From 74ae01afb9be8f1a393fd4b468966310cd55d764 Mon Sep 17 00:00:00 2001
From: George Burgess IV <gbiv@google.com>
Date: Mon, 4 Aug 2025 12:39:50 -0600
Subject: [PATCH] [utils] add `stop_at_sha` to revert_checker's API

This is useful for downstream consumers of this as a module. It's
unclear if interactive use wants this lever, but support can easily be
added if so.
---
 llvm/utils/revert_checker.py      | 18 ++++++++++++++++--
 llvm/utils/revert_checker_test.py | 14 ++++++++++++++
 2 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/llvm/utils/revert_checker.py b/llvm/utils/revert_checker.py
index 3c3ff3e6b846f..ced68bc8475d5 100755
--- a/llvm/utils/revert_checker.py
+++ b/llvm/utils/revert_checker.py
@@ -41,6 +41,7 @@
 
 import argparse
 import collections
+import itertools
 import logging
 import re
 import subprocess
@@ -246,7 +247,11 @@ def _load_pr_commit_mappings(
 # enough for the 99% case of reverts: rarely should someone land a cleanish
 # revert of a >6 month old change...
 def find_reverts(
-    git_dir: str, across_ref: str, root: str, max_pr_lookback: int = 20000
+    git_dir: str,
+    across_ref: str,
+    root: str,
+    max_pr_lookback: int = 20000,
+    stop_at_sha: Optional[str] = None,
 ) -> List[Revert]:
     """Finds reverts across `across_ref` in `git_dir`, starting from `root`.
 
@@ -260,6 +265,9 @@ def find_reverts(
             SHAs. These heuristics require that commit history from `root` to
             `some_parent_of_root` is loaded in memory. `max_pr_lookback` is how
             many commits behind `across_ref` should be loaded in memory.
+        stop_at_sha: If non-None and `stop_at_sha` is encountered while walking
+            to `across_ref` from `root`, stop checking for reverts. This allows for
+            faster incremental checking between `find_reverts` calls.
     """
     across_sha = _rev_parse(git_dir, across_ref)
     root_sha = _rev_parse(git_dir, root)
@@ -281,10 +289,16 @@ def find_reverts(
         root_sha,
     )
 
+    commit_log_stream: Iterable[_LogEntry] = _log_stream(git_dir, root_sha, across_sha)
+    if stop_at_sha:
+        commit_log_stream = itertools.takewhile(
+            lambda x: x.sha != stop_at_sha, commit_log_stream
+        )
+
     all_reverts = []
     # Lazily load PR <-> commit mappings, since it can be expensive.
     pr_commit_mappings = None
-    for sha, commit_message in _log_stream(git_dir, root_sha, across_sha):
+    for sha, commit_message in commit_log_stream:
         reverts, pr_reverts = _try_parse_reverts_from_commit_message(
             commit_message,
         )
diff --git a/llvm/utils/revert_checker_test.py b/llvm/utils/revert_checker_test.py
index c149be8dc0dd1..987bf6a97e2c9 100755
--- a/llvm/utils/revert_checker_test.py
+++ b/llvm/utils/revert_checker_test.py
@@ -130,6 +130,20 @@ def test_known_reverts_across_arbitrary_llvm_rev(self) -> None:
             ],
         )
 
+    def test_stop_at_sha_works(self) -> None:
+        reverts = revert_checker.find_reverts(
+            git_dir=get_llvm_project_path(),
+            # This SHA is a direct child of the reverted SHA expected below.
+            across_ref="2d5f3b0a61fb171617012a2c3ba05fd31fb3bb1d",
+            # This SHA is a direct child of the revert SHA listed below.
+            root="2c01b278580212914ec037bb5dd9b73702dfe7f1",
+            max_pr_lookback=50,
+            # This SHA is the first revert that would be returned, if not for
+            # `stop_at_sha`.
+            stop_at_sha="50866e84d1da8462aeb96607bf6d9e5bbd5869c5",
+        )
+        self.assertEqual(reverts, [])
+
     def test_pr_based_revert_works(self) -> None:
         reverts = revert_checker.find_reverts(
             git_dir=get_llvm_project_path(),