Fix stack organization

[rafajunio]

Wrong calculation near the limits of argv, envv and auxv

[dledda-r7]

Hello! Thanks for your PR.
I am leaving couple of questions to get a bit of context for these changes.

[dledda-r7]

Were you observing any strange behavior before making these changes? If so, what were you experiencing?
Could you provide a PoC of an unexpected result you had and how this fix is solving the problem?
This will really help us understand what is going on.

[rafajunio]

Hello @dledda-r7 , for testing it is possible to use the follow code (print_stack.c):

#include <stdio.h>

extern char **environ;

int main(int argc, char **argv)
{
	int cnt;
	size_t *stack = (size_t *)argv - 1;

	printf("argc: %d\n", argc);
	printf("\n");
	for (cnt = 0; argv[cnt] != NULL; cnt++) {
		printf("argv[%d]: %s\n", cnt, argv[cnt]);
	}
	printf("\n");

	for (cnt = 0; environ[cnt] != NULL; cnt++) {
		printf("envv[%d]: %s\n", cnt, environ[cnt]);
	}

	printf("Stack:\n");
	for (cnt = 0; stack[cnt] || stack[cnt + 1]; cnt++) {
		printf("  0x%08zx\n", stack[cnt]);
	}
	return 0;
}

Compile as follow:
gcc print_stack.c -o print_stack
Testing:
./print_stack hello this is a test

From the libreflect it is possible to use the follow code:
./noexec print_stack hello this is a test

[acammack]

Ah yes, off-by-one errors, the bane of long nights and blurry debugger sessions. This one is made even worse by the debug output looking correct as it mangles the stack it's trying to produce and that stack mostly working 😈 . Thanks for the catch, cheers!

[acammack]

Thanks, the code tests correctly and looks good to me! As a heads up, Rapid7 usually requires PRs to be made from unique topic branches, though I'll let the current team members take it from here

[adfoster-r7]

@rafajunio It'd be great to rebase this PR, we had to make some changes for CI to run: #272

[rafajunio]

Lesson learned, not use master to create PR anymore =) (my fork)

Thank you, I hope that now is everything good. I merge the commits in the rebase as well.