Re: toolchain/44102: Internal compiler error on trivial source

[christos%zoulas.com@localhost (Christos Zoulas)]

On Nov 17,  4:55pm, lacombar%gmail.com@localhost (Arnaud Lacombe) wrote:
-- Subject: Re: toolchain/44102: Internal compiler error on trivial source

| The following reply was made to PR toolchain/44102; it has been noted by 
GNATS.
| 
| From: Arnaud Lacombe <lacombar%gmail.com@localhost>
| To: Paul Koning <paul_koning%dell.com@localhost>
| Cc: gnats-bugs%netbsd.org@localhost, toolchain-manager%netbsd.org@localhost, 
|       gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost
| Subject: Re: toolchain/44102: Internal compiler error on trivial source
| Date: Wed, 17 Nov 2010 11:53:03 -0500
| 
|  Hi,
|  
|  On Wed, Nov 17, 2010 at 10:44 AM, Paul Koning 
<paul_koning%dell.com@localhost> wrote:
|  >
|  > On Nov 17, 2010, at 12:30 AM, David Holland wrote:
|  >
|  >> The following reply was made to PR toolchain/44102; it has been noted by 
GNATS.
|  >>
|  >> From: David Holland <dholland-bugs%netbsd.org@localhost>
|  >> To: gnats-bugs%NetBSD.org@localhost
|  >> Cc:
|  >> Subject: Re: toolchain/44102: Internal compiler error on trivial source
|  >> Date: Wed, 17 Nov 2010 05:29:11 +0000
|  >>
|  >> On Wed, Nov 17, 2010 at 05:20:01AM +0000, lacombar%gmail.com@localhost 
wrote:
|  >>> static inline void __attribute__((noreturn))
|  >>> __bug()
|  >>> {
|  >>> }
|  >>
|  >> Yes, so it shouldn't crash, but this is not valid code and there is
|  >> almost certainly no way to trigger the problem with valid code.
|  >>
|  >> Probably not even all that hard to find by inspecting the gcc source...
|  >
|  > I'm not sure about not that hard...
|  >
|  > Some testing shows the following:
|  >
|  > gcc 4.2.1 (on Mac OS) also has the issue
|  > gcc 4.5.0 (for MIPS/NetBSD) is ok
|  > gcc 4.6.0 (current trunk pre-release) is ok
|  >
|  > "ok" means it produces a warning "'noreturn' function does return".
|  >
|  it looks like:
|  
|  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37879
|  
|  But I'm not sure the fix will not have GPLv3 issue.
|  
|  > So one solution to this PR would be to update the toolchain to 4.5.1 (the 
latest released gcc).
|  >
|  this will clearly have a GPLv3 issue :) Which I kinda remind a lot
|  people are not fond of.

Been there, done that (binutils). Unless we want to immitate OpenBSD...

christos