diff options
Diffstat (limited to 'target-s390x/fpu_helper.c')
-rw-r--r-- | target-s390x/fpu_helper.c | 74 |
1 files changed, 20 insertions, 54 deletions
diff --git a/target-s390x/fpu_helper.c b/target-s390x/fpu_helper.c index 92458ce8da..ea61936571 100644 --- a/target-s390x/fpu_helper.c +++ b/target-s390x/fpu_helper.c @@ -212,27 +212,31 @@ uint64_t HELPER(axb)(CPUS390XState *env, uint64_t ah, uint64_t al, return RET128(ret); } -/* 32-bit FP subtraction RR */ -uint32_t HELPER(sebr)(CPUS390XState *env, uint32_t f1, uint32_t f2) +/* 32-bit FP subtraction */ +uint64_t HELPER(seb)(CPUS390XState *env, uint64_t f1, uint64_t f2) { - env->fregs[f1].l.upper = float32_sub(env->fregs[f1].l.upper, - env->fregs[f2].l.upper, - &env->fpu_status); - HELPER_LOG("%s: adding 0x%d resulting in 0x%d in f%d\n", __func__, - env->fregs[f2].l.upper, env->fregs[f1].l.upper, f1); - - return set_cc_nz_f32(env->fregs[f1].l.upper); + float32 ret = float32_sub(f1, f2, &env->fpu_status); + handle_exceptions(env, GETPC()); + return ret; } -/* 64-bit FP subtraction RR */ -uint32_t HELPER(sdbr)(CPUS390XState *env, uint32_t f1, uint32_t f2) +/* 64-bit FP subtraction */ +uint64_t HELPER(sdb)(CPUS390XState *env, uint64_t f1, uint64_t f2) { - env->fregs[f1].d = float64_sub(env->fregs[f1].d, env->fregs[f2].d, - &env->fpu_status); - HELPER_LOG("%s: subtracting 0x%ld resulting in 0x%ld in f%d\n", - __func__, env->fregs[f2].d, env->fregs[f1].d, f1); + float64 ret = float64_sub(f1, f2, &env->fpu_status); + handle_exceptions(env, GETPC()); + return ret; +} - return set_cc_nz_f64(env->fregs[f1].d); +/* 128-bit FP subtraction */ +uint64_t HELPER(sxb)(CPUS390XState *env, uint64_t ah, uint64_t al, + uint64_t bh, uint64_t bl) +{ + float128 ret = float128_sub(make_float128(ah, al), + make_float128(bh, bl), + &env->fpu_status); + handle_exceptions(env, GETPC()); + return RET128(ret); } /* 32-bit FP division RR */ @@ -446,27 +450,6 @@ uint32_t HELPER(cxb)(CPUS390XState *env, uint64_t ah, uint64_t al, return float_comp_to_cc(env, cmp); } -/* 32-bit FP subtraction RM */ -void HELPER(seb)(CPUS390XState *env, uint32_t f1, uint32_t val) -{ - float32 v1 = env->fregs[f1].l.upper; - CPU_FloatU v2; - - v2.l = val; - env->fregs[f1].l.upper = float32_sub(v1, v2.f, &env->fpu_status); -} - -/* 64-bit FP subtraction RM */ -uint32_t HELPER(sdb)(CPUS390XState *env, uint32_t f1, uint64_t a2) -{ - float64 v1 = env->fregs[f1].d; - CPU_DoubleU v2; - - v2.ll = cpu_ldq_data(env, a2); - env->fregs[f1].d = v1 = float64_sub(v1, v2.d, &env->fpu_status); - return set_cc_nz_f64(v1); -} - /* 64-bit FP multiplication RM */ void HELPER(mdb)(CPUS390XState *env, uint32_t f1, uint64_t a2) { @@ -620,23 +603,6 @@ void HELPER(lzxr)(CPUS390XState *env, uint32_t f1) env->fregs[f1 + 1].ll = x.ll.lower; } -/* 128-bit FP subtraction RR */ -uint32_t HELPER(sxbr)(CPUS390XState *env, uint32_t f1, uint32_t f2) -{ - CPU_QuadU v1; - CPU_QuadU v2; - CPU_QuadU res; - - v1.ll.upper = env->fregs[f1].ll; - v1.ll.lower = env->fregs[f1 + 2].ll; - v2.ll.upper = env->fregs[f2].ll; - v2.ll.lower = env->fregs[f2 + 2].ll; - res.q = float128_sub(v1.q, v2.q, &env->fpu_status); - env->fregs[f1].ll = res.ll.upper; - env->fregs[f1 + 2].ll = res.ll.lower; - return set_cc_nz_f128(res.q); -} - /* 32-bit FP multiplication RR */ void HELPER(meebr)(CPUS390XState *env, uint32_t f1, uint32_t f2) { |