target/arm: Implement FEAT_HBC

FEAT_HBC (Hinted conditional branches) provides a new instruction
BC.cond, which behaves exactly like the existing B.cond except
that it provides a hint to the branch predictor about the
likely behaviour of the branch.

Since QEMU does not implement branch prediction, we can treat
this identically to B.cond.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

docs/system/arm/emulation.rst

@@ -42,6 +42,7 @@ the following architecture extensions:
 - FEAT_FlagM2 (Enhancements to flag manipulation instructions)
 - FEAT_GTG (Guest translation granule size)
 - FEAT_HAFDBS (Hardware management of the access flag and dirty bit state)
+- FEAT_HBC (Hinted conditional branches)
 - FEAT_HCX (Support for the HCRX_EL2 register)
 - FEAT_HPDS (Hierarchical permission disables)
 - FEAT_HPDS2 (Translation table page-based hardware attributes)

linux-user/elfload.c

@@ -815,6 +815,7 @@ uint32_t get_elf_hwcap2(void)
     GET_FEATURE_ID(aa64_sme_f64f64, ARM_HWCAP2_A64_SME_F64F64);
     GET_FEATURE_ID(aa64_sme_i16i64, ARM_HWCAP2_A64_SME_I16I64);
     GET_FEATURE_ID(aa64_sme_fa64, ARM_HWCAP2_A64_SME_FA64);
+    GET_FEATURE_ID(aa64_hbc, ARM_HWCAP2_A64_HBC);
 
     return hwcaps;
 }

target/arm/cpu.h

@@ -4088,6 +4088,11 @@ static inline bool isar_feature_aa64_i8mm(const ARMISARegisters *id)
     return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, I8MM) != 0;
 }
 
+static inline bool isar_feature_aa64_hbc(const ARMISARegisters *id)
+{
+    return FIELD_EX64(id->id_aa64isar2, ID_AA64ISAR2, BC) != 0;
+}
+
 static inline bool isar_feature_aa64_tgran4_lpa2(const ARMISARegisters *id)
 {
     return FIELD_SEX64(id->id_aa64mmfr0, ID_AA64MMFR0, TGRAN4) >= 1;

target/arm/tcg/a64.decode

@@ -126,7 +126,8 @@ CBZ             sf:1 011010 nz:1 ................... rt:5 &cbz imm=%imm19
 
 TBZ             . 011011 nz:1 ..... .............. rt:5 &tbz  imm=%imm14 bitpos=%imm31_19
 
-B_cond          0101010 0 ................... 0 cond:4 imm=%imm19
+# B.cond and BC.cond
+B_cond          0101010 0 ................... c:1 cond:4 imm=%imm19
 
 BR              1101011 0000 11111 000000 rn:5 00000 &r
 BLR             1101011 0001 11111 000000 rn:5 00000 &r

target/arm/tcg/cpu64.c

@@ -1027,6 +1027,10 @@ void aarch64_max_tcg_initfn(Object *obj)
     t = FIELD_DP64(t, ID_AA64ISAR1, I8MM, 1);     /* FEAT_I8MM */
     cpu->isar.id_aa64isar1 = t;
 
+    t = cpu->isar.id_aa64isar2;
+    t = FIELD_DP64(t, ID_AA64ISAR2, BC, 1);      /* FEAT_HBC */
+    cpu->isar.id_aa64isar2 = t;
+
     t = cpu->isar.id_aa64pfr0;
     t = FIELD_DP64(t, ID_AA64PFR0, FP, 1);        /* FEAT_FP16 */
     t = FIELD_DP64(t, ID_AA64PFR0, ADVSIMD, 1);   /* FEAT_FP16 */

target/arm/tcg/translate-a64.c

@@ -1453,6 +1453,10 @@ static bool trans_TBZ(DisasContext *s, arg_tbz *a)
 
 static bool trans_B_cond(DisasContext *s, arg_B_cond *a)
 {
+    /* BC.cond is only present with FEAT_HBC */
+    if (a->c && !dc_isar_feature(aa64_hbc, s)) {
+        return false;
+    }
     reset_btype(s);
     if (a->cond < 0x0e) {
         /* genuinely conditional branches */