Details
-
Bug
-
Resolution: Unresolved
-
Minor
-
None
-
None
Description
I tested running our taskmgr for a couple of months on WinXPSP3 and recently noticed a bug, which looks like a race-condition to me.
I tested the following 2 taskmgr builds and both are affected the very same way:
0.4.15-dev-6640-g02883d1 GCC8.4.0 RosBEWin2.2.2 built in dbg-config
0.4.14-release-71-g55a9807 MSVC2010SP1 RosBEWin2.1.6 built in rls-config
Reproduction steps
- Build ReactOS on your host system (with RosBE2.2.2) and while doing that leave the taskmgr running in the processes-tab
- During the build several processes are created and ended very frequently. That is normal.
Observed result
the ros taskmgr shows for a second a zombie process like the highlighted line in the picture, then the zombie is gone for a second, then the next zombie shows up, then it vanishes, and so on
taskmgr_misbehaves_on_XPSP3.PNG
Expected result
the taskmgr of MS XPSP3 does not show those zombie processes ever
I do *assume* that those zombies are processes that just ended successfully some milliseconds ago, when the taskmgr already queried the running processes, but did not query all the data yet, that would belong to those processes.
In my tests I never saw more than one of such zombies at the same moment.