Thanks for the advice!  I tried the -O1 level and for my applications it
generally generates smaller code size (what I'm really looking for) and
doesn't have the same optimization bugs I ran into with the -O2 and -Os.

Robert

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of
[EMAIL PROTECTED]
Sent: Thursday, February 08, 2007 1:28 PM
To: uClinux development list
Subject: Re: [uClinux-dev] C++ O2 optimization bugs? 

Robert Daniels wrote:

> I've been using the snapgear toolchain arm-linux-tools-20051123 with
the
> gcc-3.3.2 compiler and have experienced some -O2 optimization bugs in
my
> c++ application.  I've tracked these bugs down and tweaked my code to
> eliminate them but don't know if there are any more.

Did you try other compiler optimisation levels?  I'd try at least -O1
and 
possible -Os to see if the bugs are present at those settings.  The code
speed 
differences aren't usually great.  We build most of our products with
-O1 
since it typically produces smaller and faster code for the CPUs we are 
targetting  (arm, sh-4 and previously coldfire -- all small cache).
Don't 
blindly believe higher optimisation levels mean faster.


> I'm wondering if I upgrade to a newer version of the toolchain if I'll
have
> a greater probability of reduced optimization bugs.

Usually this is true.  However, newer versions have introduced bugs on 
occasion.


> toolchain which I believe includes gcc-3.4.4 - does anyone know if
there
> are improvements to the gcc optimizer that would improve my situation?

Without knowing the details of the bugs you are running up against, this

question isn't really answerable :-)  You could check the gcc change log
but I 
doubt it would make a lot of sense.


Regards,

Pauli
-- 
Dr Paul Dale, Software Grunt, SnapGear
Leaders in embedded Linux security   http://www.SnapGear.com/


_______________________________________________
uClinux-dev mailing list
[email protected]
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by [email protected]
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev
_______________________________________________
uClinux-dev mailing list
[email protected]
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by [email protected]
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev

Reply via email to