Gitiles
Code Review Sign In
gerrit-public.fairphone.software / toolchain / llvm-project / ac1cc934b36920ad14e5a724b0d00a36e7b72d55 / llvm / lib / CodeGen / SelectionDAG / SelectionDAGBuilder.cpp
  1. 6148fe6 Optimize code a bit. No functional change intended. by Craig Topper · 14 years ago
  2. c8e2d91 Simplify code that tries to do vector extracts for shuffles when the mask width and the input vector widths don't match. No need to check the min and max are in range before calculating the start index. The range check after having the start index is sufficient. Also no need to check for an extract from the beginning differently. by Craig Topper · 14 years ago
  3. d024cef Turn avx2 vinserti128 intrinsic calls into INSERT_SUBVECTOR DAG nodes and remove patterns for selecting the intrinsic. Similar was already done for avx1. by Craig Topper · 14 years ago
  4. e09d1c5 Remove 'else' after 'if' that ends in return. by Craig Topper · 14 years ago
  5. 80c540e Teach CodeGen's version of computeMaskedBits to understand the range metadata. by Rafael Espindola · 14 years ago
  6. 24a6298 More debug output. by Eric Christopher · 14 years ago
  7. c1e2dcd Add a debug statement. by Eric Christopher · 14 years ago
  8. 79f03e9 Assign node orders to target intrinsics which do not produce results. rdar://11096639 by Evan Cheng · 14 years ago
  9. be7a101 Add another debug statement. by Eric Christopher · 14 years ago
  10. 97b02fc llvm::SwitchInst by Stepan Dyatkovskiy · 14 years ago
  11. 5b648af Taken into account Duncan's comments for r149481 dated by 2nd Feb 2012: by Stepan Dyatkovskiy · 14 years ago
  12. 7cf6db7 Fix warnings about adding a bool to a string. Patch by Sean Silva! by Bill Wendling · 14 years ago
  13. f6298e9 Fix a codegen fault in which log2 or exp2 could be dead-code eliminated even though they could have sideeffects. by James Molloy · 14 years ago
  14. 65f9d19 Re-commit r151623 with fix. Only issue special no-return calls if it's a direct call. by Evan Cheng · 14 years ago
  15. ee7b899 Revert r151623 "Some ARM implementaions, e.g. A-series, does return stack prediction. ...", it is breaking the Clang build during the Compiler-RT part. by Daniel Dunbar · 14 years ago
  16. 1d66609 Code cleanup following CR by Duncan. by Nadav Rotem · 14 years ago
  17. 875e463 Fix a bug in the code that builds SDNodes from vector GEPs. by Nadav Rotem · 14 years ago
  18. 87c7b09 Some ARM implementaions, e.g. A-series, does return stack prediction. That is, by Evan Cheng · 14 years ago
  19. 6fe3e3d SDAGBuilder: Remove register sets that were never read and prune dead code surrounding it. by Benjamin Kramer · 14 years ago
  20. 682c76b Turn avx insert intrinsic calls into INSERT_SUBVECTOR DAG nodes and remove duplicate patterns for selecting the intrinsics by Pete Cooper · 14 years ago
  21. da97054 If the Address of a variable is an argument then treat the entire by Eric Christopher · 14 years ago
  22. 38b3161 Allow an integer to be converted into an MMX type when it's used in an inline by Bill Wendling · 14 years ago
  23. 18c6be7 More newline cleanups. by Eric Christopher · 14 years ago
  24. 5c45205 Add some handy-dandy newlines. by Eric Christopher · 14 years ago
  25. 8b98bf2 Properly emit _fltused with FastISel. Refactor to share code with SDAG. by Michael J. Spencer · 14 years ago
  26. 29d6ed6 Rename getExceptionAddressRegister() to getExceptionPointerRegister() for consistency with setExceptionPointerRegister(...). by Lang Hames · 14 years ago
  27. 05d6f2f Don't reserve the R0 and R1 registers here. We don't use these registers, and by Bill Wendling · 14 years ago
  28. 0aef16a [unwind removal] Remove all of the code for the dead 'unwind' instruction. There by Bill Wendling · 14 years ago
  29. 513aaa5 SwitchInst refactoring. by Stepan Dyatkovskiy · 14 years ago
  30. 8d9d1a0 Remove the now-dead llvm.eh.exception and llvm.eh.selector intrinsics. by Bill Wendling · 14 years ago
  31. 0256be9 continue making the world safe for ConstantDataVector. At this point, by Chris Lattner · 14 years ago
  32. cf12970 eliminate the Constant::getVectorElements method. There are better (and by Chris Lattner · 14 years ago
  33. 9be5959 Use the right method to get the # elements in a CDS. by Chris Lattner · 14 years ago
  34. 00245f4 add more support for ConstantDataSequential by Chris Lattner · 14 years ago
  35. 46a9f01 More dead code removal (using -Wunreachable-code) by David Blaikie · 14 years ago
  36. e3d305a Changed flag operand of ISD::FP_ROUND to TargetConstant as it should not get checked for legalisation by Pete Cooper · 14 years ago
  37. e85b95d Changed intrinsic ID operand to a target constant as its not used in any arithmetic so should not be checked in legalisation by Pete Cooper · 14 years ago
  38. f726e15 Allow vector shuffle normalizing to use concat vector even if the sources are commuted in the shuffle mask. by Craig Topper · 14 years ago
  39. 6b77a07 Turn a few more inline asm errors into "emitErrors" instead of fatal errors. by Chris Lattner · 14 years ago
  40. fcc041e Remove the restriction that target intrinsics can only involve legal types. Targets can perfects well support intrinsics on illegal types, as long as they are prepared to perform custom expansion during type legalization. For example, a target where i64 is illegal might still support the i64 intrinsic operation using pairs of i32's. ARM already does some expansions like this for non-intrinsic operations. by Owen Anderson · 14 years ago
  41. 96f8c55 Add some constantness to BranchProbabilityInfo and BlockFrequnencyInfo. by Jakub Staszak · 14 years ago
  42. e7f329f Enable synthesis of FLOG2 and FEXP2 SelectionDAG nodes from libm calls. These are already marked as illegal by default. by Owen Anderson · 14 years ago
  43. 637cc6a Initial CodeGen support for CTTZ/CTLZ where a zero input produces an by Chandler Carruth · 14 years ago
  44. bb15fec Enhance both TargetLibraryInfo and SelectionDAGBuilder so that the latter can use the former to prevent the formation of libm SDNode's when -fno-builtin is passed. by Owen Anderson · 14 years ago
  45. 0b9b9da Teach SelectionDAG to match more calls to libm functions onto existing SDNodes. Mark these nodes as illegal by default, unless the target declares otherwise. by Owen Anderson · 14 years ago
  46. 3924cb0 by Nadav Rotem · 14 years ago
  47. 50f02cb Move global variables in TargetMachine into new TargetOptions class. As an API by Nick Lewycky · 14 years ago
  48. 5ebc95f Remove dead llvm.eh.sjlj.dispatchsetup intrinsic. by Bill Wendling · 14 years ago
  49. e2530dc Fix an obvious omission in the SelectionDAGBuilder where we were by Chandler Carruth · 14 years ago
  50. 70679df Remove some unnecessary includes of PseudoSourceValue.h. by Jay Foad · 14 years ago
  51. 82cd9e8 Added invariant field to the DAG.getLoad method and changed all calls. by Pete Cooper · 14 years ago
  52. e8261a2 Don't use floating point to do an integer's job. by Jakob Stoklund Olesen · 14 years ago
  53. d278d35 Fix a bunch of unused variable warnings when doing a release build with gcc-4.6. by Duncan Sands · 14 years ago
  54. bab6678 Fix comment to refer to correct instruction by Hal Finkel · 14 years ago
  55. 2730a00 Clear out the landing pad to call site map for each function. by Bill Wendling · 14 years ago
  56. 24abd9d Encode register class constreaints in inline asm instructions. by Jakob Stoklund Olesen · 14 years ago
  57. 979009e Use a utility from MathExtras to clarify a check and avoid undefined behavior. Based on patch by Ahmed Charles. by Eli Friedman · 14 years ago
  58. 7ecfbd9 Thread the chain through the eh.sjlj.setjmp intrinsic, like it's documented to by Bill Wendling · 14 years ago
  59. 1456cd2 Remove the old atomic instrinsics. autoupgrade functionality is included with this patch. by Eli Friedman · 14 years ago
  60. 267f323 Modify the mapping from landing pad to call sites to accept more than one call by Bill Wendling · 14 years ago
  61. 3d11aa7 Create a mapping between the landing pad basic block and the call site index for later use. by Bill Wendling · 14 years ago
  62. f40df1d Promote comment to doxycomment. Adjust whitespace. No functionality change. by Nick Lewycky · 14 years ago
  63. 95031ed Clean up uses of switch instructions so they are not dependent on the operand ordering. Patch by Stepan Dyatkovskiy. by Eli Friedman · 14 years ago
  64. d299dcc Use the local we already set up. by Eric Christopher · 14 years ago
  65. baf3941 Strip off pointer casts when looking at the eh.sjlj.functioncontext's argument. by Bill Wendling · 14 years ago
  66. 66b110f Create and use an llvm.eh.sjlj.functioncontext intrinsic. by Bill Wendling · 14 years ago
  67. f78c6a8 Fix check for unaligned load/store so it doesn't catch over-aligned load/store. by Eli Friedman · 14 years ago
  68. f151821 Error out on CodeGen of unaligned load/store. Fix test so it isn't accidentally testing that case. by Eli Friedman · 14 years ago
  69. e74e0c8 tidy up a bit by Chris Lattner · 14 years ago
  70. 9d904e1 Directly point debug info to the stack slot of the arugment, instead of trying to keep track of vreg in which it the arugment is copied. The LiveDebugVariable can keep track of variable's ranges. by Devang Patel · 14 years ago
  71. f2641e1 Add codegen support for vector select (in the IR this means a select by Duncan Sands · 14 years ago
  72. a098436 Split the init.trampoline intrinsic, which currently combines GCC's by Duncan Sands · 14 years ago
  73. 452aae6 Atomic load/store on ARM/Thumb. by Eli Friedman · 14 years ago
  74. 342e8df Basic x86 code generation for atomic load and store instructions. by Eli Friedman · 14 years ago
  75. 6b477b9 Fix 80 col violations. by Evan Cheng · 14 years ago
  76. 247fd3b Add the support in code-gen for the landingpad instruction lowering. by Bill Wendling · 14 years ago
  77. a408e5b Revert patch. Forgot a dependent commit. by Bill Wendling · 14 years ago
  78. 2a52194 Add the body of 'visitLandingPad'. by Bill Wendling · 14 years ago
  79. fae1475 Initial commit of the 'landingpad' instruction. by Bill Wendling · 14 years ago
  80. aab841c Do not drop undef debug values. These are used as range termination marker by live debug variable pass. by Devang Patel · 14 years ago
  81. 30a49e9 New approach to r136737: insert the necessary fences for atomic ops in platform-independent code, since a bunch of platforms (ARM, Mips, PPC, Alpha are the relevant targets here) need to do essentially the same thing. by Eli Friedman · 14 years ago
  82. f891bf8 Add the 'resume' instruction for the new EH rewrite. by Bill Wendling · 14 years ago
  83. ad088e6 Revert r136253, r136263, r136269, r136313, r136325, r136326, r136329, r136338, by Bill Wendling · 14 years ago
  84. 0480a8f Do not lose branch weights when lowering SwitchInst. by Jakub Staszak · 14 years ago
  85. 539db98 Remove unneeded const_cast. by Jakub Staszak · 14 years ago
  86. adec587 Misc optimizer+codegen work for 'cmpxchg' and 'atomicrmw'. They appear to be by Eli Friedman · 14 years ago
  87. 7eadbea Use the pointer type size. by Bill Wendling · 14 years ago
  88. 6a8cac7 And now something that compiles... by Bill Wendling · 14 years ago
  89. 4b0a365 Make sure to sext or trunc the result from the register. by Bill Wendling · 14 years ago
  90. 3cc8768 Visit the landingpad instruction. by Bill Wendling · 14 years ago
  91. c9a551e LangRef and basic memory-representation/reading/writing for 'cmpxchg' and by Eli Friedman · 14 years ago
  92. 4f02723 The personality function should be a Function* and not just a Value*. by Bill Wendling · 14 years ago
  93. 26a4848 Code generation for 'fence' instruction. by Eli Friedman · 14 years ago
  94. 6c923bb Merge the contents from exception-handling-rewrite to the mainline. by Bill Wendling · 14 years ago
  95. 6381c01 Explicitly cast narrowing conversions inside {}s that will become errors in by Jeffrey Yasskin · 14 years ago
  96. fee02c6 Initial implementation of 'fence' instruction, the new C++0x-style replacement for llvm.memory.barrier. by Eli Friedman · 14 years ago
  97. 229907c land David Blaikie's patch to de-constify Type, with a few tweaks. by Chris Lattner · 14 years ago
  98. 92464be Check register class matching instead of width of type matching by Eric Christopher · 14 years ago
  99. 15cd5a3 Don't emit a bit test if there is only one case the test can yield false. A simple SETNE is sufficient. by Benjamin Kramer · 14 years ago
  100. 57aa636 Convert InsertValueInst and ExtractValueInst APIs to use ArrayRef. by Jay Foad · 14 years ago