From: "Wormy" <wormy AT technologist DOT com>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Problems with MALLOC and FREE
Date: Fri, 10 Dec 1999 19:33:51 +0100
Organization: University of Economics and Business Administration, Vienna, Austria
Lines: 56
Message-ID: <82rh3j$sfu@cantine.wu-wien.ac.at>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 991209160149 DOT 22560D-100000 AT is>
NNTP-Posting-Host: 212186012098.11.wu-wien.teleweb.at
X-Newsreader: Microsoft Outlook Express 4.72.3110.1
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com


Eli Zaretskii schrieb in Nachricht ...
>
>On Thu, 9 Dec 1999, Wormy wrote:
>
>> But if, at shutdown, I do
>>
>> for (x=0; x<4096; x++)
>>   if (mem[x] != NULL)
>>     free (mem[x]);
>>
>> IT HAPPENS... WELL, MORE THAN ONLY SOMETIMES THAT I GET A SIGSEGV!!!!!
>> WHY IS THAT???
>
>This usually means that your program either overwrites the buffers that
>you allocate (like if it puts more than 64 bytes into each 64-byte
>buffer), or that it free()'s some of the buffers more than once.



BUT WHY DOES IT FREE MORE THAN ONCE??!

It even happens sometimes if I execute the following (one after the other):

char *mem[12000];

for (i=0; i<12000; i++)                                    // THIS PART
ALWAYS WORKS RIGHT ON MY COMP
                                                                         //
THE MEMORIES ARE PROPERLY ALLOCATED I CAN
                                                                        //
WRITE AND READ TO THEM
{
  if ((mem[i]=(char *)malloc(64*sizeof(char)))==NULL)
    HANDLE ERROR
  else
    memset(mem[i], 0, 64*sizeof(char));
}

// AND NOW WITHOUT DOING ANYTHING

for (i=0; i<12000; i++)
{
  if (mem[i] != NULL)
    free (mem[i]);
}

// AND SOMETIMES IT HAPPENS THAT THE PROGRAM SHUTS DOWN WITH A SIGSEGV!

Funny thing is it happens even if I check if mem[i] != NULL ----> So the
program shouldn't free any memory that wasn't freed yet...

Could it be a bug with DJGPP or is it maybe my computer?
Hm... maybe I should test it on several computers?