Index: vendor/llvm/dist/bindings/go/llvm/linker.go =================================================================== --- vendor/llvm/dist/bindings/go/llvm/linker.go (revision 279290) +++ vendor/llvm/dist/bindings/go/llvm/linker.go (revision 279291) @@ -1,32 +1,39 @@ //===- linker.go - Bindings for linker ------------------------------------===// // // The LLVM Compiler Infrastructure // // This file is distributed under the University of Illinois Open Source // License. See LICENSE.TXT for details. // //===----------------------------------------------------------------------===// // // This file defines bindings for the linker component. // //===----------------------------------------------------------------------===// package llvm /* #include "llvm-c/Linker.h" #include */ import "C" import "errors" -func LinkModules(Dest, Src Module) error { +type LinkerMode C.LLVMLinkerMode + +const ( + LinkerDestroySource = C.LLVMLinkerDestroySource + LinkerPreserveSource = C.LLVMLinkerPreserveSource +) + +func LinkModules(Dest, Src Module, Mode LinkerMode) error { var cmsg *C.char - failed := C.LLVMLinkModules(Dest.C, Src.C, 0, &cmsg) + failed := C.LLVMLinkModules(Dest.C, Src.C, C.LLVMLinkerMode(Mode), &cmsg) if failed != 0 { err := errors.New(C.GoString(cmsg)) C.LLVMDisposeMessage(cmsg) return err } return nil } Index: vendor/llvm/dist/bindings/ocaml/linker/linker_ocaml.c =================================================================== --- vendor/llvm/dist/bindings/ocaml/linker/linker_ocaml.c (revision 279290) +++ vendor/llvm/dist/bindings/ocaml/linker/linker_ocaml.c (revision 279291) @@ -1,34 +1,34 @@ /*===-- linker_ocaml.c - LLVM OCaml Glue ------------------------*- C++ -*-===*\ |* *| |* The LLVM Compiler Infrastructure *| |* *| |* This file is distributed under the University of Illinois Open Source *| |* License. See LICENSE.TXT for details. *| |* *| |*===----------------------------------------------------------------------===*| |* *| |* This file glues LLVM's OCaml interface to its C interface. These functions *| |* are by and large transparent wrappers to the corresponding C functions. *| |* *| |* Note that these functions intentionally take liberties with the CAMLparamX *| |* macros, since most of the parameters are not GC heap objects. *| |* *| \*===----------------------------------------------------------------------===*/ #include "llvm-c/Linker.h" #include "caml/alloc.h" #include "caml/memory.h" #include "caml/fail.h" #include "caml/callback.h" void llvm_raise(value Prototype, char *Message); -/* llmodule -> llmodule -> unit */ -CAMLprim value llvm_link_modules(LLVMModuleRef Dst, LLVMModuleRef Src) { +/* llmodule -> llmodule -> Mode.t -> unit */ +CAMLprim value llvm_link_modules(LLVMModuleRef Dst, LLVMModuleRef Src, value Mode) { char* Message; - if (LLVMLinkModules(Dst, Src, 0, &Message)) + if (LLVMLinkModules(Dst, Src, Int_val(Mode), &Message)) llvm_raise(*caml_named_value("Llvm_linker.Error"), Message); return Val_unit; } Index: vendor/llvm/dist/bindings/ocaml/linker/llvm_linker.ml =================================================================== --- vendor/llvm/dist/bindings/ocaml/linker/llvm_linker.ml (revision 279290) +++ vendor/llvm/dist/bindings/ocaml/linker/llvm_linker.ml (revision 279291) @@ -1,15 +1,21 @@ (*===-- llvm_linker.ml - LLVM OCaml Interface ------------------*- OCaml -*-===* * * The LLVM Compiler Infrastructure * * This file is distributed under the University of Illinois Open Source * License. See LICENSE.TXT for details. * *===----------------------------------------------------------------------===*) exception Error of string let () = Callback.register_exception "Llvm_linker.Error" (Error "") -external link_modules : Llvm.llmodule -> Llvm.llmodule -> unit +module Mode = struct + type t = + | DestroySource + | PreserveSource +end + +external link_modules : Llvm.llmodule -> Llvm.llmodule -> Mode.t -> unit = "llvm_link_modules" Index: vendor/llvm/dist/bindings/ocaml/linker/llvm_linker.mli =================================================================== --- vendor/llvm/dist/bindings/ocaml/linker/llvm_linker.mli (revision 279290) +++ vendor/llvm/dist/bindings/ocaml/linker/llvm_linker.mli (revision 279291) @@ -1,19 +1,26 @@ (*===-- llvm_linker.mli - LLVM OCaml Interface -----------------*- OCaml -*-===* * * The LLVM Compiler Infrastructure * * This file is distributed under the University of Illinois Open Source * License. See LICENSE.TXT for details. * *===----------------------------------------------------------------------===*) (** Linker. This interface provides an OCaml API for LLVM bitcode linker, the classes in the Linker library. *) exception Error of string +(** Linking mode. *) +module Mode : sig + type t = + | DestroySource + | PreserveSource +end + (** [link_modules dst src mode] links [src] into [dst], raising [Error] if the linking fails. *) -val link_modules : Llvm.llmodule -> Llvm.llmodule -> unit \ No newline at end of file +val link_modules : Llvm.llmodule -> Llvm.llmodule -> Mode.t -> unit \ No newline at end of file Index: vendor/llvm/dist/docs/ReleaseNotes.rst =================================================================== --- vendor/llvm/dist/docs/ReleaseNotes.rst (revision 279290) +++ vendor/llvm/dist/docs/ReleaseNotes.rst (revision 279291) @@ -1,577 +1,646 @@ ====================== LLVM 3.6 Release Notes ====================== .. contents:: :local: Introduction ============ This document contains the release notes for the LLVM Compiler Infrastructure, release 3.6. Here we describe the status of LLVM, including major improvements from the previous release, improvements in various subprojects of LLVM, and some of the current users of the code. All LLVM releases may be downloaded from the `LLVM releases web site `_. For more information about LLVM, including information about the latest release, please check out the `main LLVM web site `_. If you have questions or comments, the `LLVM Developer's Mailing List `_ is a good place to send them. Non-comprehensive list of changes in this release ================================================= .. NOTE For small 1-3 sentence descriptions, just add an entry at the end of this list. If your description won't fit comfortably in one bullet point (e.g. maybe you would like to give an example of the functionality, or simply have a lot to talk about), see the `NOTE` below for adding a new subsection. * Support for AuroraUX has been removed. * Added support for a `native object file-based bitcode wrapper format `_. -* ... next change ... +* Added support for MSVC's ``__vectorcall`` calling convention as + ``x86_vectorcallcc``. .. NOTE If you would like to document a larger change, then you can add a subsection about it right here. You can copy the following boilerplate and un-indent it (the indentation causes it to be inside this comment). Special New Feature ------------------- Makes programs 10x faster by doing Special New Thing. Prefix data rework ------------------ The semantics of the ``prefix`` attribute have been changed. Users that want the previous ``prefix`` semantics should instead use ``prologue``. To motivate this change, let's examine the primary usecases that these attributes aim to serve, 1. Code sanitization metadata (e.g. Clang's undefined behavior sanitizer) 2. Function hot-patching: Enable the user to insert ``nop`` operations at the beginning of the function which can later be safely replaced with a call to some instrumentation facility. 3. Language runtime metadata: Allow a compiler to insert data for use by the runtime during execution. GHC is one example of a compiler that needs this functionality for its tables-next-to-code functionality. Previously ``prefix`` served cases (1) and (2) quite well by allowing the user to introduce arbitrary data at the entrypoint but before the function body. Case (3), however, was poorly handled by this approach as it required that prefix data was valid executable code. In this release the concept of prefix data has been redefined to be data which occurs immediately before the function entrypoint (i.e. the symbol address). Since prefix data now occurs before the function entrypoint, there is no need for the data to be valid code. The previous notion of prefix data now goes under the name "prologue data" to emphasize its duality with the function epilogue. The intention here is to handle cases (1) and (2) with prologue data and case (3) with prefix data. See the language reference for further details on the semantics of these attributes. This refactoring arose out of discussions_ with Reid Kleckner in response to a proposal to introduce the notion of symbol offsets to enable handling of case (3). .. _discussions: http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-May/073235.html Metadata is not a Value ----------------------- Metadata nodes (``!{...}``) and strings (``!"..."``) are no longer values. They have no use-lists, no type, cannot RAUW, and cannot be function-local. Bridges between Value and Metadata ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ LLVM intrinsics can reference metadata using the ``metadata`` type, and metadata nodes can reference constant values. Function-local metadata is limited to direct arguments to LLVM intrinsics. Metadata is typeless ^^^^^^^^^^^^^^^^^^^^ The following old IR: .. code-block:: llvm @g = global i32 0 define void @foo(i32 %v) { entry: call void @llvm.md(metadata !{i32 %v}) call void @llvm.md(metadata !{i32* @global}) call void @llvm.md(metadata !0) call void @llvm.md(metadata !{metadata !"string"}) call void @llvm.md(metadata !{metadata !{metadata !1, metadata !"string"}}) ret void, !bar !1, !baz !2 } declare void @llvm.md(metadata) !0 = metadata !{metadata !1, metadata !2, metadata !3, metadata !"some string"} !1 = metadata !{metadata !2, null, metadata !"other", i32* @global, i32 7} !2 = metadata !{} should now be written as: .. code-block:: llvm @g = global i32 0 define void @foo(i32 %v) { entry: call void @llvm.md(metadata i32 %v) ; The only legal place for function-local ; metadata. call void @llvm.md(metadata i32* @global) call void @llvm.md(metadata !0) call void @llvm.md(metadata !{!"string"}) call void @llvm.md(metadata !{!{!1, !"string"}}) ret void, !bar !1, !baz !2 } declare void @llvm.md(metadata) !0 = !{!1, !2, !3, !"some string"} !1 = !{!2, null, !"other", i32* @global, i32 7} !2 = !{} Distinct metadata nodes ^^^^^^^^^^^^^^^^^^^^^^^ Metadata nodes can opt-out of uniquing, using the keyword ``distinct``. Distinct nodes are still owned by the context, but are stored in a side table, and not uniqued. In LLVM 3.5, metadata nodes would drop uniquing if an operand changed to ``null`` during optimizations. This is no longer true. However, if an operand change causes a uniquing collision, they become ``distinct``. Unlike LLVM 3.5, where serializing to assembly or bitcode would re-unique the nodes, they now remain ``distinct``. The following IR: .. code-block:: llvm !named = !{!0, !1, !2, !3, !4, !5, !6, !7, !8} !0 = !{} !1 = !{} !2 = distinct !{} !3 = distinct !{} !4 = !{!0} !5 = distinct !{!0} !6 = !{!4, !{}, !5} !7 = !{!{!0}, !0, !5} !8 = distinct !{!{!0}, !0, !5} is equivalent to the following: .. code-block:: llvm !named = !{!0, !0, !1, !2, !3, !4, !5, !5, !6} !0 = !{} !1 = distinct !{} !2 = distinct !{} !3 = !{!0} !4 = distinct !{!0} !5 = !{!3, !0, !4} !6 = distinct !{!3, !0, !4} Constructing cyclic graphs ^^^^^^^^^^^^^^^^^^^^^^^^^^ During graph construction, if a metadata node transitively references a forward declaration, the node itself is considered "unresolved" until the forward declaration resolves. An unresolved node can RAUW itself to support uniquing. Nodes automatically resolve once all their operands have resolved. However, cyclic graphs prevent the nodes from resolving. An API client that constructs a cyclic graph must call ``resolveCycles()`` to resolve nodes in the cycle. To save self-references from that burden, self-referencing nodes are implicitly ``distinct``. So the following IR: .. code-block:: llvm !named = !{!0, !1, !2, !3, !4} !0 = !{!0} !1 = !{!1} !2 = !{!2, !1} !3 = !{!2, !1} !4 = !{!2, !1} is equivalent to: .. code-block:: llvm !named = !{!0, !1, !2, !3, !3} !0 = distinct !{!0} !1 = distinct !{!1} !2 = distinct !{!2, !1} !3 = !{!2, !1} MDLocation (aka DebugLoc aka DILocation) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ There's a new first-class metadata construct called ``MDLocation`` (to be followed in subsequent releases by others). It's used for the locations referenced by ``!dbg`` metadata attachments. For example, if an old ``!dbg`` attachment looked like this: .. code-block:: llvm define i32 @foo(i32 %a, i32 %b) { entry: %add = add i32 %a, %b, !dbg !0 ret %add, !dbg !1 } !0 = metadata !{i32 10, i32 3, metadata !2, metadata !1) !1 = metadata !{i32 20, i32 7, metadata !3) !2 = metadata !{...} !3 = metadata !{...} the new attachment looks like this: .. code-block:: llvm define i32 @foo(i32 %a, i32 %b) { entry: %add = add i32 %a, %b, !dbg !0 ret %add, !dbg !1 } !0 = !MDLocation(line: 10, column: 3, scope: !2, inlinedAt: !1) !1 = !MDLocation(line: 20, column: 7, scope: !3) !2 = !{...} !3 = !{...} The fields are named, can be reordered, and have sane defaults if left out (although ``scope:`` is required). Alias syntax change ----------------------- The syntax for aliases is now closer to what is used for global variables .. code-block:: llvm @a = weak global ... @b = weak alias ... The order of the ``alias`` keyword and the linkage was swapped before. The old JIT has been removed ---------------------------- All users should transition to MCJIT. -object::Binary doesn't owns the file buffer +object::Binary doesn't own the file buffer ------------------------------------------- It is now just a wrapper, which simplifies using object::Binary with other users of the underlying file. + IR in object files is now supported ----------------------------------- Regular object files can contain IR in a section named ``.llvmbc``. The gold plugin has been rewritten ---------------------------------- It is now implemented directly on top of lib/Linker instead of ``lib/LTO``. The API of ``lib/LTO`` is sufficiently different from gold's view of the linking process that some cases could not be conveniently implemented. The new implementation is also lazier and has a ``save-temps`` option. Change in the representation of lazy loaded funcs ------------------------------------------------- -Lazy loaded functions are now represented is a way that ``isDeclaration`` +Lazy loaded functions are now represented in a way that ``isDeclaration`` returns the correct answer even before reading the body. The opt option -std-compile-opts was removed -------------------------------------------- It was effectively an alias of -O3. Python 2.7 is now required -------------------------- This was done to simplify compatibility with python 3. + The leak detector has been removed ---------------------------------- -In practice tools like asan and valgrind were finding way more bugs than +In practice, tools like asan and valgrind were finding way more bugs than the old leak detector, so it was removed. New comdat syntax ----------------- The syntax of comdats was changed to .. code-block:: llvm $c = comdat any @g = global i32 0, comdat($c) @c = global i32 0, comdat -The version without the parentheses is a syntatic sugar for a comdat with +The version without the parentheses is a syntactic sugar for a comdat with the same name as the global. -Diagnotic infrastructure used by lib/Linker and lib/Bitcode ------------------------------------------------------------ +Added support for Win64 unwind information +------------------------------------------ +LLVM now obeys the `Win64 prologue and epilogue conventions +`_ documented by +Microsoft. Unwind information is also emitted into the .xdata section. + +As a result of the ABI-required prologue changes, it is now no longer possible +to unwind the stack using a standard frame pointer walk on Win64. Instead, +users should call ``CaptureStackBackTrace``, or implement equivalent +functionality by consulting the unwind tables present in the binary. + + +Diagnostic infrastructure used by lib/Linker and lib/Bitcode +------------------------------------------------------------ + These libraries now use the diagnostic handler to print errors and warnings. This provides better error messages and simpler error handling. The PreserveSource linker mode was removed ------------------------------------------ It was fairly broken and was removed. +The mode is currently still available in the C API for source +compatibility, but it doesn't have any effect. -Changes to the ARM Backend --------------------------- +Garbage Collection +------------------ +A new experimental mechanism for describing a garbage collection safepoint was +added to LLVM. The new mechanism was not complete at the point this release +was branched so it is recommended that anyone interested in using this +mechanism track the ongoing development work on tip of tree. The hope is that +these intrinsics will be ready for general use by 3.7. Documentation can be +found `here `_. - During this release ... +The existing gc.root implementation is still supported and as fully featured +as it ever was. However, two features from GCStrategy will likely be removed +in the 3.7 release (performCustomLowering and findCustomSafePoints). If you +have a use case for either, please mention it on llvm-dev so that it can be +considered for future development. +We are expecting to migrate away from gc.root in the 3.8 time frame, +but both mechanisms will be supported in 3.7. + Changes to the MIPS Target -------------------------- During this release the MIPS target has reached a few major milestones. The compiler has gained support for MIPS-II and MIPS-III; become ABI-compatible with GCC for big and little endian O32, N32, and N64; and is now able to compile the Linux kernel for 32-bit targets. Additionally, LLD now supports microMIPS for the O32 ABI on little endian targets, and code generation for microMIPS is almost completely passing the test-suite. + ABI ^^^ A large number of bugs have been fixed for big-endian MIPS targets using the N32 and N64 ABI's as well as a small number of bugs affecting other ABI's. Please note that some of these bugs will still affect LLVM-IR generated by LLVM 3.5 since correct code generation depends on appropriate usage of the ``inreg``, ``signext``, and ``zeroext`` attributes on all function arguments and returns. There are far too many corrections to provide a complete list but here are a few notable ones: * Big-endian N32 and N64 now interlinks successfully with GCC compiled code. Previously this didn't work for the majority of cases. * The registers used to return a structure containing a single 128-bit floating point member on the N32/N64 ABI's have been changed from those specified by the ABI documentation to match those used by GCC. The documentation specifies that ``$f0`` and ``$f2`` should be used but GCC has used ``$f0`` and ``$f1`` for many years. * Returning a zero-byte struct no longer causes arguments to be read from the wrong registers when using the O32 ABI. * The exception personality has been changed for 64-bit MIPS targets to eliminate warnings about relocations in a read-only section. * Incorrect usage of odd-numbered single-precision floating point registers has been fixed when the fastcc calling convention is used with 64-bit FPU's and -mno-odd-spreg. + LLVMLinux ^^^^^^^^^ It is now possible to compile the Linux kernel. This currently requires a small number of kernel patches. See the `LLVMLinux project `_ for details. * Added -mabicalls and -mno-abicalls. The implementation may not be complete but works sufficiently well for the Linux kernel. * Fixed multiple compatibility issues between LLVM's inline assembly support and GCC's. * Added support for a number of directives used by Linux to the Integrated Assembler. + Miscellaneous ^^^^^^^^^^^^^ * Attempting to disassemble l[wd]c[23], s[wd]c[23], cache, and pref no longer triggers an assertion. * Added -muclibc and -mglibc to support toolchains that provide both uClibC and GLibC. * __SIZEOF_INT128__ is no longer defined for 64-bit targets since 128-bit integers do not work at this time for this target. * Using $t4-$t7 with the N32 and N64 ABI is deprecated when ``-fintegrated-as`` is in use and will be removed in LLVM 3.7. These names have never been supported by the GNU Assembler for these ABI's. + Changes to the PowerPC Target ----------------------------- There are numerous improvements to the PowerPC target in this release: * LLVM now generates the Vector-Scalar eXtension (VSX) instructions from version 2.06 of the Power ISA, for both big- and little-endian targets. * LLVM now has a POWER8 instruction scheduling description. -* Address Sanitizer (ASAN) support is now fully functional. +* AddressSanitizer (ASan) support is now fully functional. * Performance of simple atomic accesses has been greatly improved. * Atomic fences now use light-weight syncs where possible, again providing significant performance benefit. * The PowerPC target now supports PIC levels (-fPIC vs. -fpic). * PPC32 SVR4 now supports small-model PIC. +* Experimental support for the stackmap/patchpoint intrinsics has been added. + * There have been many smaller bug fixes and performance improvements. + Changes to the OCaml bindings ----------------------------- * The bindings now require OCaml >=4.00.0, ocamlfind, ctypes >=0.3.0 <0.4 and OUnit 2 if tests are enabled. * The bindings can now be built using cmake as well as autoconf. * LLVM 3.5 has, unfortunately, shipped a broken Llvm_executionengine implementation. In LLVM 3.6, the bindings now fully support MCJIT, however the interface is reworked from scratch using ctypes and is not backwards compatible. * Llvm_linker.Mode was removed following the changes in LLVM. This breaks the interface of Llvm_linker. * All combinations of ocamlc/ocamlc -custom/ocamlopt and shared/static builds of LLVM are now supported. * Absolute paths are not embedded into the OCaml libraries anymore. Either OCaml >=4.02.2 must be used, which includes an rpath-like $ORIGIN mechanism, or META file must be updated for out-of-tree installations; see r221139. * As usual, many more functions have been exposed to OCaml. + +Go bindings +----------- + +* A set of Go bindings based on `gollvm `_ + was introduced in this release. + + External Open Source Projects Using LLVM 3.6 ============================================ An exciting aspect of LLVM is that it is used as an enabling technology for a lot of other language and tools projects. This section lists some of the projects that have already been updated to work with LLVM 3.6. + Portable Computing Language (pocl) ---------------------------------- In addition to producing an easily portable open source OpenCL implementation, another major goal of `pocl `_ is improving performance portability of OpenCL programs with compiler optimizations, reducing the need for target-dependent manual optimizations. An important part of pocl is a set of LLVM passes used to statically parallelize multiple work-items with the kernel compiler, even in the presence of work-group barriers. This enables static parallelization of the fine-grained static concurrency in the work groups in multiple ways. + TTA-based Co-design Environment (TCE) ------------------------------------- `TCE `_ is a toolset for designing customized exposed datapath processors based on the Transport triggered architecture (TTA). The toolset provides a complete co-design flow from C/C++ programs down to synthesizable VHDL/Verilog and parallel program binaries. Processor customization points include the register files, function units, supported operations, and the interconnection network. TCE uses Clang and LLVM for C/C++/OpenCL C language support, target independent optimizations and also for parts of code generation. It generates new LLVM-based code generators "on the fly" for the designed processors and loads them in to the compiler backend as runtime libraries to avoid per-target recompilation of larger parts of the compiler chain. + Likely ------ `Likely `_ is an embeddable just-in-time Lisp for -image recognition and heterogenous computing. Algorithms are just-in-time +image recognition and heterogeneous computing. Algorithms are just-in-time compiled using LLVM's MCJIT infrastructure to execute on single or multi-threaded CPUs and potentially OpenCL SPIR or CUDA enabled GPUs. Likely seeks to explore new optimizations for statistical learning algorithms by moving them from an offline model generation step to the compile-time evaluation of a function (the learning algorithm) with constant arguments (the training data). + LDC - the LLVM-based D compiler ------------------------------- `D `_ is a language with C-like syntax and static typing. It pragmatically combines efficiency, control, and modeling power, with safety and programmer productivity. D supports powerful concepts like Compile-Time Function Execution (CTFE) and Template Meta-Programming, provides an innovative approach to concurrency and offers many classical paradigms. `LDC `_ uses the frontend from the reference compiler combined with LLVM as backend to produce efficient native code. LDC targets x86/x86_64 systems like Linux, OS X, FreeBSD and Windows and also Linux on PowerPC (32/64 bit). Ports to other architectures like ARM, AArch64 and MIPS64 are underway. + + +LLVMSharp & ClangSharp +---------------------- + +`LLVMSharp `_ and +`ClangSharp `_ are type-safe C# bindings for +Microsoft.NET and Mono that Platform Invoke into the native libraries. +ClangSharp is self-hosted and is used to generated LLVMSharp using the +LLVM-C API. + +`LLVMSharp Kaleidoscope Tutorials `_ +are instructive examples of writing a compiler in C#, with certain improvements +like using the visitor pattern to generate LLVM IR. + +`ClangSharp PInvoke Generator `_ is the +self-hosting mechanism for LLVM/ClangSharp and is demonstrative of using +LibClang to generate Platform Invoke (PInvoke) signatures for C APIs. + Additional Information ====================== A wide variety of additional information is available on the `LLVM web page `_, in particular in the `documentation `_ section. The web page also contains versions of the API documentation which is up-to-date with the Subversion version of the source code. You can access versions of these documents specific to this release by going into the ``llvm/docs/`` directory in the LLVM tree. If you have any questions or comments about LLVM, please feel free to contact us via the `mailing lists `_. Index: vendor/llvm/dist/include/llvm-c/Linker.h =================================================================== --- vendor/llvm/dist/include/llvm-c/Linker.h (revision 279290) +++ vendor/llvm/dist/include/llvm-c/Linker.h (revision 279291) @@ -1,35 +1,43 @@ /*===-- llvm-c/Linker.h - Module Linker C Interface -------------*- C++ -*-===*\ |* *| |* The LLVM Compiler Infrastructure *| |* *| |* This file is distributed under the University of Illinois Open Source *| |* License. See LICENSE.TXT for details. *| |* *| |*===----------------------------------------------------------------------===*| |* *| |* This file defines the C interface to the module/file/archive linker. *| |* *| \*===----------------------------------------------------------------------===*/ #ifndef LLVM_C_LINKER_H #define LLVM_C_LINKER_H #include "llvm-c/Core.h" #ifdef __cplusplus extern "C" { #endif + +/* Note: LLVMLinkerPreserveSource has no effect. */ +typedef enum { + LLVMLinkerDestroySource = 0, /* Allow source module to be destroyed. */ + LLVMLinkerPreserveSource = 1 /* Preserve the source module. */ +} LLVMLinkerMode; + + /* Links the source module into the destination module, taking ownership * of the source module away from the caller. Optionally returns a * human-readable description of any errors that occurred in linking. * OutMessage must be disposed with LLVMDisposeMessage. The return value * is true if an error occurred, false otherwise. */ LLVMBool LLVMLinkModules(LLVMModuleRef Dest, LLVMModuleRef Src, - unsigned Unused, char **OutMessage); + LLVMLinkerMode Mode, char **OutMessage); #ifdef __cplusplus } #endif #endif Index: vendor/llvm/dist/lib/Linker/LinkModules.cpp =================================================================== --- vendor/llvm/dist/lib/Linker/LinkModules.cpp (revision 279290) +++ vendor/llvm/dist/lib/Linker/LinkModules.cpp (revision 279291) @@ -1,1764 +1,1764 @@ //===- lib/Linker/LinkModules.cpp - Module Linker Implementation ----------===// // // The LLVM Compiler Infrastructure // // This file is distributed under the University of Illinois Open Source // License. See LICENSE.TXT for details. // //===----------------------------------------------------------------------===// // // This file implements the LLVM module linker. // //===----------------------------------------------------------------------===// #include "llvm/Linker/Linker.h" #include "llvm-c/Linker.h" #include "llvm/ADT/Hashing.h" #include "llvm/ADT/Optional.h" #include "llvm/ADT/SetVector.h" #include "llvm/ADT/SmallString.h" #include "llvm/ADT/Statistic.h" #include "llvm/IR/Constants.h" #include "llvm/IR/DebugInfo.h" #include "llvm/IR/DiagnosticInfo.h" #include "llvm/IR/DiagnosticPrinter.h" #include "llvm/IR/LLVMContext.h" #include "llvm/IR/Module.h" #include "llvm/IR/TypeFinder.h" #include "llvm/Support/CommandLine.h" #include "llvm/Support/Debug.h" #include "llvm/Support/raw_ostream.h" #include "llvm/Transforms/Utils/Cloning.h" #include #include using namespace llvm; //===----------------------------------------------------------------------===// // TypeMap implementation. //===----------------------------------------------------------------------===// namespace { class TypeMapTy : public ValueMapTypeRemapper { /// This is a mapping from a source type to a destination type to use. DenseMap MappedTypes; /// When checking to see if two subgraphs are isomorphic, we speculatively /// add types to MappedTypes, but keep track of them here in case we need to /// roll back. SmallVector SpeculativeTypes; SmallVector SpeculativeDstOpaqueTypes; /// This is a list of non-opaque structs in the source module that are mapped /// to an opaque struct in the destination module. SmallVector SrcDefinitionsToResolve; /// This is the set of opaque types in the destination modules who are /// getting a body from the source module. SmallPtrSet DstResolvedOpaqueTypes; public: TypeMapTy(Linker::IdentifiedStructTypeSet &DstStructTypesSet) : DstStructTypesSet(DstStructTypesSet) {} Linker::IdentifiedStructTypeSet &DstStructTypesSet; /// Indicate that the specified type in the destination module is conceptually /// equivalent to the specified type in the source module. void addTypeMapping(Type *DstTy, Type *SrcTy); /// Produce a body for an opaque type in the dest module from a type /// definition in the source module. void linkDefinedTypeBodies(); /// Return the mapped type to use for the specified input type from the /// source module. Type *get(Type *SrcTy); Type *get(Type *SrcTy, SmallPtrSet &Visited); void finishType(StructType *DTy, StructType *STy, ArrayRef ETypes); FunctionType *get(FunctionType *T) { return cast(get((Type *)T)); } /// Dump out the type map for debugging purposes. void dump() const { for (auto &Pair : MappedTypes) { dbgs() << "TypeMap: "; Pair.first->print(dbgs()); dbgs() << " => "; Pair.second->print(dbgs()); dbgs() << '\n'; } } private: Type *remapType(Type *SrcTy) override { return get(SrcTy); } bool areTypesIsomorphic(Type *DstTy, Type *SrcTy); }; } void TypeMapTy::addTypeMapping(Type *DstTy, Type *SrcTy) { assert(SpeculativeTypes.empty()); assert(SpeculativeDstOpaqueTypes.empty()); // Check to see if these types are recursively isomorphic and establish a // mapping between them if so. if (!areTypesIsomorphic(DstTy, SrcTy)) { // Oops, they aren't isomorphic. Just discard this request by rolling out // any speculative mappings we've established. for (Type *Ty : SpeculativeTypes) MappedTypes.erase(Ty); SrcDefinitionsToResolve.resize(SrcDefinitionsToResolve.size() - SpeculativeDstOpaqueTypes.size()); for (StructType *Ty : SpeculativeDstOpaqueTypes) DstResolvedOpaqueTypes.erase(Ty); } else { for (Type *Ty : SpeculativeTypes) if (auto *STy = dyn_cast(Ty)) if (STy->hasName()) STy->setName(""); } SpeculativeTypes.clear(); SpeculativeDstOpaqueTypes.clear(); } /// Recursively walk this pair of types, returning true if they are isomorphic, /// false if they are not. bool TypeMapTy::areTypesIsomorphic(Type *DstTy, Type *SrcTy) { // Two types with differing kinds are clearly not isomorphic. if (DstTy->getTypeID() != SrcTy->getTypeID()) return false; // If we have an entry in the MappedTypes table, then we have our answer. Type *&Entry = MappedTypes[SrcTy]; if (Entry) return Entry == DstTy; // Two identical types are clearly isomorphic. Remember this // non-speculatively. if (DstTy == SrcTy) { Entry = DstTy; return true; } // Okay, we have two types with identical kinds that we haven't seen before. // If this is an opaque struct type, special case it. if (StructType *SSTy = dyn_cast(SrcTy)) { // Mapping an opaque type to any struct, just keep the dest struct. if (SSTy->isOpaque()) { Entry = DstTy; SpeculativeTypes.push_back(SrcTy); return true; } // Mapping a non-opaque source type to an opaque dest. If this is the first // type that we're mapping onto this destination type then we succeed. Keep // the dest, but fill it in later. If this is the second (different) type // that we're trying to map onto the same opaque type then we fail. if (cast(DstTy)->isOpaque()) { // We can only map one source type onto the opaque destination type. if (!DstResolvedOpaqueTypes.insert(cast(DstTy)).second) return false; SrcDefinitionsToResolve.push_back(SSTy); SpeculativeTypes.push_back(SrcTy); SpeculativeDstOpaqueTypes.push_back(cast(DstTy)); Entry = DstTy; return true; } } // If the number of subtypes disagree between the two types, then we fail. if (SrcTy->getNumContainedTypes() != DstTy->getNumContainedTypes()) return false; // Fail if any of the extra properties (e.g. array size) of the type disagree. if (isa(DstTy)) return false; // bitwidth disagrees. if (PointerType *PT = dyn_cast(DstTy)) { if (PT->getAddressSpace() != cast(SrcTy)->getAddressSpace()) return false; } else if (FunctionType *FT = dyn_cast(DstTy)) { if (FT->isVarArg() != cast(SrcTy)->isVarArg()) return false; } else if (StructType *DSTy = dyn_cast(DstTy)) { StructType *SSTy = cast(SrcTy); if (DSTy->isLiteral() != SSTy->isLiteral() || DSTy->isPacked() != SSTy->isPacked()) return false; } else if (ArrayType *DATy = dyn_cast(DstTy)) { if (DATy->getNumElements() != cast(SrcTy)->getNumElements()) return false; } else if (VectorType *DVTy = dyn_cast(DstTy)) { if (DVTy->getNumElements() != cast(SrcTy)->getNumElements()) return false; } // Otherwise, we speculate that these two types will line up and recursively // check the subelements. Entry = DstTy; SpeculativeTypes.push_back(SrcTy); for (unsigned I = 0, E = SrcTy->getNumContainedTypes(); I != E; ++I) if (!areTypesIsomorphic(DstTy->getContainedType(I), SrcTy->getContainedType(I))) return false; // If everything seems to have lined up, then everything is great. return true; } void TypeMapTy::linkDefinedTypeBodies() { SmallVector Elements; for (StructType *SrcSTy : SrcDefinitionsToResolve) { StructType *DstSTy = cast(MappedTypes[SrcSTy]); assert(DstSTy->isOpaque()); // Map the body of the source type over to a new body for the dest type. Elements.resize(SrcSTy->getNumElements()); for (unsigned I = 0, E = Elements.size(); I != E; ++I) Elements[I] = get(SrcSTy->getElementType(I)); DstSTy->setBody(Elements, SrcSTy->isPacked()); } SrcDefinitionsToResolve.clear(); DstResolvedOpaqueTypes.clear(); } void TypeMapTy::finishType(StructType *DTy, StructType *STy, ArrayRef ETypes) { DTy->setBody(ETypes, STy->isPacked()); // Steal STy's name. if (STy->hasName()) { SmallString<16> TmpName = STy->getName(); STy->setName(""); DTy->setName(TmpName); } DstStructTypesSet.addNonOpaque(DTy); } Type *TypeMapTy::get(Type *Ty) { SmallPtrSet Visited; return get(Ty, Visited); } Type *TypeMapTy::get(Type *Ty, SmallPtrSet &Visited) { // If we already have an entry for this type, return it. Type **Entry = &MappedTypes[Ty]; if (*Entry) return *Entry; // These are types that LLVM itself will unique. bool IsUniqued = !isa(Ty) || cast(Ty)->isLiteral(); #ifndef NDEBUG if (!IsUniqued) { for (auto &Pair : MappedTypes) { assert(!(Pair.first != Ty && Pair.second == Ty) && "mapping to a source type"); } } #endif if (!IsUniqued && !Visited.insert(cast(Ty)).second) { StructType *DTy = StructType::create(Ty->getContext()); return *Entry = DTy; } // If this is not a recursive type, then just map all of the elements and // then rebuild the type from inside out. SmallVector ElementTypes; // If there are no element types to map, then the type is itself. This is // true for the anonymous {} struct, things like 'float', integers, etc. if (Ty->getNumContainedTypes() == 0 && IsUniqued) return *Entry = Ty; // Remap all of the elements, keeping track of whether any of them change. bool AnyChange = false; ElementTypes.resize(Ty->getNumContainedTypes()); for (unsigned I = 0, E = Ty->getNumContainedTypes(); I != E; ++I) { ElementTypes[I] = get(Ty->getContainedType(I), Visited); AnyChange |= ElementTypes[I] != Ty->getContainedType(I); } // If we found our type while recursively processing stuff, just use it. Entry = &MappedTypes[Ty]; if (*Entry) { if (auto *DTy = dyn_cast(*Entry)) { if (DTy->isOpaque()) { auto *STy = cast(Ty); finishType(DTy, STy, ElementTypes); } } return *Entry; } // If all of the element types mapped directly over and the type is not // a nomed struct, then the type is usable as-is. if (!AnyChange && IsUniqued) return *Entry = Ty; // Otherwise, rebuild a modified type. switch (Ty->getTypeID()) { default: llvm_unreachable("unknown derived type to remap"); case Type::ArrayTyID: return *Entry = ArrayType::get(ElementTypes[0], cast(Ty)->getNumElements()); case Type::VectorTyID: return *Entry = VectorType::get(ElementTypes[0], cast(Ty)->getNumElements()); case Type::PointerTyID: return *Entry = PointerType::get(ElementTypes[0], cast(Ty)->getAddressSpace()); case Type::FunctionTyID: return *Entry = FunctionType::get(ElementTypes[0], makeArrayRef(ElementTypes).slice(1), cast(Ty)->isVarArg()); case Type::StructTyID: { auto *STy = cast(Ty); bool IsPacked = STy->isPacked(); if (IsUniqued) return *Entry = StructType::get(Ty->getContext(), ElementTypes, IsPacked); // If the type is opaque, we can just use it directly. if (STy->isOpaque()) { DstStructTypesSet.addOpaque(STy); return *Entry = Ty; } if (StructType *OldT = DstStructTypesSet.findNonOpaque(ElementTypes, IsPacked)) { STy->setName(""); return *Entry = OldT; } if (!AnyChange) { DstStructTypesSet.addNonOpaque(STy); return *Entry = Ty; } StructType *DTy = StructType::create(Ty->getContext()); finishType(DTy, STy, ElementTypes); return *Entry = DTy; } } } //===----------------------------------------------------------------------===// // ModuleLinker implementation. //===----------------------------------------------------------------------===// namespace { class ModuleLinker; /// Creates prototypes for functions that are lazily linked on the fly. This /// speeds up linking for modules with many/ lazily linked functions of which /// few get used. class ValueMaterializerTy : public ValueMaterializer { TypeMapTy &TypeMap; Module *DstM; std::vector &LazilyLinkGlobalValues; public: ValueMaterializerTy(TypeMapTy &TypeMap, Module *DstM, std::vector &LazilyLinkGlobalValues) : ValueMaterializer(), TypeMap(TypeMap), DstM(DstM), LazilyLinkGlobalValues(LazilyLinkGlobalValues) {} Value *materializeValueFor(Value *V) override; }; class LinkDiagnosticInfo : public DiagnosticInfo { const Twine &Msg; public: LinkDiagnosticInfo(DiagnosticSeverity Severity, const Twine &Msg); void print(DiagnosticPrinter &DP) const override; }; LinkDiagnosticInfo::LinkDiagnosticInfo(DiagnosticSeverity Severity, const Twine &Msg) : DiagnosticInfo(DK_Linker, Severity), Msg(Msg) {} void LinkDiagnosticInfo::print(DiagnosticPrinter &DP) const { DP << Msg; } /// This is an implementation class for the LinkModules function, which is the /// entrypoint for this file. class ModuleLinker { Module *DstM, *SrcM; TypeMapTy TypeMap; ValueMaterializerTy ValMaterializer; /// Mapping of values from what they used to be in Src, to what they are now /// in DstM. ValueToValueMapTy is a ValueMap, which involves some overhead /// due to the use of Value handles which the Linker doesn't actually need, /// but this allows us to reuse the ValueMapper code. ValueToValueMapTy ValueMap; struct AppendingVarInfo { GlobalVariable *NewGV; // New aggregate global in dest module. const Constant *DstInit; // Old initializer from dest module. const Constant *SrcInit; // Old initializer from src module. }; std::vector AppendingVars; // Set of items not to link in from source. SmallPtrSet DoNotLinkFromSource; // Vector of GlobalValues to lazily link in. std::vector LazilyLinkGlobalValues; /// Functions that have replaced other functions. SmallPtrSet OverridingFunctions; DiagnosticHandlerFunction DiagnosticHandler; public: ModuleLinker(Module *dstM, Linker::IdentifiedStructTypeSet &Set, Module *srcM, DiagnosticHandlerFunction DiagnosticHandler) : DstM(dstM), SrcM(srcM), TypeMap(Set), ValMaterializer(TypeMap, DstM, LazilyLinkGlobalValues), DiagnosticHandler(DiagnosticHandler) {} bool run(); private: bool shouldLinkFromSource(bool &LinkFromSrc, const GlobalValue &Dest, const GlobalValue &Src); /// Helper method for setting a message and returning an error code. bool emitError(const Twine &Message) { DiagnosticHandler(LinkDiagnosticInfo(DS_Error, Message)); return true; } void emitWarning(const Twine &Message) { DiagnosticHandler(LinkDiagnosticInfo(DS_Warning, Message)); } bool getComdatLeader(Module *M, StringRef ComdatName, const GlobalVariable *&GVar); bool computeResultingSelectionKind(StringRef ComdatName, Comdat::SelectionKind Src, Comdat::SelectionKind Dst, Comdat::SelectionKind &Result, bool &LinkFromSrc); std::map> ComdatsChosen; bool getComdatResult(const Comdat *SrcC, Comdat::SelectionKind &SK, bool &LinkFromSrc); /// Given a global in the source module, return the global in the /// destination module that is being linked to, if any. GlobalValue *getLinkedToGlobal(const GlobalValue *SrcGV) { // If the source has no name it can't link. If it has local linkage, // there is no name match-up going on. if (!SrcGV->hasName() || SrcGV->hasLocalLinkage()) return nullptr; // Otherwise see if we have a match in the destination module's symtab. GlobalValue *DGV = DstM->getNamedValue(SrcGV->getName()); if (!DGV) return nullptr; // If we found a global with the same name in the dest module, but it has // internal linkage, we are really not doing any linkage here. if (DGV->hasLocalLinkage()) return nullptr; // Otherwise, we do in fact link to the destination global. return DGV; } void computeTypeMapping(); void upgradeMismatchedGlobalArray(StringRef Name); void upgradeMismatchedGlobals(); bool linkAppendingVarProto(GlobalVariable *DstGV, const GlobalVariable *SrcGV); bool linkGlobalValueProto(GlobalValue *GV); bool linkModuleFlagsMetadata(); void linkAppendingVarInit(const AppendingVarInfo &AVI); void linkGlobalInit(GlobalVariable &Dst, GlobalVariable &Src); bool linkFunctionBody(Function &Dst, Function &Src); void linkAliasBody(GlobalAlias &Dst, GlobalAlias &Src); bool linkGlobalValueBody(GlobalValue &Src); void linkNamedMDNodes(); void stripReplacedSubprograms(); }; } /// The LLVM SymbolTable class autorenames globals that conflict in the symbol /// table. This is good for all clients except for us. Go through the trouble /// to force this back. static void forceRenaming(GlobalValue *GV, StringRef Name) { // If the global doesn't force its name or if it already has the right name, // there is nothing for us to do. if (GV->hasLocalLinkage() || GV->getName() == Name) return; Module *M = GV->getParent(); // If there is a conflict, rename the conflict. if (GlobalValue *ConflictGV = M->getNamedValue(Name)) { GV->takeName(ConflictGV); ConflictGV->setName(Name); // This will cause ConflictGV to get renamed assert(ConflictGV->getName() != Name && "forceRenaming didn't work"); } else { GV->setName(Name); // Force the name back } } /// copy additional attributes (those not needed to construct a GlobalValue) /// from the SrcGV to the DestGV. static void copyGVAttributes(GlobalValue *DestGV, const GlobalValue *SrcGV) { DestGV->copyAttributesFrom(SrcGV); forceRenaming(DestGV, SrcGV->getName()); } static bool isLessConstraining(GlobalValue::VisibilityTypes a, GlobalValue::VisibilityTypes b) { if (a == GlobalValue::HiddenVisibility) return false; if (b == GlobalValue::HiddenVisibility) return true; if (a == GlobalValue::ProtectedVisibility) return false; if (b == GlobalValue::ProtectedVisibility) return true; return false; } /// Loop through the global variables in the src module and merge them into the /// dest module. static GlobalVariable *copyGlobalVariableProto(TypeMapTy &TypeMap, Module &DstM, const GlobalVariable *SGVar) { // No linking to be performed or linking from the source: simply create an // identical version of the symbol over in the dest module... the // initializer will be filled in later by LinkGlobalInits. GlobalVariable *NewDGV = new GlobalVariable( DstM, TypeMap.get(SGVar->getType()->getElementType()), SGVar->isConstant(), SGVar->getLinkage(), /*init*/ nullptr, SGVar->getName(), /*insertbefore*/ nullptr, SGVar->getThreadLocalMode(), SGVar->getType()->getAddressSpace()); return NewDGV; } /// Link the function in the source module into the destination module if /// needed, setting up mapping information. static Function *copyFunctionProto(TypeMapTy &TypeMap, Module &DstM, const Function *SF) { // If there is no linkage to be performed or we are linking from the source, // bring SF over. return Function::Create(TypeMap.get(SF->getFunctionType()), SF->getLinkage(), SF->getName(), &DstM); } /// Set up prototypes for any aliases that come over from the source module. static GlobalAlias *copyGlobalAliasProto(TypeMapTy &TypeMap, Module &DstM, const GlobalAlias *SGA) { // If there is no linkage to be performed or we're linking from the source, // bring over SGA. auto *PTy = cast(TypeMap.get(SGA->getType())); return GlobalAlias::create(PTy->getElementType(), PTy->getAddressSpace(), SGA->getLinkage(), SGA->getName(), &DstM); } static GlobalValue *copyGlobalValueProto(TypeMapTy &TypeMap, Module &DstM, const GlobalValue *SGV) { GlobalValue *NewGV; if (auto *SGVar = dyn_cast(SGV)) NewGV = copyGlobalVariableProto(TypeMap, DstM, SGVar); else if (auto *SF = dyn_cast(SGV)) NewGV = copyFunctionProto(TypeMap, DstM, SF); else NewGV = copyGlobalAliasProto(TypeMap, DstM, cast(SGV)); copyGVAttributes(NewGV, SGV); return NewGV; } Value *ValueMaterializerTy::materializeValueFor(Value *V) { auto *SGV = dyn_cast(V); if (!SGV) return nullptr; GlobalValue *DGV = copyGlobalValueProto(TypeMap, *DstM, SGV); if (Comdat *SC = SGV->getComdat()) { if (auto *DGO = dyn_cast(DGV)) { Comdat *DC = DstM->getOrInsertComdat(SC->getName()); DGO->setComdat(DC); } } LazilyLinkGlobalValues.push_back(SGV); return DGV; } bool ModuleLinker::getComdatLeader(Module *M, StringRef ComdatName, const GlobalVariable *&GVar) { const GlobalValue *GVal = M->getNamedValue(ComdatName); if (const auto *GA = dyn_cast_or_null(GVal)) { GVal = GA->getBaseObject(); if (!GVal) // We cannot resolve the size of the aliasee yet. return emitError("Linking COMDATs named '" + ComdatName + "': COMDAT key involves incomputable alias size."); } GVar = dyn_cast_or_null(GVal); if (!GVar) return emitError( "Linking COMDATs named '" + ComdatName + "': GlobalVariable required for data dependent selection!"); return false; } bool ModuleLinker::computeResultingSelectionKind(StringRef ComdatName, Comdat::SelectionKind Src, Comdat::SelectionKind Dst, Comdat::SelectionKind &Result, bool &LinkFromSrc) { // The ability to mix Comdat::SelectionKind::Any with // Comdat::SelectionKind::Largest is a behavior that comes from COFF. bool DstAnyOrLargest = Dst == Comdat::SelectionKind::Any || Dst == Comdat::SelectionKind::Largest; bool SrcAnyOrLargest = Src == Comdat::SelectionKind::Any || Src == Comdat::SelectionKind::Largest; if (DstAnyOrLargest && SrcAnyOrLargest) { if (Dst == Comdat::SelectionKind::Largest || Src == Comdat::SelectionKind::Largest) Result = Comdat::SelectionKind::Largest; else Result = Comdat::SelectionKind::Any; } else if (Src == Dst) { Result = Dst; } else { return emitError("Linking COMDATs named '" + ComdatName + "': invalid selection kinds!"); } switch (Result) { case Comdat::SelectionKind::Any: // Go with Dst. LinkFromSrc = false; break; case Comdat::SelectionKind::NoDuplicates: return emitError("Linking COMDATs named '" + ComdatName + "': noduplicates has been violated!"); case Comdat::SelectionKind::ExactMatch: case Comdat::SelectionKind::Largest: case Comdat::SelectionKind::SameSize: { const GlobalVariable *DstGV; const GlobalVariable *SrcGV; if (getComdatLeader(DstM, ComdatName, DstGV) || getComdatLeader(SrcM, ComdatName, SrcGV)) return true; const DataLayout *DstDL = DstM->getDataLayout(); const DataLayout *SrcDL = SrcM->getDataLayout(); if (!DstDL || !SrcDL) { return emitError( "Linking COMDATs named '" + ComdatName + "': can't do size dependent selection without DataLayout!"); } uint64_t DstSize = DstDL->getTypeAllocSize(DstGV->getType()->getPointerElementType()); uint64_t SrcSize = SrcDL->getTypeAllocSize(SrcGV->getType()->getPointerElementType()); if (Result == Comdat::SelectionKind::ExactMatch) { if (SrcGV->getInitializer() != DstGV->getInitializer()) return emitError("Linking COMDATs named '" + ComdatName + "': ExactMatch violated!"); LinkFromSrc = false; } else if (Result == Comdat::SelectionKind::Largest) { LinkFromSrc = SrcSize > DstSize; } else if (Result == Comdat::SelectionKind::SameSize) { if (SrcSize != DstSize) return emitError("Linking COMDATs named '" + ComdatName + "': SameSize violated!"); LinkFromSrc = false; } else { llvm_unreachable("unknown selection kind"); } break; } } return false; } bool ModuleLinker::getComdatResult(const Comdat *SrcC, Comdat::SelectionKind &Result, bool &LinkFromSrc) { Comdat::SelectionKind SSK = SrcC->getSelectionKind(); StringRef ComdatName = SrcC->getName(); Module::ComdatSymTabType &ComdatSymTab = DstM->getComdatSymbolTable(); Module::ComdatSymTabType::iterator DstCI = ComdatSymTab.find(ComdatName); if (DstCI == ComdatSymTab.end()) { // Use the comdat if it is only available in one of the modules. LinkFromSrc = true; Result = SSK; return false; } const Comdat *DstC = &DstCI->second; Comdat::SelectionKind DSK = DstC->getSelectionKind(); return computeResultingSelectionKind(ComdatName, SSK, DSK, Result, LinkFromSrc); } bool ModuleLinker::shouldLinkFromSource(bool &LinkFromSrc, const GlobalValue &Dest, const GlobalValue &Src) { // We always have to add Src if it has appending linkage. if (Src.hasAppendingLinkage()) { LinkFromSrc = true; return false; } bool SrcIsDeclaration = Src.isDeclarationForLinker(); bool DestIsDeclaration = Dest.isDeclarationForLinker(); if (SrcIsDeclaration) { // If Src is external or if both Src & Dest are external.. Just link the // external globals, we aren't adding anything. if (Src.hasDLLImportStorageClass()) { // If one of GVs is marked as DLLImport, result should be dllimport'ed. LinkFromSrc = DestIsDeclaration; return false; } // If the Dest is weak, use the source linkage. LinkFromSrc = Dest.hasExternalWeakLinkage(); return false; } if (DestIsDeclaration) { // If Dest is external but Src is not: LinkFromSrc = true; return false; } if (Src.hasCommonLinkage()) { if (Dest.hasLinkOnceLinkage() || Dest.hasWeakLinkage()) { LinkFromSrc = true; return false; } if (!Dest.hasCommonLinkage()) { LinkFromSrc = false; return false; } // FIXME: Make datalayout mandatory and just use getDataLayout(). DataLayout DL(Dest.getParent()); uint64_t DestSize = DL.getTypeAllocSize(Dest.getType()->getElementType()); uint64_t SrcSize = DL.getTypeAllocSize(Src.getType()->getElementType()); LinkFromSrc = SrcSize > DestSize; return false; } if (Src.isWeakForLinker()) { assert(!Dest.hasExternalWeakLinkage()); assert(!Dest.hasAvailableExternallyLinkage()); if (Dest.hasLinkOnceLinkage() && Src.hasWeakLinkage()) { LinkFromSrc = true; return false; } LinkFromSrc = false; return false; } if (Dest.isWeakForLinker()) { assert(Src.hasExternalLinkage()); LinkFromSrc = true; return false; } assert(!Src.hasExternalWeakLinkage()); assert(!Dest.hasExternalWeakLinkage()); assert(Dest.hasExternalLinkage() && Src.hasExternalLinkage() && "Unexpected linkage type!"); return emitError("Linking globals named '" + Src.getName() + "': symbol multiply defined!"); } /// Loop over all of the linked values to compute type mappings. For example, /// if we link "extern Foo *x" and "Foo *x = NULL", then we have two struct /// types 'Foo' but one got renamed when the module was loaded into the same /// LLVMContext. void ModuleLinker::computeTypeMapping() { for (GlobalValue &SGV : SrcM->globals()) { GlobalValue *DGV = getLinkedToGlobal(&SGV); if (!DGV) continue; if (!DGV->hasAppendingLinkage() || !SGV.hasAppendingLinkage()) { TypeMap.addTypeMapping(DGV->getType(), SGV.getType()); continue; } // Unify the element type of appending arrays. ArrayType *DAT = cast(DGV->getType()->getElementType()); ArrayType *SAT = cast(SGV.getType()->getElementType()); TypeMap.addTypeMapping(DAT->getElementType(), SAT->getElementType()); } for (GlobalValue &SGV : *SrcM) { if (GlobalValue *DGV = getLinkedToGlobal(&SGV)) TypeMap.addTypeMapping(DGV->getType(), SGV.getType()); } for (GlobalValue &SGV : SrcM->aliases()) { if (GlobalValue *DGV = getLinkedToGlobal(&SGV)) TypeMap.addTypeMapping(DGV->getType(), SGV.getType()); } // Incorporate types by name, scanning all the types in the source module. // At this point, the destination module may have a type "%foo = { i32 }" for // example. When the source module got loaded into the same LLVMContext, if // it had the same type, it would have been renamed to "%foo.42 = { i32 }". std::vector Types = SrcM->getIdentifiedStructTypes(); for (StructType *ST : Types) { if (!ST->hasName()) continue; // Check to see if there is a dot in the name followed by a digit. size_t DotPos = ST->getName().rfind('.'); if (DotPos == 0 || DotPos == StringRef::npos || ST->getName().back() == '.' || !isdigit(static_cast(ST->getName()[DotPos + 1]))) continue; // Check to see if the destination module has a struct with the prefix name. StructType *DST = DstM->getTypeByName(ST->getName().substr(0, DotPos)); if (!DST) continue; // Don't use it if this actually came from the source module. They're in // the same LLVMContext after all. Also don't use it unless the type is // actually used in the destination module. This can happen in situations // like this: // // Module A Module B // -------- -------- // %Z = type { %A } %B = type { %C.1 } // %A = type { %B.1, [7 x i8] } %C.1 = type { i8* } // %B.1 = type { %C } %A.2 = type { %B.3, [5 x i8] } // %C = type { i8* } %B.3 = type { %C.1 } // // When we link Module B with Module A, the '%B' in Module B is // used. However, that would then use '%C.1'. But when we process '%C.1', // we prefer to take the '%C' version. So we are then left with both // '%C.1' and '%C' being used for the same types. This leads to some // variables using one type and some using the other. if (TypeMap.DstStructTypesSet.hasType(DST)) TypeMap.addTypeMapping(DST, ST); } // Now that we have discovered all of the type equivalences, get a body for // any 'opaque' types in the dest module that are now resolved. TypeMap.linkDefinedTypeBodies(); } static void upgradeGlobalArray(GlobalVariable *GV) { ArrayType *ATy = cast(GV->getType()->getElementType()); StructType *OldTy = cast(ATy->getElementType()); assert(OldTy->getNumElements() == 2 && "Expected to upgrade from 2 elements"); // Get the upgraded 3 element type. PointerType *VoidPtrTy = Type::getInt8Ty(GV->getContext())->getPointerTo(); Type *Tys[3] = {OldTy->getElementType(0), OldTy->getElementType(1), VoidPtrTy}; StructType *NewTy = StructType::get(GV->getContext(), Tys, false); // Build new constants with a null third field filled in. Constant *OldInitC = GV->getInitializer(); ConstantArray *OldInit = dyn_cast(OldInitC); if (!OldInit && !isa(OldInitC)) // Invalid initializer; give up. return; std::vector Initializers; if (OldInit && OldInit->getNumOperands()) { Value *Null = Constant::getNullValue(VoidPtrTy); for (Use &U : OldInit->operands()) { ConstantStruct *Init = cast(U.get()); Initializers.push_back(ConstantStruct::get( NewTy, Init->getOperand(0), Init->getOperand(1), Null, nullptr)); } } assert(Initializers.size() == ATy->getNumElements() && "Failed to copy all array elements"); // Replace the old GV with a new one. ATy = ArrayType::get(NewTy, Initializers.size()); Constant *NewInit = ConstantArray::get(ATy, Initializers); GlobalVariable *NewGV = new GlobalVariable( *GV->getParent(), ATy, GV->isConstant(), GV->getLinkage(), NewInit, "", GV, GV->getThreadLocalMode(), GV->getType()->getAddressSpace(), GV->isExternallyInitialized()); NewGV->copyAttributesFrom(GV); NewGV->takeName(GV); assert(GV->use_empty() && "program cannot use initializer list"); GV->eraseFromParent(); } void ModuleLinker::upgradeMismatchedGlobalArray(StringRef Name) { // Look for the global arrays. auto *DstGV = dyn_cast_or_null(DstM->getNamedValue(Name)); if (!DstGV) return; auto *SrcGV = dyn_cast_or_null(SrcM->getNamedValue(Name)); if (!SrcGV) return; // Check if the types already match. auto *DstTy = cast(DstGV->getType()->getElementType()); auto *SrcTy = cast(TypeMap.get(SrcGV->getType()->getElementType())); if (DstTy == SrcTy) return; // Grab the element types. We can only upgrade an array of a two-field // struct. Only bother if the other one has three-fields. auto *DstEltTy = cast(DstTy->getElementType()); auto *SrcEltTy = cast(SrcTy->getElementType()); if (DstEltTy->getNumElements() == 2 && SrcEltTy->getNumElements() == 3) { upgradeGlobalArray(DstGV); return; } if (DstEltTy->getNumElements() == 3 && SrcEltTy->getNumElements() == 2) upgradeGlobalArray(SrcGV); // We can't upgrade any other differences. } void ModuleLinker::upgradeMismatchedGlobals() { upgradeMismatchedGlobalArray("llvm.global_ctors"); upgradeMismatchedGlobalArray("llvm.global_dtors"); } /// If there were any appending global variables, link them together now. /// Return true on error. bool ModuleLinker::linkAppendingVarProto(GlobalVariable *DstGV, const GlobalVariable *SrcGV) { if (!SrcGV->hasAppendingLinkage() || !DstGV->hasAppendingLinkage()) return emitError("Linking globals named '" + SrcGV->getName() + "': can only link appending global with another appending global!"); ArrayType *DstTy = cast(DstGV->getType()->getElementType()); ArrayType *SrcTy = cast(TypeMap.get(SrcGV->getType()->getElementType())); Type *EltTy = DstTy->getElementType(); // Check to see that they two arrays agree on type. if (EltTy != SrcTy->getElementType()) return emitError("Appending variables with different element types!"); if (DstGV->isConstant() != SrcGV->isConstant()) return emitError("Appending variables linked with different const'ness!"); if (DstGV->getAlignment() != SrcGV->getAlignment()) return emitError( "Appending variables with different alignment need to be linked!"); if (DstGV->getVisibility() != SrcGV->getVisibility()) return emitError( "Appending variables with different visibility need to be linked!"); if (DstGV->hasUnnamedAddr() != SrcGV->hasUnnamedAddr()) return emitError( "Appending variables with different unnamed_addr need to be linked!"); if (StringRef(DstGV->getSection()) != SrcGV->getSection()) return emitError( "Appending variables with different section name need to be linked!"); uint64_t NewSize = DstTy->getNumElements() + SrcTy->getNumElements(); ArrayType *NewType = ArrayType::get(EltTy, NewSize); // Create the new global variable. GlobalVariable *NG = new GlobalVariable(*DstGV->getParent(), NewType, SrcGV->isConstant(), DstGV->getLinkage(), /*init*/nullptr, /*name*/"", DstGV, DstGV->getThreadLocalMode(), DstGV->getType()->getAddressSpace()); // Propagate alignment, visibility and section info. copyGVAttributes(NG, DstGV); AppendingVarInfo AVI; AVI.NewGV = NG; AVI.DstInit = DstGV->getInitializer(); AVI.SrcInit = SrcGV->getInitializer(); AppendingVars.push_back(AVI); // Replace any uses of the two global variables with uses of the new // global. ValueMap[SrcGV] = ConstantExpr::getBitCast(NG, TypeMap.get(SrcGV->getType())); DstGV->replaceAllUsesWith(ConstantExpr::getBitCast(NG, DstGV->getType())); DstGV->eraseFromParent(); // Track the source variable so we don't try to link it. DoNotLinkFromSource.insert(SrcGV); return false; } bool ModuleLinker::linkGlobalValueProto(GlobalValue *SGV) { GlobalValue *DGV = getLinkedToGlobal(SGV); // Handle the ultra special appending linkage case first. if (DGV && DGV->hasAppendingLinkage()) return linkAppendingVarProto(cast(DGV), cast(SGV)); bool LinkFromSrc = true; Comdat *C = nullptr; GlobalValue::VisibilityTypes Visibility = SGV->getVisibility(); bool HasUnnamedAddr = SGV->hasUnnamedAddr(); if (const Comdat *SC = SGV->getComdat()) { Comdat::SelectionKind SK; std::tie(SK, LinkFromSrc) = ComdatsChosen[SC]; C = DstM->getOrInsertComdat(SC->getName()); C->setSelectionKind(SK); } else if (DGV) { if (shouldLinkFromSource(LinkFromSrc, *DGV, *SGV)) return true; } if (!LinkFromSrc) { // Track the source global so that we don't attempt to copy it over when // processing global initializers. DoNotLinkFromSource.insert(SGV); if (DGV) // Make sure to remember this mapping. ValueMap[SGV] = ConstantExpr::getBitCast(DGV, TypeMap.get(SGV->getType())); } if (DGV) { Visibility = isLessConstraining(Visibility, DGV->getVisibility()) ? DGV->getVisibility() : Visibility; HasUnnamedAddr = HasUnnamedAddr && DGV->hasUnnamedAddr(); } if (!LinkFromSrc && !DGV) return false; GlobalValue *NewGV; if (!LinkFromSrc) { NewGV = DGV; } else { // If the GV is to be lazily linked, don't create it just yet. // The ValueMaterializerTy will deal with creating it if it's used. if (!DGV && (SGV->hasLocalLinkage() || SGV->hasLinkOnceLinkage() || SGV->hasAvailableExternallyLinkage())) { DoNotLinkFromSource.insert(SGV); return false; } NewGV = copyGlobalValueProto(TypeMap, *DstM, SGV); if (DGV && isa(DGV)) if (auto *NewF = dyn_cast(NewGV)) OverridingFunctions.insert(NewF); } NewGV->setUnnamedAddr(HasUnnamedAddr); NewGV->setVisibility(Visibility); if (auto *NewGO = dyn_cast(NewGV)) { if (C) NewGO->setComdat(C); if (DGV && DGV->hasCommonLinkage() && SGV->hasCommonLinkage()) NewGO->setAlignment(std::max(DGV->getAlignment(), SGV->getAlignment())); } if (auto *NewGVar = dyn_cast(NewGV)) { auto *DGVar = dyn_cast_or_null(DGV); auto *SGVar = dyn_cast(SGV); if (DGVar && SGVar && DGVar->isDeclaration() && SGVar->isDeclaration() && (!DGVar->isConstant() || !SGVar->isConstant())) NewGVar->setConstant(false); } // Make sure to remember this mapping. if (NewGV != DGV) { if (DGV) { DGV->replaceAllUsesWith(ConstantExpr::getBitCast(NewGV, DGV->getType())); DGV->eraseFromParent(); } ValueMap[SGV] = NewGV; } return false; } static void getArrayElements(const Constant *C, SmallVectorImpl &Dest) { unsigned NumElements = cast(C->getType())->getNumElements(); for (unsigned i = 0; i != NumElements; ++i) Dest.push_back(C->getAggregateElement(i)); } void ModuleLinker::linkAppendingVarInit(const AppendingVarInfo &AVI) { // Merge the initializer. SmallVector DstElements; getArrayElements(AVI.DstInit, DstElements); SmallVector SrcElements; getArrayElements(AVI.SrcInit, SrcElements); ArrayType *NewType = cast(AVI.NewGV->getType()->getElementType()); StringRef Name = AVI.NewGV->getName(); bool IsNewStructor = (Name == "llvm.global_ctors" || Name == "llvm.global_dtors") && cast(NewType->getElementType())->getNumElements() == 3; for (auto *V : SrcElements) { if (IsNewStructor) { Constant *Key = V->getAggregateElement(2); if (DoNotLinkFromSource.count(Key)) continue; } DstElements.push_back( MapValue(V, ValueMap, RF_None, &TypeMap, &ValMaterializer)); } if (IsNewStructor) { NewType = ArrayType::get(NewType->getElementType(), DstElements.size()); AVI.NewGV->mutateType(PointerType::get(NewType, 0)); } AVI.NewGV->setInitializer(ConstantArray::get(NewType, DstElements)); } /// Update the initializers in the Dest module now that all globals that may be /// referenced are in Dest. void ModuleLinker::linkGlobalInit(GlobalVariable &Dst, GlobalVariable &Src) { // Figure out what the initializer looks like in the dest module. Dst.setInitializer(MapValue(Src.getInitializer(), ValueMap, RF_None, &TypeMap, &ValMaterializer)); } /// Copy the source function over into the dest function and fix up references /// to values. At this point we know that Dest is an external function, and /// that Src is not. bool ModuleLinker::linkFunctionBody(Function &Dst, Function &Src) { assert(Dst.isDeclaration() && !Src.isDeclaration()); // Materialize if needed. if (std::error_code EC = Src.materialize()) return emitError(EC.message()); // Link in the prefix data. if (Src.hasPrefixData()) Dst.setPrefixData(MapValue(Src.getPrefixData(), ValueMap, RF_None, &TypeMap, &ValMaterializer)); // Link in the prologue data. if (Src.hasPrologueData()) Dst.setPrologueData(MapValue(Src.getPrologueData(), ValueMap, RF_None, &TypeMap, &ValMaterializer)); // Go through and convert function arguments over, remembering the mapping. Function::arg_iterator DI = Dst.arg_begin(); for (Argument &Arg : Src.args()) { DI->setName(Arg.getName()); // Copy the name over. // Add a mapping to our mapping. ValueMap[&Arg] = DI; ++DI; } // Splice the body of the source function into the dest function. Dst.getBasicBlockList().splice(Dst.end(), Src.getBasicBlockList()); // At this point, all of the instructions and values of the function are now // copied over. The only problem is that they are still referencing values in // the Source function as operands. Loop through all of the operands of the // functions and patch them up to point to the local versions. for (BasicBlock &BB : Dst) for (Instruction &I : BB) RemapInstruction(&I, ValueMap, RF_IgnoreMissingEntries, &TypeMap, &ValMaterializer); // There is no need to map the arguments anymore. for (Argument &Arg : Src.args()) ValueMap.erase(&Arg); Src.Dematerialize(); return false; } void ModuleLinker::linkAliasBody(GlobalAlias &Dst, GlobalAlias &Src) { Constant *Aliasee = Src.getAliasee(); Constant *Val = MapValue(Aliasee, ValueMap, RF_None, &TypeMap, &ValMaterializer); Dst.setAliasee(Val); } bool ModuleLinker::linkGlobalValueBody(GlobalValue &Src) { Value *Dst = ValueMap[&Src]; assert(Dst); if (auto *F = dyn_cast(&Src)) return linkFunctionBody(cast(*Dst), *F); if (auto *GVar = dyn_cast(&Src)) { linkGlobalInit(cast(*Dst), *GVar); return false; } linkAliasBody(cast(*Dst), cast(Src)); return false; } /// Insert all of the named MDNodes in Src into the Dest module. void ModuleLinker::linkNamedMDNodes() { const NamedMDNode *SrcModFlags = SrcM->getModuleFlagsMetadata(); for (Module::const_named_metadata_iterator I = SrcM->named_metadata_begin(), E = SrcM->named_metadata_end(); I != E; ++I) { // Don't link module flags here. Do them separately. if (&*I == SrcModFlags) continue; NamedMDNode *DestNMD = DstM->getOrInsertNamedMetadata(I->getName()); // Add Src elements into Dest node. for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) DestNMD->addOperand(MapMetadata(I->getOperand(i), ValueMap, RF_None, &TypeMap, &ValMaterializer)); } } /// Drop DISubprograms that have been superseded. /// /// FIXME: this creates an asymmetric result: we strip losing subprograms from /// DstM, but leave losing subprograms in SrcM. Instead we should also strip /// losers from SrcM, but this requires extra plumbing in MapMetadata. void ModuleLinker::stripReplacedSubprograms() { // Avoid quadratic runtime by returning early when there's nothing to do. if (OverridingFunctions.empty()) return; // Move the functions now, so the set gets cleared even on early returns. auto Functions = std::move(OverridingFunctions); OverridingFunctions.clear(); // Drop subprograms whose functions have been overridden by the new compile // unit. NamedMDNode *CompileUnits = DstM->getNamedMetadata("llvm.dbg.cu"); if (!CompileUnits) return; for (unsigned I = 0, E = CompileUnits->getNumOperands(); I != E; ++I) { DICompileUnit CU(CompileUnits->getOperand(I)); assert(CU && "Expected valid compile unit"); DITypedArray SPs(CU.getSubprograms()); assert(SPs && "Expected valid subprogram array"); SmallVector NewSPs; NewSPs.reserve(SPs.getNumElements()); for (unsigned S = 0, SE = SPs.getNumElements(); S != SE; ++S) { DISubprogram SP = SPs.getElement(S); if (SP && SP.getFunction() && Functions.count(SP.getFunction())) continue; NewSPs.push_back(SP); } // Redirect operand to the overriding subprogram. if (NewSPs.size() != SPs.getNumElements()) CU.replaceSubprograms(DIArray(MDNode::get(DstM->getContext(), NewSPs))); } } /// Merge the linker flags in Src into the Dest module. bool ModuleLinker::linkModuleFlagsMetadata() { // If the source module has no module flags, we are done. const NamedMDNode *SrcModFlags = SrcM->getModuleFlagsMetadata(); if (!SrcModFlags) return false; // If the destination module doesn't have module flags yet, then just copy // over the source module's flags. NamedMDNode *DstModFlags = DstM->getOrInsertModuleFlagsMetadata(); if (DstModFlags->getNumOperands() == 0) { for (unsigned I = 0, E = SrcModFlags->getNumOperands(); I != E; ++I) DstModFlags->addOperand(SrcModFlags->getOperand(I)); return false; } // First build a map of the existing module flags and requirements. DenseMap> Flags; SmallSetVector Requirements; for (unsigned I = 0, E = DstModFlags->getNumOperands(); I != E; ++I) { MDNode *Op = DstModFlags->getOperand(I); ConstantInt *Behavior = mdconst::extract(Op->getOperand(0)); MDString *ID = cast(Op->getOperand(1)); if (Behavior->getZExtValue() == Module::Require) { Requirements.insert(cast(Op->getOperand(2))); } else { Flags[ID] = std::make_pair(Op, I); } } // Merge in the flags from the source module, and also collect its set of // requirements. bool HasErr = false; for (unsigned I = 0, E = SrcModFlags->getNumOperands(); I != E; ++I) { MDNode *SrcOp = SrcModFlags->getOperand(I); ConstantInt *SrcBehavior = mdconst::extract(SrcOp->getOperand(0)); MDString *ID = cast(SrcOp->getOperand(1)); MDNode *DstOp; unsigned DstIndex; std::tie(DstOp, DstIndex) = Flags.lookup(ID); unsigned SrcBehaviorValue = SrcBehavior->getZExtValue(); // If this is a requirement, add it and continue. if (SrcBehaviorValue == Module::Require) { // If the destination module does not already have this requirement, add // it. if (Requirements.insert(cast(SrcOp->getOperand(2)))) { DstModFlags->addOperand(SrcOp); } continue; } // If there is no existing flag with this ID, just add it. if (!DstOp) { Flags[ID] = std::make_pair(SrcOp, DstModFlags->getNumOperands()); DstModFlags->addOperand(SrcOp); continue; } // Otherwise, perform a merge. ConstantInt *DstBehavior = mdconst::extract(DstOp->getOperand(0)); unsigned DstBehaviorValue = DstBehavior->getZExtValue(); // If either flag has override behavior, handle it first. if (DstBehaviorValue == Module::Override) { // Diagnose inconsistent flags which both have override behavior. if (SrcBehaviorValue == Module::Override && SrcOp->getOperand(2) != DstOp->getOperand(2)) { HasErr |= emitError("linking module flags '" + ID->getString() + "': IDs have conflicting override values"); } continue; } else if (SrcBehaviorValue == Module::Override) { // Update the destination flag to that of the source. DstModFlags->setOperand(DstIndex, SrcOp); Flags[ID].first = SrcOp; continue; } // Diagnose inconsistent merge behavior types. if (SrcBehaviorValue != DstBehaviorValue) { HasErr |= emitError("linking module flags '" + ID->getString() + "': IDs have conflicting behaviors"); continue; } auto replaceDstValue = [&](MDNode *New) { Metadata *FlagOps[] = {DstOp->getOperand(0), ID, New}; MDNode *Flag = MDNode::get(DstM->getContext(), FlagOps); DstModFlags->setOperand(DstIndex, Flag); Flags[ID].first = Flag; }; // Perform the merge for standard behavior types. switch (SrcBehaviorValue) { case Module::Require: case Module::Override: llvm_unreachable("not possible"); case Module::Error: { // Emit an error if the values differ. if (SrcOp->getOperand(2) != DstOp->getOperand(2)) { HasErr |= emitError("linking module flags '" + ID->getString() + "': IDs have conflicting values"); } continue; } case Module::Warning: { // Emit a warning if the values differ. if (SrcOp->getOperand(2) != DstOp->getOperand(2)) { emitWarning("linking module flags '" + ID->getString() + "': IDs have conflicting values"); } continue; } case Module::Append: { MDNode *DstValue = cast(DstOp->getOperand(2)); MDNode *SrcValue = cast(SrcOp->getOperand(2)); SmallVector MDs; MDs.reserve(DstValue->getNumOperands() + SrcValue->getNumOperands()); for (unsigned i = 0, e = DstValue->getNumOperands(); i != e; ++i) MDs.push_back(DstValue->getOperand(i)); for (unsigned i = 0, e = SrcValue->getNumOperands(); i != e; ++i) MDs.push_back(SrcValue->getOperand(i)); replaceDstValue(MDNode::get(DstM->getContext(), MDs)); break; } case Module::AppendUnique: { SmallSetVector Elts; MDNode *DstValue = cast(DstOp->getOperand(2)); MDNode *SrcValue = cast(SrcOp->getOperand(2)); for (unsigned i = 0, e = DstValue->getNumOperands(); i != e; ++i) Elts.insert(DstValue->getOperand(i)); for (unsigned i = 0, e = SrcValue->getNumOperands(); i != e; ++i) Elts.insert(SrcValue->getOperand(i)); replaceDstValue(MDNode::get(DstM->getContext(), makeArrayRef(Elts.begin(), Elts.end()))); break; } } } // Check all of the requirements. for (unsigned I = 0, E = Requirements.size(); I != E; ++I) { MDNode *Requirement = Requirements[I]; MDString *Flag = cast(Requirement->getOperand(0)); Metadata *ReqValue = Requirement->getOperand(1); MDNode *Op = Flags[Flag].first; if (!Op || Op->getOperand(2) != ReqValue) { HasErr |= emitError("linking module flags '" + Flag->getString() + "': does not have the required value"); continue; } } return HasErr; } bool ModuleLinker::run() { assert(DstM && "Null destination module"); assert(SrcM && "Null source module"); // Inherit the target data from the source module if the destination module // doesn't have one already. if (!DstM->getDataLayout() && SrcM->getDataLayout()) DstM->setDataLayout(SrcM->getDataLayout()); // Copy the target triple from the source to dest if the dest's is empty. if (DstM->getTargetTriple().empty() && !SrcM->getTargetTriple().empty()) DstM->setTargetTriple(SrcM->getTargetTriple()); if (SrcM->getDataLayout() && DstM->getDataLayout() && *SrcM->getDataLayout() != *DstM->getDataLayout()) { emitWarning("Linking two modules of different data layouts: '" + SrcM->getModuleIdentifier() + "' is '" + SrcM->getDataLayoutStr() + "' whereas '" + DstM->getModuleIdentifier() + "' is '" + DstM->getDataLayoutStr() + "'\n"); } if (!SrcM->getTargetTriple().empty() && DstM->getTargetTriple() != SrcM->getTargetTriple()) { emitWarning("Linking two modules of different target triples: " + SrcM->getModuleIdentifier() + "' is '" + SrcM->getTargetTriple() + "' whereas '" + DstM->getModuleIdentifier() + "' is '" + DstM->getTargetTriple() + "'\n"); } // Append the module inline asm string. if (!SrcM->getModuleInlineAsm().empty()) { if (DstM->getModuleInlineAsm().empty()) DstM->setModuleInlineAsm(SrcM->getModuleInlineAsm()); else DstM->setModuleInlineAsm(DstM->getModuleInlineAsm()+"\n"+ SrcM->getModuleInlineAsm()); } // Loop over all of the linked values to compute type mappings. computeTypeMapping(); ComdatsChosen.clear(); for (const auto &SMEC : SrcM->getComdatSymbolTable()) { const Comdat &C = SMEC.getValue(); if (ComdatsChosen.count(&C)) continue; Comdat::SelectionKind SK; bool LinkFromSrc; if (getComdatResult(&C, SK, LinkFromSrc)) return true; ComdatsChosen[&C] = std::make_pair(SK, LinkFromSrc); } // Upgrade mismatched global arrays. upgradeMismatchedGlobals(); // Insert all of the globals in src into the DstM module... without linking // initializers (which could refer to functions not yet mapped over). for (Module::global_iterator I = SrcM->global_begin(), E = SrcM->global_end(); I != E; ++I) if (linkGlobalValueProto(I)) return true; // Link the functions together between the two modules, without doing function // bodies... this just adds external function prototypes to the DstM // function... We do this so that when we begin processing function bodies, // all of the global values that may be referenced are available in our // ValueMap. for (Module::iterator I = SrcM->begin(), E = SrcM->end(); I != E; ++I) if (linkGlobalValueProto(I)) return true; // If there were any aliases, link them now. for (Module::alias_iterator I = SrcM->alias_begin(), E = SrcM->alias_end(); I != E; ++I) if (linkGlobalValueProto(I)) return true; for (unsigned i = 0, e = AppendingVars.size(); i != e; ++i) linkAppendingVarInit(AppendingVars[i]); for (const auto &Entry : DstM->getComdatSymbolTable()) { const Comdat &C = Entry.getValue(); if (C.getSelectionKind() == Comdat::Any) continue; const GlobalValue *GV = SrcM->getNamedValue(C.getName()); assert(GV); MapValue(GV, ValueMap, RF_None, &TypeMap, &ValMaterializer); } // Link in the function bodies that are defined in the source module into // DstM. for (Function &SF : *SrcM) { // Skip if no body (function is external). if (SF.isDeclaration()) continue; // Skip if not linking from source. if (DoNotLinkFromSource.count(&SF)) continue; if (linkGlobalValueBody(SF)) return true; } // Resolve all uses of aliases with aliasees. for (GlobalAlias &Src : SrcM->aliases()) { if (DoNotLinkFromSource.count(&Src)) continue; linkGlobalValueBody(Src); } // Strip replaced subprograms before linking together compile units. stripReplacedSubprograms(); // Remap all of the named MDNodes in Src into the DstM module. We do this // after linking GlobalValues so that MDNodes that reference GlobalValues // are properly remapped. linkNamedMDNodes(); // Merge the module flags into the DstM module. if (linkModuleFlagsMetadata()) return true; // Update the initializers in the DstM module now that all globals that may // be referenced are in DstM. for (GlobalVariable &Src : SrcM->globals()) { // Only process initialized GV's or ones not already in dest. if (!Src.hasInitializer() || DoNotLinkFromSource.count(&Src)) continue; linkGlobalValueBody(Src); } // Process vector of lazily linked in functions. while (!LazilyLinkGlobalValues.empty()) { GlobalValue *SGV = LazilyLinkGlobalValues.back(); LazilyLinkGlobalValues.pop_back(); assert(!SGV->isDeclaration() && "users should not pass down decls"); if (linkGlobalValueBody(*SGV)) return true; } return false; } Linker::StructTypeKeyInfo::KeyTy::KeyTy(ArrayRef E, bool P) : ETypes(E), IsPacked(P) {} Linker::StructTypeKeyInfo::KeyTy::KeyTy(const StructType *ST) : ETypes(ST->elements()), IsPacked(ST->isPacked()) {} bool Linker::StructTypeKeyInfo::KeyTy::operator==(const KeyTy &That) const { if (IsPacked != That.IsPacked) return false; if (ETypes != That.ETypes) return false; return true; } bool Linker::StructTypeKeyInfo::KeyTy::operator!=(const KeyTy &That) const { return !this->operator==(That); } StructType *Linker::StructTypeKeyInfo::getEmptyKey() { return DenseMapInfo::getEmptyKey(); } StructType *Linker::StructTypeKeyInfo::getTombstoneKey() { return DenseMapInfo::getTombstoneKey(); } unsigned Linker::StructTypeKeyInfo::getHashValue(const KeyTy &Key) { return hash_combine(hash_combine_range(Key.ETypes.begin(), Key.ETypes.end()), Key.IsPacked); } unsigned Linker::StructTypeKeyInfo::getHashValue(const StructType *ST) { return getHashValue(KeyTy(ST)); } bool Linker::StructTypeKeyInfo::isEqual(const KeyTy &LHS, const StructType *RHS) { if (RHS == getEmptyKey() || RHS == getTombstoneKey()) return false; return LHS == KeyTy(RHS); } bool Linker::StructTypeKeyInfo::isEqual(const StructType *LHS, const StructType *RHS) { if (RHS == getEmptyKey()) return LHS == getEmptyKey(); if (RHS == getTombstoneKey()) return LHS == getTombstoneKey(); return KeyTy(LHS) == KeyTy(RHS); } void Linker::IdentifiedStructTypeSet::addNonOpaque(StructType *Ty) { assert(!Ty->isOpaque()); NonOpaqueStructTypes.insert(Ty); } void Linker::IdentifiedStructTypeSet::addOpaque(StructType *Ty) { assert(Ty->isOpaque()); OpaqueStructTypes.insert(Ty); } StructType * Linker::IdentifiedStructTypeSet::findNonOpaque(ArrayRef ETypes, bool IsPacked) { Linker::StructTypeKeyInfo::KeyTy Key(ETypes, IsPacked); auto I = NonOpaqueStructTypes.find_as(Key); if (I == NonOpaqueStructTypes.end()) return nullptr; return *I; } bool Linker::IdentifiedStructTypeSet::hasType(StructType *Ty) { if (Ty->isOpaque()) return OpaqueStructTypes.count(Ty); auto I = NonOpaqueStructTypes.find(Ty); if (I == NonOpaqueStructTypes.end()) return false; return *I == Ty; } void Linker::init(Module *M, DiagnosticHandlerFunction DiagnosticHandler) { this->Composite = M; this->DiagnosticHandler = DiagnosticHandler; TypeFinder StructTypes; StructTypes.run(*M, true); for (StructType *Ty : StructTypes) { if (Ty->isOpaque()) IdentifiedStructTypes.addOpaque(Ty); else IdentifiedStructTypes.addNonOpaque(Ty); } } Linker::Linker(Module *M, DiagnosticHandlerFunction DiagnosticHandler) { init(M, DiagnosticHandler); } Linker::Linker(Module *M) { init(M, [this](const DiagnosticInfo &DI) { Composite->getContext().diagnose(DI); }); } Linker::~Linker() { } void Linker::deleteModule() { delete Composite; Composite = nullptr; } bool Linker::linkInModule(Module *Src) { ModuleLinker TheLinker(Composite, IdentifiedStructTypes, Src, DiagnosticHandler); return TheLinker.run(); } //===----------------------------------------------------------------------===// // LinkModules entrypoint. //===----------------------------------------------------------------------===// /// This function links two modules together, with the resulting Dest module /// modified to be the composite of the two input modules. If an error occurs, /// true is returned and ErrorMsg (if not null) is set to indicate the problem. /// Upon failure, the Dest module could be in a modified state, and shouldn't be /// relied on to be consistent. bool Linker::LinkModules(Module *Dest, Module *Src, DiagnosticHandlerFunction DiagnosticHandler) { Linker L(Dest, DiagnosticHandler); return L.linkInModule(Src); } bool Linker::LinkModules(Module *Dest, Module *Src) { Linker L(Dest); return L.linkInModule(Src); } //===----------------------------------------------------------------------===// // C API. //===----------------------------------------------------------------------===// LLVMBool LLVMLinkModules(LLVMModuleRef Dest, LLVMModuleRef Src, - unsigned Unused, char **OutMessages) { + LLVMLinkerMode Mode, char **OutMessages) { Module *D = unwrap(Dest); std::string Message; raw_string_ostream Stream(Message); DiagnosticPrinterRawOStream DP(Stream); LLVMBool Result = Linker::LinkModules( D, unwrap(Src), [&](const DiagnosticInfo &DI) { DI.print(DP); }); if (OutMessages && Result) *OutMessages = strdup(Message.c_str()); return Result; } Index: vendor/llvm/dist/test/Bindings/OCaml/linker.ml =================================================================== --- vendor/llvm/dist/test/Bindings/OCaml/linker.ml (revision 279290) +++ vendor/llvm/dist/test/Bindings/OCaml/linker.ml (revision 279291) @@ -1,63 +1,63 @@ (* RUN: cp %s %T/linker.ml * RUN: %ocamlc -g -warn-error A -package llvm.linker -linkpkg %T/linker.ml -o %t * RUN: %t * RUN: %ocamlopt -g -warn-error A -package llvm.linker -linkpkg %T/linker.ml -o %t * RUN: %t * XFAIL: vg_leak *) (* Note: It takes several seconds for ocamlopt to link an executable with libLLVMCore.a, so it's better to write a big test than a bunch of little ones. *) open Llvm open Llvm_linker let context = global_context () let void_type = Llvm.void_type context (* Tiny unit test framework - really just to help find which line is busted *) let print_checkpoints = false let suite name f = if print_checkpoints then prerr_endline (name ^ ":"); f () (*===-- Linker -----------------------------------------------------------===*) let test_linker () = let fty = function_type void_type [| |] in let make_module name = let m = create_module context name in let fn = define_function ("fn_" ^ name) fty m in ignore (build_ret_void (builder_at_end context (entry_block fn))); m in let m1 = make_module "one" and m2 = make_module "two" in link_modules m1 m2; dispose_module m1; dispose_module m2; let m1 = make_module "one" and m2 = make_module "two" in - link_modules m1 m2; + link_modules m1 m2 Mode.DestroySource; dispose_module m1; let m1 = make_module "one" and m2 = make_module "one" in try link_modules m1 m2; failwith "must raise" with Error _ -> dispose_module m1; dispose_module m2 (*===-- Driver ------------------------------------------------------------===*) let _ = suite "linker" test_linker