<div dir="ltr"><br><br><div class="gmail_quote"><div dir="ltr">On Tue, Sep 15, 2015 at 9:30 AM Daniel Sanders <<a href="mailto:Daniel.Sanders@imgtec.com">Daniel.Sanders@imgtec.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div>
<div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt">Thanks for the reply. We should hold the main conversation in the llvmdev thread but to briefly reply here.</span></font></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt"><br>
<br>
> Briefly I don't think this is the right direction and I don't think you've explained why this is the right direction.<br>
<br></span></font></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt">
Do you at least accept that Triples are not usable as a way to describe the target? and that they are ambiguous and inconsistent in their meanings?</span></font></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt"><br>
<br></span></font></div></div></blockquote><div><br></div><div>I think that they're insufficient to fully describe a target given command line options that can change defaults is a better way of putting it.</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt">
> I don't like the idea of using global metadata for it and you haven't explained why this is acceptable<br>
<br></span></font></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt">
FWIW, my original proposal was a 'target tuple' statement to match the 'target triple' one we have today. Whatever serialization approach we choose doesn't significantly change the overall direction and has no effect on the current phase of the project. There's
no need to block the early phases of the project on small details of the later phases.<br>
</span></font><br></div></div></blockquote><div><br></div><div>Sure, as long as the general direction seems good. IR level serialization of things is ... touchy and I'd like to make sure that we get something there that we're going to be happy with long term given the stability guarantees. It's easier to define things on the side that can change rather than make a mistake at the IR level in a way that's not easy to upgrade. (Which we'll also need for any triple change as well).</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt">
<font size="2"><span style="font-size:10pt"></span></font></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt">> I actually was spending a non-trivial amount of time trying to design a way for this to work along side the assembler and main compiler (which I don't think you've bothered to try to do)<br>
<br></span></font></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt">
What makes you say I haven't bothered to design a way to work with the assembler and the main compiler? That's one of the core concepts driving the work and working compatibly with other toolchains is the objective.<br></span></font></div></div></blockquote><div><br></div><div>I think it's trying to solve a problem in the assembler interfaces, but from the wrong tack. I hope I explained it in one of my followups.</div><div><br></div><div>Thanks!</div><div><br></div><div>-eric</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><font size="2"><span style="font-size:10pt">
</span></font>
<div style="font-family:Times New Roman;color:#000000;font-size:16px">
<hr>
<div style="direction:ltr"><font color="#000000" face="Tahoma" size="2"></font></div></div></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><div style="font-family:Times New Roman;color:#000000;font-size:16px"><div style="direction:ltr"><font color="#000000" face="Tahoma" size="2"><b>From:</b> Eric Christopher [<a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>]<br>
<b>Sent:</b> 15 September 2015 16:56<br>
</font></div></div></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><div style="font-family:Times New Roman;color:#000000;font-size:16px"><div style="direction:ltr"><font color="#000000" face="Tahoma" size="2"><b>To:</b> Daniel Sanders; <a href="mailto:reviews%2BD10969%2Bpublic%2B5f3f4828b5695b54@reviews.llvm.org" target="_blank">reviews+D10969+public+5f3f4828b5695b54@reviews.llvm.org</a>; <a href="mailto:renato.golin@linaro.org" target="_blank">renato.golin@linaro.org</a><br>
<b>Cc:</b> <a href="mailto:jyknight@google.com" target="_blank">jyknight@google.com</a>; <a href="mailto:dschuff@google.com" target="_blank">dschuff@google.com</a>; <a href="mailto:Matthew.Arsenault@amd.com" target="_blank">Matthew.Arsenault@amd.com</a>; <a href="mailto:stanislav.mekhanoshin@amd.com" target="_blank">stanislav.mekhanoshin@amd.com</a>; <a href="mailto:danalbert@google.com" target="_blank">danalbert@google.com</a>; <a href="mailto:srhines@google.com" target="_blank">srhines@google.com</a>; <a href="mailto:javed.absar@arm.com" target="_blank">javed.absar@arm.com</a>; <a href="mailto:emaste@freebsd.org" target="_blank">emaste@freebsd.org</a>; <a href="mailto:jholewinski@nvidia.com" target="_blank">jholewinski@nvidia.com</a>; <a href="mailto:tberghammer@google.com" target="_blank">tberghammer@google.com</a>; <a href="mailto:ted.woodward@codeaurora.org" target="_blank">ted.woodward@codeaurora.org</a>;
<a href="mailto:jfb@chromium.org" target="_blank">jfb@chromium.org</a>; <a href="mailto:llvm-commits@lists.llvm.org" target="_blank">llvm-commits@lists.llvm.org</a></font></div></div></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><div style="font-family:Times New Roman;color:#000000;font-size:16px"><div style="direction:ltr"><font color="#000000" face="Tahoma" size="2"><br>
<b>Subject:</b> Re: [PATCH] D10969: Replace Triple with a new TargetTuple in MCTargetDesc/* and related. NFC.<br>
</font></div></div></div></div><div><div style="direction:ltr;font-family:Tahoma;color:#000000;font-size:10pt"><div style="font-family:Times New Roman;color:#000000;font-size:16px">
<div></div>
<div>
<div dir="ltr">
<p dir="ltr">I started a reply yesterday actually and I'm sorry I didn't finish it. Briefly I don't think this is the right direction and I don't think you've explained why this is the right direction. The serialization that you have planned is insufficient
and doesn't take into account portability, I don't like the idea of using global metadata for it and you haven't explained why this is acceptable. I actually was spending a non-trivial amount of time trying to design a way for this to work along side the assembler
and main compiler (which I don't think you've bothered to try to do)</p>
<p>Please revert this immediately.</p>
<p>-eric</p>
<p><br>
</p>
<br>
<div class="gmail_quote">
<div dir="ltr">On Tue, Sep 15, 2015, 6:29 AM Daniel Sanders <<a href="mailto:Daniel.Sanders@imgtec.com" target="_blank">Daniel.Sanders@imgtec.com</a>> wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
(apologies if this double-posts, it seems Phabricator just died)<br>
<br>
I've committed this (along with the trivial clang patch) in r247683 on the grounds that I have Renato's LGTM and I don't have any pending review comments. It doesn't seem reasonable to continue to wait for Eric with no indication of when he might be able to
take a look at it. For the record, I've been waiting since July and my request last week for an indication as to when he might find time has not been replied to. Of course, post-commit review comments are still welcome and I'm happy to follow up on them.<br>
<br>
> -----Original Message-----<br>
> From: Daniel Sanders<br>
> Sent: 15 September 2015 14:20<br>
> To: Daniel Sanders; <a href="mailto:renato.golin@linaro.org" target="_blank">renato.golin@linaro.org</a><br>
> Cc: <a href="mailto:jyknight@google.com" target="_blank">jyknight@google.com</a>;
<a href="mailto:dschuff@google.com" target="_blank">dschuff@google.com</a>;<br>
> <a href="mailto:Matthew.Arsenault@amd.com" target="_blank">Matthew.Arsenault@amd.com</a>;
<a href="mailto:stanislav.mekhanoshin@amd.com" target="_blank">stanislav.mekhanoshin@amd.com</a>;<br>
> <a href="mailto:danalbert@google.com" target="_blank">danalbert@google.com</a>;
<a href="mailto:srhines@google.com" target="_blank">srhines@google.com</a>; <a href="mailto:javed.absar@arm.com" target="_blank">
javed.absar@arm.com</a>;<br>
> <a href="mailto:echristo@gmail.com" target="_blank">echristo@gmail.com</a>; <a href="mailto:emaste@freebsd.org" target="_blank">
emaste@freebsd.org</a>; <a href="mailto:jholewinski@nvidia.com" target="_blank">jholewinski@nvidia.com</a>;<br>
> <a href="mailto:tberghammer@google.com" target="_blank">tberghammer@google.com</a>;
<a href="mailto:ted.woodward@codeaurora.org" target="_blank">ted.woodward@codeaurora.org</a>;<br>
> <a href="mailto:jfb@chromium.org" target="_blank">jfb@chromium.org</a>; <a href="mailto:llvm-commits@lists.llvm.org" target="_blank">
llvm-commits@lists.llvm.org</a><br>
> Subject: Re: [PATCH] D10969: Replace Triple with a new TargetTuple in<br>
> MCTargetDesc/* and related. NFC.<br>
><br>
> This revision was automatically updated to reflect the committed changes.<br>
> Closed by commit rL247683: Replace Triple with a new TargetTuple in<br>
> MCTargetDesc/* and related. NFC. (authored by dsanders).<br>
><br>
> Changed prior to commit:<br>
> <a href="http://reviews.llvm.org/D10969?vs=34223&id=34801#toc" rel="noreferrer" target="_blank">http://reviews.llvm.org/D10969?vs=34223&id=34801#toc</a><br>
><br>
> Repository:<br>
> rL LLVM<br>
><br>
> <a href="http://reviews.llvm.org/D10969" rel="noreferrer" target="_blank">http://reviews.llvm.org/D10969</a><br>
><br>
> Files:<br>
> cfe/trunk/lib/Parse/ParseStmtAsm.cpp<br>
> cfe/trunk/tools/driver/cc1as_main.cpp<br>
> llvm/trunk/include/llvm/ADT/TargetTuple.h<br>
> llvm/trunk/include/llvm/MC/MCELFObjectWriter.h<br>
> llvm/trunk/include/llvm/MC/MCSubtargetInfo.h<br>
> llvm/trunk/include/llvm/Support/TargetRegistry.h<br>
> llvm/trunk/include/llvm/Target/TargetMachine.h<br>
> llvm/trunk/include/llvm/Target/TargetSubtargetInfo.h<br>
> llvm/trunk/lib/CodeGen/LLVMTargetMachine.cpp<br>
> llvm/trunk/lib/MC/MCDisassembler/Disassembler.cpp<br>
> llvm/trunk/lib/MC/MCDisassembler/MCExternalSymbolizer.cpp<br>
> llvm/trunk/lib/MC/MCDisassembler/MCRelocationInfo.cpp<br>
> llvm/trunk/lib/MC/MCSubtargetInfo.cpp<br>
> llvm/trunk/lib/Support/CMakeLists.txt<br>
> llvm/trunk/lib/Support/TargetTuple.cpp<br>
> llvm/trunk/lib/Target/AArch64/AArch64Subtarget.cpp<br>
> llvm/trunk/lib/Target/AArch64/AArch64Subtarget.h<br>
> llvm/trunk/lib/Target/AArch64/Disassembler/AArch64Disassembler.cpp<br>
> llvm/trunk/lib/Target/AArch64/MCTargetDesc/AArch64AsmBackend.cpp<br>
> llvm/trunk/lib/Target/AArch64/MCTargetDesc/AArch64ELFStreamer.cpp<br>
> llvm/trunk/lib/Target/AArch64/MCTargetDesc/AArch64MCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/AArch64/MCTargetDesc/AArch64MCAsmInfo.h<br>
> llvm/trunk/lib/Target/AArch64/MCTargetDesc/AArch64MCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/AArch64/MCTargetDesc/AArch64MCTargetDesc.h<br>
> llvm/trunk/lib/Target/AMDGPU/AMDGPUSubtarget.cpp<br>
> llvm/trunk/lib/Target/AMDGPU/MCTargetDesc/AMDGPUAsmBackend.cpp<br>
> llvm/trunk/lib/Target/AMDGPU/MCTargetDesc/AMDGPUMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/AMDGPU/MCTargetDesc/AMDGPUMCAsmInfo.h<br>
><br>
> llvm/trunk/lib/Target/AMDGPU/MCTargetDesc/AMDGPUMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/AMDGPU/MCTargetDesc/AMDGPUMCTargetDesc.h<br>
> llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp<br>
> llvm/trunk/lib/Target/ARM/ARMSubtarget.cpp<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMAsmBackend.cpp<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMAsmBackend.h<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMAsmBackendDarwin.h<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMAsmBackendELF.h<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMAsmBackendWinCOFF.h<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMELFStreamer.cpp<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMMCAsmInfo.h<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMMCCodeEmitter.cpp<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/ARM/MCTargetDesc/ARMMCTargetDesc.h<br>
> llvm/trunk/lib/Target/BPF/BPFSubtarget.cpp<br>
> llvm/trunk/lib/Target/BPF/MCTargetDesc/BPFAsmBackend.cpp<br>
> llvm/trunk/lib/Target/BPF/MCTargetDesc/BPFMCAsmInfo.h<br>
> llvm/trunk/lib/Target/BPF/MCTargetDesc/BPFMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/BPF/MCTargetDesc/BPFMCTargetDesc.h<br>
> llvm/trunk/lib/Target/Hexagon/HexagonSubtarget.cpp<br>
> llvm/trunk/lib/Target/Hexagon/MCTargetDesc/HexagonAsmBackend.cpp<br>
> llvm/trunk/lib/Target/Hexagon/MCTargetDesc/HexagonMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/Hexagon/MCTargetDesc/HexagonMCAsmInfo.h<br>
> llvm/trunk/lib/Target/Hexagon/MCTargetDesc/HexagonMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/Hexagon/MCTargetDesc/HexagonMCTargetDesc.h<br>
> llvm/trunk/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/MSP430/MCTargetDesc/MSP430MCAsmInfo.h<br>
> llvm/trunk/lib/Target/MSP430/MCTargetDesc/MSP430MCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/MSP430/MSP430Subtarget.cpp<br>
> llvm/trunk/lib/Target/Mips/AsmParser/MipsAsmParser.cpp<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsABIInfo.cpp<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsABIInfo.h<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsAsmBackend.cpp<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsAsmBackend.h<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsMCAsmInfo.h<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/Mips/MCTargetDesc/MipsMCTargetDesc.h<br>
> llvm/trunk/lib/Target/Mips/MipsAsmPrinter.cpp<br>
> llvm/trunk/lib/Target/Mips/MipsSubtarget.cpp<br>
> llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp<br>
> llvm/trunk/lib/Target/NVPTX/MCTargetDesc/NVPTXMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/NVPTX/MCTargetDesc/NVPTXMCAsmInfo.h<br>
> llvm/trunk/lib/Target/NVPTX/MCTargetDesc/NVPTXMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/NVPTX/NVPTXSubtarget.cpp<br>
> llvm/trunk/lib/Target/PowerPC/AsmParser/PPCAsmParser.cpp<br>
> llvm/trunk/lib/Target/PowerPC/MCTargetDesc/PPCAsmBackend.cpp<br>
> llvm/trunk/lib/Target/PowerPC/MCTargetDesc/PPCMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/PowerPC/MCTargetDesc/PPCMCAsmInfo.h<br>
> llvm/trunk/lib/Target/PowerPC/MCTargetDesc/PPCMCCodeEmitter.cpp<br>
> llvm/trunk/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/PowerPC/MCTargetDesc/PPCMCTargetDesc.h<br>
> llvm/trunk/lib/Target/PowerPC/PPCSubtarget.cpp<br>
> llvm/trunk/lib/Target/Sparc/AsmParser/SparcAsmParser.cpp<br>
> llvm/trunk/lib/Target/Sparc/MCTargetDesc/SparcAsmBackend.cpp<br>
> llvm/trunk/lib/Target/Sparc/MCTargetDesc/SparcMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/Sparc/MCTargetDesc/SparcMCAsmInfo.h<br>
> llvm/trunk/lib/Target/Sparc/MCTargetDesc/SparcMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/Sparc/MCTargetDesc/SparcMCTargetDesc.h<br>
> llvm/trunk/lib/Target/Sparc/SparcSubtarget.cpp<br>
><br>
> llvm/trunk/lib/Target/SystemZ/MCTargetDesc/SystemZMCAsmBackend.cpp<br>
> llvm/trunk/lib/Target/SystemZ/MCTargetDesc/SystemZMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/SystemZ/MCTargetDesc/SystemZMCAsmInfo.h<br>
> llvm/trunk/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/SystemZ/MCTargetDesc/SystemZMCTargetDesc.h<br>
> llvm/trunk/lib/Target/SystemZ/SystemZSubtarget.cpp<br>
> llvm/trunk/lib/Target/TargetSubtargetInfo.cpp<br>
> llvm/trunk/lib/Target/X86/AsmParser/X86AsmInstrumentation.cpp<br>
> llvm/trunk/lib/Target/X86/MCTargetDesc/X86AsmBackend.cpp<br>
> llvm/trunk/lib/Target/X86/MCTargetDesc/X86MCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/X86/MCTargetDesc/X86MCAsmInfo.h<br>
> llvm/trunk/lib/Target/X86/MCTargetDesc/X86MCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/X86/MCTargetDesc/X86MCTargetDesc.h<br>
> llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp<br>
> llvm/trunk/lib/Target/X86/X86Subtarget.cpp<br>
> llvm/trunk/lib/Target/XCore/MCTargetDesc/XCoreMCAsmInfo.cpp<br>
> llvm/trunk/lib/Target/XCore/MCTargetDesc/XCoreMCAsmInfo.h<br>
> llvm/trunk/lib/Target/XCore/MCTargetDesc/XCoreMCTargetDesc.cpp<br>
> llvm/trunk/lib/Target/XCore/XCoreSubtarget.cpp<br>
> llvm/trunk/tools/dsymutil/DwarfLinker.cpp<br>
> llvm/trunk/tools/llvm-mc/llvm-mc.cpp<br>
> llvm/trunk/tools/llvm-objdump/MachODump.cpp<br>
> llvm/trunk/tools/llvm-objdump/llvm-objdump.cpp<br>
> llvm/trunk/tools/llvm-rtdyld/llvm-rtdyld.cpp<br>
> llvm/trunk/utils/TableGen/SubtargetEmitter.cpp<br>
<br>
</blockquote>
</div>
</div>
</div>
</div></div></div></blockquote></div></div>