0
votes

The following code takes an input string and puts it on the heap, then prints it:

   #include <stdio.h>
   #include <stdlib.h>
   #include <string.h>

   int main() {

      char input[50] = "BLA BLBA BLA BLA DJAIO JASJDIOA";
      char *value = (char*) calloc(1, sizeof(char));

      value[0] = '\0';


      for (int i = 0; i < strlen(input); i++) {

          value = (char*) realloc(value, sizeof(char) * (strlen(value) + 2));
          if (value == NULL)
              return 1;

          value[strlen(value)] = input[i];
          value[strlen(value) + 1] = '\0';
      }

      printf("%s\n", value);

      free(value);
      return 0;
 }

Works perfectly, but from some reason it gives these errors from valgrind:

==109423== Memcheck, a memory error detector
==109423== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==109423== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info
==109423== Command: ./write_test.o
==109423==
==109423== Conditional jump or move depends on uninitialised value(s)
==109423==    at 0x4C2A9E8: strlen (vg_replace_strmem.c:454)
==109423==    by 0x400722: main (in /home/me/testing/write_test.o)
==109423==
==109423== Invalid write of size 1
==109423==    at 0x40072E: main (in /home/me/testing/write_test.o)
==109423==  Address 0x51f4092 is 0 bytes after a block of size 2 alloc'd
==109423==    at 0x4C29B78: realloc (vg_replace_malloc.c:785)
==109423==    by 0x4006E5: main (in /home/me/testing/write_test.o)
==109423==
==109423== Conditional jump or move depends on uninitialised value(s)
==109423==    at 0x4C2A9E8: strlen (vg_replace_strmem.c:454)
==109423==    by 0x4006D2: main (in /home/me/testing/write_test.o)
==109423==
==109423== Conditional jump or move depends on uninitialised value(s)
==109423==    at 0x4C2A9E8: strlen (vg_replace_strmem.c:454)
==109423==    by 0x400703: main (in /home/me/testing/write_test.o)
==109423==
==109423== Conditional jump or move depends on uninitialised value(s)
==109423==    at 0x4C2AA08: __GI_strlen (vg_replace_strmem.c:455)
==109423==    by 0x4E9FEEB: puts (ioputs.c:36)
==109423==    by 0x40075B: main (in /home/me/testing/write_test.o)
==109423==
==109423==
==109423== HEAP SUMMARY:
==109423==     in use at exit: 0 bytes in 0 blocks
==109423==   total heap usage: 32 allocs, 32 frees, 528 bytes allocated
==109423==
==109423== All heap blocks were freed -- no leaks are possible
==109423==
==109423== For counts of detected and suppressed errors, rerun with: -v
==109423== Use --track-origins=yes to see where uninitialised values come from
==109423== ERROR SUMMARY: 123 errors from 5 contexts (suppressed: 0 from 0)

What invalid write of size 1? What Conditional jump or move depends on uninitialised value?

2
You might want to consider keeping track of the allocated string manually, instead of using strlen. This will also improve performance somewhat. Also note that calling realloc in a loop like this is horribly inefficient and leads to heap segmentation. - Lundin

2 Answers

2
votes

The problem is here:

value[strlen(value)] = input[i];  

after the previous line, value doesn't point anymore to a NUL terminated string, therefore the strlen on the following line will return an indeterminate value:

value[strlen(value) + 1] = '\0';

You need this:

...
int len = strlen(value);
value[len] = input[i];
value[len + 1] = '\0';
...
2
votes

After that line

value[strlen(value)] = input[i];

value is no longer a NUL terminated string because you have just overwritten NUL. So calling strlen(value) in the next line and using it as array index invokes UB:

Solution:

size_t len = strlen(value);
value[len] = input[i];
value[len+1] = '\0';