https://bugs.winehq.org/show_bug.cgi?id=54866
Bug ID: 54866 Summary: ieframe:webbrowser - test_SetQueryNetSessionCount() sometimes gets an unexpected session count on Windows Product: Wine Version: unspecified Hardware: x86-64 OS: Windows Status: NEW Severity: normal Priority: P2 Component: ieframe Assignee: wine-bugs@winehq.org Reporter: fgouget@codeweavers.com
ieframe:webbrowser - test_SetQueryNetSessionCount() sometimes gets an unexpected session count on Windows:
webbrowser.c:4440: init_count 1 webbrowser.c:4443: Test failed: count = 3 webbrowser.c:4446: Test failed: count = 3 webbrowser.c:4449: Test failed: count = 2 webbrowser.c:4452: Test failed: count = 2
See https://test.winehq.org/data/patterns.html#ieframe:webbrowser
These failures only happen on Windows 10 21H1 to 22H2.
On Windows 10 init_count is either 1 or 2 (on w8adm is can be 0 too), but all the failures so far happened with an init_count of 1. That said there are also a lot of successful runs with init_count == 1.
This looks a lot like the browser creates a net session for its own purposes:
* Either init_count is 1 and the extra net session is never created or gets created after test_SetQueryNetSessionCount() is done so all is fine.
* Or the extra net session is created it before test_SetQueryNetSessionCount() starts and outlives it so our tests pass.
* Or the extra net session gets created after we get init_count but before the next test so that all our tests are off by one.
* But it seems the extra net session never gets created after the first SESSION_INCREMENT test.
Note that on my Windows 10 VM when I start iexplorer.exe by hand and then run the test I always get init_count == 2. But if IE is not already running I get init_count == 0.
In any case if it's a race condition I could send a patch to retry a couple of times.