[LLVMbugs] [Bug 10802] New: Incorrect code generated for uitofp with x86 target (x86-64 is fine)

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=10802

           Summary: Incorrect code generated for uitofp with x86 target
                    (x86-64 is fine)
           Product: new-bugs
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matt at pharr.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=7178)
 --> (http://llvm.org/bugs/attachment.cgi?id=7178)
test case, part 1

The attached test program basically computes the i32 value 1 and then converts
it to a float (with enough obfuscation so that the optimizer doesn't just do
the conversion itself.)  With top of tree, when compiled with the x86-64
target, it prints 1.0, as expected:

% llc -march=x86-64 -filetype=obj -o bug.o bug.ll && g++ -m64 bug.cpp bug.o &&
./a.out
1.000000
%

With the 32-bit x86 target, it seems to generate incorrect code for the
conversion:

% llc -march=x86 -filetype=obj -o bug.o bug.ll && g++ -Wl,-no_pie -m32 bug.cpp
bug.o && ./a.out
4294967296.000000
%

-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.