[LLVMbugs] [Bug 22418] New: Some of the neon-scalar-copy.ll DUP tests have regressed at some point, to mov+fmov.

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=22418

            Bug ID: 22418
           Summary: Some of the neon-scalar-copy.ll DUP tests have
                    regressed at some point, to mov+fmov.
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Backend: AArch64
          Assignee: unassignedbugs at nondot.org
          Reporter: ahmed.bougacha at gmail.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

For some of the tests in neon-scalar-copy.ll, the CHECK lines were
inadvertently removed, and seem to have regressed at some point.  Consider:

define <1 x i8> @test_vector_dup_bv16B(<16 x i8> %v1) {
 %shuffle.i = shufflevector <16 x i8> %v1, <16 x i8> undef, <1 x i32> <i32 14> 
 ret <1 x i8> %shuffle.i
}

which used to check for:

    dup {{b[0-9]+}}, {{v[0-9]+}}.b[14]

This now generates:

    umov    w8, v0.b[14]
    fmov    s0, w8
    ret

Which looks worse to me.

-- 
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/20150130/016fa874/attachment.html>