ f"in {name if type(name) is str else '?'}\n"
+ )
+ tb = tb.tb_next
+ return lines
+
+
+def _safe_format_exception(error: BaseException) -> str:
+ """A traceback.format_exception look-alike that runs no exception code.
+
+ Frames come from the traceback objects (file, line, function; no source
+ lines, which would mean file I/O on the dying path); the message is the
+ exception's exact-``str`` args, never ``str(error)``. Chains follow
+ ``__cause__`` / ``__context__`` like the interpreter, bounded and
+ cycle-safe.
+ """
+ chain: list[tuple[BaseException, str]] = []
+ seen: set[int] = set()
+ current: BaseException | None = error
+ link = ""
+ while current is not None and len(chain) < _MAX_CHAIN:
+ if id(current) in seen:
+ break
+ seen.add(id(current))
+ chain.append((current, link))
+ cause = _EXC_CAUSE(current)
+ if cause is not None:
+ current = cause
+ link = (
+ "\nThe above exception was the direct cause of the "
+ "following exception:\n\n"
+ )
+ elif _EXC_SUPPRESS_CONTEXT(current) is not True:
+ current = _EXC_CONTEXT(current)
+ link = (
+ "\nDuring handling of the above exception, another "
+ "exception occurred:\n\n"
+ )
+ else:
+ current = None
+ out: list[str] = []
+ # Innermost first, as the interpreter prints; each outer exception's link
+ # text sits between it and the exception it chains to.
+ for index in range(len(chain) - 1, -1, -1):
+ exc, _ = chain[index]
+ out.append("Traceback (most recent call last):\n")
+ out.extend(_safe_frames(exc))
+ out.append(_safe_exception_line(exc) + "\n")
+ if index > 0:
+ out.append(chain[index][1])
+ return "".join(out)
+
def ple_fail_closed(
*,
@@ -88,9 +204,10 @@ def ple_fail_closed(
# Three-arg form: it does not depend on being inside an active except
# block, and this function is called from one frame further out than
# the raise.
- tb = "".join(
- traceback.format_exception(type(error), error, error.__traceback__)
- )
+ # _safe_format_exception, not traceback.format_exception: see the note
+ # above _EXC_ARGS -- the stdlib formatter calls str(error), which is
+ # exception-defined code and can block the dying process forever.
+ tb = _safe_format_exception(error)
raw = tb.encode("utf-8", "backslashreplace")
# Truncate from the END, not the start. format_exception puts the
# frames first and the exception TYPE AND MESSAGE last, so a head-side
--
2.43.0
Reference in New Issue View Git Blame Copy Permalink