[PATCH] Constant folding around pre-increment loads and stores

[Evan Cheng]

LGTM. The longer term plan is to move indexed mode formation out of dag isel. But I do believe this patch will provide immediate improvements.

Thanks,

Evan

On Feb 4, 2013, at 10:04 AM, Hal Finkel <hfinkel at anl.gov> wrote:

> ----- Original Message -----
>> From: "Krzysztof Parzyszek" <kparzysz at codeaurora.org>
>> To: llvm-commits at cs.uiuc.edu
>> Sent: Monday, February 4, 2013 11:47:41 AM
>> Subject: Re: [PATCH] Constant folding around pre-increment loads and stores
>> 
>> On 1/28/2013 6:40 PM, Hal Finkel wrote:
>>> 
>>> [...] code generation for pre-increment forms of loads and stores.
>> 
>> IMO, it would be a lot better to have this done late (but before RA).
>> I'm not going to try to change anybody's mind though...
> 
> Can you be more specific? Do you mean pre-inc formation in general, or just the associated constant folding? and why?
> 
> Just to be clear (to other potential reviewers), this patch does not change when the pre-inc formation is done, it just adds some additional (useful) constant folding in a way which should reduce register pressure. As DAGCombine already does this kind of folding, I think this fits with the current scheme.
> 
> Thanks again,
> Hal
> 
>> 
>> -Krzysztof
>> 
>> --
>> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
>> hosted by The Linux Foundation
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits