[v2,045/100] target/arm: Implement SVE2 WHILEGT, WHILEGE, WHILEHI, WHILEHS

Message ID 20200618042644.1685561-46-richard.henderson@linaro.org
State New
Headers show
Series
  • target/arm: Implement SVE2
Related show

Commit Message

Richard Henderson June 18, 2020, 4:25 a.m.
Rename the existing sve_while (less-than) helper to sve_whilel
to make room for a new sve_whileg helper for greater-than.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

---
v2: Use a new helper function to implement this.
---
 target/arm/helper-sve.h    |  3 +-
 target/arm/sve.decode      |  2 +-
 target/arm/sve_helper.c    | 38 +++++++++++++++++++++++++-
 target/arm/translate-sve.c | 56 ++++++++++++++++++++++++++++----------
 4 files changed, 82 insertions(+), 17 deletions(-)

-- 
2.25.1

Patch

diff --git a/target/arm/helper-sve.h b/target/arm/helper-sve.h
index acb967506d..4c4a8b27b1 100644
--- a/target/arm/helper-sve.h
+++ b/target/arm/helper-sve.h
@@ -913,7 +913,8 @@  DEF_HELPER_FLAGS_4(sve_brkns, TCG_CALL_NO_RWG, i32, ptr, ptr, ptr, i32)
 
 DEF_HELPER_FLAGS_3(sve_cntp, TCG_CALL_NO_RWG, i64, ptr, ptr, i32)
 
-DEF_HELPER_FLAGS_3(sve_while, TCG_CALL_NO_RWG, i32, ptr, i32, i32)
+DEF_HELPER_FLAGS_3(sve_whilel, TCG_CALL_NO_RWG, i32, ptr, i32, i32)
+DEF_HELPER_FLAGS_3(sve_whileg, TCG_CALL_NO_RWG, i32, ptr, i32, i32)
 
 DEF_HELPER_FLAGS_4(sve_subri_b, TCG_CALL_NO_RWG, void, ptr, ptr, i64, i32)
 DEF_HELPER_FLAGS_4(sve_subri_h, TCG_CALL_NO_RWG, void, ptr, ptr, i64, i32)
diff --git a/target/arm/sve.decode b/target/arm/sve.decode
index 5f76a95139..b7038f9f57 100644
--- a/target/arm/sve.decode
+++ b/target/arm/sve.decode
@@ -700,7 +700,7 @@  SINCDECP_z      00100101 .. 1010 d:1 u:1 10000 00 .... .....    @incdec2_pred
 CTERM           00100101 1 sf:1 1 rm:5 001000 rn:5 ne:1 0000
 
 # SVE integer compare scalar count and limit
-WHILE           00100101 esz:2 1 rm:5 000 sf:1 u:1 1 rn:5 eq:1 rd:4
+WHILE           00100101 esz:2 1 rm:5 000 sf:1 u:1 lt:1 rn:5 eq:1 rd:4
 
 ### SVE Integer Wide Immediate - Unpredicated Group
 
diff --git a/target/arm/sve_helper.c b/target/arm/sve_helper.c
index a79039ad52..5d80dc8c58 100644
--- a/target/arm/sve_helper.c
+++ b/target/arm/sve_helper.c
@@ -3711,7 +3711,7 @@  uint64_t HELPER(sve_cntp)(void *vn, void *vg, uint32_t pred_desc)
     return sum;
 }
 
-uint32_t HELPER(sve_while)(void *vd, uint32_t count, uint32_t pred_desc)
+uint32_t HELPER(sve_whilel)(void *vd, uint32_t count, uint32_t pred_desc)
 {
     uintptr_t oprsz = extract32(pred_desc, 0, SIMD_OPRSZ_BITS) + 2;
     intptr_t esz = extract32(pred_desc, SIMD_DATA_SHIFT, 2);
@@ -3737,6 +3737,42 @@  uint32_t HELPER(sve_while)(void *vd, uint32_t count, uint32_t pred_desc)
     return predtest_ones(d, oprsz, esz_mask);
 }
 
+uint32_t HELPER(sve_whileg)(void *vd, uint32_t count, uint32_t pred_desc)
+{
+    uintptr_t oprsz = extract32(pred_desc, 0, SIMD_OPRSZ_BITS) + 2;
+    intptr_t esz = extract32(pred_desc, SIMD_DATA_SHIFT, 2);
+    uint64_t esz_mask = pred_esz_masks[esz];
+    ARMPredicateReg *d = vd;
+    intptr_t i, invcount, oprbits;
+    uint64_t bits;
+
+    if (count == 0) {
+        return do_zero(d, oprsz);
+    }
+
+    oprbits = oprsz * 8;
+    tcg_debug_assert(count <= oprbits);
+
+    bits = esz_mask;
+    if (oprbits & 63) {
+        bits &= MAKE_64BIT_MASK(0, oprbits & 63);
+    }
+
+    invcount = oprbits - count;
+    for (i = (oprsz - 1) / 8; i > invcount / 64; --i) {
+        d->p[i] = bits;
+        bits = esz_mask;
+    }
+
+    d->p[i] = bits & MAKE_64BIT_MASK(invcount & 63, 64);
+
+    while (--i >= 0) {
+        d->p[i] = 0;
+    }
+
+    return predtest_ones(d, oprsz, esz_mask);
+}
+
 /* Recursive reduction on a function;
  * C.f. the ARM ARM function ReducePredicated.
  *
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 262194f163..dc6f39b5bb 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -3121,7 +3121,14 @@  static bool trans_WHILE(DisasContext *s, arg_WHILE *a)
     TCGv_ptr ptr;
     unsigned desc, vsz = vec_full_reg_size(s);
     TCGCond cond;
+    uint64_t maxval;
+    /* Note that GE/HS has a->eq == 0 and GT/HI has a->eq == 1. */
+    bool eq = a->eq == a->lt;
 
+    /* The greater-than conditions are all SVE2. */
+    if (!a->lt && !dc_isar_feature(aa64_sve2, s)) {
+        return false;
+    }
     if (!sve_access_check(s)) {
         return true;
     }
@@ -3144,22 +3151,42 @@  static bool trans_WHILE(DisasContext *s, arg_WHILE *a)
      */
     t0 = tcg_temp_new_i64();
     t1 = tcg_temp_new_i64();
-    tcg_gen_sub_i64(t0, op1, op0);
+
+    if (a->lt) {
+        tcg_gen_sub_i64(t0, op1, op0);
+        if (a->u) {
+            maxval = a->sf ? UINT64_MAX : UINT32_MAX;
+            cond = eq ? TCG_COND_LEU : TCG_COND_LTU;
+        } else {
+            maxval = a->sf ? INT64_MAX : INT32_MAX;
+            cond = eq ? TCG_COND_LE : TCG_COND_LT;
+        }
+    } else {
+        tcg_gen_sub_i64(t0, op0, op1);
+        if (a->u) {
+            maxval = 0;
+            cond = eq ? TCG_COND_GEU : TCG_COND_GTU;
+        } else {
+            maxval = a->sf ? INT64_MIN : INT32_MIN;
+            cond = eq ? TCG_COND_GE : TCG_COND_GT;
+        }
+    }
 
     tmax = tcg_const_i64(vsz >> a->esz);
-    if (a->eq) {
+    if (eq) {
         /* Equality means one more iteration.  */
         tcg_gen_addi_i64(t0, t0, 1);
 
-        /* If op1 is max (un)signed integer (and the only time the addition
-         * above could overflow), then we produce an all-true predicate by
-         * setting the count to the vector length.  This is because the
-         * pseudocode is described as an increment + compare loop, and the
-         * max integer would always compare true.
+        /*
+         * For the less-than while, if op1 is maxval (and the only time
+         * the addition above could overflow), then we produce an all-true
+         * predicate by setting the count to the vector length.  This is
+         * because the pseudocode is described as an increment + compare
+         * loop, and the maximum integer would always compare true.
+         * Similarly, the greater-than while has the same issue with the
+         * minimum integer due to the decrement + compare loop.
          */
-        tcg_gen_movi_i64(t1, (a->sf
-                              ? (a->u ? UINT64_MAX : INT64_MAX)
-                              : (a->u ? UINT32_MAX : INT32_MAX)));
+        tcg_gen_movi_i64(t1, maxval);
         tcg_gen_movcond_i64(TCG_COND_EQ, t0, op1, t1, tmax, t0);
     }
 
@@ -3168,9 +3195,6 @@  static bool trans_WHILE(DisasContext *s, arg_WHILE *a)
     tcg_temp_free_i64(tmax);
 
     /* Set the count to zero if the condition is false.  */
-    cond = (a->u
-            ? (a->eq ? TCG_COND_LEU : TCG_COND_LTU)
-            : (a->eq ? TCG_COND_LE : TCG_COND_LT));
     tcg_gen_movi_i64(t1, 0);
     tcg_gen_movcond_i64(cond, t0, op0, op1, t0, t1);
     tcg_temp_free_i64(t1);
@@ -3190,7 +3214,11 @@  static bool trans_WHILE(DisasContext *s, arg_WHILE *a)
     ptr = tcg_temp_new_ptr();
     tcg_gen_addi_ptr(ptr, cpu_env, pred_full_reg_offset(s, a->rd));
 
-    gen_helper_sve_while(t2, ptr, t2, t3);
+    if (a->lt) {
+        gen_helper_sve_whilel(t2, ptr, t2, t3);
+    } else {
+        gen_helper_sve_whileg(t2, ptr, t2, t3);
+    }
     do_pred_flags(t2);
 
     tcg_temp_free_ptr(ptr);