Changes in funcimport.ll

41d5033e - [IR] Enable opaque pointers by default

Tue, 31 May 2022 09:58:35 +0000

[IR] Enable opaque pointers by defaultThis enabled opaque pointers by default in LLVM. The effect of thisis twofold:* If IR that contains *neither* explicit ptr nor %T* types is passed to tools, we will now use opaque pointer mode, unless -opaque-pointers=0 has been explicitly passed.* Users of LLVM as a library will now default to opaque pointers. It is possible to opt-out by calling setOpaquePointers(false) on LLVMContext.A cmake option to toggle this default will not be provided. Frontendsor other tools that want to (temporarily) keep using typed pointersshould disable opaque pointers via LLVMContext.Differential Revision: https://reviews.llvm.org/D126689 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

9fbc0405 - Make GlobalValues with non-default visibilility dso_local.

Thu, 18 Jan 2018 02:08:23 +0000

Make GlobalValues with non-default visibilility dso_local.This is similar to r322317, but for visibility. It is not as neatbecause we have to special case extern_weak.The idea is the same as the previous change, make the transition toexplicit dso_local easier for the frontends. With this they only haveto add dso_local to symbols where we need some external information todecide if it is dso_local (like it being part of an ELF executable).llvm-svn: 322806 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

e4b0231c - Make internal/private GVs implicitly dso_local.

Thu, 11 Jan 2018 22:15:05 +0000

Make internal/private GVs implicitly dso_local.While updating clang tests for having clang set dso_local I noticedthat:- There are *a lot* of tests to update.- Many of the updates are redundant.They are redundant because a GV is "obviously dso_local". This patchstarts formalizing that a bit by requiring that internal and privateGVs be dso_local too. Since they all are, we don't have to printdso_local to the textual representation, making it a bit more compactand easier to read.llvm-svn: 322317 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

2f0cc477 - ThinLTO: Don't import aliases of any kind (even linkonce_odr)

Thu, 27 Jul 2017 15:09:06 +0000

ThinLTO: Don't import aliases of any kind (even linkonce_odr)Summary:Until a more advanced version of importing can be implemented foraliases (one that imports an alias as an available_externally definitionof the aliasee), skip the narrow subset of cases that was possible butcame at a cost: aliases of linkonce_odr functions could be importedbecause the linkonce_odr function could be safely duplicated from thesource module. This came/comes at the cost of not being able to 'home'imported linkonce functions (they had to be emitted linkonce_odr in allthe destination modules (even if they weren't used by an alias) ratherthan as available_externally - causing extra object size).Tangentially, this also was the only reason ThinLTO would emit multipleCUs in to the resulting DWARF - which happens to be a problem forFission (there's a fix for this in GDB but not released yet, etc).(actually it's not the only reason - but I'm sending a patch to fix theother reason shortly)There's no reason to believe this particularly narrow alias importingwas especially/meaningfully important, only that it was /possible/ toimplement in this way. When a more general solution is done, it shouldstill satisfy the DWARF concerns above, since the import will still beavailable_externally, and thus not create extra CUs.Since now all aliases are treated the same, I removed/simplified sometest cases since they were testing corner cases where there are nolonger any corners.Reviewers: tejohnson, mehdi_aminiDifferential Revision: https://reviews.llvm.org/D35875llvm-svn: 309278 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

185b4ab6 - [ThinLTO] Stop importing constant global vars as copies in the backend

Fri, 02 Dec 2016 01:02:30 +0000

[ThinLTO] Stop importing constant global vars as copies in the backendSummary:We were doing an optimization in the ThinLTO backends of importingconstant unnamed_addr globals unconditionally as a local copy (regardlessof whether the thin link decided to import them). This should be done inthe thin link instead, so that resulting exported references are markedand promoted appropriately, but will need a summary enhancement to markthese variables as constant unnamed_addr.The function import logic during the thin link was trying to handlethis proactively, by conservatively marking all values referenced inthe initializer lists of exported global variables as also exported.However, this only handled values referenced directly from theinitializer list of an exported global variable. If the value is itselfa constant unnamed_addr variable, we could end up exporting itsreferences as well. This caused multiple issues. The first is that thetransitively exported references weren't promoted. Secondly, some couldnot be promoted/renamed (e.g. they had a section or other constraint).recursively, instead of just adding the first level of initializer listreferences to the ExportList directly.Remove this optimization and the associated handling in the functionimport backend. SPEC measurements indicate we weren't getting muchfrom it in any case.Fixes PR31052.Reviewers: mehdi_aminiSubscribers: krasin, llvm-commitsDifferential Revision: https://reviews.llvm.org/D26880llvm-svn: 288446 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

bb3a1d92 - ThinLTO: never promote as external weak

Wed, 20 Apr 2016 04:18:11 +0000

ThinLTO: never promote as external weakThis linkage is *not* intended to express that a declaration refersto a weak symbol, but that the symbol might not be present at linktime. I don't believe it was the intent.From: Mehdi Amini llvm-svn: 266856 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

68da426e - Move summary creation out of llvm-as into opt

Tue, 12 Apr 2016 21:35:18 +0000

Move summary creation out of llvm-as into optSummary:Let keep llvm-as "dumb": it converts textual IR to bitcode. Thiscommit removes the dependency from llvm-as to libLLVMAnalysis.We'll add back summary in llvm-as if we get to a textualrepresentation for it at some point. In the meantime, opt seemslike a better place for that.Reviewers: tejohnsonSubscribers: joker.eph, llvm-commitsDifferential Revision: http://reviews.llvm.org/D19032From: Mehdi Amini llvm-svn: 266131 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

ae280e54 - ThinLTO renaming: use module hash instead of position in the summary

Mon, 11 Apr 2016 23:26:46 +0000

ThinLTO renaming: use module hash instead of position in the summaryThis is more robust to changes in the link ordering.Differential Revision: http://reviews.llvm.org/D18946From: Mehdi Amini llvm-svn: 266018 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

8d05185a - Rework linkInModule(), making it oblivious to ThinLTO

Sat, 19 Mar 2016 00:40:31 +0000

Rework linkInModule(), making it oblivious to ThinLTOSummary:ThinLTO is relying on linkInModule to import selected function.However a lot of "magic" was hidden in linkInModule and the IRMover,who would rename and promote global variables on the fly.This is moving to an approach where the steps are decoupled and theclient is reponsible to specify the list of globals to import.As a consequence some test are changed because they were relying onthe previous behavior which was importing the definition of *every*single global without control on the client side.Now the burden is on the client to decide if a global has to be importedor not.Reviewers: tejohnsonSubscribers: joker.eph, llvm-commitsDifferential Revision: http://reviews.llvm.org/D18122From: Mehdi Amini llvm-svn: 263863 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

26ab5772 - [ThinLTO] Renaming of function index to module summary index (NFC)

Tue, 15 Mar 2016 00:04:37 +0000

[ThinLTO] Renaming of function index to module summary index (NFC)(Resubmitting after fixing missing file issue)With the changes in r263275, there are now more than just functions inthe summary. Completed the renaming of data structures (started inr263275) to reflect the wider scope. In particular, changed theFunctionIndex* data structures to ModuleIndex*, and renamed relatedvariables and comments. Also renamed the files to reflect the changes.A companion clang patch will immediately succeed this patch to reflectthis renaming.llvm-svn: 263513 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

cec0cae3 - Revert "[ThinLTO] Renaming of function index to module summary index (NFC)"

Mon, 14 Mar 2016 21:18:10 +0000

Revert "[ThinLTO] Renaming of function index to module summary index (NFC)"This reverts commit r263490. Missed a file.llvm-svn: 263493 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

892920b3 - [ThinLTO] Renaming of function index to module summary index (NFC)

Mon, 14 Mar 2016 21:05:56 +0000

[ThinLTO] Renaming of function index to module summary index (NFC)With the changes in r263275, there are now more than just functions inthe summary. Completed the renaming of data structures (started inr263275) to reflect the wider scope. In particular, changed theFunctionIndex* data structures to ModuleIndex*, and renamed relatedvariables and comments. Also renamed the files to reflect the changes.A companion clang patch will immediately succeed this patch to reflectthis renaming.llvm-svn: 263490 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

4b5ec263 - Switch the linker to having a whitelist of GVs.

Wed, 02 Dec 2015 22:59:04 +0000

Switch the linker to having a whitelist of GVs.This replaces DoNotLinkFromSource with ValuesToLink. It also moves thecomputation of ValuesToLink earlier.It is a bit simpler and an important step in slitting the linker into anir mover and a linker proper.The test change is because we now avoid creating dead declarations.llvm-svn: 254559 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

8c04472e - Delete what is now duplicated code.

Wed, 02 Dec 2015 22:22:24 +0000

Delete what is now duplicated code.Having to import an alias as declaration is not thinlto specific.The test difference are because when we already have a decl and we arenot importing it, we just leave the decl alone.llvm-svn: 254556 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

89345771 - Disallow aliases to available_externally.

Thu, 26 Nov 2015 19:22:59 +0000

Disallow aliases to available_externally.They are as much trouble as aliases to declarations. They are requiringthe code generator to define a symbol with the same value as anothersymbol, but the second symbol is undefined.If representing this is important for some optimization, we could addsupport for available_externally aliases. They would be *required* topoint to a declaration (or available_externally definition).llvm-svn: 254170 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

6290dbc0 - [ThinLTO] Handle bitcode without function summary sections gracefully

Sat, 21 Nov 2015 21:55:48 +0000

[ThinLTO] Handle bitcode without function summary sections gracefullySummary:Several fixes to the handling of bitcode files without function summarysections so that they are skipped during ThinLTO processing in llvm-ltoand the gold plugin when appropriate instead of aborting.1 Don't assert when trying to add a FunctionInfo that doesn't have a summary attached.2 Skip FunctionInfo structures that don't have attached function summary sections when trying to create the combined function summary.3 In both llvm-lto and gold-plugin, check whether a bitcode file has a function summary section before trying to parse the index, and skip the bitcode file if it does not.4 Fix hasFunctionSummaryInMemBuffer in BitcodeReader, which had a bug where we returned to early while looking for the summary section.Also added llvm-lto and gold-plugin based tests for cases where wedon't have function summaries in the bitcode file. I verified thateither the first couple fixes described above are enough to avoid thecrashes, or fixes 1,3,4. But have combined them all here for addedrobustness.Reviewers: joker.ephSubscribers: llvm-commits, joker.ephDifferential Revision: http://reviews.llvm.org/D14903llvm-svn: 253796 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

ba5d68df - [ThinLTO] Update test to be more tolerant of ordering changes

Thu, 12 Nov 2015 19:31:46 +0000

[ThinLTO] Update test to be more tolerant of ordering changesUpdate the ThinLTO function importing test to use DAG forms of checks sothat it is more tolerant of changes to relative ordering betweenimported decls/defs. This reduces the number of changes required by thecomdat importing patch I am sending for review shortly.llvm-svn: 252932 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

3cd8161c - [ThinLTO] WeakAny fixes/cleanup

Tue, 10 Nov 2015 18:20:11 +0000

[ThinLTO] WeakAny fixes/cleanupEnsure WeakAny variables are imported as ExternalWeak declarations. Tohandle WeakAny more consistently and fix this issue:1) Update helper doImportAsDefinition to properly flag WeakAny variables and aliases as not importing defintions. Update callers of doImportAsDefinition to remove now redundant checks for WeakAny aliases, or ignore aliases, as appropriate.2) Add any !doImportAsDefinition GVs to DoNotLinkFromSource set during linking of the GV prototype, where we usually add GVs to the DoNotLinkFromSource set for other reasons. Remove now unnecessary adding of WeakAny aliases to DoNotLinkFromSource set from copyGlobalAliasProto. Remove now unnecessary guard against linking non-imported function bodies from ModuleLinker::run.llvm-svn: 252626 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

f1b0a6e3 - [ThinLTO] Always set linkage type to external when converting alias

Wed, 04 Nov 2015 16:01:16 +0000

[ThinLTO] Always set linkage type to external when converting aliasWhen converting an alias to a non-alias when the aliasee is notimported, ensure that the linkage type is set to external so that it isa valid linkage type. Added a test case that exposed this issue.llvm-svn: 252054 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll

c7ed52f2 - Restore "Support for ThinLTO function importing and symbol linking."

Tue, 03 Nov 2015 00:14:15 +0000

Restore "Support for ThinLTO function importing and symbol linking."This restores commit r251837, with the new library dependence added tollvm-link/Makefile to address bot failures.llvm-svn: 251866 List of files: /llvm-project-15.0.7/llvm/test/Linker/funcimport.ll