Re: [boost] bug in 1.35's sp_counted_base_gcc_sparc.hpp

From: Peter Dimov (pdimov_at_[hidden])
Date: 2007-02-06 17:28:10

Jerry Lawson wrote:

>> Sounds like a gcc bug. Which versions of gcc have you tried?
>
> I have gcc 4.1.1 on a solaris10 sparv9 machine (64 bit).

Can you please try reporting it on the GCC Bugzilla?

> I'm not sure how one would go about using 32-bit pointers on a 64-bit
> machine. Is that a gcc option or a hardwired thing in the processor?

I've only seen this odd memory model on some HP IA64 boxes. It probably 
won't be a problem for SPARC, so your proposed fix is a viable option. 
Another alternative is to take advantage of the __sync intrinsics for GCC 
4.1+:

http://gcc.gnu.org/onlinedocs/gcc-4.1.1/gcc/Atomic-Builtins.html