This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch: prologue on NT
- To: Richard Henderson <rth at cygnus dot com>
- Subject: Re: Patch: prologue on NT
- From: Donn Terry <donn at verinet dot com>
- Date: Sun, 18 Apr 1999 14:56:25 -0600
- CC: "egcs-patches at egcs dot cygnus dot com" <egcs-patches at egcs dot cygnus dot com>
- References: <3718B7CB.2523DC28@verinet.com> <19990417190320.A1268@cygnus.com>
Richard Henderson wrote:
> On Sat, Apr 17, 1999 at 10:33:16AM -0600, Donn Terry wrote:
> > On NT, the MS exception handling stuff is used everywhere. (The
> > OS presumes it's present, and the code is in NTDLL.DLL.) It in
> > turn relies on "reverse execution" to unwind the stack on the ALPHA.
> > (The same general stuff that gdb has to do on any Alpha, only
> > moreso.) Not all conceptually legal stack entry and exit sequences
> > are actually supported. Specifically, subq as a stack push is NOT
> > supported. If it is used, and an exception occurs, the application
> > will loop indefinitely in NTDLL.DLL.
>
> How extraordinarily lame.
I'm at home, but I've bootstrapped with your changes, and all seems
MOSTLY OK. However, I need to get to the bible to see exactly
what the rules are on large frame sizes; something seems wrong but
I need the reference to make sure I don't misspeak myself. Clearly
the code will run, but whether it will unwind needs some research I
can't do from here.
Donn
P.S. The regressions are still not in shape to tell us anything useful;
more patches coming as fast as I can put them together and make sure
they're right.