llvm-mirror/storerd-io-over-rr.ll at 1df524b3708d51ebb4432f3ed661d6a73f6994b5 - llvm-mirror - Git.je (Gitea)

RPCS3/llvm-mirror

mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 04:02:41 +01:00

Krzysztof Parzyszek 37ebf5f0df [Hexagon] Prefer _io over _rr for 64-bit store with constant offset

Identify patterns where the address is aligned to an 8-byte boundary,
but both the base address and the constant offset are both proper
multiples of 4. In such cases, extract Base+4 into a separate instruc-
tion, and use S2_storerd_io, instead of using S4_storerd_rr.

llvm-svn: 277497

2016-08-02 18:50:05 +00:00

13 lines

330 B

LLVM

Raw Blame History

 ; RUN: llc -march=hexagon < %s | FileCheck %s
 ; Check for memd(base + #offset), instead of memd(base + reg<<#c).
 ; CHECK: memd(r{{[0-9]+}}+#
 define void @fred(i32 %p, i64 %v) #0 {
   %t0 = add i32 %p, 4
   %t1 = inttoptr i32 %t0 to i64*
   store i64 %v, i64* %t1
   ret void
 }
 attributes #0 = { nounwind "target-cpu"="hexagonv60" }