Re: [RFA] [PR tree-optimization/33562] [PATCH 1/4] Byte tracking in DSE

[Jeff Law <law at redhat dot com>]

On 12/16/2016 11:35 AM, Joseph Myers wrote:
> On Thu, 15 Dec 2016, Jeff Law wrote:
>
> > This version attacks the problem by improving DSE to track stores to memory at
> > a byte level.  That allows us to determine if a series of stores completely
> > covers an earlier store (thus making the earlier store dead).
>
> Question: suppose you have an assignment for a struct with padding, then
> stores for all the elements.  Does it detect that the original assignment
> is dead (because there is no need to copy padding on struct assignment)?
It would not detect the original assignment as dead in that case as it 
does not know anything special about the padding bytes.

So we'd likely end up just trimming the original assignment, 
particularly if it was done via a CONSTRUCTOR assignment or memset.

This situation could be addressed by handling the padding bytes 
specially -- ie, just clear them from LIVE when LIVE is initialized. 
It'd likely "just work" then.

Seems like it might be a reasonable follow-up as it'd hopefully make 
more of the original assignments fully dead rather than just trimmable.

jeff