Re: New Miscompilation Into Bytecode Error

[Ranjit Mathew <rmathew at gmail dot com>]

>  > > Can you please find out which of these is causing the problem?
>  >
>  > What I have found is that we have been producing bad
>  > bytecode for some time now, but that gij was able to
>  > throw a VerifyError.
>  >
>  > But now gij is hanging while throwing this exception...

Richard has fixed EH on x86 broken by Geoff's patch.

Today's clean sources correctly throw exceptions
as well as make gij throw the verifier error for the bad
bytecode.

But why wasn't it working for gij when I had reverted 
Geoff's patch? (I do note that verify_fail() in verify.cc
has attribute "noreturn" - could it have something to
do with it?)

Strange.

It seems that there is some merit after all to the
statement "With enough procrastination, most problems
eventually sort themselves out"! ;-)

In any case, I've filed the original miscompilation 
into bytecode bug as PR16327:

  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16327

Ranjit.

-- 
Ranjit Mathew          Email: rmathew AT gmail DOT com

Bangalore, INDIA.      Web: http://ranjitmathew.tripod.com/