Trap when a process is closed

[Gregg Young]

This is caused by a message being sent to the button window and/or its menu after it has been destroyed.

[Gregg Young]

The first attached file is the (rare) widget trap. The code related to the trap is the same as 2.90. The second attachment is after the trap in the third attachment has been fixed. The trap in the third occurred after the destroy hwndItem was move after the destroy hwndSubMenu and QWL_USER was set to NULL for both window.

[Gregg Young]

This was reported by Dariusz. The fix for the above also fixed this so it appears it was the same issue.

I finally pegged this one down to a repeatable scenario.

Let me correct something first. Even though I had reported the last issue as PMMERGE, I am no longer seeing this, instead I always see SYS3175 in LSWITCH.EXE:

---

01-21-2019 15:51:53 SYS3175 PID 0039 TID 0001 Slot 0073
G:\UTIL\LSWITCHER\LSWITCH.EXE
c0000005
00032f18
P1=00000001 P2=00db0000 P3=XXXXXXXX P4=XXXXXXXXEAX=0005436c EBX=00000103 ECX=80000100 EDX=00db0000
ESI=0005436c EDI=00054470DS=0053 DSACC=f0f3 DSLIM=ffffffffES=0053 ESACC=f0f3 ESLIM=ffffffffFS=150b FSACC=00f3 FSLIM=00000030
GS=0000 GSACC= GSLIM=
CS:EIP=005b:00032f18 CSACC=f0df CSLIM=ffffffff
SS:ESP=0053:00054360 SSACC=f0f3 SSLIM=ffffffff
EBP=00054806 FLG=00010206

LSWITCH.EXE 0002:00012f18

---

Basically, it happens that if I attempt to use the TaskBar=>Popup Window=>RMB (click to attempt to issue a CLOSE to a window) I'm immediately seeing this error.

Now here is the strange part, this does not happen against all windows. Specifically, so far I've only been able to replicate this against the RDP client (RDESKTOP.EXE).

So I've attached two files, a screenshot of what the Taskbar looks like when I'm seeing this, as well as the TRP file itself. This time I've narrowed this down to only the matching popuplog.os2 entry above.

Not sure if this syncs-up with the email discussion you Gregg and Steven were having, some of the wording there, especially about the window close seemed to fit, but in my case I am not quite at that close point.

[Gregg Young]

The 4th attached file contains 2 versions of this trap from 2.91 RC 3