target/arm: Fix syndrome for FGT traps on ERET

In commit 442c9d6 when we converted the ERET, ERETAA, ERETAB
instructions to decodetree, the conversion accidentally lost the
correct setting of the syndrome register when taking a trap because
of the FEAT_FGT HFGITR_EL1.ERET bit.  Instead of reporting a correct
full syndrome value with the EC and IL bits, we only reported the low
two bits of the syndrome, because the call to syn_erettrap() got
dropped.

Fix the syndrome values for these traps by reinstating the
syn_erettrap() calls.

Fixes: 442c9d6 ("target/arm: Convert ERET, ERETAA, ERETAB to decodetree")
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20231024172438.2990945-1-peter.maydell@linaro.org

target/arm/tcg/translate-a64.c

@@ -1606,7 +1606,7 @@ static bool trans_ERET(DisasContext *s, arg_ERET *a)
         return false;
     }
     if (s->fgt_eret) {
-        gen_exception_insn_el(s, 0, EXCP_UDEF, 0, 2);
+        gen_exception_insn_el(s, 0, EXCP_UDEF, syn_erettrap(0), 2);
         return true;
     }
     dst = tcg_temp_new_i64();
@@ -1633,7 +1633,7 @@ static bool trans_ERETA(DisasContext *s, arg_reta *a)
     }
     /* The FGT trap takes precedence over an auth trap. */
     if (s->fgt_eret) {
-        gen_exception_insn_el(s, 0, EXCP_UDEF, a->m ? 3 : 2, 2);
+        gen_exception_insn_el(s, 0, EXCP_UDEF, syn_erettrap(a->m ? 3 : 2), 2);
         return true;
     }
     dst = tcg_temp_new_i64();