This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: C++ PATCH: Improve exceptions/inlining compile-time performance
- From: Mark Mitchell <mark at codesourcery dot com>
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: 26 Apr 2003 13:21:14 -0700
- Subject: Re: C++ PATCH: Improve exceptions/inlining compile-time performance
- References: <200304261354.JAA30862@makai.watson.ibm.com>
On Sat, 2003-04-26 at 06:54, David Edelsohn wrote:
> >>>>> Mark Mitchell writes:
>
> Mark> Aha! What that means is that your compiler is actually *better* than
> Mark> expected. The "bailing out" message is what happens when a SEGV occurs;
> Mark> your compiler is not crashing. There are no expected error messages in
> Mark> the testsuite for line 6 because we expect the compiler to crash on line
> Mark> 5.
>
> I am not sure about HP/UX, but AIX maps address 0 so compilers can
> produce code to hoist loads before NULL pointer checks. If some data
> structure within G++ contains a NULL pointer, it would SEGV on other
> systems but not on AIX. That is one guess about why the compiler bails
> out on Linux, but not AIX.
Yes, HP-UX does that too, so that is a likely explanation of what's
going on.
We can probably figure out what's going on in this case, but, in
general, this means that results are always going to be a bit skewed
between GNU/Linux and AIX/HP-UX; cases where the latter compilers do not
crash may show up as "excess errors".
--
Mark Mitchell
CodeSourcery, LLC
mark at codesourcery dot com