Fixup X86_TUNE_DEEP_BRANCH_PREDICTION

[H. J. Lu]

On Mon, Apr 23, 2007 at 03:15:11PM -0700, Mike Stump wrote:
> On Apr 23, 2007, at 12:47 PM, Richard Henderson wrote:
> >On Mon, Apr 23, 2007 at 12:04:30PM -0700, H. J. Lu wrote:
> >>Intel processors all have special optimization not to push onto
> >>return stack when the call is to the instruction following the call.
> >>X86_TUNE_DEEP_BRANCH_PREDICTION should be off.
> >
> >In which case, Mike, your patch is ok if you re-add GENERIC
> >to the list that do want deep-branch-prediction.
> 
> Ok, I checked in the below after a new round of regtesting on i686- 
> apple-darwin9, just to be sure.
> 
> Thanks.
> 
> mrs2 $ cat ~/diffs/deepa.diffs
> 2007-04-23  Mike Stump  <mrs@apple.com>
> 
> 	* config/i386/i386.c (ix86_tune_features
> 	[X86_TUNE_DEEP_BRANCH_PREDICTION]: Prefer call over thunks on
> 	nocona and core2.
> 
> Doing diffs in .:
> --- ./config/i386/i386.c.~1~	2007-04-21 15:48:04.000000000 -0700
> +++ ./config/i386/i386.c	2007-04-23 13:43:51.000000000 -0700
> @@ -1026,8 +1026,7 @@ unsigned int ix86_tune_features[X86_TUNE
>    m_486 | m_PENT | m_PPRO | m_ATHLON_K8_AMDFAM10 | m_K6 | m_CORE2 |  
> m_GENERIC,
> 
>    /* X86_TUNE_DEEP_BRANCH_PREDICTION */
> -  m_PPRO | m_K6_GEODE | m_ATHLON_K8_AMDFAM10 | m_PENT4
> -  | m_NOCONA | m_CORE2 | m_GENERIC,
> +  m_PPRO | m_K6_GEODE | m_ATHLON_K8_AMDFAM10 | m_PENT4 | m_GENERIC,
> 

I think we should also remove m_PENT4 and m_PPRO.


H.J.