Re: [RFC 00/15] x86_64: Optimize percpu accesses

From: H. Peter Anvin
Date: Wed Jul 09 2008 - 13:46:27 EST


Mike Travis wrote:
H. Peter Anvin wrote:
Hi Mike,

Did the suspected linker bug issue ever get resolved?

-hpa

Hi Peter,

I was not able to figure out how the two versions of the same
kernel compiled by gcc-4.2.0 and gcc-4.2.4 differed. Currently,
I'm sticking with gcc-4.2.4 as it boots much farther.

There still is a problem where if I bump THREAD_ORDER, the
problem goes away and everything so far that I've tested boots
up fine.

We tried to install a later gcc (4.3.1) that might have the
"GCC_HAS_SP" flag but our sys admin reported:

The 4.3.1 version gives me errors on the make. I had to
pre-install gmp and mpfr, but, I still get errors on the make.

I think that was the latest he found on the GNU/GCC site.


We have seen miscompilations with gcc 4.3.0 at least; not sure about 4.3.1.

-hpa

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/