<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p><tt>Hi,</tt></p>
<p><tt>I mean that </tt><tt>it seems that this implementation
restricts the tasks that participate in the reduction to be
defined in the same lexical scope. I have not been able to find
this restriction in the OpenMP specification. What the spec says
about this topic is:</tt></p>
<tt>"A list item that appears in an in_reduction clause of a
task-generating construct must appear</tt><tt><br>
</tt><tt> in a task_reduction clause of a construct corresponding to
a taskgroup region that includes</tt><tt><br>
</tt><tt> the participating task in its taskgroup set. The construct
corresponding to the innermost region</tt><tt><br>
</tt><tt> that meets this condition must specify the same
reduction-identifier as the in_reduction</tt><tt><br>
</tt><tt> clause."<br>
<br>
I tried to compile the attached file. The compile command is:<br>
<br>
clang -fopenmp task_red.cc<br>
<br>
Regards,<br>
Raúl<br>
<br>
</tt>
<div class="moz-cite-prefix">El 08/03/18 a las 15:30, Alexey Bataev
escribió:<br>
</div>
<blockquote type="cite"
cite="mid:AM5PR0902MB1923C2CE4109CA0A68FF9BF185DF0@AM5PR0902MB1923.eurprd09.prod.outlook.com">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<p>Hi,</p>
<p>>>> <tt>Anyway, this implementation of task
reductions is not fully functional, as only are allowed if the
task is created in the scope of taskgroup.</tt><tt><br>
</tt></p>
<p><tt>What do you mean?Could you provide some more detais?<br>
</tt></p>
<p><tt><br>
</tt></p>
<pre class="moz-signature" cols="72">-------------
Best regards,
Alexey Bataev</pre>
<div class="moz-cite-prefix">08.03.2018 7:47, Raúl Peñacoba пишет:<br>
</div>
<blockquote type="cite"
cite="mid:6d491557-35fb-9d45-62be-d1b33dadb424@bsc.es">
<meta http-equiv="Content-Type" content="text/html;
charset=utf-8">
<p><tt>Hi,</tt></p>
<tt> </tt>
<p><tt>I did some little research and I have realized that the
patch I attached doesn't fix the problem completely. I fixed
the storage of the data. When a using task reduction with
vla, or a class that uses omp_orig to initialize the private
copies, the size of the vla and omp_orig are stored in a
storage using __kmpc_threadprivate_cached().</tt></p>
<tt> </tt>
<p><tt>The problem comes because, since in outline parallel
function each reduction has lazy_priv flag, the
initialization will be done in
__kmpc_task_reduction_get_th_data(), called from outline
task function. This is done </tt><tt><u>before</u></tt><tt>
storing the vla size or omp_orig, so the initialization
would be wrong in these cases.</tt></p>
<tt> </tt>
<p><tt>I think that if the storage is put before the
__kmpc_task_reduction_get_th_data() it should work.</tt></p>
<tt> </tt>
<p><tt>Anyway, this implementation of task reductions is not
fully functional, as only are allowed if the task is created
in the scope of taskgroup.</tt><tt><br>
</tt></p>
<p><tt>Sorry for the mail duplicate.<br>
</tt> </p>
<tt> </tt>
<p><tt>Regards,</tt></p>
<tt> Raúl</tt><tt><br>
</tt><tt><br>
</tt>
<div class="moz-cite-prefix"><tt>El 06/03/18 a las 16:21, Alexey
Bataev escribió:</tt><tt><br>
</tt></div>
<blockquote type="cite"
cite="mid:AM5PR0902MB192332923553F3139032AB6385D90@AM5PR0902MB1923.eurprd09.prod.outlook.com">
<p><tt>Hi, thanks for the report, I'll take a look.</tt><tt><br>
</tt> </p>
<pre class="moz-signature" cols="72">-------------
Best regards,
Alexey Bataev</pre>
<div class="moz-cite-prefix"><tt>06.03.2018 4:25, Raúl
Peñacoba via cfe-dev пишет:</tt><tt><br>
</tt> </div>
<blockquote type="cite"
cite="mid:%3C8c87489c-07b9-25bc-0b3b-66d9d295bf8f@bsc.es%3E"><tt>Hi,
</tt><tt><br>
</tt> <tt><br>
</tt><tt> I've noticed that clang does not work properly
when using task reductions with non-constant arrays and/or
classes with omp_orig constructor initializers. </tt><tt><br>
</tt> <tt><br>
</tt><tt> As reduction initialize/combine/finalize functions
only receive references to whatever should be initialized,
combined..., clang generates some additional storages to
save the size of the reduction array and a pointer to the
omp_orig. That storages are generated through
__kmpc_thread_private_cached() calls in both reduction
functions and outline task function. </tt><tt><br>
</tt> <tt><br>
</tt><tt> The problem is that the storage used in outline
task function does not match the storage in reduction
functions. This happens in programs that have a task
reduction or a taskloop reduction with the in_reduction
clause. </tt><tt><br>
</tt> <tt><br>
</tt><tt> To solve this i made an ugly patch that reuses the
SourceLocation of
CodeGenFunction::EmitOMPTaskgroupDirective() in
CodeGenFunction::EmitOMPTaskBasedDirective() to match the
storages and make it work. Anyway, i suppose that there is
a better way to fix this. </tt><tt><br>
</tt> <tt><br>
</tt><tt> Index: CGStmtOpenMP.cpp </tt><tt><br>
</tt><tt>
=================================================================== </tt><tt><br>
</tt><tt> --- CGStmtOpenMP.cpp (revision 326685) </tt><tt><br>
</tt><tt> +++ CGStmtOpenMP.cpp (working copy) </tt><tt><br>
</tt><tt> @@ -2713,6 +2713,8 @@ </tt><tt><br>
</tt><tt>
emitEmptyBoundParameters); </tt><tt><br>
</tt><tt> } </tt><tt><br>
</tt> <tt><br>
</tt><tt> +static SourceLocation reductionLoc; </tt><tt><br>
</tt><tt> + </tt><tt><br>
</tt><tt> void CodeGenFunction::EmitOMPTaskBasedDirective(
</tt><tt><br>
</tt><tt> const OMPExecutableDirective &S, const
OpenMPDirectiveKind CapturedRegion, </tt><tt><br>
</tt><tt> const RegionCodeGenTy &BodyGen, const
TaskGenTy &TaskGen, </tt><tt><br>
</tt><tt> @@ -2952,7 +2954,7 @@ </tt><tt><br>
</tt><tt> // FIXME: This must removed once the
runtime library is fixed. </tt><tt><br>
</tt><tt> // Emit required threadprivate variables
for </tt><tt><br>
</tt><tt> // initilizer/combiner/finalizer. </tt><tt><br>
</tt><tt> -
CGF.CGM.getOpenMPRuntime().emitTaskReductionFixups(CGF,
S.getLocStart(), </tt><tt><br>
</tt><tt> +
CGF.CGM.getOpenMPRuntime().emitTaskReductionFixups(CGF,
reductionLoc, </tt><tt><br>
</tt><tt> RedCG, Cnt); </tt><tt><br>
</tt><tt> } </tt><tt><br>
</tt><tt> } </tt><tt><br>
</tt><tt> @@ -3180,8 +3182,9 @@ </tt><tt><br>
</tt><tt> std::advance(IRHS, 1); </tt><tt><br>
</tt><tt> } </tt><tt><br>
</tt><tt> } </tt><tt><br>
</tt><tt> + reductionLoc = S.getLocStart(); </tt><tt><br>
</tt><tt> llvm::Value *ReductionDesc = </tt><tt><br>
</tt><tt> -
CGF.CGM.getOpenMPRuntime().emitTaskReductionInit(CGF,
S.getLocStart(), </tt><tt><br>
</tt><tt> +
CGF.CGM.getOpenMPRuntime().emitTaskReductionInit(CGF,
reductionLoc, </tt><tt><br>
</tt><tt> LHSs, RHSs, Data); </tt><tt><br>
</tt><tt> const auto *VD =
cast<VarDecl>(cast<DeclRefExpr>(E)->getDecl());
</tt><tt><br>
</tt><tt> CGF.EmitVarDecl(*VD); </tt><tt><br>
</tt> <tt><br>
</tt><tt> Regards, </tt><tt><br>
</tt> <tt><br>
</tt><tt> Raúl </tt><tt><br>
</tt> <tt><br>
</tt> <tt><br>
</tt> <tt><a class="moz-txt-link-freetext"
href="https://nam03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fbsc.es%2Fdisclaimer&data=02%7C01%7C%7C2a787fae16d1409fb12808d584f2b8da%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636561100369995188&sdata=dRdsyHQ2zmAHZbnAdxiGhTMO%2ByEbHnhIo5nKgWIu9fA%3D&reserved=0"
originalsrc="http://bsc.es/disclaimer"
shash="j7hdeiLsvAbUdw0iIrDJrZvuOyMdvlonX8weB4d1Q1uT9nkNF8ym02ddIvdyzK7FAOuXCFVJzjRkaiJ9zH1CBUZFsKgf5NG0Siz9LQbDKueL1BcoJZMWVHW9SJnq5L7rUdtGZTpVxjReN/KriqmiXm1t3L6Qps7ptWLQOE7RHVM="
moz-do-not-send="true">http://bsc.es/disclaimer</a></tt><tt>
</tt><tt><br>
</tt> <tt><br>
</tt> </blockquote>
<tt><br>
</tt> </blockquote>
<tt><br>
</tt> <br>
<br>
WARNING / LEGAL TEXT: This message is intended only for the use
of the individual or entity to which it is addressed and may
contain information which is privileged, confidential,
proprietary, or exempt from disclosure under applicable law. If
you are not the intended recipient or the person responsible for
delivering the message to the intended recipient, you are
strictly prohibited from disclosing, distributing, copying, or
in any way using this message. If you have received this
communication in error, please notify the sender and destroy and
delete any copies you may have received. <br>
<br>
<a
href="https://nam03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.bsc.es%2Fdisclaimer&data=02%7C01%7C%7C2a787fae16d1409fb12808d584f2b8da%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636561100369995188&sdata=Qg460eDN4kZxnL4NIiP06MjaE85rnbyx2M%2FyVR7jif8%3D&reserved=0"
originalsrc="http://www.bsc.es/disclaimer"
shash="Hb30Ww/zrerjodaCGq1SFTIRJezQFUmTWNWr/UFbLxxJfNfu6z1IztPoXzhGacw7SmfcPYX+a6fFrIKMiVKNxv7ZLVJKNr9xrc+OgJ6FFQA6iDsDFMVSka/MMXYNW3NcFARpWH9FKzAeMmnb7mlJJQXMD/PXdSJztISceF1zH0c="
moz-do-not-send="true">http://www.bsc.es/disclaimer</a> <br>
</blockquote>
<br>
</blockquote>
<br>
<br>
<br>
WARNING / LEGAL TEXT: This message is intended only for the use of the
individual or entity to which it is addressed and may contain
information which is privileged, confidential, proprietary, or exempt
from disclosure under applicable law. If you are not the intended
recipient or the person responsible for delivering the message to the
intended recipient, you are strictly prohibited from disclosing,
distributing, copying, or in any way using this message. If you have
received this communication in error, please notify the sender and
destroy and delete any copies you may have received.
<br><br>
<a href="http://www.bsc.es/disclaimer">http://www.bsc.es/disclaimer</a>
<br>
</body>
</html>