[LLVMbugs] [Bug 15388] New: 32bit Linux, the assignment operator of vector value cause FPU stack over-flow.

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=15388

            Bug ID: 15388
           Summary: 32bit Linux, the assignment operator of vector value
                    cause FPU stack over-flow.
           Product: clang
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: LLVM Codegen
          Assignee: unassignedclangbugs at nondot.org
          Reporter: qiancong at multicorewareinc.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Created attachment 10104
  --> http://llvm.org/bugs/attachment.cgi?id=10104&action=edit
A bug test code

In 32bit Linux, asignment operator or pass argument for the vector type value 
defined by ext_vector, may cause FPU register stack over-flow, thus some
following float operator get not correct result!

analyze the attached demo code(int 32bit Linux Box):
from the disassembled code,  we found that clang generate asm code like
"movq mm0, [esp+4]" for char8/uchar8  asignment operator, but no "EMMS"
instruction to reset, so cause FPU register stack over-flow , if any float
operator was followed,  maybe cause the float result not correct..


I have test this demo code by clang3.2 and clang3.3, can represent the bug..
Any fix or update of the bug, please tell me. I will very appreciate　for your
help.

-- 
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/20130228/9c385a18/attachment.html>