search for: isinbounds

...>(BEValueV)) { // If the increment is an inbounds GEP, then we know the address // space cannot be wrapped around. We cannot make any guarantee // about signed or unsigned overflow because pointers are // unsigned but we may have a negative index from the base // pointer. if (GEP->isInBounds()) Flags = setFlags(Flags, SCEV::FlagNW); } //-------------------------------------------------------------------------// Is there any reason to not check also Sub operator in a similar way to the Add operator? //-------------------------------------------------------------------------// if (...

...non-edge conditions. You could teach it that anything that compares it against null are false after an inbounds: IE // If the GEP is inbounds, and this is address space 0, we know the pointer is // not null. if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(I)) if (GEP->isInBounds() && GEP->getAddressSpace() == 0) { // Need a propagateInequality that replaces any equality checks against // null with false } However, you want to go further here: What you really want to do in this case is teach GVN that any inbounds gep that post-dominates a compa...

...// If the increment is an inbounds GEP, then we know the address > // space cannot be wrapped around. We cannot make any guarantee > // about signed or unsigned overflow because pointers are > // unsigned but we may have a negative index from the base > // pointer. > if (GEP->isInBounds()) > Flags = setFlags(Flags, SCEV::FlagNW); > } > //-------------------------------------------------------------------------// > > Is there any reason to not check also Sub operator in a similar way to the Add > operator? > > //-----------------------------------------...

...9 Replacement = ConstantExpr::getGetElementPtr(Pointer, 2130 &Indices[0], Indices.size()); -------------------------------------------------------------------------------------------> modified 2129 if(cast<GEPOperator>(this)->isInBounds()) 2130 Replacement = ConstantExpr::getInBoundsGetElementPtr(Pointer, 2131 &Indices[0], Indices.size()); 2132 else 2133 Replacement = ConstantExpr::getGetElementPtr(Pointer, 2134...

On Mon, May 4, 2015 at 9:40 AM, Nicholas White <n.j.white at gmail.com> wrote: > Thanks - that makes sense. It's interesting that at -O3 the optimizer > can't reduce the below though - I'll dig into it a bit and see if I > can make a patch that fixes it: I'm unsure what you expect to happen below. It's not quite the same testcase. GVN will PRE the loads, so you

I can't upload my program due to confidentiality, but the problem is obvious. At lib/Analysis/ConstantFolding.cpp:646 Constant *C = ConstantExpr::getGetElementPtr(Ops[0], NewIdxs); if (ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) { if (Constant *Folded = ConstantFoldConstantExpression(CE, TD, TLI)) C = Folded; } The generated ConstantExpr C doesn't inherit the

...inbounds GEP, then we know the address > > // space cannot be wrapped around. We cannot make any guarantee > > // about signed or unsigned overflow because pointers are > > // unsigned but we may have a negative index from the base > > // pointer. > > if (GEP->isInBounds()) > > Flags = setFlags(Flags, SCEV::FlagNW); > > } > > //-------------------------------------------------------------------------// > > > > Is there any reason to not check also Sub operator in a similar way > > to the Add > > operator? > > &gt...

...n for large strides? Hi Renato, Did you figure this out? The code you’re pointing to should preserve NSW and NUW. It is only clearing NW. Here’s what I think should happen, without actually trying it myself: - createNodeForGEP notices that inbounds implies NSW SCEV::NoWrapFlags Wrap = GEP->isInBounds() ? SCEV::FlagNSW : SCEV::FlagAnyWrap - The array index should already be an AddRecExpr with NSW const SCEV *IndexS = getSCEV(Index); - createNodeForGEP attempts to scale the index using the NSW flag we picked up earlier const SCEV *LocalOffset = getMulExpr(IndexS, ElementSize, Wra...

Hi folks, I'm trying to analyse this piece of IR: for.body: ; preds = %for.body, %entry %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ] %0 = shl nsw i64 %indvars.iv, 1 %arrayidx = getelementptr inbounds i32* %b, i64 %0 %1 = load i32* %arrayidx, align 4, !tbaa !1 %add = add nsw i32 %1, %I %arrayidx3 = getelementptr