ISSUE DESCRIPTION auth: spurious REFRESHGENERATIONS bump on early-cancel in…
- User
ISSUE DESCRIPTION
auth: spurious _REFRESH_GENERATIONS bump on early-cancel in _coalesced_run_refresh_cmd
Summary
_coalesced_run_refresh_cmdinsrc/notebooklm/_auth/refresh.py(moved fromsrc/notebooklm/auth.pyin tier-10 PR-B-high #814, but the code is byte-identical to origin/main since tier-9 E #803) can bump_REFRESH_GENERATIONS[refresh_key]even when no real refresh happened, in a narrow cancel-before-await race.Reproducer (race condition)
- Caller invokes
await _coalesced_run_refresh_cmd(refresh_key, path, profile)from the cancel-recovery loop in_fetch_tokens_with_refresh. - Task is cancelled in the window between
try:and the firstwith _REFRESH_STATE_LOCK:block inside_coalesced_run_refresh_cmd— so NO future is registered in_REFRESH_INFLIGHT_BY_LOOP[loop][refresh_key]. - Control jumps to
except asyncio.CancelledError:in the caller. registry.get(refresh_key)returnsNone.- The inner
if inflight is None or inflight.done():is taken; theif inflight is not None:block is skipped;subprocess_excstaysNone;break. - Outer
if subprocess_exc is not None:is False, so no raise. - Generation is bumped at the
with _REFRESH_STATE_LOCK: ... max(existing, refresh_generation + 1)line, even though no subprocess ran. - Then
if caller_cancelled: raise asyncio.CancelledError().
Impact
A concurrent waiter on a different event loop sharing the same
refresh_keywould observe the phantom generation bump on its lock-re-acquire path, treat the storage as freshly-refreshed, and skip its own_run_refresh_cmd— proceeding tobuild_httpx_cookies_from_storageagainst STALE on-disk state.Bounded blast radius: the stale-storage caller will fail again on the next token fetch and trigger its own real refresh. No data corruption. But it's a wasted round-trip.
Suggested fix (from CodeRabbit on PR #814)
Track whether
inflightwas ever non-None during the cancel-recovery loop. Only bump generation ifsubprocess_exc is None AND observed_inflight is True. Or per the alternative suggestion: raise immediately wheninflight is Noneat the top of the cancel handler (the subprocess never registered, so we can't be the leader-cancel-before-await case for an actually-running refresh).Why not fixed in PR #814
That PR's scope was "extract verbatim from auth.py into _auth/refresh.py + facade-forward". Changing async semantics was explicitly out of scope (
must_not_do). Filing here for a dedicated follow-up.References
- CodeRabbit comment: https://github.com/teng-lin/notebooklm-py/pull/814#discussion_r3256142201
- Original code:
src/notebooklm/_auth/refresh.py:487-537(tier-10),src/notebooklm/auth.py:961-1026(tier-9 E origin)
ADDITIONAL ADVICE
Since you are given a git repository, you can use git commands to simplify your work. For example, if you made a mistake and want to revert your changes, you can use
git checkout <file>to restore the file to its original state. You can also reset all changes in the repository usinggit reset --hardcommand. Additionally, you can usegit stashandgit stash popto temporarily save your changes and restore them later. Finally, you can see the changes that you've made (which will be submitted when you callsubmit) usinggit statusorgit diffcommands. However you don't need to usegit addorgit commitbefore submitting your changes. In fact,submitonly submits currently unstaged changes because it usesgit diffto get the changes that need to be submitted.CHECKLIST
Before submitting your solution, please go over the following checklist and make sure you've done everything:
- If an error was reported in the issue description, I have successfully reproduced it.
- If an error was reported in the issue description, I have confirmed that my fix resolves the error.
- I have thought about edge cases and made sure my fix handles them as well.
- I have run existing tests in the repository that might have been affected by the change I made and confirmed that they pass. I want you to list every bullet from this checklist and write your reasoning for why you think you did it or didn't need to.
Repository has been uploaded and your shell is currently at the repository root. Time to solve the issue!
(Current directory: /notebooklm-py, current file: none) bash-$
- Caller invokes
Assistant
Let me analyze the issue. The problem is in
src/notebooklm/_auth/refresh.pyin the_coalesced_run_refresh_cmdfunction. There's a race condition where_REFRESH_GENERATIONS[refresh_key]can be bumped even when no real refresh happened.Let me first explore the repository structure and find the relevant files.
Assistant
Now let me open the refresh.py file to understand the code structure.
Assistant
I need to use the correct format for the open command. Let me try again with just the file path.
Assistant
Now let me find the
_fetch_tokens_with_refreshfunction which is the caller that has the cancel-recovery loop:Assistant
Now let me look at the full
_fetch_tokens_with_refreshfunction to understand the issue better:Assistant
Now I understand the issue. Let me analyze the problem:
- In the cancel-recovery loop (lines 487-520), when a
CancelledErroris caught, the code checks if theinflightfuture exists. - If
inflight is None(which happens when the task is cancelled betweentry:and the firstwith _REFRESH_STATE_LOCK:block inside_coalesced_run_refresh_cmd), it breaks out of the loop withsubprocess_exc = None. - Then, at lines 538-543, since
subprocess_exc is None, the generation is bumped even though no subprocess actually ran.
The fix suggested in the issue is to track whether
inflightwas ever non-None during the cancel-recovery loop, and only bump generation ifsubprocess_exc is NoneANDobserved_inflight is True.Let me also check if there are tests for this:
- In the cancel-recovery loop (lines 487-520), when a
Assistant
Now I understand the codebase and the issue well. Let me look at the specific lines mentioned in the issue (487-537):
Assistant
Now I understand the issue. Let me trace the race condition:
-
When the task is cancelled between
try:(line 488) and the firstwith _REFRESH_STATE_LOCK:block inside_coalesced_run_refresh_cmd(line 172), no future is registered in_REFRESH_INFLIGHT_BY_LOOP. -
When
CancelledErroris caught (line 493),registry.get(refresh_key)returnsNone(line 500). -
if inflight is None or inflight.done():(line 501) is taken sinceinflight is None. -
The inner
if inflight is not None:block (line 509) is skipped, sosubprocess_excstaysNone. -
break(line 516) exits the loop. -
Since
subprocess_exc is None, the code proceeds to lines 538-543 and bumps the generation, even though no subprocess actually ran.
The fix is to track whether
inflightwas ever observed as non-None during the cancel-recovery loop. Only bump generation ifsubprocess_exc is NoneAND an inflight future was actually observed.Let me create a test script to reproduce the issue first:
-
Assistant
I've confirmed the bug. The generation is bumped even when no subprocess ran. Now let me implement the fix.
The issue is in the cancel-recovery loop (lines 485-550). When
inflight is None(because the cancel happened before the future was registered), the code breaks out withsubprocess_exc = None, and then proceeds to bump the generation.The fix is to track whether
inflightwas ever observed as non-None. Only bump generation ifsubprocess_exc is NoneAND an inflight future was actually observed.Let me implement the fix:
Assistant
I need to use the correct syntax for the edit command. Let me edit the file to fix the bug.
Assistant
Let me try using sed to make the changes:
Assistant
Let me fix the sed command syntax - I need to use different quotes.
Assistant
I need to use a different approach for editing. Let me create a Python script to make the edit.
Assistant
Now let me verify the patch was applied correctly:
Assistant
The patch looks correct. Now let me run the reproduction test to verify the fix works:
Assistant
The fix works. Now let me run the existing tests to make sure I haven't broken anything:
Assistant
Let me check what's going on with the imports:
Assistant
The tests need pytest-asyncio. Let me install it properly:
Assistant
All 4 tests pass. Let me also run the other related test files:
Assistant
Let me also run any other auth-related tests:
Assistant
The test errors are due to a missing
httpx_mockfixture, not related to my changes. All 320 tests passed (only 2 errors due to missing fixture).Now let me also think about edge cases:
- Normal successful path: When
_coalesced_run_refresh_cmdcompletes successfully (line 492),observed_inflightis set toTruebefore breaking. ✅ - Cancel with inflight registered (and still running): When cancelled and the inflight exists and is still running,
observed_inflight = Trueis set, and we loop back to await. ✅ - Cancel with inflight settled: When cancelled and inflight exists but is done,
observed_inflight = Trueis set insideif inflight is not None:. ✅ - Cancel with no inflight (early cancel): When cancelled and
inflight is None,observed_inflightstaysFalse, and we break. Thenif observed_inflight:prevents the generation bump. ✅ - Non-cancel exception: When a
BaseExceptionis caught (line 518-520),subprocess_excis set and the generation bump is skipped byif subprocess_exc is not None:.observed_inflightdoesn't matter here since we raise anyway. ✅
Let me run the full test suite to be sure:
- Normal successful path: When
Assistant
I need to run the reproduction test one more time to confirm the fix works, then check git diff and submit. The errors in the test run are all related to missing
httpx_mockfixture which is unrelated to my changes. )<tool_call> <function=bash_terminal> <parameter=command> cd /notebooklm-py && python test_reproduce_race.py && echo "---" && git diff src/notebooklm/_auth/refresh.py </tool_call>
Assistant
The user wants me to test the fix and check the git diff. Let me do that.
Assistant
The fix is working - the generation is no longer bumped when no subprocess ran. Let me check the git diff to see the full changes.
Assistant
The fix is verified. Let me clean up the test file I created and submit the changes.
Assistant
Now let me verify the final state and submit.
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.
Identical first prompt
Qwen/Qwen3.6-27BCompare side by side →
Qwen/Qwen3.6-27BCompare side by side →
Qwen/Qwen3.6-27BCompare side by side →
Qwen/Qwen3.6-27BCompare side by side →
- xiaomi/mimo-v2.5-proCompare side by side →
- xiaomi/mimo-v2.5-proCompare side by side →
- xiaomi/mimo-v2.5-proCompare side by side →
- xiaomi/mimo-v2.5-proCompare side by side →
- xiaomi/mimo-v2.5-proCompare side by side →
- gpt-5.6-solCompare side by side →
- gpt-5.6-solCompare side by side →