[llvm-bugs] [Bug 32715] New: Assertion failure: /lib/Analysis/ScopInfo.cpp:797: void polly::MemoryAccess::computeBoundsOnAccessRelation(unsigned int): Assertion `M in.sle(Max) && "Minimum expected to be less or equal than max"' failed.

via llvm-bugs llvm-bugs at lists.llvm.org
Wed Apr 19 12:48:19 PDT 2017


https://bugs.llvm.org/show_bug.cgi?id=32715

            Bug ID: 32715
           Summary: Assertion failure: /lib/Analysis/ScopInfo.cpp:797:
                    void
                    polly::MemoryAccess::computeBoundsOnAccessRelation(uns
                    igned int): Assertion `M in.sle(Max) && "Minimum
                    expected to be less or equal than max"' failed.
           Product: Polly
           Version: unspecified
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: Other
          Assignee: polly-dev at googlegroups.com
          Reporter: sabuasal at codeaurora.org
                CC: llvm-bugs at lists.llvm.org

Created attachment 18315
  --> https://bugs.llvm.org/attachment.cgi?id=18315&action=edit
reduced ll file for the test case

This bug causes a crash in ScopInfo with an assertion failure:

"opt: lib/Analysis/ScopInfo.cpp:797: void
polly::MemoryAccess::computeBoundsOnAccessRelation(unsigned int): Assertion `M
in.sle(Max) && "Minimum expected to be less or equal than max"' failed."



Reduced IR attached. 

can recrete the crash with 


opt -S -O3 -polly  -polly-scops  -polly-tiling=false -polly-allow-nonaffine
-polly-process-unprofitable  ./test_case.ll

My investigation shows this was introduced by r294891 :

Author: Tobias Grosser <tobias at grosser.es>
Date:   Sun Feb 12 08:11:12 2017 +0000
    Do not use wrapping ranges to bound non-affine accesses

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20170419/f65fa25d/attachment.html>


More information about the llvm-bugs mailing list