Q59895: Why Can’t a Stand-Alone DLL Use the Emulator Math Library?

Article: Q59895
Product(s): See article
Version(s): 5.10 6.00
Operating System(s): OS/2
Keyword(s): ENDUSER | | mspl13_c
Last Modified: 18-APR-1990

Question:

In MTDYNA.DOC, it states that a stand-alone DLL should use LLIBCDLL,
and therefore, the alternate math library. In the C Version 6.00
"Advanced Programming Techniques" manual, it states that the alternate
math library must be used because of the following:

   For a DLL to use the 80x87 coprocessor of the emulation point
   library, the DLL and all of its client programs must agree on which
   process is going to handle floating-point exceptions and on which
   process is going to handle emulation if the machine does not have a
   coprocessor.

I don't understand why exception handling and/or task switching causes
more problems for using an 80x87 in a DLL context than in normal
applications. The DLL can always find the PID of the client and could
signal it, register a DosExitList function, or do several other things
in the case of an exception. Given that the DLL C run-time databases
could be per-instance data segments, why is there a problem?

Response:

The problem with handling 80x87 floating-point exceptions and using
DLLs is that when we call DosSetVec() to set a handler for 80x87
exceptions, the handler is set for the entire process. Whoever calls
DosSetVec() last will get control of all 80x87 exceptions.

Suppose we are lucky and the DLL calls DosSetVec() last and an
exception occurs. The DLL has no way of knowing whether the exception
occurred in the DLL's code or the EXE's code (or perhaps some other
DLL). Thus, the DLL has no way of knowing what to do. Perhaps the
exception should be ignored or maybe the process should be terminated.

Calling DosExitList() is not a solution because the DLL has to decide
what to do right now. Also, finding out the PID and then signaling the
process is not very useful because this means that the EXE and all of
the DLLs in the process must set up special signal handlers to handle
exceptions. In other words, the EXE and the DLL must all use a special
library to cooperate when a floating point exception occurs, and that
is what CDLLOBJS.LIB is for.

LLIBCDLL.LIB is for writing stand-alone DLLs, that is, DLLs that can
be called from any EXE. Because LLIBCDLL.LIB uses alternate math, it
is completely self-contained and doesn't need any special code in the
EXE.

THE INFORMATION PROVIDED IN THE MICROSOFT KNOWLEDGE BASE IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. MICROSOFT DISCLAIMS ALL WARRANTIES, EITHER EXPRESS OR IMPLIED, INCLUDING THE WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT SHALL MICROSOFT CORPORATION OR ITS SUPPLIERS BE LIABLE FOR ANY DAMAGES WHATSOEVER INCLUDING DIRECT, INDIRECT, INCIDENTAL, CONSEQUENTIAL, LOSS OF BUSINESS PROFITS OR SPECIAL DAMAGES, EVEN IF MICROSOFT CORPORATION OR ITS SUPPLIERS HAVE BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. SOME STATES DO NOT ALLOW THE EXCLUSION OR LIMITATION OF LIABILITY FOR CONSEQUENTIAL OR INCIDENTAL DAMAGES SO THE FOREGOING LIMITATION MAY NOT APPLY.

Copyright Microsoft Corporation 1986-2002.