[Sbcl-devel] More on NetBSD segfaults

[Richard M K. <kr...@pr...>]

Hello,

About the stack overflow segfaulting on NetBSD, Christophe suggested I
check the following:

1) that the memory faults are being taken on the sigaltstack,
2) that the guard page violation is detected,
3) that the guard page is being unprotected,
4) that the sigaltstack call in bsd-os.c succeeds.

I've done the last three, not sure how to do the first. Here's what
gdb tells me is going on:

http://progn.net/static/tmp/sbcl-0.8.14.30-7-7-+lt+ldb-run-explode_stack_with_gdb-20040930.html

The backtrace at that URL is lightly colorized. There's a flat-text
version next to it

http://progn.net/static/tmp/sbcl-0.8.14.30-7-7-+lt+ldb-run-explode_stack_with_gdb-20040930.txt

I'm really not sure what's going on, but it looks as though the first
unhandled segfaults occur after the call back into lisp, after the
stack is full.

Any advice?

Thanks in advance,
Richard