[Bug 49656] New: SAP GUI - Ws2_32.getaddrinfo crashes with non-standard service (/etc/services)

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=49656

Bug ID: 49656
           Summary: SAP GUI - Ws2_32.getaddrinfo crashes with non-standard
                    service (/etc/services)
           Product: Wine
           Version: 5.14
          Hardware: x86
                OS: Mac OS X
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: winsock
          Assignee: [e-mail]
          Reporter: [e-mail]

Created attachment 67890
  --> https://bugs.winehq.org/attachment.cgi?id=67890
Terminal output (both success and crash)

Calling getaddrinfo with pNodeName = NULL and a non-standard pServiceName that
is not present in /etc/services causes an Unhandled page fault.

I noticed this issue using SAP GUI, a Windows client for a corporate ERP system
that uses unusual TCP ports for some of its connections to servers.
These port numbers are registered in %WINDIR%\system32\drivers\etc\services
during installation, and after going through some traces I noticed the error
stopped after adding the same entries to the host's (macOS) /etc/services file.

Since SAP GUI is licensed software, I created a little C program that simulates
the same problem.
It simply calls getaddrinfo with a service name "I_dont_exist".
If this entry is not present in /etc/services, the program crashes.

Tested on Wine 5.14 upstream binaries downloaded from PlayOnLinux/Phoenicis
Wine building service.
I also tested in other Wine version I use, namely 5.14-staging, 5.0 (WineHQ
Homebrew), 5.7-staging (WineHQ Homebrew) and CrossOver 19.0.1. All had the same
result.

Sample C program with .exe and source code:
https://abap.ninja/bugreport/wine_getaddrinfo_crash.zip

[Bug 49656] SAP GUI - Ws2_32.getaddrinfo crashes with non-standard service (/etc/services)

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=49656

--- Comment #1 from Hans Leidekker  ---
Created attachment 67897
  --> https://bugs.winehq.org/attachment.cgi?id=67897
patch

Thanks for the bug report. Here's a patch that should fix the crash. It doesn't
address the issue of getaddrinfo ignoring entries added to
%WINDIR%\system32\drivers\etc\services.

[Bug 49656] SAP GUI - Ws2_32.getaddrinfo crashes with non-standard service (/etc/services)

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=49656

Hans Leidekker  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
      Fixed by SHA1|                            |a6d9f46b25a481648a2a06ed8d1
                   |                            |594fc59a4e313
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #2 from Hans Leidekker  ---
Fixed with a6d9f46b25a481648a2a06ed8d1594fc59a4e313.

[Bug 49656] SAP GUI - Ws2_32.getaddrinfo crashes with non-standard service (/etc/services)

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=49656

Alexandre Julliard  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #3 from Alexandre Julliard  ---
Closing bugs fixed in 5.15.

[Bug 49656] SAP GUI - Ws2_32.getaddrinfo crashes with non-standard service (/etc/services)

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=49656

Michael Stefaniuc  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |5.0.x

[Bug 49656] SAP GUI - Ws2_32.getaddrinfo crashes with non-standard service (/etc/services)

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=49656

Michael Stefaniuc  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|5.0.x                       |---

--- Comment #4 from Michael Stefaniuc  ---
Removing the 5.0.x milestone from bug fixes included in 5.0.3.