diff options
| author | Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com> | 2020-03-20 09:07:10 +0000 |
|---|---|---|
| committer | Andre Vieira <andre.simoesdiasvieira@arm.com> | 2020-03-20 09:08:39 +0000 |
| commit | 0efe7d8796e00a5737017fe472680b653bd83d90 (patch) | |
| tree | bf319ea475f48f72c4e48adb57e67b7edeef34b2 /gcc/config | |
| parent | 4119cd693d27e9dd87c547de75283edd45bf6dce (diff) | |
| download | gcc-0efe7d8796e00a5737017fe472680b653bd83d90.zip gcc-0efe7d8796e00a5737017fe472680b653bd83d90.tar.gz gcc-0efe7d8796e00a5737017fe472680b653bd83d90.tar.bz2 | |
gcc, Arm: Fix MVE move from GPR -> GPR
This patch fixes the pattern mve_mov for the case where both MVE vectors are in
R registers and the move does not get optimized away. I use the same approach
as we do for NEON, where we use four register moves.
gcc/ChangeLog:
2020-03-20 Andre Vieira <andre.simoesdiasvieira@arm.com>
* config/arm/mve.md (mve_mov<mode>): Fix R->R case.
gcc/testsuite/ChangeLog:
2020-03-20 Andre Vieira <andre.simoesdiasvieira@arm.com>
* gcc.target/arm/mve/intrinsics/mve_move_gpr_to_gpr.c: New test.
Diffstat (limited to 'gcc/config')
| -rw-r--r-- | gcc/config/arm/mve.md | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/gcc/config/arm/mve.md b/gcc/config/arm/mve.md index 5667882..b80a2a6 100644 --- a/gcc/config/arm/mve.md +++ b/gcc/config/arm/mve.md @@ -663,7 +663,7 @@ else return "vldrb.8 %q0, %E1"; case 5: - return output_move_neon (operands); + return output_move_quad (operands); case 7: return "vstrb.8 %q1, %E0"; default: @@ -671,7 +671,7 @@ return ""; } } - [(set_attr "type" "mve_move,mve_move,mve_move,mve_move,mve_load,mve_move,mve_move,mve_store") + [(set_attr "type" "mve_move,mve_move,mve_move,mve_move,mve_load,multiple,mve_move,mve_store") (set_attr "length" "4,8,8,4,8,8,4,4") (set_attr "thumb2_pool_range" "*,*,*,*,1018,*,*,*") (set_attr "neg_pool_range" "*,*,*,*,996,*,*,*")]) |