Just a sample of the Echomail archive
Cooperative anarchy at its finest, still active today. Darkrealms is the Zone 1 Hub.
|    MAKENL_NG    |    MakeNL Next Generation.    |    1,725 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 685 of 1,725    |
|    Ulrich Schroeter to Andrew Leary    |
|    Debugged OS2-32bit bug (Release of v3.3.    |
|    13 Jul 13 23:59:18    |
      Hi Andrew,              Tuesday June 25 2013 12:09, you wrote to me:               AL> Hello Ulrich!        AL> 24 Jun 13 00:39, you wrote to me:        AL>>> This appears to be an OpenWatcom issue. I've already modified        AL>>> the code to use version b, with no effect on the SearchMaxMSG        AL>>> crash under OS/2.        AL> ...               US>> based on the oswatfnd.c, oswatful.c, oswatxxx.h, osgenaps.c and        US>> os.c an ASM routine _dos_findnext() starts moving around some        US>> memory in the debugger it identifies itself as: Assembly: findos2        US>> 005B:0002AF6A findos2@copydir_+0000002A        AL> ..        AL> Yes, that is definitely a problem.        AL>        US>> Solution(s) ?        US>> I don't realy know ...        US>> I assume, that the findfirst() / findnext() code requires a        US>> defined memory allocation ?!? -or- further investigation is        US>> needed to discover the reason, why _dos_findnext() writes into a        US>> memory area that is allocated by the active process        AL>        AL> It appears to be a problem with OpenWatcom 1.9. I will attempt to        AL> compile the code with an older version to see if this resolves the        AL> issue for MakeNL.              ok, I've worked on a workaround for the _dos_findfirst() / _dos_findnext()       routines together with Markus the last 2? 3? weeks.              Today I've got an alternate bundle of DosFindFirst() / DosFindNext()       routines compiled.              The program started running ... and crashed too :(              Of special interest is, that the program crashed unrelated to       the SearchMaxMSG routine.              In the preparation steps, makenl collects the files defined in the makenl       control file, while scanning update, inbound and working directory       through the myfnmerge() routine.              With the new routines included (findfirs.c, findnext.c from watcom ow19 src       Bld/Clib/File/C)       the program probably crashes, while trying to transfer the results       found in the low-level clib routines to the makenl data working area.              As seen in my debug run of the original makenl code, the _dos_findnext()       information transfer started overwriting areas of the codes local data area.              It looks like, that there is not enough memory available, so the       segments started overlapping that results in the crash       that brings me back to below still known report              (I still did not a debug run, but probably the result is similar       to the previously published debug result ... that the low level routine       starts overwriting local data area of the running code)                      US>> As least similar findfirst() findnext() routines        US>> as shown under        US>> http://www.openwatcom.org:4000/@md=d&cd=//&cdf=//depot/public/4os        US>> 2/c/w rappers.c        US>> &ra=s&c=rGG@//depot/public/4os2/c/wrappers.c?ac=64 has some more        US>> code included.        AL>        US>> here 2 comments are of interest:        US>> 20 Aug 07 GKY Move #pragma alloc_text to end for OpenWatcom        US>> compat 01 Sep 07 GKY Add xDosSetPathInfo to fix case where FS3        US>> buffer crosses 64k boundry               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^               US>> maybe something similar is required for the makenl_ng code ?!?        AL> Possibly; if so it will take some time to research and implement.        AL> Thanks for the help in debugging this; it is greatly appreciated.              my C experiences are far from such experiences required, to get such       a task running ...       I still have no idea how such extended memory usage monitoring       can be implemented, to prevent overlapping of data areas              From other non-C compile and link projects, I know that its possible       to receive a map of memory layout for the compiled program, that may help in       analysing the memory requirements by the compiled program. But I'm currently       don't know how I can add this to the watcom compile and link configuration              I still start the compile with the following commandline       wmake -f makefos2.wat DEBUG=1 OS=OS2 >> comp_os2.log       where makefos2.wat is the copied watcom.compile file from the makenl_ng       distribution package              within the watcom.compile file I don't know which flags I can set       so that a map layout file will be written in the compile run               AL> Andrew              regards, uli ;-)              ---        * Origin: AMBROSIA - Frankfurt/Main - Germany (2:244/1120)    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca