[llvm-commits] FW: [PATCH] Fix a bug in GenWidenVectorLoads

[David A. Greene]

"Rotem, Nadav" <nadav.rotem at intel.com> writes:

> Ping.
>
> From: llvm-commits-bounces at cs.uiuc.edu [mailto:llvm-commits-bounces at cs.uiuc.edu] On Behalf Of Rotem, Nadav
> Sent: Wednesday, July 13, 2011 10:30
> To: llvm-commits at cs.uiuc.edu
> Subject: [llvm-commits] [PATCH] Fix a bug in GenWidenVectorLoads
>
> On AVX, we can load 8xi32 at a time. The bug happens when two uneven loads are used: When we load the v12i32 type, the GenWidenVectorLoads method
> generates two loads: v8xi32 and v4xi32 and attempts to use CONCAT_VECTORS to join them. In this fix I concat undef values to fix the smaller
> value. The test “widen_load-2.ll” exposes this bug  (if you implement wide loads…).

Looks fine as far as I can tell.

                            -Dave