[LLVMbugs] [Bug 11033] New: SSE4.2 instruction (pcmpgtq) is generated even with -mattr=-sse42

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Wed Sep 28 10:51:02 PDT 2011


http://llvm.org/bugs/show_bug.cgi?id=11033

           Summary: SSE4.2 instruction (pcmpgtq) is generated even with
                    -mattr=-sse42
           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=7371)
 --> (http://llvm.org/bugs/attachment.cgi?id=7371)
test case

With the attached test case, a pcmpgtq is generated with top-of-tree when
-mattr=-sse42 is specified, even though pcmpgtq seems to have been introduced
in SSE4.2.  (And a macbook air from last year, with a CPU that supports SSE 4.1
but not SSE4.2 agreed with this, as it generates an illegal instruction trap if
it tries to run the resulting code.)

% llc x.ll -o - -mattr=-sse42 | grep pcmpgtq
    pcmpgtq    %xmm0, %xmm1
    pcmpgtq    %xmm5, %xmm3

git bisect tells me that this behavior started with this commit, but it's
unclear to me why this would be.


354efd88db96c9662d41c1e154fdee37324802db is the first bad commit
commit 354efd88db96c9662d41c1e154fdee37324802db
Author: Nadav Rotem <nadav.rotem at intel.com>
Date:   Sun Sep 18 14:57:03 2011 +0000

    setOperationAction should be done on the return value of the type, not the
operands.



    git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@140001
91177308-0d34-0410-b5e6-96231b3b80d8

-- 
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