This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Fix DCE REG_LIBCALL note moving from noop move insns (PR rtl-optimization/33644)

On 10/19/07, Jakub Jelinek <> wrote:
> On Fri, Oct 19, 2007 at 07:23:05PM +0200, Paolo Bonzini wrote:
> >
> > >If we want to change the meaning or the interpretation to clean this up,
> > >I'm all for it, but I think it's too late to do that for 4.3.
> >
> > I don't think that this is 4.3 material anyway, except for Eric's DCE patch.
> The removal of libcall notes from TLS sequences surely is not 4.3 material.


> But not DCEing CONST_OR_PURE_CALL_P calls or


> insns with UNSPECs inside of
> PARALLELs is IMHO a regression from 4.2 and as such should be fixed.

I'm not sure on this. During the current DCE development,
we've asked for what UNSPEC means,
and the collective answer we got was what's described in rtl.def.

As the FIXME in rtl.def suggests, the longer term direction
we agreed to is to allow deleting UNSPEC regardless of where it appears
(as you suggests)
but last I heard, there are backends that are using UNSPEC
for places that it should have used USE
and for what it's worth, this was considered the existing practice.

If we think we should fix this in 4.3, that's fine by me
but certainly it would require some cleanup efforts from various backends.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]