Benoit Belley via llvm-dev
2017-Jun-20 15:38 UTC
[llvm-dev] JIT, LTO and @llvm.global_ctors: Looking for advise
Thanks for the hindsight. I am currently working on a patch/potential fix which introduces a new Linker::ImportIntrinsicGlobalVariables flag. The patch includes a unit test reproducing the problem. Hopefully, that will help getting more feedback. Note that it might take a while before I am allowed to upload the patch since I need approval from Autodesk Legal department. Cheers, Benoit Benoit Belley Sr Principal Developer M&E-Product Development Group MAIN +1 514 393 1616 DIRECT +1 438 448 6304 FAX +1 514 393 0110 Twitter <http://twitter.com/autodesk> Facebook <https://www.facebook.com/Autodesk> Autodesk, Inc. 10 Duke Street Montreal, Quebec, Canada H3C 2L7 www.autodesk.com <http://www.autodesk.com/> From: David Blaikie <dblaikie at gmail.com> Date: mardi 20 juin 2017 à 11:12 To: Benoit Belley <benoit.belley at autodesk.com>, llvm-dev <llvm-dev at lists.llvm.org>, Lang Hames <lhames at gmail.com>, Teresa Johnson <tejohnson at google.com> Subject: Re: [llvm-dev] JIT, LTO and @llvm.global_ctors: Looking for advise>+Lang (for JIT) & Teresa (for LTO/ThinLTO). > >Sounds like maybe the LinkOnlyNeeded got reused for a bit more than the >original intent & maybe there should be more than one flag here - not >sure. > >On Mon, Jun 19, 2017 at 9:16 AM Benoit Belley via llvm-dev ><llvm-dev at lists.llvm.org> wrote: > > >Hi Everyone, > >We are looking for advise regarding the proper use of LTO in >conjunction with just-in time generated code. Our usage scenario goes >as follows. > > 1. Our front-end generates an LLVM module. > > 2. A small runtime support library is linked-in. The runtime > library is distributed as bitcode. It is generated using "clang++ > -emit-llvm' and 'llvm-link'. This allows LTO to kick-in and > functions from the runtime support library become candidates for > inlining. This is the desired effect that we are looking for. > > 3. The resulting LLVM module is compiled and dynamically loaded. We > are currently using the MCJIT API, but are planning to move to ORC > very soon. > >Our LLVM module linking code goes roughly as follows: > > Linker linker(jittedModule); > std::unique_ptr<llvm::Module> moduleToLink( > getLazyIRFileModule(bcFileName, error, context)); > linker.linkInModule(std::move(module), > Linker::LinkOnlyNeeded | > Linker::InternalizeLinkedSymbol); > >Our issue is with the Linker::LinkOnlyNeeded flag. Using it has a huge >positive impact on link and compilation time :-). But, it causes the >@llvm.global_ctors and @llvm.global_dtors references from the >linked-in modules to be discarded :-(. AFAICT, the Linker code assumes >ThinLTO when the LinkOnlyNeeded flags is specified, and full-LTO >otherwise. > >To resolve this, we have locally patched >llvm/lib/Linker/LinkModules.cpp with: > > bool ModuleLinker::run() { > > // .... > > if (shouldImportIntrinsicGlobalVariables()) { > auto addIntrinsicGlobalVariable = [ValuesToLink, > srcM](llvm::StringRef name) { > if (GlobalValue *GV = SrcM->getGlobalVariable(name)) { > ValuesToLink.insert(GV); > } > }; > > // These are added here, because they must not be internalized. > addIntrinsicGlobalVariable("llvm.used"); > addIntrinsicGlobalVariable("llvm.compiler.used"); > addIntrinsicGlobalVariable("llvm.global_ctors"); > addIntrinsicGlobalVariable("llvm.global_dtors"); > } > > // ... > > } > >Questions: > > 1. Is attempting to use llvm::Linker appropriate for our usage > pattern ? Should we directly use llvm::IRMover instead ? > > 2. Or, is our patch to ModuleLinker::run() the way to go ? Should > we submit back a patch along these lines ? > > 3. Other suggestions ? > >[Note] We are currently using LLVM 4.0.1-rc2. > >Cheers, >Benoit > > > >Benoit Belley >Sr Principal Developer >M&E-Product Development Group > >MAIN +1 514 393 1616 <tel:(514)%20393-1616> >DIRECT +1 438 448 6304 <tel:(438)%20448-6304> >FAX +1 514 393 0110 <tel:(514)%20393-0110> > >Twitter <http://twitter.com/autodesk> >Facebook <https://www.facebook.com/Autodesk> > >Autodesk, Inc. >10 Duke Street >Montreal, Quebec, Canada H3C 2L7 >www.autodesk.com <http://www.autodesk.com> <http://www.autodesk.com/> > > >_______________________________________________ >LLVM Developers mailing list >llvm-dev at lists.llvm.org >http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
Peter Collingbourne via llvm-dev
2017-Jun-20 17:51 UTC
[llvm-dev] JIT, LTO and @llvm.global_ctors: Looking for advise
Hi Benoit, It seems to me that LinkOnlyNeeded failing to link the llvm.* special variables is a bug. I think we should probably just change the behaviour of LinkOnlyNeeded so that it links them. Note that ThinLTO does not use the Linker class, it uses IRMover directly. That might not be appropriate for your use case, though, because IRMover does not follow references when linking (for example, if your module defines two functions f and g, where f calls g, it will not link g if you only ask for f). Peter On Tue, Jun 20, 2017 at 8:38 AM, Benoit Belley via llvm-dev < llvm-dev at lists.llvm.org> wrote:> Thanks for the hindsight. > > I am currently working on a patch/potential fix which introduces a new > Linker::ImportIntrinsicGlobalVariables flag. The patch includes a unit > test reproducing the problem. Hopefully, that will help getting more > feedback. > > Note that it might take a while before I am allowed to upload the patch > since I need approval from Autodesk Legal department. > > Cheers, > Benoit > > Benoit Belley > Sr Principal Developer > M&E-Product Development Group > > MAIN +1 514 393 1616 > DIRECT +1 438 448 6304 > FAX +1 514 393 0110 > > Twitter <http://twitter.com/autodesk> > Facebook <https://www.facebook.com/Autodesk> > > Autodesk, Inc. > 10 Duke Street > Montreal, Quebec, Canada H3C 2L7 > www.autodesk.com <http://www.autodesk.com/> > > > > > > From: David Blaikie <dblaikie at gmail.com> > Date: mardi 20 juin 2017 à 11:12 > To: Benoit Belley <benoit.belley at autodesk.com>, llvm-dev > <llvm-dev at lists.llvm.org>, Lang Hames <lhames at gmail.com>, Teresa Johnson > <tejohnson at google.com> > Subject: Re: [llvm-dev] JIT, LTO and @llvm.global_ctors: Looking for > advise > > > >+Lang (for JIT) & Teresa (for LTO/ThinLTO). > > > >Sounds like maybe the LinkOnlyNeeded got reused for a bit more than the > >original intent & maybe there should be more than one flag here - not > >sure. > > > >On Mon, Jun 19, 2017 at 9:16 AM Benoit Belley via llvm-dev > ><llvm-dev at lists.llvm.org> wrote: > > > > > >Hi Everyone, > > > >We are looking for advise regarding the proper use of LTO in > >conjunction with just-in time generated code. Our usage scenario goes > >as follows. > > > > 1. Our front-end generates an LLVM module. > > > > 2. A small runtime support library is linked-in. The runtime > > library is distributed as bitcode. It is generated using "clang++ > > -emit-llvm' and 'llvm-link'. This allows LTO to kick-in and > > functions from the runtime support library become candidates for > > inlining. This is the desired effect that we are looking for. > > > > 3. The resulting LLVM module is compiled and dynamically loaded. We > > are currently using the MCJIT API, but are planning to move to ORC > > very soon. > > > >Our LLVM module linking code goes roughly as follows: > > > > Linker linker(jittedModule); > > std::unique_ptr<llvm::Module> moduleToLink( > > getLazyIRFileModule(bcFileName, error, context)); > > linker.linkInModule(std::move(module), > > Linker::LinkOnlyNeeded | > > Linker::InternalizeLinkedSymbol); > > > >Our issue is with the Linker::LinkOnlyNeeded flag. Using it has a huge > >positive impact on link and compilation time :-). But, it causes the > >@llvm.global_ctors and @llvm.global_dtors references from the > >linked-in modules to be discarded :-(. AFAICT, the Linker code assumes > >ThinLTO when the LinkOnlyNeeded flags is specified, and full-LTO > >otherwise. > > > >To resolve this, we have locally patched > >llvm/lib/Linker/LinkModules.cpp with: > > > > bool ModuleLinker::run() { > > > > // .... > > > > if (shouldImportIntrinsicGlobalVariables()) { > > auto addIntrinsicGlobalVariable = [ValuesToLink, > > srcM](llvm::StringRef name) { > > if (GlobalValue *GV = SrcM->getGlobalVariable(name)) { > > ValuesToLink.insert(GV); > > } > > }; > > > > // These are added here, because they must not be internalized. > > addIntrinsicGlobalVariable("llvm.used"); > > addIntrinsicGlobalVariable("llvm.compiler.used"); > > addIntrinsicGlobalVariable("llvm.global_ctors"); > > addIntrinsicGlobalVariable("llvm.global_dtors"); > > } > > > > // ... > > > > } > > > >Questions: > > > > 1. Is attempting to use llvm::Linker appropriate for our usage > > pattern ? Should we directly use llvm::IRMover instead ? > > > > 2. Or, is our patch to ModuleLinker::run() the way to go ? Should > > we submit back a patch along these lines ? > > > > 3. Other suggestions ? > > > >[Note] We are currently using LLVM 4.0.1-rc2. > > > >Cheers, > >Benoit > > > > > > > >Benoit Belley > >Sr Principal Developer > >M&E-Product Development Group > > > >MAIN +1 514 393 1616 <tel:(514)%20393-1616> > >DIRECT +1 438 448 6304 <tel:(438)%20448-6304> > >FAX +1 514 393 0110 <tel:(514)%20393-0110> > > > >Twitter <http://twitter.com/autodesk> > >Facebook <https://www.facebook.com/Autodesk> > > > >Autodesk, Inc. > >10 Duke Street > >Montreal, Quebec, Canada H3C 2L7 > >www.autodesk.com <http://www.autodesk.com> <http://www.autodesk.com/> > > > > > >_______________________________________________ > >LLVM Developers mailing list > >llvm-dev at lists.llvm.org > >http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev > > _______________________________________________ > LLVM Developers mailing list > llvm-dev at lists.llvm.org > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev >-- -- Peter -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170620/689ce8f3/attachment.html>
Benoit Belley via llvm-dev
2017-Jun-20 22:45 UTC
[llvm-dev] JIT, LTO and @llvm.global_ctors: Looking for advise
Thanks Peter, this is very useful feedback. I did manage to change the behavior of LinkOnlyNeeded to correctly import all variables with AppendingLinkage. In fact, I discovered that there was already something fishy. A variable with AppendingLinkage would get imported correctly from the source module if the destination module already contained a definition for that variable and wouldn't be imported otherwiseŠ My local fix ensures that it correctly gets imported in both cases. You are right; ThinLTO no longer uses the Linker class. I was able to remove the useless include of Linker.h from ThinLTOCodeGenerator.cpp. That being said, Linker.h and LinkModules.cpp still have a few comments about ThinLTO, namely: /// \brief Link \p Src into the composite. /// /// Passing OverrideSymbols as true will have symbols from Src /// shadow those in the Dest. /// For ThinLTO function importing/exporting the \p ModuleSummaryIndex /// is passed. If \p GlobalsToImport is provided, only the globals that /// are part of the set will be imported from the source module. /// /// Returns true on error. // Don't want to append to global_ctors list, for example, when we // are importing for ThinLTO, otherwise the global ctors and dtors // get executed multiple times for local variables (the latter causing // double frees). // For ThinLTO we don't import more than what was required. // The client has to guarantee that the linkonce will be availabe at link // time (by promoting it to weak for instance). Are these obsolete comments ? Should these be cleaned-up ? If so, how ? Thanks again, Benoit From: Peter Collingbourne <peter at pcc.me.uk> Date: mardi 20 juin 2017 à 13:51 To: Benoit Belley <benoit.belley at autodesk.com> Cc: David Blaikie <dblaikie at gmail.com>, llvm-dev <llvm-dev at lists.llvm.org>, Lang Hames <lhames at gmail.com>, Teresa Johnson <tejohnson at google.com> Subject: Re: [llvm-dev] JIT, LTO and @llvm.global_ctors: Looking for advise>Hi Benoit, > >It seems to me that LinkOnlyNeeded failing to link the llvm.* special >variables is a bug. I think we should probably just change the behaviour >of LinkOnlyNeeded so that it links them. > >Note that ThinLTO does not use the Linker class, it uses IRMover >directly. That might not be appropriate for your use case, though, >because IRMover does not follow references when linking (for example, if >your module defines two functions f and g, where > f calls g, it will not link g if you only ask for f). > >Peter > > >On Tue, Jun 20, 2017 at 8:38 AM, Benoit Belley via llvm-dev ><llvm-dev at lists.llvm.org> wrote: > >Thanks for the hindsight. > >I am currently working on a patch/potential fix which introduces a new >Linker::ImportIntrinsicGlobalVariables flag. The patch includes a unit >test reproducing the problem. Hopefully, that will help getting more >feedback. > >Note that it might take a while before I am allowed to upload the patch >since I need approval from Autodesk Legal department. > >Cheers, >Benoit > >Benoit Belley >Sr Principal Developer >M&E-Product Development Group > >MAIN +1 514 393 1616 <tel:%2B1%20514%20393%201616> >DIRECT +1 438 448 6304 <tel:%2B1%20438%20448%206304> >FAX +1 514 393 0110 <tel:%2B1%20514%20393%200110> > >Twitter <http://twitter.com/autodesk> >Facebook <https://www.facebook.com/Autodesk> > >Autodesk, Inc. >10 Duke Street >Montreal, Quebec, Canada H3C 2L7 >www.autodesk.com <http://www.autodesk.com> <http://www.autodesk.com/> > > > > > >From: David Blaikie <dblaikie at gmail.com> >Date: mardi 20 juin 2017 à 11:12 >To: Benoit Belley <benoit.belley at autodesk.com>, llvm-dev ><llvm-dev at lists.llvm.org>, Lang Hames <lhames at gmail.com>, Teresa Johnson ><tejohnson at google.com> >Subject: Re: [llvm-dev] JIT, LTO and @llvm.global_ctors: Looking for >advise > > >>+Lang (for JIT) & Teresa (for LTO/ThinLTO). >> >>Sounds like maybe the LinkOnlyNeeded got reused for a bit more than the >>original intent & maybe there should be more than one flag here - not >>sure. >> >>On Mon, Jun 19, 2017 at 9:16 AM Benoit Belley via llvm-dev >><llvm-dev at lists.llvm.org> wrote: >> >> >>Hi Everyone, >> >>We are looking for advise regarding the proper use of LTO in >>conjunction with just-in time generated code. Our usage scenario goes >>as follows. >> >> 1. Our front-end generates an LLVM module. >> >> 2. A small runtime support library is linked-in. The runtime >> library is distributed as bitcode. It is generated using "clang++ >> -emit-llvm' and 'llvm-link'. This allows LTO to kick-in and >> functions from the runtime support library become candidates for >> inlining. This is the desired effect that we are looking for. >> >> 3. The resulting LLVM module is compiled and dynamically loaded. We >> are currently using the MCJIT API, but are planning to move to ORC >> very soon. >> >>Our LLVM module linking code goes roughly as follows: >> >> Linker linker(jittedModule); >> std::unique_ptr<llvm::Module> moduleToLink( >> getLazyIRFileModule(bcFileName, error, context)); >> linker.linkInModule(std::move(module), >> Linker::LinkOnlyNeeded | >> Linker::InternalizeLinkedSymbol); >> >>Our issue is with the Linker::LinkOnlyNeeded flag. Using it has a huge >>positive impact on link and compilation time :-). But, it causes the >>@llvm.global_ctors and @llvm.global_dtors references from the >>linked-in modules to be discarded :-(. AFAICT, the Linker code assumes >>ThinLTO when the LinkOnlyNeeded flags is specified, and full-LTO >>otherwise. >> >>To resolve this, we have locally patched >>llvm/lib/Linker/LinkModules.cpp with: >> >> bool ModuleLinker::run() { >> >> // .... >> >> if (shouldImportIntrinsicGlobalVariables()) { >> auto addIntrinsicGlobalVariable = [ValuesToLink, >> srcM](llvm::StringRef name) { >> if (GlobalValue *GV = SrcM->getGlobalVariable(name)) { >> ValuesToLink.insert(GV); >> } >> }; >> >> // These are added here, because they must not be internalized. >> addIntrinsicGlobalVariable("llvm.used"); >> addIntrinsicGlobalVariable("llvm.compiler.used"); >> addIntrinsicGlobalVariable("llvm.global_ctors"); >> addIntrinsicGlobalVariable("llvm.global_dtors"); >> } >> >> // ... >> >> } >> >>Questions: >> >> 1. Is attempting to use llvm::Linker appropriate for our usage >> pattern ? Should we directly use llvm::IRMover instead ? >> >> 2. Or, is our patch to ModuleLinker::run() the way to go ? Should >> we submit back a patch along these lines ? >> >> 3. Other suggestions ? >> >>[Note] We are currently using LLVM 4.0.1-rc2. >> >>Cheers, >>Benoit >> >> >> >>Benoit Belley >>Sr Principal Developer >>M&E-Product Development Group >> > > >>MAIN +1 514 393 1616 <tel:%2B1%20514%20393%201616> <tel:(514)%20393-1616> >>DIRECT +1 438 448 6304 <tel:%2B1%20438%20448%206304> >><tel:(438)%20448-6304> >>FAX +1 514 393 0110 <tel:%2B1%20514%20393%200110> <tel:(514)%20393-0110> >> >>Twitter <http://twitter.com/autodesk> >>Facebook <https://www.facebook.com/Autodesk> >> >>Autodesk, Inc. >>10 Duke Street >>Montreal, Quebec, Canada H3C 2L7 >>www.autodesk.com <http://www.autodesk.com> <http://www.autodesk.com> >><http://www.autodesk.com/> >> >> >>_______________________________________________ >>LLVM Developers mailing list >>llvm-dev at lists.llvm.org >>http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev > >_______________________________________________ >LLVM Developers mailing list >llvm-dev at lists.llvm.org >http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev > > > > > > > > >-- >-- >Peter