gh-104584: Fix error handling from backedge optimization (#106484)

When `_PyOptimizer_BackEdge` returns `NULL`, we should restore `next_instr` (and `stack_pointer`). To accomplish this we should jump to `resume_with_error` instead of just `error`.

The problem this causes is subtle -- the only repro I have is in PR gh-106393, at commit d7df54b139bcc47f5ea094bfaa9824f79bc45adc. But the fix is real (as shown later in that PR).

While we're at it, also improve the debug output: the offsets at which traces are identified are now measured in bytes, and always show the start offset. This makes it easier to correlate executor calls with optimizer calls, and either with `dis` output.

<!-- gh-issue-number: gh-104584 -->
* Issue: gh-104584
<!-- /gh-issue-number -->
This commit is contained in:
Guido van Rossum 2023-07-06 11:39:53 -07:00 committed by GitHub
parent 56353b1002
commit 003ba71dcb
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 15 additions and 11 deletions

View file

@ -2737,11 +2737,11 @@ _PyUopExecute(_PyExecutorObject *executor, _PyInterpreterFrame *frame, PyObject
#endif
DPRINTF(3,
"Entering _PyUopExecute for %s (%s:%d) at offset %ld\n",
"Entering _PyUopExecute for %s (%s:%d) at byte offset %ld\n",
PyUnicode_AsUTF8(_PyFrame_GetCode(frame)->co_qualname),
PyUnicode_AsUTF8(_PyFrame_GetCode(frame)->co_filename),
_PyFrame_GetCode(frame)->co_firstlineno,
(long)(frame->prev_instr + 1 -
2 * (long)(frame->prev_instr + 1 -
(_Py_CODEUNIT *)_PyFrame_GetCode(frame)->co_code_adaptive));
PyThreadState *tstate = _PyThreadState_GET();