diff mbox

[v2,4/5] target-ppc: add vector bit permute doubleword instruction

Message ID 1470737540-5526-5-git-send-email-raji@linux.vnet.ibm.com
State New
Headers show

Commit Message

Rajalakshmi Srinivasaraghavan Aug. 9, 2016, 10:12 a.m. UTC 
Add vbpermd instruction from ISA 3.0.

Signed-off-by: Rajalakshmi Srinivasaraghavan <raji@linux.vnet.ibm.com>
---
 target-ppc/helper.h             |    1 +
 target-ppc/int_helper.c         |   20 ++++++++++++++++++++
 target-ppc/translate/vmx-impl.c |    1 +
 target-ppc/translate/vmx-ops.c  |    1 +
 4 files changed, 23 insertions(+), 0 deletions(-)

Comments

Richard Henderson Aug. 10, 2016, 5:04 a.m. UTC | #1 
On 08/09/2016 03:42 PM, Rajalakshmi Srinivasaraghavan wrote:
> +void helper_vbpermd(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
> +{
> +    int i, j;
> +    uint64_t perm = 0;
> +
> +    VECTOR_FOR_INORDER_I(i, u64) {
> +        perm = 0;
> +        for (j = 0; j < 8; j++) {
> +            int index = VBPERMQ_INDEX(b, (i * 8) + j);
> +            if (index < 64) {
> +                uint64_t mask = (1ull << (63 - (index & 0x3F)));
> +                if (a->u64[VBPERMQ_DW(index)] & mask) {
> +                    perm |= (0x80 >> j);
> +                }
> +            }
> +        }
> +        r->u64[i] = perm;
> +    }

You need to care for overlap between R vs {A,B}.


r~
diff mbox

Patch

diff --git a/target-ppc/helper.h b/target-ppc/helper.h
index 6e6e7b3..d1d9418 100644
--- a/target-ppc/helper.h
+++ b/target-ppc/helper.h
@@ -335,6 +335,7 @@  DEF_HELPER_2(vpopcntb, void, avr, avr)
 DEF_HELPER_2(vpopcnth, void, avr, avr)
 DEF_HELPER_2(vpopcntw, void, avr, avr)
 DEF_HELPER_2(vpopcntd, void, avr, avr)
+DEF_HELPER_3(vbpermd, void, avr, avr, avr)
 DEF_HELPER_3(vbpermq, void, avr, avr, avr)
 DEF_HELPER_2(vgbbd, void, avr, avr)
 DEF_HELPER_3(vpmsumb, void, avr, avr, avr)
diff --git a/target-ppc/int_helper.c b/target-ppc/int_helper.c
index 188ac6f..d6f26bb 100644
--- a/target-ppc/int_helper.c
+++ b/target-ppc/int_helper.c
@@ -1134,6 +1134,26 @@  void helper_vperm(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b,
 #define VBPERMQ_DW(index) (((index) & 0x40) == 0)
 #endif
 
+void helper_vbpermd(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
+{
+    int i, j;
+    uint64_t perm = 0;
+
+    VECTOR_FOR_INORDER_I(i, u64) {
+        perm = 0;
+        for (j = 0; j < 8; j++) {
+            int index = VBPERMQ_INDEX(b, (i * 8) + j);
+            if (index < 64) {
+                uint64_t mask = (1ull << (63 - (index & 0x3F)));
+                if (a->u64[VBPERMQ_DW(index)] & mask) {
+                    perm |= (0x80 >> j);
+                }
+            }
+        }
+        r->u64[i] = perm;
+    }
+}
+
 void helper_vbpermq(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
 {
     int i;
diff --git a/target-ppc/translate/vmx-impl.c b/target-ppc/translate/vmx-impl.c
index 2cf8c8f..5ddff58 100644
--- a/target-ppc/translate/vmx-impl.c
+++ b/target-ppc/translate/vmx-impl.c
@@ -754,6 +754,7 @@  GEN_VXFORM_DUAL(vclzw, PPC_NONE, PPC2_ALTIVEC_207, \
                 vpopcntw, PPC_NONE, PPC2_ALTIVEC_207)
 GEN_VXFORM_DUAL(vclzd, PPC_NONE, PPC2_ALTIVEC_207, \
                 vpopcntd, PPC_NONE, PPC2_ALTIVEC_207)
+GEN_VXFORM(vbpermd, 6, 23);
 GEN_VXFORM(vbpermq, 6, 21);
 GEN_VXFORM_NOA(vgbbd, 6, 20);
 GEN_VXFORM(vpmsumb, 4, 16)
diff --git a/target-ppc/translate/vmx-ops.c b/target-ppc/translate/vmx-ops.c
index 5b2826e..32bd533 100644
--- a/target-ppc/translate/vmx-ops.c
+++ b/target-ppc/translate/vmx-ops.c
@@ -261,6 +261,7 @@  GEN_VXFORM_DUAL(vclzh, vpopcnth, 1, 29, PPC_NONE, PPC2_ALTIVEC_207),
 GEN_VXFORM_DUAL(vclzw, vpopcntw, 1, 30, PPC_NONE, PPC2_ALTIVEC_207),
 GEN_VXFORM_DUAL(vclzd, vpopcntd, 1, 31, PPC_NONE, PPC2_ALTIVEC_207),
 
+GEN_VXFORM_300(vbpermd, 6, 23),
 GEN_VXFORM_207(vbpermq, 6, 21),
 GEN_VXFORM_207(vgbbd, 6, 20),
 GEN_VXFORM_207(vpmsumb, 4, 16),