Persistent startup crash when NetDrive control program is active

[Lewis Rosenthal]

I finally chased this one down and figured out why it was coming and going.

Issue:

When ndctl.exe (the NetDrive daemon) is running and the following plugins are loaded:

ndpftp ndpsmb ndpntfs ndpdav ndpfat ndpiso

upon starting NewView, the following crash is reported and recorded:

App Version: V2.19.4
Components Version: v1.11.26
Library Version: V1.5.14

Running as process: C:\ECS\BIN\NEWVIEW.EXE
31 January 2017, 20:39:04
Exception type: EGPFault
Description: Access violation exception (EGPFault) occured
at CS:EIP  =$0000005B:$000E5B67
Location: $000E5B67
Callstack:
 $00010740
System version:20.45.0
  OS/2 4.5
RAM: 1982 MB
Boot drive: C
Video resolution: 1400x1050
Color depth: 32 bits
Video driver: GRADD: SDDGRADD,WOVERLAY
Loaded files (2):

Further, when selecting No (do not exit the program) following the startup crash, and subsequently searching all help files, attempting to open the directory selection window fails (this is what put me onto NetDrive in the first place).

I suspect it is the Samba plugin, as that is the plugin currently seeing most development activity, and this issue did not occur until I was in the middle of testing new Samba builds. I will confirm by disabling Samba and rebooting.

[Lewis Rosenthal]

I suspect this issue is system-specific and likely related to some ancillary library installed on the system (or the version of the Samba plugin for NetDrive). This is not happening on this system now, though it did as recently as last week.

Resolving as invalid because I have no evidence that this is really a NewView issue.

[Lewis Rosenthal]

Reopening, following discussion at the Arca Noae staff meeting today. Here's a current log entry:

App Version: V2.19.7
Components Version: v1.11.26
Library Version: V1.5.14

Running as process: C:\ECS\BIN\NEWVIEW.EXE
13 November 2021, 09:33:12
Exception type: EGPFault
Description: Access violation exception (EGPFault) occured
at CS:EIP  =$0000005B:$00045505
Location: $00045505
Callstack:
 $00045505
 $000E5785
 $000E5984
 $00038187
 $00038113
 $0003ABDD
 $0003AD6A
 $0003AE1E
 $00010339
 $1FFEC183
System version:20.45.0
  OS/2 4.5
RAM: 1982 MB
Boot drive: C
Video resolution: 1400x1050
Color depth: 32 bits
Video driver: GRADD: SDDGRADD
Loaded files (0):
Top-level open windows: 0

If I stop ndctl.exe, NewView opens normally (no trap). If I restart the control program, I get this same trap.

Interestingly, ndctl.log says:

[   local] internal plugin has been initialized.
[  ndpftp] (level 2) plugin has been loaded.
[  ndpsmb] (level 2) plugin has been loaded.
[ ndpntfs] (level 2) plugin has been loaded.
[  ndpdav] (level 2) plugin has been loaded.
[  ndpfat] (level 2) plugin has been loaded.
[  ndpiso] (level 2) plugin has been loaded.
Initialization completed. The control program is running.
NDF0045: Resource ISOFS (W) specified in file volumes.cfg, line 21 was not mounted
NDF0007: The system error is SYS0087

I'll once again try to narrow down the list of plugins to see which one may be causing a directory enumeration problem. NewView seems to be the only app to complain about this.

[Lewis Rosenthal]

Upon further testing, I can confirm the following:

1. The crash can occur when one or more plugins are loaded. The plugin mix is immaterial.
2. The crash can occur when more and one NetDrive volume has been attached and yet has nothing mounted under it, i.e.:

{0}[c:\] dir k:

 Volume in drive K is NETDRIVE
SYS0002: The system cannot find the file specified. "K:\*"
               0 bytes in 0 files and 0 dirs
      67,108,864 bytes (64MB) free

In this case, K: is a volume which has been restored in my volumes.cfg file but which generally is not mounted. Likewise, I: (used for mounting ISOs) is generally not mounted.

However, it is possible to have one volume attached with no mountpoints, and it is possible to have only one volume attached with active mountpoints and still get the crash. Consider X:, which uses the webdav plugin:

{0}[c:\] dir x:

 Volume in drive X is NETDRIVE
 Directory of  X:\*

11-14-21  23:47         <DIR>        0  gcc
11-14-21  23:47         <DIR>        0  kerberos
11-14-21  23:47         <DIR>        0  lucide
               0 bytes in 0 files and 3 dirs
   1,073,741,824 bytes (1,024MB) free

If I delete all other volumes from my configuration, leaving only X: and its three mountpoints below, NewView will crash at the same address.

My normal volume layout is:

G: (always mounted to Samba share)
H: (always mounted to FTP)
I: (usually unmounted)
K: (usually unmounted)
M: (usually unmounted)
X: (always 3 mountpoints to webdav locations)

I can remove all but X:, and get the crash (as stated above). I can remove K:, M:, X: and not get the crash.

[stevenhl]

Reporter resolved issue locally by ensuring that MUGLIB DLLs where a matched set.

The symptom of the mismatched DLLs was that the WebDAV NetDrive plugin reported that the NetewUseGetInfo API completed without error, but returned 0 data bytes. The 0 data byte return was not checked for by the NewView code and triggered an exception.