Init df for split pass since some target use REG_NOTE in split pattern

Jeff Law law@redhat.com
Mon Aug 8 19:45:00 GMT 2016


On 07/25/2016 08:31 PM, Kito Cheng wrote:
> Hi Jeff:
>
> Oop, patch in attachment, and I hit this bug in gcc.dg/torture/vshuf-v2di.c
> with our nds32 internal branch.
>
>
> Hi Richard:
> I think we really need reg dead note for some optimization, and btw,
> here is our split pattern:
So I see a real need to be able to look at those notes -- I think the 
big question in my mind is whether or not we want to pay the price for 
all the splitter passes or if we can/should restrict it to a subset of 
the passes.  Note that I think we have 4 distinct passes that call 
split_all_insns.

For the H8, we could restrict it to cases where epilogue_completed is 
true.  I haven't looked closely at the other targets.


I'm pretty sure we do _not_ want it for split_all_insns_noflow since 
that's used when the CFG is not necessarily correct and thus I don't see 
how we can reliably have death/unused notes.


Jeff



More information about the Gcc-patches mailing list