[LLVMbugs] [Bug 8747] New: [x86] error in definition of SSE 4.1 round_ps intrinsic?

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Mon Dec 6 13:46:13 PST 2010


           Summary: [x86] error in definition of SSE 4.1 round_ps
           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

The round_ps intrinsic seems to be defined to take 2 float vector parameters in
addition to the desired rounding mode.  This doesn't seem to match the
definition of the instruction, which does take two SSE registers, but the first
operand is the register where the result is returned.

As a repro case, when I run the attached test program, I get as output:

% ./a.out
<4 x float> (<4 x float>, <4 x float>, i32)

whereas I'd expect it to give "<4 x float> (<4 x float>, i32)"

I'm not enough of an llvm internals expert to provide a suggested patch, but I
assume this is an easy fix (assuming my bug report is correct.)


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.

More information about the llvm-bugs mailing list