<div>+BUILTIN_TYPE(Image1d_T, Image1dTTy)</div><div>+BUILTIN_TYPE(Image1dArray_T, Image1dArrayTTy)</div><div>+BUILTIN_TYPE(Image1dBuffer_T, Image1dBufferTTy)</div><div>+BUILTIN_TYPE(Image2d_T, Image2dTTy)</div><div>+BUILTIN_TYPE(Image2dArray_T, Image2dArrayTTy)</div>
<div>+BUILTIN_TYPE(Image3d_T, Image3dTTy)</div><div><br></div><div>The other builtin types corresponding to *_t keywords don't have the '_T' as part of the builtin name (Char16, Char32, WChar_S, WChar_U). Including the _T here is inconsistent. Also, how would you feel about giving these identifiers an OpenCL prefix, as we do for the ObjC builtin types?</div>
<div><br></div><div><div>+ bool isImage1dT() const; // OpenCL image1d_t</div><div>+ bool isImage1dArrayT() const; // OpenCL image1d_t</div><div>+ bool isImage1dBufferT() const; // OpenCL image1d_t</div>
<div>+ bool isImage2dT() const; // OpenCL image2d_t</div><div>+ bool isImage2dArrayT() const; // OpenCL image2d_t</div><div>+ bool isImage3dT() const; // OpenCL image3d_t</div>
</div><div><br></div><div>Some of these comments are wrong.</div><div><br></div><div><div>+inline bool Type::isImage1dT() const {</div><div>+ if (const BuiltinType *BT = dyn_cast<BuiltinType>(CanonicalType))</div><div>
+ return BT->getKind() == BuiltinType::Image1d_T;</div><div>+ return false;</div><div>+}</div></div><div><br></div><div>Please write this as "return isSpecificBuiltinType(BuiltinType::Image1d_T);"</div><div>
<br></div><div>+ Width = Target->getPointerWidth(0); // Currently these types are pointers</div><div><div>+ Align = Target->getPointerAlign(0); // to opaque types</div></div><div><br></div><div>Please put the comment by itself on the line before this (and add a full stop).</div>
<div><br></div><div>+ case BuiltinType::Event_T: Encoding = llvm::dwarf::DW_ATE_unsigned;</div><div><br></div><div>It seems that you're giving these types the size and alignment of a pointer, but emitting debug info as if they're 'unsigned int'. Is that really the way this is supposed to work?</div>
<div><br></div><div><div>+ llvm::Type *ResultType;</div><div>+</div><div>+ switch (cast<BuiltinType>(T)->getKind()) {</div><div>+ default: break;</div><div>+ case BuiltinType::Image1d_T:</div><div>+ ResultType = llvm::PointerType::get(llvm::StructType::create(</div>
<div>+ CGM.getLLVMContext(),"opencl.image1d_t"), 0);</div><div>+ break;</div></div><div><br></div><div>In your 'default' case, you return an uninitialized value. Please change this to return the PointerType directly, rather than pushing it through a local variable, and add an llvm_unreachable to your default case.</div>
<div><br></div><div>+ friend class CodeGenTypes;</div><div>[...]</div><div>+ ResultType = CGM.OpenCLRuntime->convertOpenCLSpecificType(Ty);</div><div><br></div><div>Please remove the 'friend' declaration and instead use CGM.getOpenCLRuntime() here.</div>
<div><br></div><div>Please also add tests for the PCH serialization/deserialization of these types.</div><div><br></div><div class="gmail_quote">On Sun, Oct 7, 2012 at 2:07 AM, Benyei, Guy <span dir="ltr"><<a href="mailto:guy.benyei@intel.com" target="_blank">guy.benyei@intel.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div lang="EN-US" link="blue" vlink="purple">
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">Anton,
<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">I’ve fixed my patch according to your comments. I’ve also did some changes. See the patch attached.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">Tanya,<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">I think the OpenCL specific types, including the sampler and the event types should be well distinguishable in IR format. The fact, that a 32 bit integer exposes
a superset of the sampler functionality in some perspective is just a coincidence. The sampler type itself could contain additional information (for optimizations and other architecture specifics), and it doesn’t have to hold the initializer value in the bitwise
or integer format.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">The backend should be able to recognize the sampler type from the IR.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">In the attached patch, I’ve moved the OpenCL specific type conversion to the CGOpenCLRuntime class – this class was meant to be overloaded by OpenCL vendors
to implement vendor specific behavior (codegen for OpenCL locals is implemented there), and so if a vendor prefers using the i32 representation, he could implement it there.
<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">In the next increment this class should also contain a sampler initialization function, so sampler initialization could be implemented also in a vendor specific
way.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">Please review this new patch<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d">Thanks<u></u><u></u></span></p>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><img width="420" height="116" src="https://mail.google.com/mail/u/0/?ui=2&ik=a3c008e2d4&view=att&th=13a3c708e42297ae&attid=0.1&disp=emb&realattid=6c037c059be5f933_0.0.1&zw&atsh=1" alt="email_signature_guy_new2"><u></u><u></u></span></p>
</div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1f497d"><u></u> <u></u></span></p>
<div>
<div style="border:none;border-top:solid #b5c4df 1.0pt;padding:3.0pt 0in 0in 0in">
<p class="MsoNormal"><b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif"">From:</span></b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif""> Villmow, Micah [mailto:<a href="mailto:Micah.Villmow@amd.com" target="_blank">Micah.Villmow@amd.com</a>]
<br>
<b>Sent:</b> Thursday, October 04, 2012 22:15<br>
<b>To:</b> Tanya Lattner; Benyei, Guy<br>
<b>Cc:</b> <a href="mailto:cfe-dev@cs.uiuc.edu" target="_blank">cfe-dev@cs.uiuc.edu</a>; <a href="mailto:Anton.Lokhmotov@arm.com" target="_blank">Anton.Lokhmotov@arm.com</a><br>
<b>Subject:</b> RE: [cfe-dev] OpenCL & SPIR specific types - proposal and patch<u></u><u></u></span></p>
</div>
</div><div><div>
<p class="MsoNormal"><u></u> <u></u></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">There needs to be a way to differentiate between an integer and a sampler by only looking at the type. The sampler itself is an opaque type in OpenCL. The only requirement
is that it is initialized with a 32bit unsigned integer, not that the type itself is an integer.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><u></u> <u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">Micah<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><u></u> <u></u></span></p>
<div style="border:none;border-left:solid blue 1.5pt;padding:0in 0in 0in 4.0pt">
<div>
<div style="border:none;border-top:solid #b5c4df 1.0pt;padding:3.0pt 0in 0in 0in">
<p class="MsoNormal"><b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif"">From:</span></b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif""> Tanya Lattner [<a href="mailto:lattner@apple.com" target="_blank">mailto:lattner@apple.com</a>]
<br>
<b>Sent:</b> Thursday, October 04, 2012 11:49 AM<br>
<b>To:</b> Benyei, Guy<br>
<b>Cc:</b> <a href="mailto:cfe-dev@cs.uiuc.edu" target="_blank">cfe-dev@cs.uiuc.edu</a>; Villmow, Micah;
<a href="mailto:Anton.Lokhmotov@arm.com" target="_blank">Anton.Lokhmotov@arm.com</a><br>
<b>Subject:</b> Re: [cfe-dev] OpenCL & SPIR specific types - proposal and patch<u></u><u></u></span></p>
</div>
</div>
<p class="MsoNormal"><u></u> <u></u></p>
<div>
<p class="MsoNormal">I'm hoping to have comments on this patch tomorrow, but since I have proposed several patches to Clang for the sampler type (and have another in revision).. can you explain why you want to change the type from an integer to a pointer? <u></u><u></u></p>
</div>
<div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
<div>
<p class="MsoNormal">-Tanya<u></u><u></u></p>
</div>
<p class="MsoNormal"><u></u> <u></u></p>
<div>
<div>
<p class="MsoNormal">On Oct 3, 2012, at 8:06 AM, "Benyei, Guy" <<a href="mailto:guy.benyei@intel.com" target="_blank">guy.benyei@intel.com</a>> wrote:<u></u><u></u></p>
</div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><u></u> <u></u></p>
<div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">I’d like to renew the discussion about making the OpenCL specific types first class citizens in Clang.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">I think this change is required by the OpenCL specifications, since these type names are keywords of the OpenCL C language.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">This change is also needed in order to enable efficient checking of OpenCL restrictions on these types (OpenCL 1.2 spec, section 6.9).<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">Furthermore, the proposed change will turn these types to pointers to opaque types, which means that it will hide the actual (vendor specific) implementation, so the OpenCL
vendors using Clang will be able to implement these types in their own way.<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">This change would also be a basis for the implementation of SPIR generation by Clang. The SPIR discussion and spec can be found here:<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><a href="http://lists.cs.uiuc.edu/pipermail/cfe-dev/2012-September/024132.html" target="_blank"><span style="color:purple">http://lists.cs.uiuc.edu/pipermail/cfe-dev/2012-September/024132.html</span></a><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><a href="http://lists.cs.uiuc.edu/pipermail/cfe-dev/2012-September/024178.html" target="_blank"><span style="color:purple">http://lists.cs.uiuc.edu/pipermail/cfe-dev/2012-September/024178.html</span></a><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif"">Earlier discussion about the OpenCL types was started by Anton Lokhmotov:<u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><a href="http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-May/015297.html" target="_blank"><span style="color:purple">http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-May/015297.html</span></a><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><a href="http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-April/014741.html" target="_blank"><span style="color:purple">http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-April/014741.html</span></a><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><a href="http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-March/014118.html" target="_blank"><span style="color:purple">http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-March/014118.html</span></a><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><a href="http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-March/014121.html" target="_blank"><span style="color:purple">http://lists.cs.uiuc.edu/pipermail/cfe-dev/2011-March/014121.html</span></a><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""> <u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""><image001.png><u></u><u></u></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif""> <u></u><u></u></span></p>
</div>
<p class="MsoNormal"><span style="font-size:13.5pt;font-family:"Helvetica","sans-serif""><opencl_types.patch><u></u><u></u></span></p>
<p><span style="font-size:13.5pt;font-family:"Helvetica","sans-serif"">---------------------------------------------------------------------<br>
Intel Israel (74) Limited<u></u><u></u></span></p>
<p><span style="font-size:13.5pt;font-family:"Helvetica","sans-serif"">This e-mail and any attachments may contain confidential material for<br>
the sole use of the intended recipient(s). Any review or distribution<br>
by others is strictly prohibited. If you are not the intended<br>
recipient, please contact the sender and delete all copies.<u></u><u></u></span></p>
<p class="MsoNormal"><span style="font-size:13.5pt;font-family:"Helvetica","sans-serif"">_______________________________________________<br>
cfe-dev mailing list<br>
<a href="mailto:cfe-dev@cs.uiuc.edu" target="_blank"><span style="color:purple">cfe-dev@cs.uiuc.edu</span></a><br>
<a href="http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev" target="_blank"><span style="color:purple">http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev</span></a><u></u><u></u></span></p>
</div>
</div>
<p class="MsoNormal"><u></u> <u></u></p>
</div>
</div></div></div>
</div>
<br><p>---------------------------------------------------------------------<br>
Intel Israel (74) Limited</p>
<p>This e-mail and any attachments may contain confidential material for<br>
the sole use of the intended recipient(s). Any review or distribution<br>
by others is strictly prohibited. If you are not the intended<br>
recipient, please contact the sender and delete all copies.</p><br>_______________________________________________<br>
cfe-dev mailing list<br>
<a href="mailto:cfe-dev@cs.uiuc.edu" target="_blank">cfe-dev@cs.uiuc.edu</a><br>
<a href="http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev" target="_blank">http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev</a><br>
<br></blockquote></div><br>