<div dir="ltr">The new run line is what you need.<div><br></div><div>opt < %s -slp-schedule-budget=0 -slp-min-tree-size=0 -slp-threshold=-30 -slp-vectorizer -S<br></div><div><br></div><div><span style="font-size:12.8px">r279369 added slp-min-tree-size as a command line option. The other flag changes here ensure that we still vectorize the test case. It was previously only being vectorized by chance due to undefined behavior in the cost calculation.</span><br></div><div><span style="font-size:12.8px"><br></span></div><div><span style="font-size:12.8px">-- Matt</span></div><div><span style="font-size:12.8px"><br></span></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Aug 22, 2016 at 4:58 PM, Hans Wennborg via llvm-commits <span dir="ltr"><<a href="mailto:llvm-commits@lists.llvm.org" target="_blank">llvm-commits@lists.llvm.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">No, I didn't realize that the flag change was necessary..<br>
<br>
The old test had this run line:<br>
<span class="">opt < %s -slp-recursion-max-depth=0 -slp-vectorizer -S<br>
<br>
</span>The new one has:<br>
opt < %s -slp-schedule-budget=0 -slp-min-tree-size=0<br>
-slp-threshold=-30 -slp-vectorizer -S<br>
<br>
So it seems -slp-recursion-max-depth was removed and<br>
-slp-schedule-budget=0 -slp-min-tree-size=0 -slp-threshold=-30 were<br>
added.<br>
<br>
Which ones do I need for it to test the right thing on the branch and<br>
not run into the UB?<br>
<br>
Thanks,<br>
Hans<br>
<br>
<br>
On Mon, Aug 22, 2016 at 1:48 PM, Matthew Simpson<br>
<div class="HOEnZb"><div class="h5"><<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>> wrote:<br>
> Thanks Hans,<br>
><br>
> Did you also get r279369? It's needed by the test case change.<br>
><br>
> --Matt<br>
><br>
> -----Original Message-----<br>
> From: <a href="mailto:hwennborg@google.com">hwennborg@google.com</a> [mailto:<a href="mailto:hwennborg@google.com">hwennborg@google.com</a>] On Behalf Of Hans Wennborg<br>
> Sent: Monday, August 22, 2016 4:36 PM<br>
> To: Matthew Simpson <<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>><br>
> Cc: Michael Kuperstein <<a href="mailto:mkuper@google.com">mkuper@google.com</a>>; Kostya Serebryany <<a href="mailto:kcc@google.com">kcc@google.com</a>>; LLVM Commits <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>><br>
> Subject: Re: [llvm] r279125 - [SLP] Initialize VectorizedValue when gathering<br>
><br>
> Since the UB was already there, it's not something we're regressing and so I won't block 3.9 waiting on a fix for that.<br>
><br>
> I have committed Matt's modification of the test from r279370 to 3.9 in r279468.<br>
><br>
> Thanks,<br>
> Hans<br>
><br>
> On Fri, Aug 19, 2016 at 2:54 PM, Matthew Simpson <<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>> wrote:<br>
>> I'm also fine reverting the original patch in the meantime so the ubsan bot can go back to normal. Let me know if there's a preference here. Thanks!<br>
>><br>
>> -- Matt<br>
>><br>
>> -----Original Message-----<br>
>> From: <a href="mailto:hwennborg@google.com">hwennborg@google.com</a> [mailto:<a href="mailto:hwennborg@google.com">hwennborg@google.com</a>] On Behalf Of<br>
>> Hans Wennborg<br>
>> Sent: Friday, August 19, 2016 5:17 PM<br>
>> To: Michael Kuperstein <<a href="mailto:mkuper@google.com">mkuper@google.com</a>><br>
>> Cc: Matthew Simpson <<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>>; LLVM Commits<br>
>> <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>><br>
>> Subject: Re: [llvm] r279125 - [SLP] Initialize VectorizedValue when<br>
>> gathering<br>
>><br>
>> I probably won't treat it as a blocker, but if it's practical to pick up the fix (e.g. if we have to do a third release candidate anyway), I'll take it.<br>
>><br>
>> Thanks,<br>
>> Hans<br>
>><br>
>> On Fri, Aug 19, 2016 at 2:08 PM, Michael Kuperstein <<a href="mailto:mkuper@google.com">mkuper@google.com</a>> wrote:<br>
>>> Regardless of the fix, I don't think this should block 3.9 - the UB<br>
>>> was not introduced by this commit, and is not a regression. The only<br>
>>> thing that changed is that now we have an in-tree test that actually exercises the UB.<br>
>>><br>
>>> Hans, what do you think?<br>
>>><br>
>>><br>
>>> On Fri, Aug 19, 2016 at 12:04 PM, Matthew Simpson<br>
>>> <<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>><br>
>>> wrote:<br>
>>>><br>
>>>> I've submitted a fix for review here:<br>
>>>> <a href="https://reviews.llvm.org/D23723" rel="noreferrer" target="_blank">https://reviews.llvm.org/<wbr>D23723</a><br>
>>>><br>
>>>><br>
>>>><br>
>>>> -- Matt<br>
>>>><br>
>>>><br>
>>>><br>
>>>> From: Matthew Simpson [mailto:<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.<wbr>org</a>]<br>
>>>> Sent: Friday, August 19, 2016 1:30 PM<br>
>>>> To: 'Michael Kuperstein' <<a href="mailto:mkuper@google.com">mkuper@google.com</a>><br>
>>>> Cc: 'LLVM Commits' <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>><br>
>>>> Subject: RE: [llvm] r279125 - [SLP] Initialize VectorizedValue when<br>
>>>> gathering<br>
>>>><br>
>>>><br>
>>>><br>
>>>> No, we don’t. Performing the add in int64_t should work, though.<br>
>>>> I’ll also probably need to update the test case since we’ll have a different cost.<br>
>>>><br>
>>>><br>
>>>><br>
>>>> From: Michael Kuperstein [mailto:<a href="mailto:mkuper@google.com">mkuper@google.com</a>]<br>
>>>> Sent: Friday, August 19, 2016 1:20 PM<br>
>>>><br>
>>>><br>
>>>> To: Matthew Simpson <<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>><br>
>>>> Cc: LLVM Commits <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>><br>
>>>> Subject: Re: [llvm] r279125 - [SLP] Initialize VectorizedValue when<br>
>>>> gathering<br>
>>>><br>
>>>><br>
>>>><br>
>>>> I don't think we have a SaturatingAdd for signed types.<br>
>>>><br>
>>>><br>
>>>><br>
>>>> On Fri, Aug 19, 2016 at 10:05 AM, Matthew Simpson<br>
>>>> <<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>> wrote:<br>
>>>><br>
>>>> That's right. Can we just change this to a SaturatingAdd, then?<br>
>>>><br>
>>>><br>
>>>><br>
>>>> -- Matt<br>
>>>><br>
>>>><br>
>>>><br>
>>>> On Fri, Aug 19, 2016 at 12:58 PM, Michael Kuperstein via<br>
>>>> llvm-commits <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>> wrote:<br>
>>>><br>
>>>> getTreeCost() can return INT_MAX, and tryToReduce() adds<br>
>>>> getReductionCost() to that.<br>
>>>><br>
>>>> Looks like the UB has been there for a while, the new test just<br>
>>>> exposes it.<br>
>>>><br>
>>>><br>
>>>><br>
>>>> On Fri, Aug 19, 2016 at 9:53 AM, Matthew Simpson via llvm-commits<br>
>>>> <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>> wrote:<br>
>>>><br>
>>>> Sorry about that, I’ll take a look!<br>
>>>><br>
>>>><br>
>>>><br>
>>>> -- Matt<br>
>>>><br>
>>>><br>
>>>><br>
>>>> From: Kostya Serebryany [mailto:<a href="mailto:kcc@google.com">kcc@google.com</a>]<br>
>>>> Sent: Friday, August 19, 2016 12:43 PM<br>
>>>> To: Matthew Simpson <<a href="mailto:mssimpso@codeaurora.org">mssimpso@codeaurora.org</a>><br>
>>>> Cc: LLVM Commits <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>><br>
>>>> Subject: Re: [llvm] r279125 - [SLP] Initialize VectorizedValue when<br>
>>>> gathering<br>
>>>><br>
>>>><br>
>>>><br>
>>>> I think this is causing the ubsan bot failure. please fix:<br>
>>>><br>
>>>> lib/Transforms/Vectorize/<wbr>SLPVectorizer.cpp:4177:34: runtime error:<br>
>>>> signed integer overflow: <a href="tel:2147483647" value="+12147483647">2147483647</a> + 11 cannot be represented in type 'int'<br>
>>>><br>
>>>><br>
>>>> <a href="http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds" rel="noreferrer" target="_blank">http://lab.llvm.org:8011/<wbr>builders/sanitizer-x86_64-<wbr>linux-fast/builds</a><br>
>>>> / 15655/steps/check-llvm%<wbr>20ubsan/logs/stdio<br>
>>>><br>
>>>><br>
>>>><br>
>>>> On Thu, Aug 18, 2016 at 12:50 PM, Matthew Simpson via llvm-commits<br>
>>>> <<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a>> wrote:<br>
>>>><br>
>>>> Author: mssimpso<br>
>>>> Date: Thu Aug 18 14:50:32 2016<br>
>>>> New Revision: 279125<br>
>>>><br>
>>>> URL: <a href="http://llvm.org/viewvc/llvm-project?rev=279125&view=rev" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-<wbr>project?rev=279125&view=rev</a><br>
>>>> Log:<br>
>>>> [SLP] Initialize VectorizedValue when gathering<br>
>>>><br>
>>>> We abort building vectorizable trees in some cases (e.g., if the<br>
>>>> maximum recursion depth is reached, if the region size is too large,<br>
>>>> etc.). If this happens for a reduction, we can be left with a root<br>
>>>> entry that needs to be gathered. For these cases, we need make sure<br>
>>>> we actually set VectorizedValue to the resulting vector.<br>
>>>><br>
>>>> This patch ensures we properly set VectorizedValue, and it also<br>
>>>> ensures the insertelement sequence generated for the gathers is<br>
>>>> inserted at the correct location.<br>
>>>><br>
>>>> Reference: <a href="https://llvm.org/bugs/show_bug.cgi?id=28330" rel="noreferrer" target="_blank">https://llvm.org/bugs/show_<wbr>bug.cgi?id=28330</a><br>
>>>> Differential Revison: <a href="https://reviews.llvm.org/D23410" rel="noreferrer" target="_blank">https://reviews.llvm.org/<wbr>D23410</a><br>
>>>><br>
>>>> Added:<br>
>>>> llvm/trunk/test/Transforms/<wbr>SLPVectorizer/AArch64/gather-<wbr>root.ll<br>
>>>> Modified:<br>
>>>> llvm/trunk/lib/Transforms/<wbr>Vectorize/SLPVectorizer.cpp<br>
>>>><br>
>>>> Modified: llvm/trunk/lib/Transforms/<wbr>Vectorize/SLPVectorizer.cpp<br>
>>>> URL:<br>
>>>> <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Vector" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-<wbr>project/llvm/trunk/lib/<wbr>Transforms/Vector</a><br>
>>>> i ze/SLPVectorizer.cpp?rev=<wbr>279125&r1=279124&r2=279125&<wbr>view=diff<br>
>>>><br>
>>>> ==============================<wbr>==============================<wbr>========<br>
>>>> =<br>
>>>> =========<br>
>>>> --- llvm/trunk/lib/Transforms/<wbr>Vectorize/SLPVectorizer.cpp (original)<br>
>>>> +++ llvm/trunk/lib/Transforms/<wbr>Vectorize/SLPVectorizer.cpp Thu Aug 18<br>
>>>> 14:50:32 2016<br>
>>>> @@ -2153,11 +2153,61 @@ void BoUpSLP::<wbr>reorderInputsAccordingToOp<br>
>>>> }<br>
>>>><br>
>>>> void BoUpSLP::<wbr>setInsertPointAfterBundle(<wbr>ArrayRef<Value *> VL) {<br>
>>>> - Instruction *VL0 = cast<Instruction>(VL[0]);<br>
>>>> - BasicBlock::iterator NextInst(VL0);<br>
>>>> - ++NextInst;<br>
>>>> - Builder.SetInsertPoint(VL0-><wbr>getParent(), NextInst);<br>
>>>> - Builder.<wbr>SetCurrentDebugLocation(VL0-><wbr>getDebugLoc());<br>
>>>> +<br>
>>>> + // Get the basic block this bundle is in. All instructions in the<br>
>>>> bundle<br>
>>>> + // should be in this block.<br>
>>>> + auto *Front = cast<Instruction>(VL.front()); auto *BB =<br>
>>>> + Front->getParent(); assert(all_of(make_range(VL.<wbr>begin(),<br>
>>>> + VL.end()), [&](Value *V) -> bool {<br>
>>>> + return cast<Instruction>(V)-><wbr>getParent() == BB; }));<br>
>>>> +<br>
>>>> + // The last instruction in the bundle in program order.<br>
>>>> + Instruction *LastInst = nullptr;<br>
>>>> +<br>
>>>> + // Find the last instruction. The common case should be that BB<br>
>>>> + has<br>
>>>> been<br>
>>>> + // scheduled, and the last instruction is VL.back(). So we start<br>
>>>> + with // VL.back() and iterate over schedule data until we reach<br>
>>>> + the end of<br>
>>>> the<br>
>>>> + // bundle. The end of the bundle is marked by null ScheduleData.<br>
>>>> + if (BlocksSchedules.count(BB)) {<br>
>>>> + auto *Bundle = BlocksSchedules[BB]-><wbr>getScheduleData(VL.back());<br>
>>>> + if (Bundle && Bundle->isPartOfBundle())<br>
>>>> + for (; Bundle; Bundle = Bundle->NextInBundle)<br>
>>>> + LastInst = Bundle->Inst;<br>
>>>> + }<br>
>>>> +<br>
>>>> + // LastInst can still be null at this point if there's either not<br>
>>>> + an<br>
>>>> entry<br>
>>>> + // for BB in BlocksSchedules or there's no ScheduleData available<br>
>>>> + for // VL.back(). This can be the case if buildTree_rec aborts<br>
>>>> + for various // reasons (e.g., the maximum recursion depth is<br>
>>>> + reached, the maximum<br>
>>>> region<br>
>>>> + // size is reached, etc.). ScheduleData is initialized in the<br>
>>>> scheduling<br>
>>>> + // "dry-run".<br>
>>>> + //<br>
>>>> + // If this happens, we can still find the last instruction by<br>
>>>> + brute<br>
>>>> force. We<br>
>>>> + // iterate forwards from Front (inclusive) until we either see<br>
>>>> + all // instructions in the bundle or reach the end of the block.<br>
>>>> + If Front<br>
>>>> is the<br>
>>>> + // last instruction in program order, LastInst will be set to<br>
>>>> + Front,<br>
>>>> and we<br>
>>>> + // will visit all the remaining instructions in the block.<br>
>>>> + //<br>
>>>> + // One of the reasons we exit early from buildTree_rec is to<br>
>>>> + place an<br>
>>>> upper<br>
>>>> + // bound on compile-time. Thus, taking an additional compile-time<br>
>>>> + hit<br>
>>>> here is<br>
>>>> + // not ideal. However, this should be exceedingly rare since it<br>
>>>> requires that<br>
>>>> + // we both exit early from buildTree_rec and that the bundle be<br>
>>>> out-of-order<br>
>>>> + // (causing us to iterate all the way to the end of the block).<br>
>>>> + if (!LastInst) {<br>
>>>> + SmallPtrSet<Value *, 16> Bundle(VL.begin(), VL.end());<br>
>>>> + for (auto &I : make_range(BasicBlock::<wbr>iterator(Front), BB->end())) {<br>
>>>> + if (Bundle.erase(&I))<br>
>>>> + LastInst = &I;<br>
>>>> + if (Bundle.empty())<br>
>>>> + break;<br>
>>>> + }<br>
>>>> + }<br>
>>>> +<br>
>>>> + // Set the insertion point after the last instruction in the bundle.<br>
>>>> Set the<br>
>>>> + // debug location to Front.<br>
>>>> + Builder.SetInsertPoint(BB, next(BasicBlock::iterator(<wbr>LastInst)));<br>
>>>> + Builder.<wbr>SetCurrentDebugLocation(Front-<wbr>>getDebugLoc());<br>
>>>> }<br>
>>>><br>
>>>> Value *BoUpSLP::Gather(ArrayRef<<wbr>Value *> VL, VectorType *Ty) { @@<br>
>>>> -2235,7 +2285,9 @@ Value *BoUpSLP::vectorizeTree(<wbr>TreeEntry<br>
>>>><br>
>>>> if (E->NeedToGather) {<br>
>>>> setInsertPointAfterBundle(E-><wbr>Scalars);<br>
>>>> - return Gather(E->Scalars, VecTy);<br>
>>>> + auto *V = Gather(E->Scalars, VecTy);<br>
>>>> + E->VectorizedValue = V;<br>
>>>> + return V;<br>
>>>> }<br>
>>>><br>
>>>> unsigned Opcode = getSameOpcode(E->Scalars); @@ -2282,7 +2334,10<br>
>>>> @@ Value *BoUpSLP::vectorizeTree(<wbr>TreeEntry<br>
>>>> E->VectorizedValue = V;<br>
>>>> return V;<br>
>>>> }<br>
>>>> - return Gather(E->Scalars, VecTy);<br>
>>>> + setInsertPointAfterBundle(E-><wbr>Scalars);<br>
>>>> + auto *V = Gather(E->Scalars, VecTy);<br>
>>>> + E->VectorizedValue = V;<br>
>>>> + return V;<br>
>>>> }<br>
>>>> case Instruction::ExtractValue: {<br>
>>>> if (canReuseExtract(E->Scalars, Instruction::ExtractValue)) {<br>
>>>> @@ -2294,7 +2349,10 @@ Value *BoUpSLP::vectorizeTree(<wbr>TreeEntry<br>
>>>> E->VectorizedValue = V;<br>
>>>> return propagateMetadata(V, E->Scalars);<br>
>>>> }<br>
>>>> - return Gather(E->Scalars, VecTy);<br>
>>>> + setInsertPointAfterBundle(E-><wbr>Scalars);<br>
>>>> + auto *V = Gather(E->Scalars, VecTy);<br>
>>>> + E->VectorizedValue = V;<br>
>>>> + return V;<br>
>>>> }<br>
>>>> case Instruction::ZExt:<br>
>>>> case Instruction::SExt:<br>
>>>><br>
>>>> Added:<br>
>>>> llvm/trunk/test/Transforms/<wbr>SLPVectorizer/AArch64/gather-<wbr>root.ll<br>
>>>> URL:<br>
>>>> <a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/SLPVe" rel="noreferrer" target="_blank">http://llvm.org/viewvc/llvm-<wbr>project/llvm/trunk/test/<wbr>Transforms/SLPVe</a><br>
>>>> c torizer/AArch64/gather-root.<wbr>ll?rev=279125&view=auto<br>
>>>><br>
>>>> ==============================<wbr>==============================<wbr>========<br>
>>>> =<br>
>>>> =========<br>
>>>> --- llvm/trunk/test/Transforms/<wbr>SLPVectorizer/AArch64/gather-<wbr>root.ll<br>
>>>> (added)<br>
>>>> +++ llvm/trunk/test/Transforms/<wbr>SLPVectorizer/AArch64/gather-<wbr>root.ll<br>
>>>> +++ Thu<br>
>>>> Aug 18 14:50:32 2016<br>
>>>> @@ -0,0 +1,95 @@<br>
>>>> +; RUN: opt < %s -slp-vectorizer -S | FileCheck %s<br>
>>>> +--check-prefix=DEFAULT ; RUN: opt < %s -slp-recursion-max-depth=0<br>
>>>> +-slp-vectorizer -S | FileCheck<br>
>>>> %s --check-prefix=GATHER<br>
>>>> +<br>
>>>> +target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:<wbr>64-i128:128-n32:64-S128"<br>
>>>> +target triple = "aarch64--linux-gnu"<br>
>>>> +<br>
>>>> +@a = common global [80 x i8] zeroinitializer, align 16<br>
>>>> +<br>
>>>> +; DEFAULT-LABEL: @PR28330(<br>
>>>> +; DEFAULT: %tmp17 = phi i32 [ %tmp34, %for.body ], [ 0, %entry ] ;<br>
>>>> +DEFAULT: %tmp18 = phi i32 [ %tmp35, %for.body ], [ %n, %entry ] ;<br>
>>>> +DEFAULT: %[[S0:.+]] = select <8 x i1> %1, <8 x i32> <i32 -720, i32<br>
>>>> -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720, i32 -720>,<br>
>>>> <8 x i32><br>
>>>> <i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32 -80, i32<br>
>>>> -80><br>
>>>> +; DEFAULT: %[[R0:.+]] = shufflevector <8 x i32> %[[S0]], <8 x i32><br>
>>>> +undef,<br>
>>>> <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32<br>
>>>> undef, i32<br>
>>>> undef><br>
>>>> +; DEFAULT: %[[R1:.+]] = add <8 x i32> %[[S0]], %[[R0]] ; DEFAULT:<br>
>>>> +%[[R2:.+]] = shufflevector <8 x i32> %[[R1]], <8 x i32> undef,<br>
>>>> <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef,<br>
>>>> i32 undef, i32 undef><br>
>>>> +; DEFAULT: %[[R3:.+]] = add <8 x i32> %[[R1]], %[[R2]] ; DEFAULT:<br>
>>>> +%[[R4:.+]] = shufflevector <8 x i32> %[[R3]], <8 x i32> undef,<br>
>>>> <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32<br>
>>>> undef, i32 undef, i32 undef><br>
>>>> +; DEFAULT: %[[R5:.+]] = add <8 x i32> %[[R3]], %[[R4]] ; DEFAULT:<br>
>>>> +%[[R6:.+]] = extractelement <8 x i32> %[[R5]], i32 0 ; DEFAULT:<br>
>>>> +%tmp34 = add i32 %[[R6]], %tmp17 ; ; GATHER-LABEL: @PR28330( ;<br>
>>>> +GATHER: %tmp17 = phi i32 [ %tmp34, %for.body ], [ 0, %entry ] ;<br>
>>>> +GATHER: %tmp18 = phi i32 [ %tmp35, %for.body ], [ %n, %entry ] ;<br>
>>>> +GATHER: %tmp19 = select i1 %tmp1, i32 -720, i32 -80 ; GATHER:<br>
>>>> +%tmp21 = select i1 %tmp3, i32 -720, i32 -80 ; GATHER: %tmp23 =<br>
>>>> +select i1 %tmp5, i32 -720, i32 -80 ; GATHER: %tmp25 = select i1<br>
>>>> +%tmp7, i32 -720, i32 -80 ; GATHER: %tmp27 = select i1 %tmp9, i32<br>
>>>> +-720, i32 -80 ; GATHER: %tmp29 = select i1 %tmp11, i32 -720, i32 -80 ; GATHER:<br>
>>>> +%tmp31 = select i1 %tmp13, i32 -720, i32 -80 ; GATHER: %tmp33 =<br>
>>>> +select i1 %tmp15, i32 -720, i32 -80 ; GATHER: %[[I0:.+]] =<br>
>>>> +insertelement <8 x i32> undef, i32 %tmp19, i32 0 ; GATHER:<br>
>>>> +%[[I1:.+]] = insertelement <8 x i32> %[[I0]], i32 %tmp21, i32 1 ;<br>
>>>> +GATHER: %[[I2:.+]] = insertelement <8 x i32> %[[I1]], i32 %tmp23,<br>
>>>> +i32 2 ; GATHER: %[[I3:.+]] = insertelement <8 x i32> %[[I2]], i32<br>
>>>> +%tmp25, i32 3 ; GATHER: %[[I4:.+]] = insertelement <8 x i32><br>
>>>> +%[[I3]], i32 %tmp27, i32 4 ; GATHER: %[[I5:.+]] = insertelement <8<br>
>>>> +x<br>
>>>> +i32> %[[I4]], i32 %tmp29, i32 5 ; GATHER: %[[I6:.+]] =<br>
>>>> +i32> insertelement<br>
>>>> +<8 x i32> %[[I5]], i32 %tmp31, i32 6 ; GATHER: %[[I7:.+]] =<br>
>>>> +insertelement <8 x i32> %[[I6]], i32 %tmp33, i32 7 ; GATHER:<br>
>>>> +%[[R0:.+]] = shufflevector <8 x i32> %[[I7]], <8 x i32> undef,<br>
>>>> <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32<br>
>>>> undef, i32<br>
>>>> undef><br>
>>>> +; GATHER: %[[R1:.+]] = add <8 x i32> %[[I7]], %[[R0]] ; GATHER:<br>
>>>> +%[[R2:.+]] = shufflevector <8 x i32> %[[R1]], <8 x i32> undef,<br>
>>>> <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef,<br>
>>>> i32 undef, i32 undef><br>
>>>> +; GATHER: %[[R3:.+]] = add <8 x i32> %[[R1]], %[[R2]] ; GATHER:<br>
>>>> +%[[R4:.+]] = shufflevector <8 x i32> %[[R3]], <8 x i32> undef,<br>
>>>> <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32<br>
>>>> undef, i32 undef, i32 undef><br>
>>>> +; GATHER: %[[R5:.+]] = add <8 x i32> %[[R3]], %[[R4]] ; GATHER:<br>
>>>> +%[[R6:.+]] = extractelement <8 x i32> %[[R5]], i32 0 ; GATHER:<br>
>>>> +%tmp34 = add i32 %[[R6]], %tmp17<br>
>>>> +<br>
>>>> +define void @PR28330(i32 %n) {<br>
>>>> +entry:<br>
>>>> + %tmp0 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> +i8]* @a,<br>
>>>> i64 0, i64 1), align 1<br>
>>>> + %tmp1 = icmp eq i8 %tmp0, 0<br>
>>>> + %tmp2 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> + i8]* @a,<br>
>>>> i64 0, i64 2), align 2<br>
>>>> + %tmp3 = icmp eq i8 %tmp2, 0<br>
>>>> + %tmp4 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> + i8]* @a,<br>
>>>> i64 0, i64 3), align 1<br>
>>>> + %tmp5 = icmp eq i8 %tmp4, 0<br>
>>>> + %tmp6 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> + i8]* @a,<br>
>>>> i64 0, i64 4), align 4<br>
>>>> + %tmp7 = icmp eq i8 %tmp6, 0<br>
>>>> + %tmp8 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> + i8]* @a,<br>
>>>> i64 0, i64 5), align 1<br>
>>>> + %tmp9 = icmp eq i8 %tmp8, 0<br>
>>>> + %tmp10 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> + i8]* @a,<br>
>>>> i64 0, i64 6), align 2<br>
>>>> + %tmp11 = icmp eq i8 %tmp10, 0<br>
>>>> + %tmp12 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> + i8]* @a,<br>
>>>> i64 0, i64 7), align 1<br>
>>>> + %tmp13 = icmp eq i8 %tmp12, 0<br>
>>>> + %tmp14 = load i8, i8* getelementptr inbounds ([80 x i8], [80 x<br>
>>>> + i8]* @a,<br>
>>>> i64 0, i64 8), align 8<br>
>>>> + %tmp15 = icmp eq i8 %tmp14, 0<br>
>>>> + br label %for.body<br>
>>>> +<br>
>>>> +for.body:<br>
>>>> + %tmp17 = phi i32 [ %tmp34, %for.body ], [ 0, %entry ]<br>
>>>> + %tmp18 = phi i32 [ %tmp35, %for.body ], [ %n, %entry ]<br>
>>>> + %tmp19 = select i1 %tmp1, i32 -720, i32 -80<br>
>>>> + %tmp20 = add i32 %tmp17, %tmp19<br>
>>>> + %tmp21 = select i1 %tmp3, i32 -720, i32 -80<br>
>>>> + %tmp22 = add i32 %tmp20, %tmp21<br>
>>>> + %tmp23 = select i1 %tmp5, i32 -720, i32 -80<br>
>>>> + %tmp24 = add i32 %tmp22, %tmp23<br>
>>>> + %tmp25 = select i1 %tmp7, i32 -720, i32 -80<br>
>>>> + %tmp26 = add i32 %tmp24, %tmp25<br>
>>>> + %tmp27 = select i1 %tmp9, i32 -720, i32 -80<br>
>>>> + %tmp28 = add i32 %tmp26, %tmp27<br>
>>>> + %tmp29 = select i1 %tmp11, i32 -720, i32 -80<br>
>>>> + %tmp30 = add i32 %tmp28, %tmp29<br>
>>>> + %tmp31 = select i1 %tmp13, i32 -720, i32 -80<br>
>>>> + %tmp32 = add i32 %tmp30, %tmp31<br>
>>>> + %tmp33 = select i1 %tmp15, i32 -720, i32 -80<br>
>>>> + %tmp34 = add i32 %tmp32, %tmp33<br>
>>>> + %tmp35 = add nsw i32 %tmp18, -1<br>
>>>> + %tmp36 = icmp eq i32 %tmp35, 0<br>
>>>> + br i1 %tmp36, label %for.end, label %for.body<br>
>>>> +<br>
>>>> +for.end:<br>
>>>> + ret void<br>
>>>> +}<br>
>>>><br>
>>>><br>
>>>> ______________________________<wbr>_________________<br>
>>>> llvm-commits mailing list<br>
>>>> <a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a><br>
>>>> <a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/<wbr>mailman/listinfo/llvm-commits</a><br>
>>>><br>
>>>><br>
>>>><br>
>>>><br>
>>>> ______________________________<wbr>_________________<br>
>>>> llvm-commits mailing list<br>
>>>> <a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a><br>
>>>> <a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/<wbr>mailman/listinfo/llvm-commits</a><br>
>>>><br>
>>>><br>
>>>><br>
>>>><br>
>>>> ______________________________<wbr>_________________<br>
>>>> llvm-commits mailing list<br>
>>>> <a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a><br>
>>>> <a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/<wbr>mailman/listinfo/llvm-commits</a><br>
>>>><br>
>>>><br>
>>>><br>
>>>><br>
>>><br>
>>><br>
>><br>
><br>
______________________________<wbr>_________________<br>
llvm-commits mailing list<br>
<a href="mailto:llvm-commits@lists.llvm.org">llvm-commits@lists.llvm.org</a><br>
<a href="http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits" rel="noreferrer" target="_blank">http://lists.llvm.org/cgi-bin/<wbr>mailman/listinfo/llvm-commits</a><br>
</div></div></blockquote></div><br></div>