Closed GoogleCodeExporter closed 9 years ago
Original comment by Charles....@gmail.com
on 12 Feb 2010 at 2:06
Original comment by Charles....@gmail.com
on 12 Feb 2010 at 2:07
Original comment by Charles....@gmail.com
on 14 Jul 2010 at 9:39
of course, having looked at it more closely, it's really more than
"portability".
the tricky bit is that we don't want control to be transferred to a catch-all
exception handler at the top of the stack, popping the stack. otherwise we
could just use the normal Windows C/C++ syntax. instead we want to keep the
stack
just below the point of failure, and change the state only just enough to
invoke the Dis exception handling code at that point. on other platforms,
the signal handler can call disfault, which will eventually do an error/longjmp
to pop the stack itself. here, it apparently has to return through
the Windows exception handler, presumably because it has its own state to
remove.
Original comment by Charles....@gmail.com
on 11 Aug 2010 at 4:57
Original comment by Charles....@gmail.com
on 11 Aug 2010 at 4:58
i've given someone a test version of os.c to try, that reverts to the earlier
and portable - within Windows - use of SetUnhandledExceptionFilter. see the
sflag == 0 test. if that seems fine, we'll proceed to use that once more, with
any extra changes needed to fix up the floating-point state on fp traps.
Original comment by Charles....@gmail.com
on 11 Aug 2010 at 5:30
CHANGES
emu/Nt/os.c
include/version.h
mkfiles/mkfile-Linux-386
committed changeset 461:d8cf126ad631
pushing to https://Charles.Forsyth:***@inferno-os.googlecode.com/hg
searching for changes
1 changesets found
Original comment by Charles....@gmail.com
on 14 Sep 2010 at 12:24
Original issue reported on code.google.com by
Charles....@gmail.com
on 5 Feb 2010 at 3:28