commit e70c96c3be3a87fd64c825f248f8a460155c59d4
author Mark Mentovai <mark@chromium.org> Wed Apr 15 12:37:32 2020 -0400
committer Copybara-Service <copybara-worker@google.com> Tue Jan 19 06:50:57 2021 -0800
tree 72b5f016a1fe9a16b3d68134480bfb0a051fb249
parent 68b526829a138b59a6c5f0f55b614091a4cf7f10

win: Remove disabled and flaky end-to-end tests

The “and not memory after range” follow-up to the “extra memory range
captured” test is very flaky in debug builds
(https://ci.chromium.org/p/crashpad/builders/ci/crashpad_win_dbg/388),
and occasionally flaky in release builds too
(https://ci.chromium.org/p/crashpad/builders/ci/crashpad_win_rel/363).
As with the already-disabled “extra memory removal” test, it’s likely
that Crashpad is discovering a pointer that references the memory
immediately beyond the intended range, and capturing that too. Unless
the environment is very tightly controlled, it’s difficult to prevent
valid memory from being pointed to, and thus to guarantee that Crashpad
won’t capture a particular block of memory.

These are probably never going to be fixed. The flaky tests are removed.

Bug: crashpad:101
Change-Id: Iab8b021f97165ff7ff710bd7cdfd8495b67a87e3
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2150055
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
GitOrigin-RevId: 7fa69e7e4383a265d8b46da1001783790197de8a

snapshot/win/end_to_end_test.py

diff --git a/snapshot/win/end_to_end_test.py b/snapshot/win/end_to_end_test.py
index fd602fb..8823785 100755
--- a/snapshot/win/end_to_end_test.py
+++ b/snapshot/win/end_to_end_test.py
@@ -363,23 +363,6 @@
                'L8')
   out.Check(r'0000655e 0000656b 00006578 00006585',
             'extra memory range captured')
-  out.Check(r'\?\?\?\?\?\?\?\? \?\?\?\?\?\?\?\? '
-            r'\?\?\?\?\?\?\?\? \?\?\?\?\?\?\?\?',
-            '  and not memory after range')
-
-  if False:
-    # TODO(scottmg): This is flakily capturing too much memory in Debug builds,
-    # possibly because a stale pointer is being captured via the stack.
-    # See: https://bugs.chromium.org/p/crashpad/issues/detail?id=101.
-    out = CdbRun(cdb_path, dump_path,
-                'dd poi(crashy_program!crashpad::g_extra_memory_not_saved)'
-                '+0x1f30 L4')
-    # We save only the pointer, not the pointed-to data. If the pointer itself
-    # wasn't saved, then we won't get any memory printed, so here we're
-    # confirming the pointer was saved but the memory wasn't.
-    out.Check(r'\?\?\?\?\?\?\?\? \?\?\?\?\?\?\?\? '
-              r'\?\?\?\?\?\?\?\? \?\?\?\?\?\?\?\?',
-              'extra memory removal')
 
   out = CdbRun(cdb_path, dump_path, '.dumpdebug')
   out.Check(r'type \?\?\? \(333333\), size 00001000',