search

go-python / gpython

gpython is a python interpreter written in go "batteries not included"
BSD 3-Clause "New" or "Revised" License
869 stars 95 forks source link

Recursive call crashes gpython #143

Open xiaxinmeng opened 2 years ago

xiaxinmeng commented 2 years ago

We define a function and recursive call the function in test.py crashing gpython. This case cannot crash CPython and it reports a RecursionError.

test.py:

def f():
    f()
f()

Expected output: RecursionError

Actual output: runtime: goroutine stack exceeds 1000000000-byte limit fatal error: stack overflow

runtime stack: runtime.throw(0x66c609, 0xe) /opt/go/go1.11/src/runtime/panic.go:608 +0x72 runtime.newstack() /opt/go/go1.11/src/runtime/stack.go:1008 +0x729 runtime.morestack() /opt/go/go1.11/src/runtime/asm_amd64.s:429 +0x8f goroutine 1 [running]: runtime.heapBitsSetType(0xc0027a3440, 0x30, 0x30, 0x658180) /opt/go/go1.11/src/runtime/mbitmap.go:911 +0xa30 fp=0xc0226e2390 sp=0xc0226e2388 pc=0x416c80 runtime.mallocgc(0x30, 0x658180, 0x1, 0x0) /home/ncw/go/src/github.com/go-python/gpython/vm/eval.go:1428 +0x40 fp=0xc0226e83c8 sp=0xc0226e8378 pc=0x5b33d0 ...additional frames elided... ....................... goroutine 19 [syscall]: os/signal.signal_recv(0x0) /opt/go/go1.11/src/runtime/sigqueue.go:139 +0x9c os/signal.loop() /opt/go/go1.11/src/os/signal/signal_unix.go:23 +0x22 created by os/signal.init.0 /opt/go/go1.11/src/os/signal/signal_unix.go:29 +0x41

version: Gpython3.4.0 on Ubuntu 16.04

raff commented 2 years ago

A couple of way to fix this: 1) recover from the panic and generate the RecursionError exception 2) implement the recursion check in Vm.Call, similar to CPython (add a counter for pending calls in the Vm struct, increment/decrement in Vm.Call and generate the RecursionError when the limit is reached).

raff commented 2 years ago

The 2nd method would probably be better since CPython has a limit of 1000 nested calls (by default) while gpython runs until it runs out of memory.

drew-512 commented 2 years ago

Out of curiosity, this is a lot of effort for an issue that seems a second place to many other areas of potential improvement for gpython. Perhaps there something I'm missing? I do agree it's not helpful to be able to run python in gpython that induces a crash, but there are many things that may do that.

If @sbinet is interested, I can work on a PR for this if we agree this is worth adding.