| Commit message (Collapse) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | There is no error message to print out, end sentence with `!' | Misha Brukman | 2004-07-07 | 1 | -1/+1 |
| | | | | | llvm-svn: 14678 | ||||
| * | Fix for bug 391. | Reid Spencer | 2004-07-07 | 1 | -1/+9 |
| | | | | | | | Improve exeception handling around bcreader invocations. llvm-svn: 14674 | ||||
| * | - #include <iostream> since its not in Value.h any more. | Reid Spencer | 2004-07-04 | 1 | -0/+1 |
| | | | | | llvm-svn: 14617 | ||||
| * | Fix -debug-pass=Executions, which relied on Function, Module, and BasicBlock | Chris Lattner | 2004-02-29 | 1 | -13/+22 |
| | | | | | | | being annotable llvm-svn: 12013 | ||||
| * | Fix typo in comment | Brian Gaeke | 2004-02-28 | 1 | -1/+1 |
| | | | | | llvm-svn: 11966 | ||||
| * | Two problems with these lines of code: | Chris Lattner | 2004-02-09 | 1 | -2/+0 |
| | | | | | | | | | | | | 1. The "work" was not in the assert, so it was punishing the optimized release 2. getNamedFunction is _very_ expensive in large programs. It is not designed to be used like this, and was taking 7% of the execution time of the code generator on perlbmk. Since the assert "can never fail", I'm just killing it. llvm-svn: 11214 | ||||
| * | Finegrainify namespacification | Chris Lattner | 2003-11-21 | 1 | -4/+1 |
| | | | | | llvm-svn: 10131 | ||||
| * | Put all LLVM code into the llvm namespace, as per bug 109. | Brian Gaeke | 2003-11-11 | 1 | -0/+4 |
| | | | | | llvm-svn: 9903 | ||||
| * | Added LLVM project notice to the top of every C++ source file. | John Criswell | 2003-10-20 | 1 | -0/+7 |
| | | | | | | | Header files will be on the way. llvm-svn: 9298 | ||||
| * | Enabling incremental bytecode loading in the JIT: | Misha Brukman | 2003-10-14 | 1 | -3/+10 |
| | | | | | | | | * FunctionPassManager ctor now takes in a ModuleProvider * run() materializes function before running passes on it llvm-svn: 9126 | ||||
| * | * Remove extraneous #includes | Chris Lattner | 2003-10-13 | 1 | -7/+9 |
| | | | | | | | | | | | | * Fix a nasty initializer ordering bug. Any only-CFG passes which registered themselves before the CFGOnlyAnalysis vector initialized got forgotten and thus got invalidated and recomputed. In particular, in my compiled version of gccas, the Loop information pass was being recomputed unnecessarily. llvm-svn: 9074 | ||||
| * | Rename method | Chris Lattner | 2003-10-12 | 1 | -1/+1 |
| | | | | | llvm-svn: 9055 | ||||
| * | Don't include Config/stdio.h or <stdio.h>. | Brian Gaeke | 2003-10-10 | 1 | -1/+0 |
| | | | | | llvm-svn: 9031 | ||||
| * | Spell `necessary' correctly. | Misha Brukman | 2003-08-18 | 1 | -1/+1 |
| | | | | | llvm-svn: 7944 | ||||
| * | Add new method to FunctionPassManager to add ImmutablePasses. | Brian Gaeke | 2003-08-14 | 1 | -0/+1 |
| | | | | | llvm-svn: 7838 | ||||
| * | Dramatically simplify how -time-passes works. | Chris Lattner | 2003-08-14 | 1 | -5/+11 |
| | | | | | | | | This also enables -time-passes for FunctionPassManagers, which allows it to be used for the JIT llvm-svn: 7834 | ||||
| * | Add FunctionPassManager - it's like a PassManager, but it only deals in | Brian Gaeke | 2003-08-12 | 1 | -0/+10 |
| | | | | | | | FunctionPasses. llvm-svn: 7778 | ||||
| * | Merged in autoconf branch. This provides configuration via the autoconf | John Criswell | 2003-06-30 | 1 | -4/+4 |
| | | | | | | | system. llvm-svn: 7014 | ||||
| * | Remove usage of sys/unistd.h | Chris Lattner | 2003-06-19 | 1 | -1/+1 |
| | | | | | llvm-svn: 6788 | ||||
| * | Add helper method | Chris Lattner | 2003-03-21 | 1 | -0/+4 |
| | | | | | llvm-svn: 5753 | ||||
| * | - Rename AnalysisUsage::preservesAll to getPreservesAll & preservesCFG to | Chris Lattner | 2002-10-21 | 1 | -2/+2 |
| | | | | | | | setPreservesCFG to be less confusing. llvm-svn: 4255 | ||||
| * | Factor timer code out of PassManager implementation, into a generic interface | Chris Lattner | 2002-10-01 | 1 | -130/+0 |
| | | | | | | | exposed by Support/Timer.h. llvm-svn: 3993 | ||||
| * | Add support for ImmutablePasses, which are not run, and cannot be | Chris Lattner | 2002-09-25 | 1 | -0/+9 |
| | | | | | | | invalidated. llvm-svn: 3921 | ||||
| * | Another portability fix provided via Casey Carter: | Chris Lattner | 2002-09-13 | 1 | -0/+1 |
| | | | | | | | | | ISSUE: getTimeRecord in lib/VMCore/Pass.cpp uses timeval and gettimeofday() without including sys/time.h. ACTION: Include sys/time.h. llvm-svn: 3699 | ||||
| * | Add a new BasicBlockPass::doInitialization/Finalization(Function &) pair of | Chris Lattner | 2002-09-12 | 1 | -4/+6 |
| | | | | | | | methods that may be useful for BasicBlockPasses. llvm-svn: 3688 | ||||
| * | - PassManager prints analysis sets nicer | Chris Lattner | 2002-08-30 | 1 | -30/+9 |
| | | | | | | | - Use Pass::AnalysisImpls instead of findAnalysisGroupMemeber llvm-svn: 3534 | ||||
| * | - Fighting with linking problem due to removing the ::ID elements. Now the | Chris Lattner | 2002-08-21 | 1 | -0/+3 |
| | | | | | | | | | | implementation .cpp files for analyses are not being included into gccas and friends because it is linking to the .a file and there is no explicit symbol reference to bring in the .o file. The new IncludeFile hack is the result. llvm-svn: 3436 | ||||
| * | Remove debugging code accidentally checked in. | Chris Lattner | 2002-08-21 | 1 | -4/+0 |
| | | | | | llvm-svn: 3430 | ||||
| * | - Implement the new AnalysisGroup feature, neccesary for Value#ing and ↵ | Chris Lattner | 2002-08-21 | 1 | -4/+121 |
| | | | | | | | pointer analysis llvm-svn: 3426 | ||||
| * | - Eliminate the need for analyses to expose an ::ID member. | Chris Lattner | 2002-08-21 | 1 | -2/+5 |
| | | | | | llvm-svn: 3414 | ||||
| * | Sort -time-passes report first by user+system, then by Wall clock time. | Chris Lattner | 2002-08-20 | 1 | -0/+11 |
| | | | | | llvm-svn: 3407 | ||||
| * | Add missing space | Chris Lattner | 2002-08-19 | 1 | -1/+1 |
| | | | | | llvm-svn: 3387 | ||||
| * | Fix the -time-passes option to not print NaN when there is zero execution time | Chris Lattner | 2002-08-19 | 1 | -6/+13 |
| | | | | | llvm-svn: 3382 | ||||
| * | Enhance -time-passes to be more detailed and useful | Chris Lattner | 2002-08-01 | 1 | -19/+89 |
| | | | | | llvm-svn: 3199 | ||||
| * | We need iostream in PassManagerT.h to use cerr | Chris Lattner | 2002-07-31 | 1 | -1/+0 |
| | | | | | llvm-svn: 3183 | ||||
| * | Implement new -debug-pass=Arguments option that causes PassManager to | Chris Lattner | 2002-07-30 | 1 | -0/+13 |
| | | | | | | | print out the command line options for the optimizations it is running. llvm-svn: 3165 | ||||
| * | * Move some code from Pass.cpp to PassManagerT.h | Chris Lattner | 2002-07-30 | 1 | -53/+9 |
| | | | | | | | | * Implement stuff so that code can declare that they only depend on the CFG of a function, not on anything else. This speeds up GCCAS a lot. llvm-svn: 3155 | ||||
| * | * PassInfo is allowed to be missing now (ie, not all passes need be registered) | Chris Lattner | 2002-07-29 | 1 | -6/+9 |
| | | | | | | | | * getPassName uses PassInfo if it's available * PassInfo is now cached in the pass so that it is only looked up once, maximum llvm-svn: 3123 | ||||
| * | * Standardize how analysis results/passes as printed with the print() virtual | Chris Lattner | 2002-07-27 | 1 | -2/+17 |
| | | | | | | | | | | | methods * Eliminate AnalysisID: Now it is just a typedef for const PassInfo* * Simplify how AnalysisID's are initialized * Eliminate Analysis/Writer.cpp/.h: incorporate printing functionality into the analyses themselves. llvm-svn: 3116 | ||||
| * | *** empty log message *** | Chris Lattner | 2002-07-24 | 1 | -1/+1 |
| | | | | | llvm-svn: 3065 | ||||
| * | Add support for pass registration | Chris Lattner | 2002-07-23 | 1 | -0/+93 |
| | | | | | llvm-svn: 3018 | ||||
| * | *** empty log message *** | Chris Lattner | 2002-07-22 | 1 | -13/+18 |
| | | | | | llvm-svn: 2985 | ||||
| * | changes for 64bit gcc | Anand Shukla | 2002-06-25 | 1 | -4/+4 |
| | | | | | llvm-svn: 2801 | ||||
| * | MEGAPATCH checkin. | Chris Lattner | 2002-06-25 | 1 | -14/+12 |
| | | | | | | | For details, See: docs/2002-06-25-MegaPatchInfo.txt llvm-svn: 2779 | ||||
| * | Make functions that preserve the CFG not invalidate analyses that only depend | Chris Lattner | 2002-05-06 | 1 | -1/+28 |
| | | | | | | | on the CFG of a function llvm-svn: 2506 | ||||
| * | Add new optional getPassName() virtual function that a Pass can override | Chris Lattner | 2002-04-29 | 1 | -5/+10 |
| | | | | | | | to make debugging output a lot nicer. llvm-svn: 2395 | ||||
| * | Add a new command line option for PassManager using utilities. | Chris Lattner | 2002-04-29 | 1 | -0/+60 |
| | | | | | | | | | Now for llc, gccas, analyze, opt, etc you can specify the -time-passes command line option that outputs a timing summary report that indicates how long each pass takes to execute. llvm-svn: 2394 | ||||
| * | Move FunctionPass::doesNotModifyCFG to AnalysisUsage::preservesCFG() | Chris Lattner | 2002-04-28 | 1 | -14/+15 |
| | | | | | llvm-svn: 2384 | ||||
| * | * Add a stub to FunctionPass so that subclasses can declare that they do not | Chris Lattner | 2002-04-28 | 1 | -4/+28 |
| | | | | | | | | | | | modify the CFG. It currently does nothing, but will in the future. * Changes to make the public PassManager.h be MUCH smaller, and devoid of implementation details. Now PassManager is just a Pimpl class that wraps PassManagerT<Module>, but lib/VMCore/Pass.cpp is now the only class that has to #include PassManagerT.h llvm-svn: 2383 | ||||
| * | Minor changes to allow Modules (which are no longer Values) to work | Chris Lattner | 2002-04-28 | 1 | -10/+9 |
| | | | | | llvm-svn: 2361 | ||||

