Segfault with too deep recursion

[michelboaventura]

This code:

require "big_int"                                                        

def fib(n)                                                               
  n > 1 ? n * fib(n - 1) : n                                             
end                                                                      

fib(100_000.to_big_i) 

Just gives a segfault. Reducing the value to 1000 solves. Compiling it with -d and running in gdb gives:

Starting program: /tmp/test
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7ffff3649700 (LWP 3833)]
[New Thread 0x7ffff3e4a700 (LWP 3832)]
[New Thread 0x7ffff464b700 (LWP 3831)]
[New Thread 0x7ffff4e4c700 (LWP 3830)]
[New Thread 0x7ffff564d700 (LWP 3829)]
[New Thread 0x7ffff5e4e700 (LWP 3828)]
[New Thread 0x7ffff664f700 (LWP 3827)]

Program received signal SIGSEGV, Segmentation fault.
0x0000000000401991 in __crystal_realloc ()

Not very useful. Running on valgrind gives a LOT of messages related with GC_* functions and this one, which may be the cause of segfault:

==3255== Stack overflow in thread 1: can't grow stack to 0xffe801c48

So I think Crystal isn't checking if it can allocate more memory on the stack. The same code on ruby gives:

test.cr:4:in `fib': stack level too deep (SystemStackError)

[jhass]

This is essentially a duplicate of #271

[michelboaventura]

Hum, sorry about that. I should have looked before posting. Closing it now, thanks!