<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<meta name="Generator" content="Microsoft Word 15 (filtered medium)">
<style><!--
/* Font Definitions */
@font-face
{font-family:Helvetica;
panose-1:2 11 6 4 2 2 2 2 2 4;}
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
{font-family:Kartika;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
span.EmailStyle19
{mso-style-type:personal-reply;
font-family:"Calibri",sans-serif;
color:windowtext;}
.MsoChpDefault
{mso-style-type:export-only;
font-size:10.0pt;}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]-->
</head>
<body lang="EN-US" link="blue" vlink="purple">
<div class="WordSection1">
<p class="MsoNormal">Thanks for looking into it Stefanos and Florian.<o:p></o:p></p>
<p class="MsoNormal">Truly appreciate your effort with the analysis and initiating the fix.<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">Regards,<o:p></o:p></p>
<p class="MsoNormal">CD<o:p></o:p></p>
<p class="MsoNormal"><o:p> </o:p></p>
<div>
<div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in 0in 0in">
<p class="MsoNormal"><b>From:</b> Stefanos Baziotis <stefanos.baziotis@gmail.com>
<br>
<b>Sent:</b> Monday, July 27, 2020 4:17 AM<br>
<b>To:</b> Florian Hahn <florian_hahn@apple.com><br>
<b>Cc:</b> Devadasan, Christudasan <Christudasan.Devadasan@amd.com>; LLVM Dev <llvm-dev@lists.llvm.org><br>
<b>Subject:</b> Re: [llvm-dev] [LAA] RtCheck on pointers of different address spaces.<o:p></o:p></p>
</div>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal">[CAUTION: External Email] <o:p></o:p></p>
<div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt">Hi,<br>
<br>
Florian thanks for creating the bug report. Actually, I couldn't reproduce the bug with -O3, but with <span style="color:black">-globals-aa -loop-load-elim I can. Not sure why.</span><br>
<br>
> Attached the testcase.<br>
<br>
Thanks for that!<o:p></o:p></p>
<p class="MsoNormal" style="mso-margin-top-alt:auto;mso-margin-bottom-alt:auto">> > `groupChecks()` will only try to group pointers that are on the same alias set.<o:p></o:p></p>
<p class="MsoNormal" style="mso-margin-top-alt:auto;margin-bottom:12.0pt">> If that’s true, the RT check should have been prevented for these pointers in question.<br>
<br>
Exactly, that was my point :)<br>
<br>
Now, a little info from my trying to find the source of the bug for anyone interested:<br>
<br>
In the test case we have 4 pointers as you said:<o:p></o:p></p>
<p class="MsoNormal">%tmp4 = bitcast %struct.barney* %tmp3 to i64* -- AS1<br>
%tmp12 = bitcast %struct.wombat* %tmp11 to i64* -- AS1<br>
%tmp16 = getelementptr inbounds [4000 x float], [4000 x float] addrspace(3)* @global.1, i32 0, i32 %tmp15 -- AS2<br>
%tmp = getelementptr inbounds [4000 x float], [4000 x float] addrspace(3)* @global.1, i32 0, i32 %arg -- AS2<br>
<br>
First of all, the claim above holds for this case too: groupChecks() won't try to group pointers that are<br>
not in the same alias set because they won't be in the same EqClass in DepCands. While groupChecks()<br>
here tries to group %tmp4 and %tmp16 together, this is _not because_ they're in the same EqClass<br>
but because of a bug.<br>
<br>
You see, groupChecks() starts by assigning an index to each pointer in Pointers (using PositionMap).<br>
But if you put a debug print in that loop, you'll see that %tmp16 is _not_ in Pointers. So, there's no<br>
index assigned to it.<br>
<br>
But, %tmp16 is in the same EqClass with %tmp. And %tmp _is_ in Pointers. Because of the latter,<br>
it means we'll come across it in the next loop. Which in turn means that we'll come across any pointer<br>
in %tmp's EqClass (will try to `addPointer()` it to some Group). So, at some point, we'll try to handle %tmp16.<br>
But remember, %tmp16 does not have an entry in PositionMap. So, this line:<br>
unsigned Pointer = PositionMap[MI->getPointer()];<br>
<br>
will make Pointer = 0 (you get 0 when it's not found in this DenseMap).<br>
<br>
Now, then we proceed by calling addPointer() which has this line:<br>
const SCEV *Start = RtCheck.Pointers[Index].Start;<br>
<br>
Remember, Index here is equal to Pointer previously. So, 0. But if you remember, when<br>
we assigned indexes to pointers (in the PositionMap loop), we started from 0. So, _some pointer_<br>
has an index of 0. And that happens to be %tmp4. Calling then getMinFromExprs() eventually<br>
results in the crash you saw.<br>
<br>
I'm not sure how we should solve that. I mean, it would be good if we used<br>
1-based indexing somehow and assert that we never get a 0 back when <o:p></o:p></p>
<div>
<p class="MsoNormal">we query PositionMap but that's not the root of the problem.<br>
<br>
P.S. The reason that %tmp16 doesn't make it to Pointers is because in createCheckForAccess(),<br>
which is called from canCheckPtrAtRT(), we couldn't find its bounds.<br>
<br>
Cheers,<br>
Stefanos<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
<div>
<div>
<p class="MsoNormal">Στις Δευ, 27 Ιουλ 2020 στις 12:34 π.μ., ο/η Florian Hahn <<a href="mailto:florian_hahn@apple.com">florian_hahn@apple.com</a>> έγραψε:<o:p></o:p></p>
</div>
<blockquote style="border:none;border-left:solid #CCCCCC 1.0pt;padding:0in 0in 0in 6.0pt;margin-left:4.8pt;margin-right:0in">
<div>
<p class="MsoNormal">Thanks for sharing the reproducer. I reduced it a bit and filed a bug report <a href="https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbugs.llvm.org%2Fshow_bug.cgi%3Fid%3D46854&data=02%7C01%7CChristudasan.Devadasan%40amd.com%7C3c166e3fc5d145ab590508d831b5c58e%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637314004075060708&sdata=4BJLUkH4Ez%2BmRwnk%2FZE0kWZWCcsBxczn0MgbTqFhN0E%3D&reserved=0" target="_blank">https://bugs.llvm.org/show_bug.cgi?id=46854</a>
<o:p></o:p></p>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
<div>
<p class="MsoNormal">Cheers,<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">Florian<o:p></o:p></p>
<div>
<p class="MsoNormal"><br>
<br>
<o:p></o:p></p>
<blockquote style="margin-top:5.0pt;margin-bottom:5.0pt">
<div>
<p class="MsoNormal">On Jul 26, 2020, at 18:52, Devadasan, Christudasan via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>> wrote:<o:p></o:p></p>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
<div>
<div>
<div>
<p class="MsoNormal">Hi Stefanos,<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">Attached the testcase. I tried to reduce it further, but the problem goes away when I remove the instructions further.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">There is a nested loop and the fault occurs while processing the inner loop (for.body)<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">To reproduce the crash: <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> opt -O3 testcase.ll -o out.ll<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">> `groupChecks()` will only try to group pointers that are on the same alias set.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">If that’s true, the RT check should have been prevented for these pointers in question.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"><b>Additional details about the problem:</b><o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> The crash first appeared after the following llvm commit to preserve the Global AA.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">commit e6cf796bab7e02d2b8ac7fd495f14f5e21494270<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">Author: Ryan Santhiraraja <<a href="mailto:rsanthir@quicinc.com" target="_blank">rsanthir@quicinc.com</a>><o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">Date: Thu Jul 2 13:53:20 2020 +0100<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> Preserve GlobalsAA analysis result in LowerConstantIntrinsics<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">The Global AA now finds out that these pointers refer different objects and won’t alias each other, and eventually ended up in different alias sets.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">Prior to this commit, none of the alias analysis could appropriate the noAlias property for these pointers. (If you revert this patch locally, the testcase will compile successfully).<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">The address-space validation (given below) prevented the RT check for these pointers earlier which I expected even to get triggered with the upstream compiler today.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">But it didn’t occur as they have the same DependenceSetId value. (The DependenceSetId starts from 1 for each Alias Set and hence pointers of<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">different AS can have the same Id. If it is intended, I am not sure that the early continue here, only based on the DependenceSetId, handled all cases)<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> // Only need to check pointers between two different dependency sets.<o:p></o:p></span></p>
<p style="margin-left:.5in"><b><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> </span></b><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">if (RtCheck.Pointers[i].DependencySetId ==<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> RtCheck.Pointers[j].DependencySetId)<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> continue;<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> // Only need to check pointers in the same alias set.<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> if (RtCheck.Pointers[i].AliasSetId != RtCheck.Pointers[j].AliasSetId)<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> continue;<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> <o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> Value *PtrI = RtCheck.Pointers[i].PointerValue;<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> Value *PtrJ = RtCheck.Pointers[j].PointerValue;<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> <o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> unsigned ASi = PtrI->getType()->getPointerAddressSpace();<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> unsigned ASj = PtrJ->getType()->getPointerAddressSpace();<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> if (ASi != ASj) {<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> LLVM_DEBUG(<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> dbgs() << "LAA: Runtime check would require comparison between"<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> " different address spaces\n");<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"> return false;<o:p></o:p></span></p>
<div>
<p class="MsoNormal"> }<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">Regards,<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">CD<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<div style="border:none;border-top:solid #E1E1E1 1.0pt;padding:3.0pt 0in 0in 0in">
<div>
<p class="MsoNormal"><b>From:</b> Stefanos Baziotis <<a href="mailto:stefanos.baziotis@gmail.com" target="_blank">stefanos.baziotis@gmail.com</a>> <br>
<b>Sent:</b> Sunday, July 26, 2020 6:09 PM<br>
<b>To:</b> Devadasan, Christudasan <<a href="mailto:Christudasan.Devadasan@amd.com" target="_blank">Christudasan.Devadasan@amd.com</a>><br>
<b>Cc:</b> LLVM Dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>><br>
<b>Subject:</b> Re: [llvm-dev] [LAA] RtCheck on pointers of different address spaces.<o:p></o:p></p>
</div>
</div>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">[CAUTION: External Email] <o:p></o:p></p>
</div>
<div>
<div>
<div>
<p class="MsoNormal">Hi,<br>
<br>
There are a lot of things going on here, but given this:<br>
<br>
> The crash occurs with the pointers 1 & 4 which are from AS1 and AS2 respectively.<br>
<br>
and the trace, I'm not sure how that can happen. `groupChecks()` will only try to group pointers that<br>
are on the same alias set (because it will only try to group pointers that are in the same Eq class<o:p></o:p></p>
</div>
<div>
<div>
<p class="MsoNormal">in DepCands, which if you see its construction in `processMemAccesses()`, won't put two pointers<br>
from different alias sets in the same Eq Class because _theoretically_, two such pointers can't<br>
share an underlying object).<br>
<br>
Do maybe have a simplified but complete IR ? Is that a possibility?<br>
<br>
Kind regards,<br>
Stefanos Baziotis<o:p></o:p></p>
</div>
</div>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<div>
<div>
<p class="MsoNormal">Στις Κυρ, 26 Ιουλ 2020 στις 1:06 μ.μ., ο/η Devadasan, Christudasan via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a>> έγραψε:<o:p></o:p></p>
</div>
</div>
<blockquote style="border:none;border-left:solid #CCCCCC 1.0pt;padding:0in 0in 0in 6.0pt;margin-left:4.8pt;margin-top:5.0pt;margin-right:0in;margin-bottom:5.0pt">
<div>
<div>
<div>
<p class="MsoNormal">Hello,<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">I Have a question related to the RT check on pointers during Loop Access Analysis pass.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">There is a testcase with loop code that consist of 4 different memory operations referring two global objects of different address spaces.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">One from global constant (address space 4, addr_size = 64) and the other from local, LDS (address space 3, addr_size= 32).<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">(Details of various address spaces available for AMDGPU backend: <a href="https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fllvm.org%2Fdocs%2FAMDGPUUsage.html%23address-spaces&data=02%7C01%7CChristudasan.Devadasan%40amd.com%7C3c166e3fc5d145ab590508d831b5c58e%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637314004075070704&sdata=3Fx51NaDT1WAwLXeWXR1%2FoOx%2FpTCfbVAgOh85MhU4Ks%3D&reserved=0" target="_blank">https://llvm.org/docs/AMDGPUUsage.html#address-spaces</a>)<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">With upstream compiler, the testcase fails with a crash (given at the end of the e-mail) in the opt while trying to generate the RT check for these pointers. Precisely, with two pointers of different address spaces.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">The operand type check fails while trying to insert a ‘AddExpr’ SCEV node as their effective type differs for these pointers (One with 32-bit and the other with 64-bit)<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"><b>Question: Is this intended to try for the RtCheck on pointers from different address spaces?</b><o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> The comments given in the code snippet (below) hints they aren’t.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">Code snippet from LoopAccessAnalysis.cpp:<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">-----------------------------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> bool AccessAnalysis::canCheckPtrAtRT(...) {<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> ----------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> // If the pointers that we would use for the bounds comparison have different<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> // address spaces, assume the values aren't directly comparable, so we can't<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> // use them for the runtime check. We also have to assume they could<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> // overlap. In the future there should be metadata for whether address spaces<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> // are disjoint.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> unsigned NumPointers = RtCheck.Pointers.size();<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> for (unsigned i = 0; i < NumPointers; ++i) {<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> for (unsigned j = i + 1; j < NumPointers; ++j) {<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> // Only need to check pointers between two different dependency sets.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> if (RtCheck.Pointers[i].DependencySetId ==<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> RtCheck.Pointers[j].DependencySetId)<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> continue;<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> // Only need to check pointers in the same alias set.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> if (RtCheck.Pointers[i].AliasSetId != RtCheck.Pointers[j].AliasSetId)<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> continue;<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> Value *PtrI = RtCheck.Pointers[i].PointerValue;<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> Value *PtrJ = RtCheck.Pointers[j].PointerValue;<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> unsigned ASi = PtrI->getType()->getPointerAddressSpace();<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> unsigned ASj = PtrJ->getType()->getPointerAddressSpace();<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> if (ASi != ASj) {<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> LLVM_DEBUG(<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> dbgs() << "LAA: Runtime check would require comparison between"<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> " different address spaces\n");<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> return false;<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> }<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> }<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">----------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">More details about the objects, the pointers and the memory operations:<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">----------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">%struct_var1 = type { <2 x float> }<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">%struct_var2 = type { %struct_var1 }<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">%class_var1 = type { i32, i32, i32, %struct_var2*, i32, i32, i32}<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">%class_var2 = type { %class_var1, i8, i8*, i32 }<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">Objects:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">@Obj1 = external protected local_unnamed_addr addrspace(4) externally_initialized global %class_var2, align 8<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">@Obj2 = internal unnamed_addr addrspace(3) constant [4000 x float] undef, align 16<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">Pointers:<o:p></o:p></p>
</div>
<p style="margin-left:1.0in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">1.</span><span style="font-size:7.0pt;font-family:"Times New Roman",serif"> </span><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">%struct_var1.cast
= bitcast %struct_var1* %struct_var2.gep to i64* (write) // AS1<o:p></o:p></span></p>
<p style="margin-left:1.0in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">2.</span><span style="font-size:7.0pt;font-family:"Times New Roman",serif"> </span><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">%struct_var2.cast
= bitcast %struct_var2* %arrayidx74 to i64* (read-only) // AS1<o:p></o:p></span></p>
<p style="margin-left:1.0in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">3.</span><span style="font-size:7.0pt;font-family:"Times New Roman",serif"> </span><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">%arrayidx1705
= getelementptr inbounds [4000 x float], [4000 x float] addrspace(3)* @Obj2, i32 0, i32 %add125 (read-only) // AS2<o:p></o:p></span></p>
<p style="margin-left:1.0in"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">4.</span><span style="font-size:7.0pt;font-family:"Times New Roman",serif"> </span><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">%arrayidx274
= getelementptr inbounds [4000 x float], [4000 x float] addrspace(3)* @Obj2, i32 0, i32 %arg1 (read-only) // AS2<o:p></o:p></span></p>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">While the pointers 1 & 2 belong to one Alias Set (AS1), pointers 3 & 4 belong to a different set (AS2). It is because, the Global Alias Analysis found there is no alias between these two sets of pointers.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">The crash occurs with the pointers 1 & 4 which are from AS1 and AS2 respectively.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">The DependenceSetId will be reset to 1 before processing a new AS and the pointers in question, ended up having the same DependenceSetId (1), though they are from different AS.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> ----------------------------------------------------------------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> The load/store (memory operations) using these pointers in the loop. (extracted only the relevant instructions):<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> ----------------------------------------------------------------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> define protected amdgpu_kernel void @test_func(i32 %arg1, i32 %arg2) {<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> entry:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <b>%arrayidx274 = getelementptr inbounds [4000 x float], [4000 x float] addrspace(3)* @Obj2, i32 0, i32 %arg1</b><o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"><b> </b>---------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> br i1 %cmp1, label %header, label %for.end<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">header:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> % <b>struct_var2.ld</b> = load %struct_var2*, %struct_var2* addrspace(4)* getelementptr inbounds (%class_var2, %class_var2 addrspace(4)* @Obj1, i64 0, i32 0, i32 3), align 8<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> %struct_var2.gep = getelementptr inbounds %struct_var2, %struct_var2* %struct_var2.ld, i64 undef, i32 0<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"><b> %struct_var1.cast = bitcast %struct_var1* %struct_var2.gep to i64*</b><o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> br label %for.body<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">for.body:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> --------- <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> %arrayidx74 = getelementptr inbounds %struct_var2, %struct_var2* %struct_var2.ld, i64 %idxprom73<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"><b> %struct_var2.cast = bitcast %struct_var2* %arrayidx74 to i64*</b><o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"><b> </b>%for.body.ld = load i64, i64* %<b>struct_var2.cast</b>, align 8<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> ---------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> br i1 %cmp2, label %if.then, label %if.end<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">if.then:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> %rem = srem i32 1, %arg2<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> %add125 = add nuw nsw i32 %rem, 1<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"><b> %arrayidx1705 = getelementptr inbounds [4000 x float], [4000 x float] addrspace(3)* @Obj2, i32 0, i32 %add125</b><o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> %arrayidx1705.ld = load float, float addrspace(3)* %<b>arrayidx1705</b>, align 4<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> %arrayidx274.ld = load float, float addrspace(3)* %<b>arrayidx274</b>, align 4<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> ---------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> br label %if.end<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">if.end:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> store i64 %for.body.ld, i64* %<b>struct_var1.cast</b>, align 8<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> ---------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> br i1 %cmp3, label %for.body, label %for.end<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">for.end:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> br exit<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">exit:<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> }<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">----------------------------------------------------------------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">The actual crash and the back-trace:<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">----------------------------------------------------------------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">opt: $SRC/llvm-project/llvm/lib/Analysis/ScalarEvolution.cpp:2165: const llvm::SCEV *llvm::ScalarEvolution::getAddExpr(SmallVectorImpl<const llvm::SCEV *> &, SCEV::NoWrapFlags, unsigned int): Assertion `getEffectiveSCEVType(Ops[i]->getType())
== ETy && "SCEVAddExpr operand types don't match!"' failed.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">PLEASE submit a bug report to <a href="https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbugs.llvm.org%2F&data=02%7C01%7CChristudasan.Devadasan%40amd.com%7C3c166e3fc5d145ab590508d831b5c58e%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637314004075070704&sdata=zkdapYGZIcF9y9LZm2wEZi0Dnge8dzHTb6qjjU%2BEaFQ%3D&reserved=0" target="_blank">https://bugs.llvm.org/</a> and
include the crash backtrace.<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">Stack dump:<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">0. Program arguments: $Tools/bin/opt -O3 test.ll -o out.ll<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::ScalarEvolution::getAddExpr(llvm::SmallVectorImpl<llvm::SCEV const*>&, llvm::SCEV::NoWrapFlags, unsigned int) $SRC/llvm-project/llvm/lib/Analysis/ScalarEvolution.cpp:2164:5<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::ScalarEvolution::getAddExpr(llvm::SCEV const*, llvm::SCEV const*, llvm::SCEV::NoWrapFlags, unsigned int) $SRC/llvm-project/llvm/include/llvm/Analysis/ScalarEvolution.h:526:3<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::ScalarEvolution::getMinusSCEV(llvm::SCEV const*, llvm::SCEV const*, llvm::SCEV::NoWrapFlags, unsigned int) $SRC/llvm-project/llvm/lib/Analysis/ScalarEvolution.cpp:3833:3<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">getMinFromExprs(llvm::SCEV const*, llvm::SCEV const*, llvm::ScalarEvolution*) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:275:15<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::RuntimeCheckingPtrGroup::addPointer(unsigned int) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:292:15<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::RuntimePointerChecking::groupChecks(llvm::EquivalenceClasses<llvm::PointerIntPair<llvm::Value*, 1u, bool, llvm::PointerLikeTypeTraits<llvm::Value*>, llvm::PointerIntPairInfo<llvm::Value*, 1u, llvm::PointerLikeTypeTraits<llvm::Value*>
> > >&, bool) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:413:13<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::RuntimePointerChecking::generateChecks(llvm::EquivalenceClasses<llvm::PointerIntPair<llvm::Value*, 1u, bool, llvm::PointerLikeTypeTraits<llvm::Value*>, llvm::PointerIntPairInfo<llvm::Value*, 1u, llvm::PointerLikeTypeTraits<llvm::Value*>
> > >&, bool) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:259:12<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">(anonymous namespace)::AccessAnalysis::canCheckPtrAtRT(llvm::RuntimePointerChecking&, llvm::ScalarEvolution*, llvm::Loop*, llvm::DenseMap<llvm::Value const*, llvm::Value*, llvm::DenseMapInfo<llvm::Value const*>, llvm::detail::DenseMapPair<llvm::Value
const*, llvm::Value*> > const&, bool) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:830:3<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::LoopAccessInfo::analyzeLoop(llvm::AAResults*, llvm::LoopInfo*, llvm::TargetLibraryInfo const*, llvm::DominatorTree*) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:2038:8<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::LoopAccessInfo::LoopAccessInfo(llvm::Loop*, llvm::ScalarEvolution*, llvm::TargetLibraryInfo const*, llvm::AAResults*, llvm::DominatorTree*, llvm::LoopInfo*) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:2222:1<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">std::_MakeUniq<llvm::LoopAccessInfo>::__single_object std::make_unique<llvm::LoopAccessInfo, llvm::Loop*&, llvm::ScalarEvolution*&, llvm::TargetLibraryInfo const*&, llvm::AAResults*&, llvm::DominatorTree*&, llvm::LoopInfo*&>(llvm::Loop*&,
llvm::ScalarEvolution*&, llvm::TargetLibraryInfo const*&, llvm::AAResults*&, llvm::DominatorTree*&, llvm::LoopInfo*&) /usr/bin/../lib/gcc/x86_64-linux-gnu/7.5.0/../../../../include/c++/7.5.0/bits/unique_ptr.h:821:34<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::LoopAccessLegacyAnalysis::getInfo(llvm::Loop*) $SRC/llvm-project/llvm/lib/Analysis/LoopAccessAnalysis.cpp:2275:5<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">(anonymous namespace)::LoopLoadElimination::runOnFunction(llvm::Function&)::'lambda'(llvm::Loop&)::operator()(llvm::Loop&) const $SRC/llvm-project/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp:651:53<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::LoopAccessInfo const& llvm::function_ref<llvm::LoopAccessInfo const& (llvm::Loop&)>::callback_fn<(anonymous namespace)::LoopLoadElimination::runOnFunction(llvm::Function&)::'lambda'(llvm::Loop&)>(long, llvm::Loop&) $SRC/llvm-project/llvm/include/llvm/ADT/STLExtras.h:185:5<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::function_ref<llvm::LoopAccessInfo const& (llvm::Loop&)>::operator()(llvm::Loop&) const $SRC/llvm-project/llvm/include/llvm/ADT/STLExtras.h:203:5<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">eliminateLoadsAcrossLoops(llvm::Function&, llvm::LoopInfo&, llvm::DominatorTree&, llvm::BlockFrequencyInfo*, llvm::ProfileSummaryInfo*, llvm::function_ref<llvm::LoopAccessInfo const& (llvm::Loop&)>) $SRC/llvm-project/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp:618:53<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">(anonymous namespace)::LoopLoadElimination::runOnFunction(llvm::Function&) $SRC/llvm-project/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp:649:5<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::FPPassManager::runOnFunction(llvm::Function&) $SRC/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1587:23<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::FPPassManager::runOnModule(llvm::Module&) $SRC/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1629:16<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">(anonymous namespace)::MPPassManager::runOnModule(llvm::Module&) $SRC/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1698:23<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::legacy::PassManagerImpl::run(llvm::Module&) $SRC/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:614:16<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">llvm::legacy::PassManager::run(llvm::Module&) $SRC/llvm-project/llvm/lib/IR/LegacyPassManager.cpp:1824:3<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">main $SRC/llvm-project/llvm/tools/opt/opt.cpp:955:3<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">__libc_start_main /build/glibc-2ORdQG/glibc-2.27/csu/../csu/libc-start.c:344:0<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">_start ($Tools/bin/opt+0xc98b6a)<o:p></o:p></p>
</div>
<div style="margin-left:.5in">
<p class="MsoNormal">Aborted (core dumped)<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">-----------------------------------------------------------------------------------------------------------------------------------------------------------------<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">Regards,<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">CD<o:p></o:p></p>
</div>
</div>
</div>
<div>
<p class="MsoNormal">_______________________________________________<br>
LLVM Developers mailing list<br>
<a href="mailto:llvm-dev@lists.llvm.org" target="_blank">llvm-dev@lists.llvm.org</a><br>
<a href="https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.llvm.org%2Fcgi-bin%2Fmailman%2Flistinfo%2Fllvm-dev&data=02%7C01%7CChristudasan.Devadasan%40amd.com%7C3c166e3fc5d145ab590508d831b5c58e%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637314004075080700&sdata=9XGk7h3B%2BfEjfSaAzl9CxKDx%2FD47EjvK2CnvvWQ7wJk%3D&reserved=0" target="_blank">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><o:p></o:p></p>
</div>
</blockquote>
</div>
</div>
</div>
<p class="MsoNormal"><testcase.ll><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">_______________________________________________<br>
LLVM Developers mailing list<br>
</span><a href="mailto:llvm-dev@lists.llvm.org" target="_blank"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">llvm-dev@lists.llvm.org</span></a><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif"><br>
</span><a href="https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.llvm.org%2Fcgi-bin%2Fmailman%2Flistinfo%2Fllvm-dev&data=02%7C01%7CChristudasan.Devadasan%40amd.com%7C3c166e3fc5d145ab590508d831b5c58e%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637314004075090693&sdata=S23M8X7lB5Q0o9lZnwAtTzIsKdcyZK%2BGCZz6xpfmxyg%3D&reserved=0" target="_blank"><span style="font-size:9.0pt;font-family:"Helvetica",sans-serif">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</span></a><o:p></o:p></p>
</div>
</blockquote>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
</div>
</blockquote>
</div>
</div>
</div>
</body>
</html>