diff options
author | Chandler Carruth <chandlerc@gmail.com> | 2011-12-24 10:55:54 +0000 |
---|---|---|
committer | Chandler Carruth <chandlerc@gmail.com> | 2011-12-24 10:55:54 +0000 |
commit | acc068e873a1a2afa1edef20452722d97eec8f71 (patch) | |
tree | bcff5586d3e4b64e301ef6ae94c63946abd1f68e /test/CodeGen/X86/clz.ll | |
parent | c08e57c7c9ebba27360d5e27f56a44bcaa963a52 (diff) | |
download | llvm-acc068e873a1a2afa1edef20452722d97eec8f71.tar.gz llvm-acc068e873a1a2afa1edef20452722d97eec8f71.tar.bz2 llvm-acc068e873a1a2afa1edef20452722d97eec8f71.tar.xz |
Switch the lowering of CTLZ_ZERO_UNDEF from a .td pattern back to the
X86ISelLowering C++ code. Because this is lowered via an xor wrapped
around a bsr, we want the dagcombine which runs after isel lowering to
have a chance to clean things up. In particular, it is very common to
see code which looks like:
(sizeof(x)*8 - 1) ^ __builtin_clz(x)
Which is trying to compute the most significant bit of 'x'. That's
actually the value computed directly by the 'bsr' instruction, but if we
match it too late, we'll get completely redundant xor instructions.
The more naive code for the above (subtracting rather than using an xor)
still isn't handled correctly due to the dagcombine getting confused.
Also, while here fix an issue spotted by inspection: we should have been
expanding the zero-undef variants to the normal variants when there is
an 'lzcnt' instruction. Do so, and test for this. We don't want to
generate unnecessary 'bsr' instructions.
These two changes fix some regressions in encoding and decoding
benchmarks. However, there is still a *lot* to be improve on in this
type of code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@147244 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/clz.ll')
-rw-r--r-- | test/CodeGen/X86/clz.ll | 27 |
1 files changed, 26 insertions, 1 deletions
diff --git a/test/CodeGen/X86/clz.ll b/test/CodeGen/X86/clz.ll index 4e080309b6..ad47bde854 100644 --- a/test/CodeGen/X86/clz.ll +++ b/test/CodeGen/X86/clz.ll @@ -31,7 +31,7 @@ entry: ; CHECK: t3: ; CHECK: bsrw ; CHECK-NOT: cmov -; CHECK: xorw $15, +; CHECK: xorl $15, ; CHECK: ret } @@ -63,3 +63,28 @@ entry: %tmp1 = tail call i32 @llvm.ctlz.i32(i32 %or, i1 false) ret i32 %tmp1 } + +define i32 @t6(i32 %n) nounwind { +entry: +; Don't generate any xors when a 'ctlz' intrinsic is actually used to compute +; the most significant bit, which is what 'bsr' does natively. +; CHECK: t6: +; CHECK: bsrl +; CHECK-NOT: xorl +; CHECK: ret + %ctlz = tail call i32 @llvm.ctlz.i32(i32 %n, i1 true) + %bsr = xor i32 %ctlz, 31 + ret i32 %bsr +} + +define i32 @t7(i32 %n) nounwind { +entry: +; Same as t6, but ensure this happens even when there is a potential zero. +; CHECK: t7: +; CHECK: bsrl +; CHECK-NOT: xorl +; CHECK: ret + %ctlz = tail call i32 @llvm.ctlz.i32(i32 %n, i1 false) + %bsr = xor i32 %ctlz, 31 + ret i32 %bsr +} |