IRA conflict graph, multiple alternatives and commutative operands

Bernd Schmidt bernds@codesourcery.com
Fri Mar 19 18:46:00 GMT 2010


On 03/19/2010 01:52 PM, Vladimir N. Makarov wrote:
> After some reasoning I got a conclusion that this patch has more
> advantages than disadvantages.  The patch should not reproduce PR22072
> and will result in worse code only when clobbered classes are different
> from matched classes and pseudos get one of the clobbered classes (I
> think the probability of this is very small).

> So I guess the patch is ok.  It needs a blank line after alt1
> declaration.  It would be nice to add some comment too.

I've also added the testcase from the PR, bootstrapped & regression
tested on i686-linux, ran arm-elf regression tests and committed the
patch below.  Thanks!


Bernd
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ira-42258.diff
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20100319/3c6a3fe1/attachment.ksh>


More information about the Gcc-patches mailing list