arch-riscv,isa: Fix for compressed jump (c_j) imm

c_j(al) has a special format, called CJ.
The jump offset format is instbits[12:2] --> offset[11|4|9:8|10|6|7|3:1|5]
Currently in decoder.isa, c_j format is JOp, the imm and branchTarget are incorrect
In the execute section (decoder.isa:228), the imm fields is ignored and the offset is calculated correctlly.
As a result, we get decoder flush for each c_j instance
I've added CJOp format in compressed.isa, and use it in execute section.
In addition, c_j is mappped to jal zero, cj_imm, and actually is neither indirect control nor a function call
I fixed the flags accordently.
I'll fix all IsRet, IsCall and IsIndirectControl flags for rest of (c_)jal(r) in my next commit.
I ran coremark -O0 before my fix and I got 37.7% branch miss-rate, after the fix the branch miss-rate is <13%

Change-Id: I608d5894a78a1ebefe36f21e21aaea68b42bccfc
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/17808
Maintainer: Jason Lowe-Power <jason@lowepower.com>
Tested-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Jason Lowe-Power <jason@lowepower.com>
Reviewed-by: Alec Roelke <alec.roelke@gmail.com>

3 files changed, 34 insertions, 13 deletions

diff --git a/src/arch/riscv/isa/bitfields.isa b/src/arch/riscv/isa/bitfields.isa
index 903fce385..20f1fc0de 100644
--- a/src/arch/riscv/isa/bitfields.isa
+++ b/src/arch/riscv/isa/bitfields.isa
@@ -104,6 +104,14 @@ def bitfield FC1 <11:7>;
 def bitfield FC2 <6:2>;
 def bitfield FP2 <4:2>;
 def bitfield CJUMPIMM <12:2>;
+def bitfield CJUMPIMM3TO1 <5:3>;
+def bitfield CJUMPIMM4TO4 <11:11>;
+def bitfield CJUMPIMM5TO5 <2:2>;
+def bitfield CJUMPIMM6TO6 <7:7>;
+def bitfield CJUMPIMM7TO7 <6:6>;
+def bitfield CJUMPIMM9TO8 <10:9>;
+def bitfield CJUMPIMM10TO10 <8:8>;
+def bitfield CJUMPIMMSIGN <12:12>;
 def bitfield CIMM8 <12:5>;
 def bitfield CIMM6 <12:7>;
 def bitfield CIMM5 <6:2>;
diff --git a/src/arch/riscv/isa/decoder.isa b/src/arch/riscv/isa/decoder.isa
index 8de4829a6..8fcfba6ca 100644
--- a/src/arch/riscv/isa/decoder.isa
+++ b/src/arch/riscv/isa/decoder.isa
@@ -224,18 +224,9 @@ decode QUADRANT default Unknown::unknown() {
                 }
             }
         }
-        0x5: JOp::c_j({{
-            int64_t offset = CJUMPIMM<3:1> << 1 |
-                             CJUMPIMM<9:9> << 4 |
-                             CJUMPIMM<0:0> << 5 |
-                             CJUMPIMM<5:5> << 6 |
-                             CJUMPIMM<4:4> << 7 |
-                             CJUMPIMM<8:7> << 8 |
-                             CJUMPIMM<6:6> << 10;
-            if (CJUMPIMM<10:10> > 0)
-                offset |= ~((int64_t)0x7FF);
-            NPC = PC + offset;
-        }}, IsIndirectControl, IsUncondControl, IsCall);
+        0x5: CJOp::c_j({{
+            NPC = PC + imm;
+        }}, IsDirectControl, IsUncondControl);
         format CBOp {
             0x6: c_beqz({{
                 if (Rp1 == 0)
diff --git a/src/arch/riscv/isa/formats/compressed.isa b/src/arch/riscv/isa/formats/compressed.isa
index 3ebc1c6ae..b520d53eb 100644
--- a/src/arch/riscv/isa/formats/compressed.isa
+++ b/src/arch/riscv/isa/formats/compressed.isa
@@ -47,6 +47,28 @@ def format CIOp(imm_code, code, imm_type='int64_t', *opt_flags) {{
     exec_output = ImmExecute.subst(iop)
 }};
 
+def format CJOp(code, *opt_flags) {{
+    regs = ['_destRegIdx[0]', '_srcRegIdx[0]']
+    imm_code = """
+           imm =             CJUMPIMM3TO1 << 1 |
+                             CJUMPIMM4TO4 << 4 |
+                             CJUMPIMM5TO5 << 5 |
+                             CJUMPIMM6TO6 << 6 |
+                             CJUMPIMM7TO7 << 7 |
+                             CJUMPIMM9TO8 << 8 |
+                             CJUMPIMM10TO10 << 10;
+            if (CJUMPIMMSIGN)
+                imm |= ~((int64_t)0x7FF);
+    """
+    iop = InstObjParams(name, Name, 'ImmOp<int64_t>',
+        {'code': code, 'imm_code': imm_code,
+         'regs': ','.join(regs)}, opt_flags)
+    header_output = BranchDeclare.subst(iop)
+    decoder_output = ImmConstructor.subst(iop)
+    decode_block = BasicDecode.subst(iop)
+    exec_output = BranchExecute.subst(iop)
+}};
+
 def format CBOp(code, *opt_flags) {{
     imm_code = """
                 imm = CIMM5<2:1> << 1 |
@@ -78,4 +100,4 @@ def format CompressedStore(sdisp_code, memacc_code,
     (header_output, decoder_output, decode_block, exec_output) = \
         LoadStoreBase(name, Name, sdisp_code, ea_code, memacc_code, mem_flags,
         inst_flags, 'Store', exec_template_base='Store')
-}};
\ No newline at end of file
+}};