[Python-Dev] stack check on Unix: any suggestions? (original) (raw)

M.-A. Lemburg mal@lemburg.com
Tue, 29 Aug 2000 21:42:08 +0200

Jeremy Hylton wrote:

>>>>> "MAL" == M -A Lemburg <mal@lemburg.com> writes: >> I don't see an obvious solution. Is there any way to implement >> PyOSCheckStack on Unix? I imagine that each platform would have >> its own variant and that there is no hope of getting them >> debugged before 2.0b1. MAL> I've looked around in the include files for Linux but haven't MAL> found any APIs which could be used to check the stack size. MAL> Not even getrusage() returns anything useful for the current MAL> stack size. Right. MAL> For the foo() example I found that on my machine the core dump MAL> happens at depth 9821 (counted from 0), so setting the MAL> recursion limit to something around 9000 should fix it at least MAL> for Linux2. Right. I had forgotten about the MAXRECURSIONLIMIT. It would probably be better to set the limit lower on Linux only, right? If so, what's the cleanest was to make the value depend on the platform.

Perhaps a naive test in the configure script might help. I used the following script to determine the limit:

import resource i = 0
def foo(x): global i print i,resource.getrusage(resource.RUSAGE_SELF)
i = i + 1 foo(x) foo(None)

Perhaps a configure script could emulate the stack requirements of eval_code2() by using declaring a buffer of a certain size. The script would then run in a similar way as the one above printing the current stack depth and then dump core at some point. The configure script would then have to remove the core file and use the last written depth number as basis for setting the MAX_RECURSION_LIMIT.

E.g. for the above Python script I get:

9818 (4.2199999999999998, 0.48999999999999999, 0, 0, 0, 0, 1432, 627, 0, 0, 0, 0, 0, 0, 0, 0) 9819 (4.2199999999999998, 0.48999999999999999, 0, 0, 0, 0, 1432, 627, 0, 0, 0, 0, 0, 0, 0, 0) 9820 (4.2199999999999998, 0.48999999999999999, 0, 0, 0, 0, 1432, 627, 0, 0, 0, 0, 0, 0, 0, 0) 9821 (4.2199999999999998, 0.48999999999999999, 0, 0, 0, 0, 1432, 627, 0, 0, 0, 0, 0, 0, 0, 0) Segmentation fault (core dumped)

-- Marc-Andre Lemburg

Business: http://www.lemburg.com/ Python Pages: http://www.lemburg.com/python/