[LLVMbugs] [Bug 22745] New: generator crashes with -mfpmath=387 -mno-sse

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=22745

            Bug ID: 22745
           Summary: generator crashes with -mfpmath=387 -mno-sse
           Product: clang
           Version: 3.5
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: LLVM Codegen
          Assignee: unassignedclangbugs at nondot.org
          Reporter: dan.a.ibanez at gmail.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Created attachment 13966
  --> http://llvm.org/bugs/attachment.cgi?id=13966&action=edit
preprocessed source and run script

I'm working on a meshing code and noticed different results on MacBook and
Lenovo machines. I think I've traced it to slightly different floating point
arithmetic. Per this StackOverflow answer:

http://stackoverflow.com/questions/17663780/is-there-a-document-describing-how-clang-handles-excess-floating-point-precision

I tried compiling with "-O2 -mfpmath=387 -mno-sse", which crashed the clang
code generator. Attached is a tarball with the clang output and the temporary
files it requested.

I don't actually care so much whether clang can support this obscure
combination of flags, what I'm really interested in is what flags do I use to
get exactly the same answers on both machines.

I did see that bug 3604 is a possible duplicate, but clang still crashes so
I'll submit this bug as a request that it not crash with these flags, instead
exit cleanly with a descriptive error.

-- 
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/20150301/4961eb90/attachment.html>