| Age | Commit message (Expand) | Author |
|---|
| 2014-10-27 | Prune CRLF. | NAKAMURA Takumi |
| 2014-07-30 | Update for llvm change. | Rafael Espindola |
| 2014-06-04 | Fix alias generation and bring back the tests for it. | Rafael Espindola |
| 2014-03-30 | Disable a couple of tests for now, corresponding to r204934. | NAKAMURA Takumi |
| 2013-11-14 | Handle weakref in cases where target is not an IDENTIFIER_NODE. | Rafael Espindola |
| 2013-11-14 | Allow for a bitcast that shows up in some versions of gcc. | Rafael Espindola |
| 2013-11-14 | Handle weakref the clang way. | Rafael Espindola |
| 2013-09-19 | Some linux distributions have patched their system gcc to turn on the | Duncan Sands |
| 2013-07-24 | Fix the build after removal of the SSPBufferSize codegen option. | Duncan Sands |
| 2013-07-15 | GCC doesn't generate FMA_EXPR on 32 bit x86. | Duncan Sands |
| 2013-06-28 | Implement __builtin_signbit. | Peter Collingbourne |
| 2013-06-27 | Implement support for FMA_EXPR. | Peter Collingbourne |
| 2013-06-26 | Prevent validator tests from writing to the current directory. | Peter Collingbourne |
| 2013-06-06 | When LLVM optimization is explicitly turned off, don't run the always-inliner | Duncan Sands |
| 2013-06-06 | Don't run the full inliner when passed -O3 -fplugin-arg-dragonegg-ir-optimize=0. | Duncan Sands |
| 2013-05-25 | Force SSE on so that the vectorizer will do something on 32 bit platforms. | Duncan Sands |
| 2013-05-25 | Add support for __builtin_init_heap_trampoline, fixing PR16015. While there, | Duncan Sands |
| 2013-05-24 | Enable fast math IR optimizations when passed -ffast-math. | Duncan Sands |
| 2013-05-24 | Enable the loop vectorizer. | Duncan Sands |
| 2013-05-23 | Add support for the iround builtin, which gfortran-4.8 generates. | Duncan Sands |
| 2013-05-20 | Like gcc-4.7, gcc-4.8 uses the optimal alignment rather than the maximal | Duncan Sands |
| 2013-05-20 | Functions are no longer allowed to alias globals (and vice versa) in gcc-4.8. | Duncan Sands |
| 2013-04-17 | If the user gave an alignment explicitly, keep it as an explicit alignment even | Duncan Sands |
| 2013-04-17 | Turn off libcall simplification when -fno-builtin is specified. Patch by Peter | Duncan Sands |
| 2013-03-31 | For gcc 4.7, use gimple_call_fntype to extract a GIMPLE_CALL function type. | Peter Collingbourne |
| 2013-03-18 | Set TLS model on external variables. | Peter Collingbourne |
| 2013-03-16 | Fix assertion failure when compiling stdin with debug info. | Peter Collingbourne |
| 2013-01-17 | Revert the previous commit. It didn't help because the compilator testsuite | Duncan Sands |
| 2013-01-16 | This test crashes gcc-4.6 on x86-32, but passes with dragonegg. Arrange for the | Duncan Sands |
| 2013-01-14 | Get this test passing on 32 bit machines. | Duncan Sands |
| 2013-01-10 | Add support for -m96bit-long-double and clean up the handling of the memory type | Duncan Sands |
| 2013-01-06 | Arrange for this to pass on 32 bit platforms as well as 64 bit. | Duncan Sands |
| 2013-01-06 | This tests the 64 bit ABI so isn't relevant on 32 bit. | Duncan Sands |
| 2013-01-06 | This test requires SSE. | Duncan Sands |
| 2013-01-06 | The 64 bit builtin is not available on 32 bit platforms, adjust the test. | Duncan Sands |
| 2013-01-05 | Recognize and ignore builtin_assume_aligned. | Duncan Sands |
| 2012-12-05 | The GCC optimizers can generate code that requires getting the address of a | Duncan Sands |
| 2012-11-28 | Add support for GCC's vector copysign builtins, fixing PR12664. | Duncan Sands |
| 2012-11-06 | Add support for the iceil builtin (new in gcc-4.7) fixing PR14270. | Duncan Sands |
| 2012-10-23 | Check the optimizers make nice getelementptrs out of POINTER_PLUS_EXPR. | Duncan Sands |
| 2012-10-17 | Specifying -mno-sse (or similar) wasn't actually turning off SSE! | Duncan Sands |
| 2012-10-01 | GCC's builtin alloca (the version without an explicit alignment) should be | Duncan Sands |
| 2012-09-30 | There is no reason for a basic block that finished with an unsupported GCC | Duncan Sands |
| 2012-09-29 | In some situations (eg unsupported builtin) we may produce no code whatsoever | Duncan Sands |
| 2012-09-29 | Don't bother outputting unreachable return blocks. | Duncan Sands |
| 2012-09-10 | Dragonegg's multiple return value (MRV) code returns this struct of three i32's | Duncan Sands |
| 2012-09-08 | Correctly compute the alignment of struct fields, in particular bitfields: | Duncan Sands |
| 2012-08-21 | Unbreak the use of "--param ssp-buffer-size=" in dragonegg after Chad's changes. | Duncan Sands |
| 2012-07-12 | Add support for rdrand builtins. One issue here is that gcc-4.5 doesn't | Duncan Sands |
| 2012-06-26 | Pass TLS model through to LLVM. | Hans Wennborg |
