[PATCH v3 59/70] tcg/tci: Emit setcond before brcond

[Richard Henderson <richard.henderson@linaro.org>]

The encoding planned for tci does not have enough room for
brcond2, with 4 registers and a condition as input as well
as the label.  Resolve the condition into TCG_REG_TMP, and
relax brcond to one register plus a label, considering the
condition to always be reg != 0.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/tci.c                | 68 ++++++++++------------------------------
 tcg/tci/tcg-target.c.inc | 52 +++++++++++-------------------
 2 files changed, 35 insertions(+), 85 deletions(-)

diff --git a/tcg/tci.c b/tcg/tci.c
index d27db9f720..e7268b13e1 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -141,6 +141,16 @@ static void tci_args_nl(const uint8_t **tb_ptr, uint8_t *n0, void **l1)
     check_size(start, tb_ptr);
 }
 
+static void tci_args_rl(const uint8_t **tb_ptr, TCGReg *r0, void **l1)
+{
+    const uint8_t *start = *tb_ptr;
+
+    *r0 = tci_read_r(tb_ptr);
+    *l1 = (void *)tci_read_label(tb_ptr);
+
+    check_size(start, tb_ptr);
+}
+
 static void tci_args_rr(const uint8_t **tb_ptr,
                         TCGReg *r0, TCGReg *r1)
 {
@@ -212,19 +222,6 @@ static void tci_args_rrs(const uint8_t **tb_ptr,
     check_size(start, tb_ptr);
 }
 
-static void tci_args_rrcl(const uint8_t **tb_ptr,
-                          TCGReg *r0, TCGReg *r1, TCGCond *c2, void **l3)
-{
-    const uint8_t *start = *tb_ptr;
-
-    *r0 = tci_read_r(tb_ptr);
-    *r1 = tci_read_r(tb_ptr);
-    *c2 = tci_read_b(tb_ptr);
-    *l3 = (void *)tci_read_label(tb_ptr);
-
-    check_size(start, tb_ptr);
-}
-
 static void tci_args_rrrc(const uint8_t **tb_ptr,
                           TCGReg *r0, TCGReg *r1, TCGReg *r2, TCGCond *c3)
 {
@@ -293,21 +290,6 @@ static void tci_args_rrrr(const uint8_t **tb_ptr,
     check_size(start, tb_ptr);
 }
 
-static void tci_args_rrrrcl(const uint8_t **tb_ptr, TCGReg *r0, TCGReg *r1,
-                            TCGReg *r2, TCGReg *r3, TCGCond *c4, void **l5)
-{
-    const uint8_t *start = *tb_ptr;
-
-    *r0 = tci_read_r(tb_ptr);
-    *r1 = tci_read_r(tb_ptr);
-    *r2 = tci_read_r(tb_ptr);
-    *r3 = tci_read_r(tb_ptr);
-    *c4 = tci_read_b(tb_ptr);
-    *l5 = (void *)tci_read_label(tb_ptr);
-
-    check_size(start, tb_ptr);
-}
-
 static void tci_args_rrrrrc(const uint8_t **tb_ptr, TCGReg *r0, TCGReg *r1,
                             TCGReg *r2, TCGReg *r3, TCGReg *r4, TCGCond *c5)
 {
@@ -723,8 +705,8 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
             break;
 #endif
         case INDEX_op_brcond_i32:
-            tci_args_rrcl(&tb_ptr, &r0, &r1, &condition, &ptr);
-            if (tci_compare32(regs[r0], regs[r1], condition)) {
+            tci_args_rl(&tb_ptr, &r0, &ptr);
+            if ((uint32_t)regs[r0]) {
                 tb_ptr = ptr;
             }
             break;
@@ -741,15 +723,6 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
             T2 = tci_uint64(regs[r5], regs[r4]);
             tci_write_reg64(regs, r1, r0, T1 - T2);
             break;
-        case INDEX_op_brcond2_i32:
-            tci_args_rrrrcl(&tb_ptr, &r0, &r1, &r2, &r3, &condition, &ptr);
-            T1 = tci_uint64(regs[r1], regs[r0]);
-            T2 = tci_uint64(regs[r3], regs[r2]);
-            if (tci_compare64(T1, T2, condition)) {
-                tb_ptr = ptr;
-                continue;
-            }
-            break;
         case INDEX_op_mulu2_i32:
             tci_args_rrrr(&tb_ptr, &r0, &r1, &r2, &r3);
             tci_write_reg64(regs, r1, r0, (uint64_t)regs[r2] * regs[r3]);
@@ -877,8 +850,8 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
             break;
 #endif
         case INDEX_op_brcond_i64:
-            tci_args_rrcl(&tb_ptr, &r0, &r1, &condition, &ptr);
-            if (tci_compare64(regs[r0], regs[r1], condition)) {
+            tci_args_rl(&tb_ptr, &r0, &ptr);
+            if (regs[r0]) {
                 tb_ptr = ptr;
             }
             break;
@@ -1188,9 +1161,9 @@ int print_insn_tci(bfd_vma addr, disassemble_info *info)
 
     case INDEX_op_brcond_i32:
     case INDEX_op_brcond_i64:
-        tci_args_rrcl(&tb_ptr, &r0, &r1, &c, &ptr);
-        info->fprintf_func(info->stream, "%-12s  %s,%s,%s,%p",
-                           op_name, str_r(r0), str_r(r1), str_c(c), ptr);
+        tci_args_rl(&tb_ptr, &r0, &ptr);
+        info->fprintf_func(info->stream, "%-12s  %s,0,ne,%p",
+                           op_name, str_r(r0), ptr);
         break;
 
     case INDEX_op_setcond_i32:
@@ -1315,13 +1288,6 @@ int print_insn_tci(bfd_vma addr, disassemble_info *info)
                            str_r(r3), str_r(r4), str_c(c));
         break;
 
-    case INDEX_op_brcond2_i32:
-        tci_args_rrrrcl(&tb_ptr, &r0, &r1, &r2, &r3, &c, &ptr);
-        info->fprintf_func(info->stream, "%-12s  %s,%s,%s,%s,%s,%p",
-                           op_name, str_r(r0), str_r(r1),
-                           str_r(r2), str_r(r3), str_c(c), ptr);
-        break;
-
     case INDEX_op_mulu2_i32:
         tci_args_rrrr(&tb_ptr, &r0, &r1, &r2, &r3);
         info->fprintf_func(info->stream, "%-12s  %s,%s,%s,%s",
diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc
index b29e75425d..e06d4e9380 100644
--- a/tcg/tci/tcg-target.c.inc
+++ b/tcg/tci/tcg-target.c.inc
@@ -349,6 +349,17 @@ static void tcg_out_op_rI(TCGContext *s, TCGOpcode op,
 }
 #endif
 
+static void tcg_out_op_rl(TCGContext *s, TCGOpcode op, TCGReg r0, TCGLabel *l1)
+{
+    uint8_t *old_code_ptr = s->code_ptr;
+
+    tcg_out_op_t(s, op);
+    tcg_out_r(s, r0);
+    tci_out_label(s, l1);
+
+    old_code_ptr[1] = s->code_ptr - old_code_ptr;
+}
+
 static void tcg_out_op_rr(TCGContext *s, TCGOpcode op, TCGReg r0, TCGReg r1)
 {
     uint8_t *old_code_ptr = s->code_ptr;
@@ -400,20 +411,6 @@ static void tcg_out_op_rrs(TCGContext *s, TCGOpcode op,
     old_code_ptr[1] = s->code_ptr - old_code_ptr;
 }
 
-static void tcg_out_op_rrcl(TCGContext *s, TCGOpcode op,
-                            TCGReg r0, TCGReg r1, TCGCond c2, TCGLabel *l3)
-{
-    uint8_t *old_code_ptr = s->code_ptr;
-
-    tcg_out_op_t(s, op);
-    tcg_out_r(s, r0);
-    tcg_out_r(s, r1);
-    tcg_out8(s, c2);
-    tci_out_label(s, l3);
-
-    old_code_ptr[1] = s->code_ptr - old_code_ptr;
-}
-
 static void tcg_out_op_rrrc(TCGContext *s, TCGOpcode op,
                             TCGReg r0, TCGReg r1, TCGReg r2, TCGCond c3)
 {
@@ -487,23 +484,6 @@ static void tcg_out_op_rrrr(TCGContext *s, TCGOpcode op,
     old_code_ptr[1] = s->code_ptr - old_code_ptr;
 }
 
-static void tcg_out_op_rrrrcl(TCGContext *s, TCGOpcode op,
-                              TCGReg r0, TCGReg r1, TCGReg r2, TCGReg r3,
-                              TCGCond c4, TCGLabel *l5)
-{
-    uint8_t *old_code_ptr = s->code_ptr;
-
-    tcg_out_op_t(s, op);
-    tcg_out_r(s, r0);
-    tcg_out_r(s, r1);
-    tcg_out_r(s, r2);
-    tcg_out_r(s, r3);
-    tcg_out8(s, c4);
-    tci_out_label(s, l5);
-
-    old_code_ptr[1] = s->code_ptr - old_code_ptr;
-}
-
 static void tcg_out_op_rrrrrc(TCGContext *s, TCGOpcode op,
                               TCGReg r0, TCGReg r1, TCGReg r2,
                               TCGReg r3, TCGReg r4, TCGCond c5)
@@ -704,7 +684,10 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args,
         break;
 
     CASE_32_64(brcond)
-        tcg_out_op_rrcl(s, opc, args[0], args[1], args[2], arg_label(args[3]));
+        tcg_out_op_rrrc(s, (opc == INDEX_op_brcond_i32
+                            ? INDEX_op_setcond_i32 : INDEX_op_setcond_i64),
+                        TCG_REG_TMP, args[0], args[1], args[2]);
+        tcg_out_op_rl(s, opc, TCG_REG_TMP, arg_label(args[3]));
         break;
 
     CASE_32_64(neg)      /* Optional (TCG_TARGET_HAS_neg_*). */
@@ -730,8 +713,9 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args,
                           args[3], args[4], args[5]);
         break;
     case INDEX_op_brcond2_i32:
-        tcg_out_op_rrrrcl(s, opc, args[0], args[1], args[2],
-                          args[3], args[4], arg_label(args[5]));
+        tcg_out_op_rrrrrc(s, INDEX_op_setcond2_i32, TCG_REG_TMP,
+                          args[0], args[1], args[2], args[3], args[4]);
+        tcg_out_op_rl(s, INDEX_op_brcond_i32, TCG_REG_TMP, arg_label(args[5]));
         break;
     case INDEX_op_mulu2_i32:
         tcg_out_op_rrrr(s, opc, args[0], args[1], args[2], args[3]);
-- 
2.25.1