Re: "stack smashing detected"

On 6/12/2024 7:36 PM, Ben Bacarisse wrote:

DFS <nospam@dfs.com> writes:
 

Same C program I just made a post about 'undefined behavior' on.
>
I just noticed I get "stack smashing detected" only when I run the program
using a dataset of 75+ consecutive integers.
>
set of 2 to 74 consecutive values: no problem.
set of random values of any size: no problem.
75+ consecutive values: problem
>
Any easy way to find out what's causing this issue (probably a buffer
overflow)?

 I compiled using -fsanitize=undefined and at runtime I get:
 $ ./dfs 70 -c
70 Consecutive:
No commas     : 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70
dfs.c:126:18: runtime error: index 70 out of bounds for type 'int [*]'
...
 Line 126:
    if(nums[i]-nums[i+1]!=0) {ucnt=1;} else {ucnt++;}
 nums is a VLA declared int nums[N];

int nums[N]  is OK isn't it, if I'm not running huge datasets?
or should you always malloc:
int *nums = malloc(N * sizeof(int));

and i runs from 0 to i < N (70 in
this run).  When i == 69 (the max) nums[i+1] is undefined (it's out of
bounds).

Good catch.  Fixed.
The code worked even with that bug - it always correctly identifies the mode(s), according to that website.  But I probably got lucky and the mode(s) weren't the last data point.
I also no longer run that mode-analysis routine for consecutive datasets (because they have no repeating values).

If I run
 $ ./dfs 1 -c
1 Consecutive:
No commas     : 1
dfs.c:126:18: runtime error: index 1 out of bounds for type 'int [*]'
dfs.c:161:14: runtime error: index -1 out of bounds for type 'int [*]'
dfs.c:162:32: runtime error: index 1 out of bounds for type 'int [*]'

Yeah, I did very little arg validation.

I see two further problems. 

What are they?

With some stats you will have to set a
minimum for N, but the first problem is just an incorrect index.
 BTW, when I was starting out, I'd give my right arm for gcc's
-fsanitize=undefined and/or valgrind.

I just tried -fsanitize=undefined and got that error.  Is that new in gcc?
valgrind is too hardcore for me (for the hobby C programs I write anyway).
my new compile string is:
$ gcc -Wall -Wextra -O3 -fsanitize=undefined mmv2.c -o mmv2 -lm
Thanks for looking into it.