- User Since
- Aug 11 2014, 12:00 PM (265 w, 6 d)
Oct 26 2015
Improved comment and added check in visitSDIV and visitUDIV in response to feedback from Artyom Skrobov.
Moved test to X86 directory and added new test functions without minsize attribute. X86 is a better home for the test since X86 returns different values for isIntDivCheap() depending on the minsize attribute.
Verified the test catches the crash bug when the s/urem fix is not in place.
Verified all tests pass with the fix.
Oct 23 2015
Aug 31 2015
Aug 24 2015
Added new comment to tests.
Aug 19 2015
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@245385 91177308-0d34-0410-b5e6-96231b3b80d8
Aug 18 2015
Aug 13 2015
Updated with improved comment from Hal Finkel.
Aug 12 2015
Jul 28 2015
Fix more old style comments caught by Jonathan. NFC.
The meaning of one of the original comments is mysterious. This patch does not clarify it.
Tried to address review comments from Jonathan Roelofs. NFC from previous patch. Updated patch title text and summary to remove statement about asserts.
Jul 15 2015
Mar 12 2015
Removed the assert() due to cases that call isTruncateFree without regard to type size. Improved comment for these odd cases.
Verified regression checks pass.
Fixed patch file direction.
Oops, patch is in wrong direction. Will upload new patch.
Nov 3 2014
Use fixed point font for include path.
Oct 30 2014
This patch is abandoned. See D6046.
Oct 8 2014
Sep 26 2014
Operator error: uploaded old patch again. Here is the new patch with renamed regression test file.
Changes from previous patch:
- Use more succinct test file name "BitOffsetDecoder.td"
- Update test file comments to use "offset" instead of "shift"
Sep 25 2014
Sep 10 2014
New patch file: reduce source churn by not alphabetizing llvm-objdump command line option definitions.
'make check' for all targets passes.
The -bare option keeps objdump output consistent with test expectations regardless of host platform. As discussed on the list, default llvm-objdump output could change per host: GNU style for Linux, otools for Apple, etc. Diagnostic tests cannot depend on the per-host default. Tests only need updating (a tedious process) if -bare output changes.
Sep 9 2014
Aug 13 2014
New patch uploaded:
- Moved test file to X86 subdirectory
- Ungrouped code comments.
Aug 12 2014
I also wondered about the best location for the input file. The bug is not specific to x86 and there is precedence for x86 files in the Inputs directory, so I just chose that location. I don't care where the test input ultimately lands, but I'll leave it as-is unless somebody insists to move it.
Resubmitting with proper Phabricator fields.