diff --git a/include/fpu/softfloat-types.h b/include/fpu/softfloat-types.h
index 8210a94ea1..4e378cb612 100644
--- a/include/fpu/softfloat-types.h
+++ b/include/fpu/softfloat-types.h
@@ -80,6 +80,12 @@ this code that are retained.
 #ifndef SOFTFLOAT_TYPES_H
 #define SOFTFLOAT_TYPES_H
 
+/* This 'flag' type must be able to hold at least 0 and 1. It should
+ * probably be replaced with 'bool' but the uses would need to be audited
+ * to check that they weren't accidentally relying on it being a larger type.
+ */
+typedef uint8_t flag;
+
 /*
  * Software IEC/IEEE floating-point types.
  */
@@ -112,4 +118,62 @@ typedef struct {
 #define make_float128(high_, low_) ((float128) { .high = high_, .low = low_ })
 #define make_float128_init(high_, low_) { .high = high_, .low = low_ }
 
+/*
+ * Software IEC/IEEE floating-point underflow tininess-detection mode.
+ */
+
+enum {
+    float_tininess_after_rounding  = 0,
+    float_tininess_before_rounding = 1
+};
+
+/*
+ *Software IEC/IEEE floating-point rounding mode.
+ */
+
+enum {
+    float_round_nearest_even = 0,
+    float_round_down         = 1,
+    float_round_up           = 2,
+    float_round_to_zero      = 3,
+    float_round_ties_away    = 4,
+    /* Not an IEEE rounding mode: round to the closest odd mantissa value */
+    float_round_to_odd       = 5,
+};
+
+/*
+ * Software IEC/IEEE floating-point exception flags.
+ */
+
+enum {
+    float_flag_invalid   =  1,
+    float_flag_divbyzero =  4,
+    float_flag_overflow  =  8,
+    float_flag_underflow = 16,
+    float_flag_inexact   = 32,
+    float_flag_input_denormal = 64,
+    float_flag_output_denormal = 128
+};
+
+
+/*
+ * Floating Point Status. Individual architectures may maintain
+ * several versions of float_status for different functions. The
+ * correct status for the operation is then passed by reference to
+ * most of the softfloat functions.
+ */
+
+typedef struct float_status {
+    signed char float_detect_tininess;
+    signed char float_rounding_mode;
+    uint8_t     float_exception_flags;
+    signed char floatx80_rounding_precision;
+    /* should denormalised results go to zero and set the inexact flag? */
+    flag flush_to_zero;
+    /* should denormalised inputs go to zero and set the input_denormal flag? */
+    flag flush_inputs_to_zero;
+    flag default_nan_mode;
+    flag snan_bit_is_one;
+} float_status;
+
 #endif /* SOFTFLOAT_TYPES_H */
diff --git a/include/fpu/softfloat.h b/include/fpu/softfloat.h
index 4e16e22e58..f3b9008f78 100644
--- a/include/fpu/softfloat.h
+++ b/include/fpu/softfloat.h
@@ -82,12 +82,6 @@ this code that are retained.
 #ifndef SOFTFLOAT_H
 #define SOFTFLOAT_H
 
-/* This 'flag' type must be able to hold at least 0 and 1. It should
- * probably be replaced with 'bool' but the uses would need to be audited
- * to check that they weren't accidentally relying on it being a larger type.
- */
-typedef uint8_t flag;
-
 #define LIT64( a ) a##LL
 
 /*----------------------------------------------------------------------------
@@ -102,53 +96,6 @@ enum {
 
 #include "fpu/softfloat-types.h"
 
-/*----------------------------------------------------------------------------
-| Software IEC/IEEE floating-point underflow tininess-detection mode.
-*----------------------------------------------------------------------------*/
-enum {
-    float_tininess_after_rounding  = 0,
-    float_tininess_before_rounding = 1
-};
-
-/*----------------------------------------------------------------------------
-| Software IEC/IEEE floating-point rounding mode.
-*----------------------------------------------------------------------------*/
-enum {
-    float_round_nearest_even = 0,
-    float_round_down         = 1,
-    float_round_up           = 2,
-    float_round_to_zero      = 3,
-    float_round_ties_away    = 4,
-    /* Not an IEEE rounding mode: round to the closest odd mantissa value */
-    float_round_to_odd       = 5,
-};
-
-/*----------------------------------------------------------------------------
-| Software IEC/IEEE floating-point exception flags.
-*----------------------------------------------------------------------------*/
-enum {
-    float_flag_invalid   =  1,
-    float_flag_divbyzero =  4,
-    float_flag_overflow  =  8,
-    float_flag_underflow = 16,
-    float_flag_inexact   = 32,
-    float_flag_input_denormal = 64,
-    float_flag_output_denormal = 128
-};
-
-typedef struct float_status {
-    signed char float_detect_tininess;
-    signed char float_rounding_mode;
-    uint8_t     float_exception_flags;
-    signed char floatx80_rounding_precision;
-    /* should denormalised results go to zero and set the inexact flag? */
-    flag flush_to_zero;
-    /* should denormalised inputs go to zero and set the input_denormal flag? */
-    flag flush_inputs_to_zero;
-    flag default_nan_mode;
-    flag snan_bit_is_one;
-} float_status;
-
 static inline void set_float_detect_tininess(int val, float_status *status)
 {
     status->float_detect_tininess = val;
diff --git a/target/alpha/cpu.h b/target/alpha/cpu.h
index 09720c2f3b..a79fc2e780 100644
--- a/target/alpha/cpu.h
+++ b/target/alpha/cpu.h
@@ -33,8 +33,6 @@
 
 #include "exec/cpu-defs.h"
 
-#include "fpu/softfloat.h"
-
 #define ICACHE_LINE_SIZE 32
 #define DCACHE_LINE_SIZE 32
 
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index d796085be9..1b3ae62db6 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -34,6 +34,7 @@
 #include "sysemu/hw_accel.h"
 #include "kvm_arm.h"
 #include "disas/capstone.h"
+#include "fpu/softfloat.h"
 
 static void arm_cpu_set_pc(CPUState *cs, vaddr value)
 {
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index de62df091c..8c839faa8f 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -39,8 +39,6 @@
 #include "cpu-qom.h"
 #include "exec/cpu-defs.h"
 
-#include "fpu/softfloat.h"
-
 #define EXCP_UDEF            1   /* undefined instruction */
 #define EXCP_SWI             2   /* software interrupt */
 #define EXCP_PREFETCH_ABORT  3
diff --git a/target/arm/helper-a64.c b/target/arm/helper-a64.c
index 06fd321fae..10e08bdc1f 100644
--- a/target/arm/helper-a64.c
+++ b/target/arm/helper-a64.c
@@ -31,6 +31,7 @@
 #include "exec/cpu_ldst.h"
 #include "qemu/int128.h"
 #include "tcg.h"
+#include "fpu/softfloat.h"
 #include <zlib.h> /* For crc32 */
 
 /* C2.4.7 Multiply and divide */
diff --git a/target/arm/helper.c b/target/arm/helper.c
index e7586fcf6c..32e4fd4732 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -15,6 +15,7 @@
 #include <zlib.h> /* For crc32 */
 #include "exec/semihost.h"
 #include "sysemu/kvm.h"
+#include "fpu/softfloat.h"
 
 #define ARM_CPU_FREQ 1000000000 /* FIXME: 1 GHz, should be configurable */
 
diff --git a/target/arm/neon_helper.c b/target/arm/neon_helper.c
index 689491cad3..a1ec6537eb 100644
--- a/target/arm/neon_helper.c
+++ b/target/arm/neon_helper.c
@@ -11,6 +11,7 @@
 #include "cpu.h"
 #include "exec/exec-all.h"
 #include "exec/helper-proto.h"
+#include "fpu/softfloat.h"
 
 #define SIGNBIT (uint32_t)0x80000000
 #define SIGNBIT64 ((uint64_t)1 << 63)
diff --git a/target/hppa/cpu.c b/target/hppa/cpu.c
index 7b635cc4ac..969f628f0a 100644
--- a/target/hppa/cpu.c
+++ b/target/hppa/cpu.c
@@ -23,6 +23,7 @@
 #include "cpu.h"
 #include "qemu-common.h"
 #include "exec/exec-all.h"
+#include "fpu/softfloat.h"
 
 
 static void hppa_cpu_set_pc(CPUState *cs, vaddr value)
diff --git a/target/hppa/cpu.h b/target/hppa/cpu.h
index 7640c81221..c88d844938 100644
--- a/target/hppa/cpu.h
+++ b/target/hppa/cpu.h
@@ -51,7 +51,6 @@
 #define CPUArchState struct CPUHPPAState
 
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
 
 #define TARGET_PAGE_BITS 12
 
diff --git a/target/hppa/op_helper.c b/target/hppa/op_helper.c
index 4ee936bf86..a3af62daf7 100644
--- a/target/hppa/op_helper.c
+++ b/target/hppa/op_helper.c
@@ -24,7 +24,7 @@
 #include "exec/cpu_ldst.h"
 #include "sysemu/sysemu.h"
 #include "qemu/timer.h"
-
+#include "fpu/softfloat.h"
 
 void QEMU_NORETURN HELPER(excp)(CPUHPPAState *env, int excp)
 {
diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index f91e37d25d..faf39ec1ce 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -52,10 +52,6 @@
 
 #define CPUArchState struct CPUX86State
 
-#ifdef CONFIG_TCG
-#include "fpu/softfloat.h"
-#endif
-
 enum {
     R_EAX = 0,
     R_ECX = 1,
diff --git a/target/i386/fpu_helper.c b/target/i386/fpu_helper.c
index 9014b6f88a..ea5a0c4861 100644
--- a/target/i386/fpu_helper.c
+++ b/target/i386/fpu_helper.c
@@ -24,6 +24,7 @@
 #include "qemu/host-utils.h"
 #include "exec/exec-all.h"
 #include "exec/cpu_ldst.h"
+#include "fpu/softfloat.h"
 
 #define FPU_RC_MASK         0xc00
 #define FPU_RC_NEAR         0x000
diff --git a/target/m68k/cpu.c b/target/m68k/cpu.c
index 3026714471..a4ed8770aa 100644
--- a/target/m68k/cpu.c
+++ b/target/m68k/cpu.c
@@ -24,7 +24,7 @@
 #include "qemu-common.h"
 #include "migration/vmstate.h"
 #include "exec/exec-all.h"
-
+#include "fpu/softfloat.h"
 
 static void m68k_cpu_set_pc(CPUState *cs, vaddr value)
 {
diff --git a/target/m68k/cpu.h b/target/m68k/cpu.h
index 1d79885222..65f4fb95cb 100644
--- a/target/m68k/cpu.h
+++ b/target/m68k/cpu.h
@@ -28,7 +28,6 @@
 #include "qemu-common.h"
 #include "exec/cpu-defs.h"
 #include "cpu-qom.h"
-#include "fpu/softfloat.h"
 
 #define OS_BYTE     0
 #define OS_WORD     1
diff --git a/target/m68k/fpu_helper.c b/target/m68k/fpu_helper.c
index 665e7609af..3c5a82aaa0 100644
--- a/target/m68k/fpu_helper.c
+++ b/target/m68k/fpu_helper.c
@@ -23,6 +23,7 @@
 #include "exec/helper-proto.h"
 #include "exec/exec-all.h"
 #include "exec/cpu_ldst.h"
+#include "fpu/softfloat.h"
 
 /* Undefined offsets may be different on various FPU.
  * On 68040 they return 0.0 (floatx80_zero)
diff --git a/target/m68k/helper.c b/target/m68k/helper.c
index 20155c7801..917d46efcc 100644
--- a/target/m68k/helper.c
+++ b/target/m68k/helper.c
@@ -24,6 +24,7 @@
 #include "exec/gdbstub.h"
 
 #include "exec/helper-proto.h"
+#include "fpu/softfloat.h"
 
 #define SIGNBIT (1u << 31)
 
diff --git a/target/m68k/translate.c b/target/m68k/translate.c
index 70c7583621..93cd38950e 100644
--- a/target/m68k/translate.c
+++ b/target/m68k/translate.c
@@ -32,6 +32,8 @@
 
 #include "trace-tcg.h"
 #include "exec/log.h"
+#include "fpu/softfloat.h"
+
 
 //#define DEBUG_DISPATCH 1
 
diff --git a/target/microblaze/cpu.c b/target/microblaze/cpu.c
index d8df2fb07e..4dc1404800 100644
--- a/target/microblaze/cpu.c
+++ b/target/microblaze/cpu.c
@@ -28,6 +28,7 @@
 #include "hw/qdev-properties.h"
 #include "migration/vmstate.h"
 #include "exec/exec-all.h"
+#include "fpu/softfloat.h"
 
 static const struct {
     const char *name;
diff --git a/target/microblaze/cpu.h b/target/microblaze/cpu.h
index f3e7405a62..1fe21c8539 100644
--- a/target/microblaze/cpu.h
+++ b/target/microblaze/cpu.h
@@ -28,7 +28,7 @@
 #define CPUArchState struct CPUMBState
 
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
+#include "fpu/softfloat-types.h"
 struct CPUMBState;
 typedef struct CPUMBState CPUMBState;
 #if !defined(CONFIG_USER_ONLY)
diff --git a/target/microblaze/op_helper.c b/target/microblaze/op_helper.c
index 869072a2d1..1b4fe796e7 100644
--- a/target/microblaze/op_helper.c
+++ b/target/microblaze/op_helper.c
@@ -24,6 +24,7 @@
 #include "qemu/host-utils.h"
 #include "exec/exec-all.h"
 #include "exec/cpu_ldst.h"
+#include "fpu/softfloat.h"
 
 #define D(x)
 
diff --git a/target/moxie/cpu.h b/target/moxie/cpu.h
index a01f480821..d85e1fc061 100644
--- a/target/moxie/cpu.h
+++ b/target/moxie/cpu.h
@@ -34,7 +34,6 @@
 #define MOXIE_EX_BREAK      16
 
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
 
 #define TARGET_PAGE_BITS 12     /* 4k */
 
diff --git a/target/nios2/cpu.h b/target/nios2/cpu.h
index 204b39add7..cd4e40d1b4 100644
--- a/target/nios2/cpu.h
+++ b/target/nios2/cpu.h
@@ -27,7 +27,6 @@
 #define CPUArchState struct CPUNios2State
 
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
 #include "qom/cpu.h"
 struct CPUNios2State;
 typedef struct CPUNios2State CPUNios2State;
diff --git a/target/openrisc/cpu.h b/target/openrisc/cpu.h
index fb46cc9986..5050b1135c 100644
--- a/target/openrisc/cpu.h
+++ b/target/openrisc/cpu.h
@@ -29,7 +29,6 @@ struct OpenRISCCPU;
 
 #include "qemu-common.h"
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
 #include "qom/cpu.h"
 
 #define TYPE_OPENRISC_CPU "or1k-cpu"
diff --git a/target/openrisc/fpu_helper.c b/target/openrisc/fpu_helper.c
index 1375cea948..977a1e8e55 100644
--- a/target/openrisc/fpu_helper.c
+++ b/target/openrisc/fpu_helper.c
@@ -22,6 +22,7 @@
 #include "cpu.h"
 #include "exec/helper-proto.h"
 #include "exception.h"
+#include "fpu/softfloat.h"
 
 static inline uint32_t ieee_ex_to_openrisc(OpenRISCCPU *cpu, int fexcp)
 {
diff --git a/target/ppc/cpu.h b/target/ppc/cpu.h
index 9f8cbbe7aa..7bde1884a1 100644
--- a/target/ppc/cpu.h
+++ b/target/ppc/cpu.h
@@ -79,7 +79,6 @@
 
 #include "exec/cpu-defs.h"
 #include "cpu-qom.h"
-#include "fpu/softfloat.h"
 
 #if defined (TARGET_PPC64)
 #define PPC_ELF_MACHINE     EM_PPC64
diff --git a/target/ppc/fpu_helper.c b/target/ppc/fpu_helper.c
index c4dab159e4..9ae418a577 100644
--- a/target/ppc/fpu_helper.c
+++ b/target/ppc/fpu_helper.c
@@ -21,6 +21,7 @@
 #include "exec/helper-proto.h"
 #include "exec/exec-all.h"
 #include "internal.h"
+#include "fpu/softfloat.h"
 
 static inline float128 float128_snan_to_qnan(float128 x)
 {
diff --git a/target/ppc/int_helper.c b/target/ppc/int_helper.c
index 3a50f1e1b7..35bdf09773 100644
--- a/target/ppc/int_helper.c
+++ b/target/ppc/int_helper.c
@@ -23,6 +23,7 @@
 #include "qemu/host-utils.h"
 #include "exec/helper-proto.h"
 #include "crypto/aes.h"
+#include "fpu/softfloat.h"
 
 #include "helper_regs.h"
 /*****************************************************************************/
diff --git a/target/ppc/translate_init.c b/target/ppc/translate_init.c
index cbaa343e04..17a87df654 100644
--- a/target/ppc/translate_init.c
+++ b/target/ppc/translate_init.c
@@ -38,6 +38,7 @@
 #include "sysemu/qtest.h"
 #include "qemu/cutils.h"
 #include "disas/capstone.h"
+#include "fpu/softfloat.h"
 
 //#define PPC_DUMP_CPU
 //#define PPC_DEBUG_SPR
diff --git a/target/s390x/cpu.c b/target/s390x/cpu.c
index da7cb9c278..a665b9e60e 100644
--- a/target/s390x/cpu.c
+++ b/target/s390x/cpu.c
@@ -42,6 +42,7 @@
 #include "sysemu/arch_init.h"
 #include "sysemu/sysemu.h"
 #endif
+#include "fpu/softfloat.h"
 
 #define CR0_RESET       0xE0UL
 #define CR14_RESET      0xC2000000UL;
diff --git a/target/s390x/cpu.h b/target/s390x/cpu.h
index 21ce40d5b6..96df2fe5c9 100644
--- a/target/s390x/cpu.h
+++ b/target/s390x/cpu.h
@@ -41,8 +41,6 @@
 
 #include "exec/cpu-all.h"
 
-#include "fpu/softfloat.h"
-
 #define NB_MMU_MODES 4
 #define TARGET_INSN_START_EXTRA_WORDS 1
 
diff --git a/target/s390x/fpu_helper.c b/target/s390x/fpu_helper.c
index 334159119f..43f8bf1c94 100644
--- a/target/s390x/fpu_helper.c
+++ b/target/s390x/fpu_helper.c
@@ -24,6 +24,7 @@
 #include "exec/exec-all.h"
 #include "exec/cpu_ldst.h"
 #include "exec/helper-proto.h"
+#include "fpu/softfloat.h"
 
 /* #define DEBUG_HELPER */
 #ifdef DEBUG_HELPER
diff --git a/target/sh4/cpu.c b/target/sh4/cpu.c
index e37c187ca2..6302cfda3a 100644
--- a/target/sh4/cpu.c
+++ b/target/sh4/cpu.c
@@ -25,6 +25,7 @@
 #include "qemu-common.h"
 #include "migration/vmstate.h"
 #include "exec/exec-all.h"
+#include "fpu/softfloat.h"
 
 
 static void superh_cpu_set_pc(CPUState *cs, vaddr value)
diff --git a/target/sh4/cpu.h b/target/sh4/cpu.h
index 52a4568dd5..a649b68d78 100644
--- a/target/sh4/cpu.h
+++ b/target/sh4/cpu.h
@@ -40,8 +40,6 @@
 
 #include "exec/cpu-defs.h"
 
-#include "fpu/softfloat.h"
-
 #define TARGET_PAGE_BITS 12	/* 4k XXXXX */
 
 #define TARGET_PHYS_ADDR_SPACE_BITS 32
diff --git a/target/sh4/op_helper.c b/target/sh4/op_helper.c
index 4b8bbf63b4..4f825bae5a 100644
--- a/target/sh4/op_helper.c
+++ b/target/sh4/op_helper.c
@@ -21,6 +21,7 @@
 #include "exec/helper-proto.h"
 #include "exec/exec-all.h"
 #include "exec/cpu_ldst.h"
+#include "fpu/softfloat.h"
 
 #ifndef CONFIG_USER_ONLY
 
diff --git a/target/sparc/cpu.h b/target/sparc/cpu.h
index 3eaffb354e..9724134a5b 100644
--- a/target/sparc/cpu.h
+++ b/target/sparc/cpu.h
@@ -29,8 +29,6 @@
 
 #include "exec/cpu-defs.h"
 
-#include "fpu/softfloat.h"
-
 /*#define EXCP_INTERRUPT 0x100*/
 
 /* trap definitions */
diff --git a/target/sparc/fop_helper.c b/target/sparc/fop_helper.c
index c7fb176e4c..b6642fd1d7 100644
--- a/target/sparc/fop_helper.c
+++ b/target/sparc/fop_helper.c
@@ -21,6 +21,7 @@
 #include "cpu.h"
 #include "exec/exec-all.h"
 #include "exec/helper-proto.h"
+#include "fpu/softfloat.h"
 
 #define QT0 (env->qt0)
 #define QT1 (env->qt1)
diff --git a/target/tricore/cpu.h b/target/tricore/cpu.h
index f41d2ceb69..e7dfe4bcc6 100644
--- a/target/tricore/cpu.h
+++ b/target/tricore/cpu.h
@@ -24,7 +24,6 @@
 #include "qemu-common.h"
 #include "cpu-qom.h"
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
 
 #define CPUArchState struct CPUTriCoreState
 
diff --git a/target/tricore/fpu_helper.c b/target/tricore/fpu_helper.c
index 7979bb6692..df162902d6 100644
--- a/target/tricore/fpu_helper.c
+++ b/target/tricore/fpu_helper.c
@@ -20,6 +20,7 @@
 #include "qemu/osdep.h"
 #include "cpu.h"
 #include "exec/helper-proto.h"
+#include "fpu/softfloat.h"
 
 #define QUIET_NAN 0x7fc00000
 #define ADD_NAN   0x7fc00001
diff --git a/target/tricore/helper.c b/target/tricore/helper.c
index 378c2a4a76..45276d3782 100644
--- a/target/tricore/helper.c
+++ b/target/tricore/helper.c
@@ -19,6 +19,7 @@
 
 #include "cpu.h"
 #include "exec/exec-all.h"
+#include "fpu/softfloat.h"
 
 enum {
     TLBRET_DIRTY = -4,
diff --git a/target/unicore32/cpu.c b/target/unicore32/cpu.c
index fb837aab4c..29d160a88d 100644
--- a/target/unicore32/cpu.c
+++ b/target/unicore32/cpu.c
@@ -18,6 +18,7 @@
 #include "qemu-common.h"
 #include "migration/vmstate.h"
 #include "exec/exec-all.h"
+#include "fpu/softfloat.h"
 
 static void uc32_cpu_set_pc(CPUState *cs, vaddr value)
 {
diff --git a/target/unicore32/cpu.h b/target/unicore32/cpu.h
index a3cc71416d..42e1d52478 100644
--- a/target/unicore32/cpu.h
+++ b/target/unicore32/cpu.h
@@ -23,7 +23,6 @@
 #include "qemu-common.h"
 #include "cpu-qom.h"
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
 
 #define NB_MMU_MODES            2
 
diff --git a/target/unicore32/ucf64_helper.c b/target/unicore32/ucf64_helper.c
index 6c919010c3..fad3fa6618 100644
--- a/target/unicore32/ucf64_helper.c
+++ b/target/unicore32/ucf64_helper.c
@@ -11,6 +11,7 @@
 #include "qemu/osdep.h"
 #include "cpu.h"
 #include "exec/helper-proto.h"
+#include "fpu/softfloat.h"
 
 /*
  * The convention used for UniCore-F64 instructions:
diff --git a/target/xtensa/cpu.h b/target/xtensa/cpu.h
index f300c02c07..49c2e3cf9a 100644
--- a/target/xtensa/cpu.h
+++ b/target/xtensa/cpu.h
@@ -36,7 +36,6 @@
 #include "qemu-common.h"
 #include "cpu-qom.h"
 #include "exec/cpu-defs.h"
-#include "fpu/softfloat.h"
 #include "xtensa-isa.h"
 
 #define NB_MMU_MODES 4
diff --git a/target/xtensa/op_helper.c b/target/xtensa/op_helper.c
index 43182b113e..7486b99799 100644
--- a/target/xtensa/op_helper.c
+++ b/target/xtensa/op_helper.c
@@ -34,6 +34,7 @@
 #include "exec/cpu_ldst.h"
 #include "exec/address-spaces.h"
 #include "qemu/timer.h"
+#include "fpu/softfloat.h"
 
 void xtensa_cpu_do_unaligned_access(CPUState *cs,
         vaddr addr, MMUAccessType access_type,