Subject: port-vax/11698: pth doesn't work on vax because longjmp too strict
To: None <gnats-bugs@gnats.netbsd.org>
From: None <fredette@mit.edu>
List: netbsd-bugs
Date: 12/11/2000 16:04:11
>Number:         11698
>Category:       port-vax
>Synopsis:       pth doesn't work on vax because longjmp too strict
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    port-vax-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Dec 11 16:04:00 PST 2000
>Closed-Date:
>Last-Modified:
>Originator:     Matthew Fredette
>Release:        NetBSD/vax 1.5_ALPHA
>Organization:
self
>Environment:
NetBSD chimes-of-freedom.home 1.5_ALPHA NetBSD 1.5_ALPHA (CHIMES-OF-FREEDOM) #23: Wed Sep 27 20:58:52 EDT 2000     fredette@chimes-of-freedom.home:/data/home/frfredette/project/vax/wscons/sys/arch/vax/compile/CHIMES-OF-FREEDOM vax
>Description:
pth (the GNU portable thread library) doesn't work on the vax
because pth assumes that it can use longjmp to switch to a new
stack entirely.  The current vax longjmp insists on only jumping
up frames in the current stack.  The m68k and i386 versions of
longjmp do not impose this restriction.
>How-To-Repeat:
Try to compile pth and run its test suite.  The tests should
fail with "longjmp botch." error messages visible.
>Fix:
This is a patch against lib/libc/arch/vax/gen/__setjmp14.S:
>Release-Note:
>Audit-Trail:
>Unformatted: