search for: memoryoperand

...ets (indexes), would return noAlias. However, I'm somewhat confused as there is no notion of offset for the Location object. I would also like to call ScheduleDAGInstr::buildSchedGraph() with this AliasAnalysis and have MIsNeedsChainEdge() return false in this case. What should I do? Adding a MemoryOperand to such an instruction seems right, but it doesn't seem to fit quite. What Value would be referenced? BasicAliasAnalysis returns MustAlias for the same Value, e g 'Stack'. Should I implement a target AliasAnalysis, perhaps derived from BasicAliasAnalysis, and make it required for my pa...

...hat > confused as there is no notion of offset for the Location object. > > > > I would also like to call ScheduleDAGInstr::buildSchedGraph() with > this AliasAnalysis and have MIsNeedsChainEdge() return false in this > case. > > > > What should I do? Adding a MemoryOperand to such an instruction seems > right, but it doesn’t seem to fit quite. What Value would be > referenced? I think that they should have pseudo-source values, see: include/llvm/CodeGen/PseudoSourceValue.h I was under the impression that different pseudo source values from different frame ind...

...ffset for the Location object. > > > > > > > > I would also like to call ScheduleDAGInstr::buildSchedGraph() with > > this AliasAnalysis and have MIsNeedsChainEdge() return false in this > > case. > > > > > > > > What should I do? Adding a MemoryOperand to such an instruction seems > > right, but it doesn't seem to fit quite. What Value would be > > referenced? > > I think that they should have pseudo-source values, see: > include/llvm/CodeGen/PseudoSourceValue.h > > I was under the impression that different pseudo...

...tle transforms to clean things up. I know of the RFC for a new Combiner with its own syntax (https://reviews.llvm.org/D54286 is the latest I can find of it), but after manually adding my Nth manually coded pass for a niggling important transform, and then needing to add more cases to FoldImmediate/MemoryOperand/OptimizeLoad after it. I wondered how hard it would be to allow GlobalISel to reselect machine patterns, eg after they've been made available by other passes. What I was thinking is in addition to anything else that's coming, allowing Instructions to exist on the input side of Pat<>,...

...also like to call ScheduleDAGInstr::buildSchedGraph() >>>> with >>>> this AliasAnalysis and have MIsNeedsChainEdge() return false in >>>> this >>>> case. >>>> >>>> >>>> >>>> What should I do? Adding a MemoryOperand to such an instruction >>>> seems >>>> right, but it doesn't seem to fit quite. What Value would be >>>> referenced? >>> >>> I think that they should have pseudo-source values, see: >>> include/llvm/CodeGen/PseudoSourceValue.h >...

...> > I would also like to call ScheduleDAGInstr::buildSchedGraph() > > > with > > > this AliasAnalysis and have MIsNeedsChainEdge() return false in > > > this > > > case. > > > > > > > > > > > > What should I do? Adding a MemoryOperand to such an instruction > > > seems > > > right, but it doesn't seem to fit quite. What Value would be > > > referenced? > > > > I think that they should have pseudo-source values, see: > > include/llvm/CodeGen/PseudoSourceValue.h > > > &gt...

...dSchedGraph() > >>>> with > >>>> this AliasAnalysis and have MIsNeedsChainEdge() return false in > >>>> this > >>>> case. > >>>> > >>>> > >>>> > >>>> What should I do? Adding a MemoryOperand to such an instruction > >>>> seems > >>>> right, but it doesn't seem to fit quite. What Value would be > >>>> referenced? > >>> > >>> I think that they should have pseudo-source values, see: > >>> include/llvm/Co...

Hi As i didn't get any respons I am reformulating my question in the hope to make myself more clear. Has anyone an idea how to use metadata present in the llvm ir to guide the if-conversion in llc (as metadata get's dropped during SelectionDAG) ? -Michael 2013/4/13 Michael D'Hont <michael.dhont at ugent.be> > The project I am working on is to use the llvm toolchain for

...dn’t have an > MachineMemOperand’s, despite being marked as loading from memory. > > Will file a PR so that we can track some of these issues. Coming late to this discussion since I just noticed this was talking about MemRefs... We have deep and horrible problems around our handling of MemoryOperands. As it stands, there does not appear to be a single invariant which is documented and accepted throughout the various backends. There are conflicting assumptions in various bits of code. I tried to clean up some of this a while back and made some forward progress, but there's a lot of w...

...hineMemOperand’s, despite being marked as loading from memory. >> >> Will file a PR so that we can track some of these issues. > Coming late to this discussion since I just noticed this was talking about MemRefs... > > We have deep and horrible problems around our handling of MemoryOperands. As it stands, there does not appear to be a single invariant which is documented and accepted throughout the various backends. Yeah, i’ve been noticing the same thing :( > There are conflicting assumptions in various bits of code. I tried to clean up some of this a while back and made some...

The project I am working on is to use the llvm toolchain for embedded CGRA processors . This however poses some restrictions on the block formation, because modulo scheduling is applied in a later stage. For this reason the idea was to create custom pragma's to generate metadata and attach it to de branches of loops we wanted to map on a cgra module. It is a lot similar to the loop parallell

Is this something that will happen in future? Or there is no plan for it? Thanks, Ali -----Original Message----- From: llvmdev-bounces at cs.uiuc.edu [mailto:llvmdev-bounces at cs.uiuc.edu] On Behalf Of Evan Cheng Sent: Wednesday, April 02, 2008 10:30 AM To: LLVM Developers Mailing List Subject: Re: [LLVMdev] Retrieving local variable names. It's not currently available. LLVM would have to

> On Mar 28, 2016, at 6:02 PM, Pete Cooper via llvm-dev <llvm-dev at lists.llvm.org> wrote: > >> >> On Mar 28, 2016, at 3:12 PM, Chandler Carruth <chandlerc at google.com <mailto:chandlerc at google.com>> wrote: >> >> FWIW, I agree with Mehdi that we should just assert that our types don't get called with size zero. > Yeah, I agree. I’ve