Skip to contentSkip to conversation

overall: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

PATCH In review Philippe Mathieu-Daudé 9 participants 29 messages Oct 10, 2025 – Oct 16, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover overall: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

Philippe Mathieu-Daudé (16):
linux-user/arm: Checkpatch style cleanups
linux-user/arm: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check
ui: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check
net: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check
disas: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check hw/core/loader: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check hw/display: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check hw/virtio: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/alpha: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/arm: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/mips: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/ppc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/riscv: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/s390x: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/sparc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check util/bitmap: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

 include/hw/virtio/virtio-access.h       |   6 +-
 include/hw/virtio/virtio-gpu-bswap.h    |   6 +-
 target/arm/cpu.h                        |   8 +-
 target/arm/tcg/translate-a64.h          |   5 +-
 disas/disas-host.c                      |   6 +-
 hw/core/loader.c                        |   5 +-
 hw/display/artist.c                     |  12 +-
 hw/display/vga.c                        |  20 +-
 hw/display/virtio-gpu-gl.c              |   8 +-
 hw/display/virtio-gpu-rutabaga.c        |   8 +-
 hw/virtio/vhost.c                       |   7 +-
 linux-user/arm/nwfpe/double_cpdo.c      |  12 +-
 linux-user/arm/nwfpe/fpa11_cpdt.c       |  77 +-
 net/net.c                               |  14 +-
 target/alpha/translate.c                |  11 +-
 target/arm/tcg/sve_helper.c             |  10 +-
 target/arm/tcg/translate-sve.c          |  22 +-
 target/arm/tcg/translate-vfp.c          |   8 +-
 target/arm/tcg/translate.c              |   6 +-
 target/mips/tcg/msa_helper.c            | 889 ++++++++++++------------
 target/ppc/arch_dump.c                  |   9 +-
 target/ppc/int_helper.c                 |  28 +-
 target/ppc/kvm.c                        |  25 +-
 target/riscv/vector_helper.c            |  32 +-
 target/s390x/tcg/translate.c            |   6 +-
 target/sparc/vis_helper.c               |  18 +-
 ui/vdagent.c                            |  16 +-
 ui/vnc.c                                |   6 +-
 util/bitmap.c                           |  19 +-
 target/ppc/translate/vmx-impl.c.inc     |  14 +-
 target/ppc/translate/vsx-impl.c.inc     |   6 +-
 target/riscv/insn_trans/trans_rvv.c.inc |  16 +-
 target/s390x/tcg/translate_vx.c.inc     |   6 +-
 tcg/ppc/tcg-target.c.inc                |  24 +-
 34 files changed, 656 insertions(+), 709 deletions(-)
Signature · 1 line
2.51.0
#1 lore raw

PATCH 1/16 linux-user/arm: Checkpatch style cleanups

Update style in a pair of methods we are going to modify in the following commit. No functional change.

---
 linux-user/arm/nwfpe/fpa11_cpdt.c | 66 ++++++++++++++++---------------
 1 file changed, 35 insertions(+), 31 deletions(-)
linux-user/arm/nwfpe/fpa11_cpdt.c +35 -31
index fee525937c5..6b0317883a7 100644
@@ -40,18 +40,18 @@ void loadSingle(const unsigned int Fn, target_ulong addr)
 static inline
 void loadDouble(const unsigned int Fn, target_ulong addr)
 {
-   FPA11 *fpa11 = GET_FPA11();
-   unsigned int *p;
-   p = (unsigned int*)&fpa11->fpreg[Fn].fDouble;
-   fpa11->fType[Fn] = typeDouble;
+    FPA11 *fpa11 = GET_FPA11();
+    unsigned int *p;
+    p = (unsigned int *)&fpa11->fpreg[Fn].fDouble;
+    fpa11->fType[Fn] = typeDouble;
 #if HOST_BIG_ENDIAN
-   /* FIXME - handle failure of get_user() */
-   get_user_u32(p[0], addr); /* sign & exponent */
-   get_user_u32(p[1], addr + 4);
+    /* FIXME - handle failure of get_user() */
+    get_user_u32(p[0], addr);       /* sign & exponent */
+    get_user_u32(p[1], addr + 4);
 #else
-   /* FIXME - handle failure of get_user() */
-   get_user_u32(p[0], addr + 4);
-   get_user_u32(p[1], addr); /* sign & exponent */
+    /* FIXME - handle failure of get_user() */
+    get_user_u32(p[0], addr + 4);
+    get_user_u32(p[1], addr);       /* sign & exponent */
 #endif
 }
 
@@ -108,16 +108,18 @@ void storeSingle(const unsigned int Fn, target_ulong addr)
 {
    FPA11 *fpa11 = GET_FPA11();
    float32 val;
-   register unsigned int *p = (unsigned int*)&val;
+   register unsigned int *p = (unsigned int *)&val;
 
    switch (fpa11->fType[Fn])
    {
       case typeDouble:
-         val = float64_to_float32(fpa11->fpreg[Fn].fDouble, &fpa11->fp_status);
+         val = float64_to_float32(fpa11->fpreg[Fn].fDouble,
+                                  &fpa11->fp_status);
       break;
 
       case typeExtended:
-         val = floatx80_to_float32(fpa11->fpreg[Fn].fExtended, &fpa11->fp_status);
+         val = floatx80_to_float32(fpa11->fpreg[Fn].fExtended,
+                                   &fpa11->fp_status);
       break;
 
       default: val = fpa11->fpreg[Fn].fSingle;
@@ -130,29 +132,31 @@ void storeSingle(const unsigned int Fn, target_ulong addr)
 static inline
 void storeDouble(const unsigned int Fn, target_ulong addr)
 {
-   FPA11 *fpa11 = GET_FPA11();
-   float64 val;
-   register unsigned int *p = (unsigned int*)&val;
+    FPA11 *fpa11 = GET_FPA11();
+    float64 val;
+    register unsigned int *p = (unsigned int *)&val;
 
-   switch (fpa11->fType[Fn])
-   {
-      case typeSingle:
-         val = float32_to_float64(fpa11->fpreg[Fn].fSingle, &fpa11->fp_status);
-      break;
+    switch (fpa11->fType[Fn]) {
+    case typeSingle:
+        val = float32_to_float64(fpa11->fpreg[Fn].fSingle,
+                                 &fpa11->fp_status);
+        break;
 
-      case typeExtended:
-         val = floatx80_to_float64(fpa11->fpreg[Fn].fExtended, &fpa11->fp_status);
-      break;
+    case typeExtended:
+        val = floatx80_to_float64(fpa11->fpreg[Fn].fExtended,
+                                  &fpa11->fp_status);
+        break;
 
-      default: val = fpa11->fpreg[Fn].fDouble;
-   }
-   /* FIXME - handle put_user() failures */
+    default:
+        val = fpa11->fpreg[Fn].fDouble;
+    }
+    /* FIXME - handle put_user() failures */
 #if HOST_BIG_ENDIAN
-   put_user_u32(p[0], addr);	/* msw */
-   put_user_u32(p[1], addr + 4);	/* lsw */
+    put_user_u32(p[0], addr);           /* msw */
+    put_user_u32(p[1], addr + 4);       /* lsw */
 #else
-   put_user_u32(p[1], addr);	/* msw */
-   put_user_u32(p[0], addr + 4);	/* lsw */
+    put_user_u32(p[1], addr);           /* msw */
+    put_user_u32(p[0], addr + 4);       /* lsw */
 #endif
 }
 
Signature · 1 line
2.51.0
#2 lore raw

PATCH 2/16 linux-user/arm: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 linux-user/arm/nwfpe/double_cpdo.c | 12 ++----------
 linux-user/arm/nwfpe/fpa11_cpdt.c  | 29 ++++++++++++++---------------
 2 files changed, 16 insertions(+), 25 deletions(-)
linux-user/arm/nwfpe/double_cpdo.c +2 -10
index d45ece2e2fe..e29b38aada0 100644
@@ -150,11 +150,7 @@ unsigned int DoubleCPDO(const unsigned int opcode)
       case MNF_CODE:
       {
          unsigned int *p = (unsigned int*)&rFm;
-#if HOST_BIG_ENDIAN
-         p[0] ^= 0x80000000;
-#else
-         p[1] ^= 0x80000000;
-#endif
+         p[!HOST_BIG_ENDIAN] ^= 0x80000000;
          fpa11->fpreg[Fd].fDouble = rFm;
       }
       break;
@@ -162,11 +158,7 @@ unsigned int DoubleCPDO(const unsigned int opcode)
       case ABS_CODE:
       {
          unsigned int *p = (unsigned int*)&rFm;
-#if HOST_BIG_ENDIAN
-         p[0] &= 0x7fffffff;
-#else
-         p[1] &= 0x7fffffff;
-#endif
+         p[!HOST_BIG_ENDIAN] &= 0x7fffffff;
          fpa11->fpreg[Fd].fDouble = rFm;
       }
       break;
linux-user/arm/nwfpe/fpa11_cpdt.c +14 -15
index 6b0317883a7..7b0a715d403 100644
@@ -44,15 +44,14 @@ void loadDouble(const unsigned int Fn, target_ulong addr)
     unsigned int *p;
     p = (unsigned int *)&fpa11->fpreg[Fn].fDouble;
     fpa11->fType[Fn] = typeDouble;
-#if HOST_BIG_ENDIAN
     /* FIXME - handle failure of get_user() */
-    get_user_u32(p[0], addr);       /* sign & exponent */
-    get_user_u32(p[1], addr + 4);
-#else
-    /* FIXME - handle failure of get_user() */
-    get_user_u32(p[0], addr + 4);
-    get_user_u32(p[1], addr);       /* sign & exponent */
-#endif
+    if (HOST_BIG_ENDIAN) {
+        get_user_u32(p[0], addr);   /* sign & exponent */
+        get_user_u32(p[1], addr + 4);
+    } else {
+        get_user_u32(p[0], addr + 4);
+        get_user_u32(p[1], addr);   /* sign & exponent */
+    }
 }
 
 static inline
@@ -151,13 +150,13 @@ void storeDouble(const unsigned int Fn, target_ulong addr)
         val = fpa11->fpreg[Fn].fDouble;
     }
     /* FIXME - handle put_user() failures */
-#if HOST_BIG_ENDIAN
-    put_user_u32(p[0], addr);           /* msw */
-    put_user_u32(p[1], addr + 4);       /* lsw */
-#else
-    put_user_u32(p[1], addr);           /* msw */
-    put_user_u32(p[0], addr + 4);       /* lsw */
-#endif
+    if (HOST_BIG_ENDIAN) {
+        put_user_u32(p[0], addr);       /* msw */
+        put_user_u32(p[1], addr + 4);   /* lsw */
+    } else {
+        put_user_u32(p[1], addr);       /* msw */
+        put_user_u32(p[0], addr + 4);   /* lsw */
+    }
 }
 
 static inline
Signature · 1 line
2.51.0
#3 lore raw

PATCH 3/16 ui: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 ui/vdagent.c | 16 ++++++++--------
 ui/vnc.c     |  6 +-----
 2 files changed, 9 insertions(+), 13 deletions(-)
ui/vdagent.c +8 -8
index ddb91e75c64..66dc33567df 100644
@@ -660,14 +660,14 @@ static void vdagent_chr_open(Chardev *chr,
     VDAgentChardev *vd = QEMU_VDAGENT_CHARDEV(chr);
     ChardevQemuVDAgent *cfg = backend->u.qemu_vdagent.data;
 
-#if HOST_BIG_ENDIAN
-    /*
-     * TODO: vdagent protocol is defined to be LE,
-     * so we have to byteswap everything on BE hosts.
-     */
-    error_setg(errp, "vdagent is not supported on bigendian hosts");
-    return;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        /*
+         * TODO: vdagent protocol is defined to be LE,
+         * so we have to byteswap everything on BE hosts.
+         */
+        error_setg(errp, "vdagent is not supported on bigendian hosts");
+        return;
+    }
 
     vd->mouse = VDAGENT_MOUSE_DEFAULT;
     if (cfg->has_mouse) {
ui/vnc.c +1 -5
index 77c823bf2e8..e93b5335690 100644
@@ -2348,11 +2348,7 @@ static void pixel_format_message (VncState *vs) {
     vnc_write_u8(vs, vs->client_pf.bits_per_pixel); /* bits-per-pixel */
     vnc_write_u8(vs, vs->client_pf.depth); /* depth */
 
-#if HOST_BIG_ENDIAN
-    vnc_write_u8(vs, 1);             /* big-endian-flag */
-#else
-    vnc_write_u8(vs, 0);             /* big-endian-flag */
-#endif
+    vnc_write_u8(vs, HOST_BIG_ENDIAN);         /* big-endian-flag */
     vnc_write_u8(vs, 1);             /* true-color-flag */
     vnc_write_u16(vs, vs->client_pf.rmax);     /* red-max */
     vnc_write_u16(vs, vs->client_pf.gmax);     /* green-max */
Signature · 1 line
2.51.0
Reviewed-by on v1 3/16
#4 lore raw

Re: [PATCH 03/16] ui: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 8 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

Quoted 53 lines
---
 ui/vdagent.c | 16 ++++++++--------
 ui/vnc.c     |  6 +-----
 2 files changed, 9 insertions(+), 13 deletions(-)
ui/vdagent.c +8 -8
index ddb91e75c64..66dc33567df 100644
@@ -660,14 +660,14 @@ static void vdagent_chr_open(Chardev *chr,
     VDAgentChardev *vd = QEMU_VDAGENT_CHARDEV(chr);
     ChardevQemuVDAgent *cfg = backend->u.qemu_vdagent.data;
 
-#if HOST_BIG_ENDIAN
-    /*
-     * TODO: vdagent protocol is defined to be LE,
-     * so we have to byteswap everything on BE hosts.
-     */
-    error_setg(errp, "vdagent is not supported on bigendian hosts");
-    return;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        /*
+         * TODO: vdagent protocol is defined to be LE,
+         * so we have to byteswap everything on BE hosts.
+         */
+        error_setg(errp, "vdagent is not supported on bigendian hosts");
+        return;
+    }
 
     vd->mouse = VDAGENT_MOUSE_DEFAULT;
     if (cfg->has_mouse) {
ui/vnc.c +1 -6
index 77c823bf2e8..e93b5335690 100644
@@ -2348,11 +2348,7 @@ static void pixel_format_message (VncState *vs) {
     vnc_write_u8(vs, vs->client_pf.bits_per_pixel); /* bits-per-pixel */
     vnc_write_u8(vs, vs->client_pf.depth); /* depth */
 
-#if HOST_BIG_ENDIAN
-    vnc_write_u8(vs, 1);             /* big-endian-flag */
-#else
-    vnc_write_u8(vs, 0);             /* big-endian-flag */
-#endif
+    vnc_write_u8(vs, HOST_BIG_ENDIAN);         /* big-endian-flag */
     vnc_write_u8(vs, 1);             /* true-color-flag */
     vnc_write_u16(vs, vs->client_pf.rmax);     /* red-max */
     vnc_write_u16(vs, vs->client_pf.gmax);     /* green-max */
--

2.51.0

#5 lore raw

PATCH 4/16 net: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 net/net.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)
net/net.c +6 -8
index 27e0d278071..11d1f7616a6 100644
@@ -593,28 +593,26 @@ bool qemu_get_vnet_hash_supported_types(NetClientState *nc, uint32_t *types)
 
 int qemu_set_vnet_le(NetClientState *nc, bool is_le)
 {
-#if HOST_BIG_ENDIAN
+    if (!HOST_BIG_ENDIAN) {
+        return 0;
+    }
     if (!nc || !nc->info->set_vnet_le) {
         return -ENOSYS;
     }
 
     return nc->info->set_vnet_le(nc, is_le);
-#else
-    return 0;
-#endif
 }
 
 int qemu_set_vnet_be(NetClientState *nc, bool is_be)
 {
-#if HOST_BIG_ENDIAN
-    return 0;
-#else
+    if (HOST_BIG_ENDIAN) {
+        return 0;
+    }
     if (!nc || !nc->info->set_vnet_be) {
         return -ENOSYS;
     }
 
     return nc->info->set_vnet_be(nc, is_be);
-#endif
 }
 
 int qemu_can_receive_packet(NetClientState *nc)
Signature · 1 line
2.51.0
#6 lore raw

PATCH 5/16 disas: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 disas/disas-host.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)
disas/disas-host.c +1 -5
index 8146fafe804..54be5ef18e4 100644
@@ -37,11 +37,7 @@ static void initialize_debug_host(CPUDebug *s)
 
     s->info.read_memory_func = host_read_memory;
     s->info.print_address_func = host_print_address;
-#if HOST_BIG_ENDIAN
-    s->info.endian = BFD_ENDIAN_BIG;
-#else
-    s->info.endian = BFD_ENDIAN_LITTLE;
-#endif
+    s->info.endian = HOST_BIG_ENDIAN ? BFD_ENDIAN_BIG : BFD_ENDIAN_LITTLE;
 #if defined(CONFIG_TCG_INTERPRETER)
     s->info.print_insn = print_insn_tci;
 #elif defined(__i386__)
Signature · 1 line
2.51.0
#7 lore raw

PATCH 6/16 hw/core/loader: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 hw/core/loader.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
hw/core/loader.c +3 -2
index 477661a0255..aef10cb3f2b 100644
@@ -491,7 +491,9 @@ ssize_t load_elf_ram_sym(const char *filename,
 
 static void bswap_uboot_header(uboot_image_header_t *hdr)
 {
-#if !HOST_BIG_ENDIAN
+    if (HOST_BIG_ENDIAN) {
+        return;
+    }
     bswap32s(&hdr->ih_magic);
     bswap32s(&hdr->ih_hcrc);
     bswap32s(&hdr->ih_time);
@@ -499,7 +501,6 @@ static void bswap_uboot_header(uboot_image_header_t *hdr)
     bswap32s(&hdr->ih_load);
     bswap32s(&hdr->ih_ep);
     bswap32s(&hdr->ih_dcrc);
-#endif
 }
 
 
Signature · 1 line
2.51.0
Reviewed-by on v1 6/16
#8 lore raw

Re: [PATCH 06/16] hw/core/loader: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 8 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

Alistair

Quoted 31 lines
---
 hw/core/loader.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
hw/core/loader.c +3 -3
index 477661a0255..aef10cb3f2b 100644
@@ -491,7 +491,9 @@ ssize_t load_elf_ram_sym(const char *filename,
 
 static void bswap_uboot_header(uboot_image_header_t *hdr)
 {
-#if !HOST_BIG_ENDIAN
+    if (HOST_BIG_ENDIAN) {
+        return;
+    }
     bswap32s(&hdr->ih_magic);
     bswap32s(&hdr->ih_hcrc);
     bswap32s(&hdr->ih_time);
@@ -499,7 +501,6 @@ static void bswap_uboot_header(uboot_image_header_t *hdr)
     bswap32s(&hdr->ih_load);
     bswap32s(&hdr->ih_ep);
     bswap32s(&hdr->ih_dcrc);
-#endif
 }
 
 
--

2.51.0

#9 lore raw

PATCH 7/16 hw/display: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 include/hw/virtio/virtio-gpu-bswap.h |  6 +++---
 hw/display/artist.c                  | 12 ++++++------
 hw/display/vga.c                     | 20 ++++++++------------
 hw/display/virtio-gpu-gl.c           |  8 ++++----
 hw/display/virtio-gpu-rutabaga.c     |  8 ++++----
 5 files changed, 25 insertions(+), 29 deletions(-)
include/hw/virtio/virtio-gpu-bswap.h +3 -3
index dd1975e2d40..279734231e2 100644
@@ -29,7 +29,9 @@ virtio_gpu_ctrl_hdr_bswap(struct virtio_gpu_ctrl_hdr *hdr)
 static inline void
 virtio_gpu_bswap_32(void *ptr, size_t size)
 {
-#if HOST_BIG_ENDIAN
+    if (!HOST_BIG_ENDIAN) {
+        return;
+    }
 
     size_t i;
     struct virtio_gpu_ctrl_hdr *hdr = (struct virtio_gpu_ctrl_hdr *) ptr;
@@ -41,8 +43,6 @@ virtio_gpu_bswap_32(void *ptr, size_t size)
         le32_to_cpus((uint32_t *)(ptr + i));
         i = i + sizeof(uint32_t);
     }
-
-#endif
 }
 
 static inline void
hw/display/artist.c +6 -6
index 3c884c92437..8d73f2d894d 100644
@@ -738,9 +738,9 @@ static void combine_write_reg(hwaddr addr, uint64_t val, int size, void *out)
      * FIXME: is there a qemu helper for this?
      */
 
-#if !HOST_BIG_ENDIAN
-    addr ^= 3;
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        addr ^= 3;
+    }
 
     switch (size) {
     case 1:
@@ -1132,9 +1132,9 @@ static uint64_t combine_read_reg(hwaddr addr, int size, void *in)
      * FIXME: is there a qemu helper for this?
      */
 
-#if !HOST_BIG_ENDIAN
-    addr ^= 3;
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        addr ^= 3;
+    }
 
     switch (size) {
     case 1:
hw/display/vga.c +8 -12
index 90b89cf4044..01b1cfa98d3 100644
@@ -1307,13 +1307,13 @@ static void vga_draw_text(VGACommonState *s, int full_update)
                 if (cx > cx_max)
                     cx_max = cx;
                 *ch_attr_ptr = ch_attr;
-#if HOST_BIG_ENDIAN
-                ch = ch_attr >> 8;
-                cattr = ch_attr & 0xff;
-#else
-                ch = ch_attr & 0xff;
-                cattr = ch_attr >> 8;
-#endif
+                if (HOST_BIG_ENDIAN) {
+                    ch = ch_attr >> 8;
+                    cattr = ch_attr & 0xff;
+                } else {
+                    ch = ch_attr & 0xff;
+                    cattr = ch_attr >> 8;
+                }
                 font_ptr = font_base[(cattr >> 3) & 1];
                 font_ptr += 32 * 4 * ch;
                 bgcol = palette[cattr >> 4];
@@ -1489,11 +1489,7 @@ static void vga_draw_graphic(VGACommonState *s, int full_update)
     vga_draw_line_func *vga_draw_line = NULL;
     bool allocate_surface, force_shadow = false;
     pixman_format_code_t format;
-#if HOST_BIG_ENDIAN
-    bool byteswap = !s->big_endian_fb;
-#else
-    bool byteswap = s->big_endian_fb;
-#endif
+    bool byteswap = s->big_endian_fb ^ HOST_BIG_ENDIAN;
 
     full_update |= update_basic_params(s);
 
hw/display/virtio-gpu-gl.c +4 -4
index c06a078fb36..c29152a3c7a 100644
@@ -123,10 +123,10 @@ static void virtio_gpu_gl_device_realize(DeviceState *qdev, Error **errp)
     ERRP_GUARD();
     VirtIOGPU *g = VIRTIO_GPU(qdev);
 
-#if HOST_BIG_ENDIAN
-    error_setg(errp, "virgl is not supported on bigendian platforms");
-    return;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        error_setg(errp, "virgl is not supported on bigendian platforms");
+        return;
+    }
 
     if (!object_resolve_path_type("", TYPE_VIRTIO_GPU_GL, NULL)) {
         error_setg(errp, "at most one %s device is permitted", TYPE_VIRTIO_GPU_GL);
hw/display/virtio-gpu-rutabaga.c +4 -4
index ed5ae52acbe..f82a84b53aa 100644
@@ -1074,10 +1074,10 @@ static void virtio_gpu_rutabaga_realize(DeviceState *qdev, Error **errp)
     VirtIOGPUBase *bdev = VIRTIO_GPU_BASE(qdev);
     VirtIOGPU *gpudev = VIRTIO_GPU(qdev);
 
-#if HOST_BIG_ENDIAN
-    error_setg(errp, "rutabaga is not supported on bigendian platforms");
-    return;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        error_setg(errp, "rutabaga is not supported on bigendian platforms");
+        return;
+    }
 
     if (!virtio_gpu_rutabaga_init(gpudev, errp)) {
         return;
Signature · 1 line
2.51.0
Reviewed-by on v1 7/16
#10 lore raw

Re: [PATCH 07/16] hw/display: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 7 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

#11 lore raw

PATCH 8/16 hw/virtio: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 include/hw/virtio/virtio-access.h | 6 +-----
 hw/virtio/vhost.c                 | 7 +++----
 2 files changed, 4 insertions(+), 9 deletions(-)
include/hw/virtio/virtio-access.h +1 -5
index 07aae69042a..80328912ad3 100644
@@ -149,11 +149,7 @@ static inline uint64_t virtio_ldq_p(VirtIODevice *vdev, const void *ptr)
 
 static inline uint16_t virtio_tswap16(VirtIODevice *vdev, uint16_t s)
 {
-#if HOST_BIG_ENDIAN
-    return virtio_access_is_big_endian(vdev) ? s : bswap16(s);
-#else
-    return virtio_access_is_big_endian(vdev) ? bswap16(s) : s;
-#endif
+    return HOST_BIG_ENDIAN ^ virtio_access_is_big_endian(vdev) ? s : bswap16(s);
 }
 
 static inline uint16_t virtio_lduw_phys_cached(VirtIODevice *vdev,
hw/virtio/vhost.c +3 -4
index 266a11514a1..6343477b42f 100644
@@ -1168,11 +1168,10 @@ static inline bool vhost_needs_vring_endian(VirtIODevice *vdev)
     if (virtio_vdev_has_feature(vdev, VIRTIO_F_VERSION_1)) {
         return false;
     }
-#if HOST_BIG_ENDIAN
-    return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_LITTLE;
-#else
+    if (HOST_BIG_ENDIAN) {
+        return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_LITTLE;
+    }
     return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_BIG;
-#endif
 }
 
 static int vhost_virtqueue_set_vring_endian_legacy(struct vhost_dev *dev,
Signature · 1 line
2.51.0
#12 lore raw

Re: [PATCH 08/16] hw/virtio: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 17 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
  include/hw/virtio/virtio-access.h | 6 +-----
  hw/virtio/vhost.c                 | 7 +++----
  2 files changed, 4 insertions(+), 9 deletions(-)
include/hw/virtio/virtio-access.h +0 -0
index 07aae69042a..80328912ad3 100644

include/hw/virtio/virtio-access.h static inline uint64_t virtio_ldq_p(VirtIODevice *vdev, const void *ptr) · line 152

Quoted 9 lines
include/hw/virtio/virtio-access.h +1 -5
@@ -149,11 +149,7 @@ static inline uint64_t virtio_ldq_p(VirtIODevice *vdev, const void *ptr)
 
  static inline uint16_t virtio_tswap16(VirtIODevice *vdev, uint16_t s)
  {
-#if HOST_BIG_ENDIAN
-    return virtio_access_is_big_endian(vdev) ? s : bswap16(s);
-#else
-    return virtio_access_is_big_endian(vdev) ? bswap16(s) : s;
-#endif
+    return HOST_BIG_ENDIAN ^ virtio_access_is_big_endian(vdev) ? s : bswap16(s);

This patch breaks virtio devices(at least input/net devices) on s390x. I am not sure if ^ is the right check here? Changing the logic back to how it was fixes it for me.

Thanks

Farhan

Quoted 22 lines
include/hw/virtio/virtio-access.h +0 -0
  }
 
  static inline uint16_t virtio_lduw_phys_cached(VirtIODevice *vdev,
hw/virtio/vhost.c +3 -4
index 266a11514a1..6343477b42f 100644
@@ -1168,11 +1168,10 @@ static inline bool vhost_needs_vring_endian(VirtIODevice *vdev)
      if (virtio_vdev_has_feature(vdev, VIRTIO_F_VERSION_1)) {
          return false;
      }
-#if HOST_BIG_ENDIAN
-    return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_LITTLE;
-#else
+    if (HOST_BIG_ENDIAN) {
+        return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_LITTLE;
+    }
      return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_BIG;
-#endif
  }
 
  static int vhost_virtqueue_set_vring_endian_legacy(struct vhost_dev *dev,
#13 lore raw

Re: [PATCH 08/16] hw/virtio: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Hi Philippe Mathieu-Daudé

According to my test result this change introduced a minor bug, qemu output error messages "qemu-system-x86_64: virtio-blk missing headers" reproduced command:
/usr/local/bin/qemu-system-x86_64 -M q35 -m 8G -smp 8 -cpu host -enable-kvm -device VGA,bus=pcie.0,addr=0x2 -drive file=///home/fedora-40.qcow2,media=disk,if=virtio -device virtio-net-pci,mac=00:11:22:33:44:00,netdev=netdev1,id=net1,bus=pcie.0,addr=0x7
-netdev tap,id=netdev1,script=/etc/qemu-ifup,downscript=/etc/qemu-ifdown
-vnc :0 -monitor stdio
QEMU 10.1.50 monitor - type 'help' for more information
(qemu) qemu-system-x86_64: virtio-blk missing headers

Thanks
Lei

Quoted 61 lines · Farhan Ali
Quoted 17 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
  include/hw/virtio/virtio-access.h | 6 +-----
  hw/virtio/vhost.c                 | 7 +++----
  2 files changed, 4 insertions(+), 9 deletions(-)
include/hw/virtio/virtio-access.h +0 -0
index 07aae69042a..80328912ad3 100644

include/hw/virtio/virtio-access.h static inline uint64_t virtio_ldq_p(VirtIODevice *vdev, const void *ptr) · line 152

Quoted 9 lines
include/hw/virtio/virtio-access.h +1 -5
@@ -149,11 +149,7 @@ static inline uint64_t virtio_ldq_p(VirtIODevice *vdev, const void *ptr)
 
  static inline uint16_t virtio_tswap16(VirtIODevice *vdev, uint16_t s)
  {
-#if HOST_BIG_ENDIAN
-    return virtio_access_is_big_endian(vdev) ? s : bswap16(s);
-#else
-    return virtio_access_is_big_endian(vdev) ? bswap16(s) : s;
-#endif
+    return HOST_BIG_ENDIAN ^ virtio_access_is_big_endian(vdev) ? s : bswap16(s);

This patch breaks virtio devices(at least input/net devices) on s390x. I am not sure if ^ is the right check here? Changing the logic back to how it was fixes it for me.

Thanks

Farhan

Quoted 22 lines
include/hw/virtio/virtio-access.h +0 -0
  }
 
  static inline uint16_t virtio_lduw_phys_cached(VirtIODevice *vdev,
hw/virtio/vhost.c +3 -4
index 266a11514a1..6343477b42f 100644
@@ -1168,11 +1168,10 @@ static inline bool vhost_needs_vring_endian(VirtIODevice *vdev)
      if (virtio_vdev_has_feature(vdev, VIRTIO_F_VERSION_1)) {
          return false;
      }
-#if HOST_BIG_ENDIAN
-    return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_LITTLE;
-#else
+    if (HOST_BIG_ENDIAN) {
+        return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_LITTLE;
+    }
      return vdev->device_endian == VIRTIO_DEVICE_ENDIAN_BIG;
-#endif
  }
 
  static int vhost_virtqueue_set_vring_endian_legacy(struct vhost_dev *dev,
#14 lore raw

PATCH 9/16 target/alpha: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 target/alpha/translate.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
target/alpha/translate.c +6 -5
index a492520075e..b93cbe3b61f 100644
@@ -235,11 +235,12 @@ static TCGv dest_fpr(DisasContext *ctx, unsigned reg)
 static int get_flag_ofs(unsigned shift)
 {
     int ofs = offsetof(CPUAlphaState, flags);
-#if HOST_BIG_ENDIAN
-    ofs += 3 - (shift / 8);
-#else
-    ofs += shift / 8;
-#endif
+
+    if (HOST_BIG_ENDIAN) {
+        ofs += 3 - (shift / 8);
+    } else {
+        ofs += shift / 8;
+    }
     return ofs;
 }
 
Signature · 1 line
2.51.0
#15 lore raw

PATCH 10/16 target/arm: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 target/arm/cpu.h               |  8 ++++----
 target/arm/tcg/translate-a64.h |  5 ++---
 target/arm/tcg/sve_helper.c    | 10 ++--------
 target/arm/tcg/translate-sve.c | 22 +++++++++++-----------
 target/arm/tcg/translate-vfp.c |  8 +-------
 target/arm/tcg/translate.c     |  6 +++---
 6 files changed, 23 insertions(+), 36 deletions(-)
target/arm/cpu.h +4 -4
index 41414ac22b8..3f0a578a72a 100644
@@ -1249,17 +1249,17 @@ void aarch64_set_svcr(CPUARMState *env, uint64_t new, uint64_t mask);
  */
 static inline uint64_t *sve_bswap64(uint64_t *dst, uint64_t *src, int nr)
 {
-#if HOST_BIG_ENDIAN
     int i;
 
+    if (!HOST_BIG_ENDIAN) {
+        return src;
+    }
+
     for (i = 0; i < nr; ++i) {
         dst[i] = bswap64(src[i]);
     }
 
     return dst;
-#else
-    return src;
-#endif
 }
 
 void aarch64_sync_32_to_64(CPUARMState *env);
target/arm/tcg/translate-a64.h +2 -3
index 9c45f89305b..ce8303286ef 100644
@@ -82,7 +82,7 @@ static inline int vec_reg_offset(DisasContext *s, int regno,
 {
     int element_size = 1 << size;
     int offs = element * element_size;
-#if HOST_BIG_ENDIAN
+
     /* This is complicated slightly because vfp.zregs[n].d[0] is
      * still the lowest and vfp.zregs[n].d[15] the highest of the
      * 256 byte vector, even on big endian systems.
@@ -97,10 +97,9 @@ static inline int vec_reg_offset(DisasContext *s, int regno,
      * operations will have to special case loading and storing from
      * the zregs array.
      */
-    if (element_size < 8) {
+    if (HOST_BIG_ENDIAN && element_size < 8) {
         offs ^= 8 - element_size;
     }
-#endif
     offs += offsetof(CPUARMState, vfp.zregs[regno]);
     assert_fp_access_checked(s);
     return offs;
target/arm/tcg/sve_helper.c +2 -8
index c442fcb540d..dd829666cc2 100644
@@ -2861,12 +2861,9 @@ static void swap_memmove(void *vd, void *vs, size_t n)
 {
     uintptr_t d = (uintptr_t)vd;
     uintptr_t s = (uintptr_t)vs;
-    uintptr_t o = (d | s | n) & 7;
+    uintptr_t o = HOST_BIG_ENDIAN ? (d | s | n) & 7 : 0;
     size_t i;
 
-#if !HOST_BIG_ENDIAN
-    o = 0;
-#endif
     switch (o) {
     case 0:
         memmove(vd, vs, n);
@@ -2918,7 +2915,7 @@ static void swap_memmove(void *vd, void *vs, size_t n)
 static void swap_memzero(void *vd, size_t n)
 {
     uintptr_t d = (uintptr_t)vd;
-    uintptr_t o = (d | n) & 7;
+    uintptr_t o = HOST_BIG_ENDIAN ? (d | n) & 7 : 0;
     size_t i;
 
     /* Usually, the first bit of a predicate is set, so N is 0.  */
@@ -2926,9 +2923,6 @@ static void swap_memzero(void *vd, size_t n)
         return;
     }
 
-#if !HOST_BIG_ENDIAN
-    o = 0;
-#endif
     switch (o) {
     case 0:
         memset(vd, 0, n);
target/arm/tcg/translate-sve.c +11 -11
index 07b827fa8e8..60a99fbe15e 100644
@@ -2770,12 +2770,12 @@ static TCGv_i64 load_last_active(DisasContext *s, TCGv_i32 last,
      * The final adjustment for the vector register base
      * is added via constant offset to the load.
      */
-#if HOST_BIG_ENDIAN
-    /* Adjust for element ordering.  See vec_reg_offset.  */
-    if (esz < 3) {
-        tcg_gen_xori_i32(last, last, 8 - (1 << esz));
+    if (HOST_BIG_ENDIAN) {
+        /* Adjust for element ordering.  See vec_reg_offset.  */
+        if (esz < 3) {
+            tcg_gen_xori_i32(last, last, 8 - (1 << esz));
+        }
     }
-#endif
     tcg_gen_ext_i32_ptr(p, last);
     tcg_gen_add_ptr(p, p, tcg_env);
 
@@ -5394,9 +5394,9 @@ static void do_ldrq(DisasContext *s, int zt, int pg, TCGv_i64 addr, int dtype)
          * for this load operation.
          */
         TCGv_i64 tmp = tcg_temp_new_i64();
-#if HOST_BIG_ENDIAN
-        poff += 6;
-#endif
+        if (HOST_BIG_ENDIAN) {
+            poff += 6;
+        }
         tcg_gen_ld16u_i64(tmp, tcg_env, poff);
 
         poff = offsetof(CPUARMState, vfp.preg_tmp);
@@ -5478,9 +5478,9 @@ static void do_ldro(DisasContext *s, int zt, int pg, TCGv_i64 addr, int dtype)
          * for this load operation.
          */
         TCGv_i64 tmp = tcg_temp_new_i64();
-#if HOST_BIG_ENDIAN
-        poff += 4;
-#endif
+        if (HOST_BIG_ENDIAN) {
+            poff += 4;
+        }
         tcg_gen_ld32u_i64(tmp, tcg_env, poff);
 
         poff = offsetof(CPUARMState, vfp.preg_tmp);
target/arm/tcg/translate-vfp.c +1 -7
index 8d9d1ab877a..cf3ca4b33c4 100644
@@ -95,15 +95,9 @@ uint64_t vfp_expand_imm(int size, uint8_t imm8)
 static inline long vfp_f16_offset(unsigned reg, bool top)
 {
     long offs = vfp_reg_offset(false, reg);
-#if HOST_BIG_ENDIAN
-    if (!top) {
+    if (top ^ HOST_BIG_ENDIAN) {
         offs += 2;
     }
-#else
-    if (top) {
-        offs += 2;
-    }
-#endif
     return offs;
 }
 
target/arm/tcg/translate.c +3 -3
index 51654b0b91d..8ba0622489c 100644
@@ -1148,15 +1148,15 @@ long neon_element_offset(int reg, int element, MemOp memop)
 {
     int element_size = 1 << (memop & MO_SIZE);
     int ofs = element * element_size;
-#if HOST_BIG_ENDIAN
+
     /*
      * Calculate the offset assuming fully little-endian,
      * then XOR to account for the order of the 8-byte units.
      */
-    if (element_size < 8) {
+    if (HOST_BIG_ENDIAN && element_size < 8) {
         ofs ^= 8 - element_size;
     }
-#endif
+
     return neon_full_reg_offset(reg) + ofs;
 }
 
Signature · 1 line
2.51.0
#16 lore raw

PATCH 11/16 target/mips: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 target/mips/tcg/msa_helper.c | 889 +++++++++++++++++------------------
 1 file changed, 444 insertions(+), 445 deletions(-)
target/mips/tcg/msa_helper.c +444 -445
index f554b3d10ee..9f2d2bd6adc 100644
@@ -4148,41 +4148,41 @@ void helper_msa_ilvev_b(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->b[8]  = pws->b[9];
-    pwd->b[9]  = pwt->b[9];
-    pwd->b[10] = pws->b[11];
-    pwd->b[11] = pwt->b[11];
-    pwd->b[12] = pws->b[13];
-    pwd->b[13] = pwt->b[13];
-    pwd->b[14] = pws->b[15];
-    pwd->b[15] = pwt->b[15];
-    pwd->b[0]  = pws->b[1];
-    pwd->b[1]  = pwt->b[1];
-    pwd->b[2]  = pws->b[3];
-    pwd->b[3]  = pwt->b[3];
-    pwd->b[4]  = pws->b[5];
-    pwd->b[5]  = pwt->b[5];
-    pwd->b[6]  = pws->b[7];
-    pwd->b[7]  = pwt->b[7];
-#else
-    pwd->b[15] = pws->b[14];
-    pwd->b[14] = pwt->b[14];
-    pwd->b[13] = pws->b[12];
-    pwd->b[12] = pwt->b[12];
-    pwd->b[11] = pws->b[10];
-    pwd->b[10] = pwt->b[10];
-    pwd->b[9]  = pws->b[8];
-    pwd->b[8]  = pwt->b[8];
-    pwd->b[7]  = pws->b[6];
-    pwd->b[6]  = pwt->b[6];
-    pwd->b[5]  = pws->b[4];
-    pwd->b[4]  = pwt->b[4];
-    pwd->b[3]  = pws->b[2];
-    pwd->b[2]  = pwt->b[2];
-    pwd->b[1]  = pws->b[0];
-    pwd->b[0]  = pwt->b[0];
-#endif
+if (HOST_BIG_ENDIAN) {
+        pwd->b[8]  = pws->b[9];
+        pwd->b[9]  = pwt->b[9];
+        pwd->b[10] = pws->b[11];
+        pwd->b[11] = pwt->b[11];
+        pwd->b[12] = pws->b[13];
+        pwd->b[13] = pwt->b[13];
+        pwd->b[14] = pws->b[15];
+        pwd->b[15] = pwt->b[15];
+        pwd->b[0]  = pws->b[1];
+        pwd->b[1]  = pwt->b[1];
+        pwd->b[2]  = pws->b[3];
+        pwd->b[3]  = pwt->b[3];
+        pwd->b[4]  = pws->b[5];
+        pwd->b[5]  = pwt->b[5];
+        pwd->b[6]  = pws->b[7];
+        pwd->b[7]  = pwt->b[7];
+    } else {
+        pwd->b[15] = pws->b[14];
+        pwd->b[14] = pwt->b[14];
+        pwd->b[13] = pws->b[12];
+        pwd->b[12] = pwt->b[12];
+        pwd->b[11] = pws->b[10];
+        pwd->b[10] = pwt->b[10];
+        pwd->b[9]  = pws->b[8];
+        pwd->b[8]  = pwt->b[8];
+        pwd->b[7]  = pws->b[6];
+        pwd->b[6]  = pwt->b[6];
+        pwd->b[5]  = pws->b[4];
+        pwd->b[4]  = pwt->b[4];
+        pwd->b[3]  = pws->b[2];
+        pwd->b[2]  = pwt->b[2];
+        pwd->b[1]  = pws->b[0];
+        pwd->b[0]  = pwt->b[0];
+    }
 }
 
 void helper_msa_ilvev_h(CPUMIPSState *env,
@@ -4192,25 +4192,25 @@ void helper_msa_ilvev_h(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->h[4] = pws->h[5];
-    pwd->h[5] = pwt->h[5];
-    pwd->h[6] = pws->h[7];
-    pwd->h[7] = pwt->h[7];
-    pwd->h[0] = pws->h[1];
-    pwd->h[1] = pwt->h[1];
-    pwd->h[2] = pws->h[3];
-    pwd->h[3] = pwt->h[3];
-#else
-    pwd->h[7] = pws->h[6];
-    pwd->h[6] = pwt->h[6];
-    pwd->h[5] = pws->h[4];
-    pwd->h[4] = pwt->h[4];
-    pwd->h[3] = pws->h[2];
-    pwd->h[2] = pwt->h[2];
-    pwd->h[1] = pws->h[0];
-    pwd->h[0] = pwt->h[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->h[4] = pws->h[5];
+        pwd->h[5] = pwt->h[5];
+        pwd->h[6] = pws->h[7];
+        pwd->h[7] = pwt->h[7];
+        pwd->h[0] = pws->h[1];
+        pwd->h[1] = pwt->h[1];
+        pwd->h[2] = pws->h[3];
+        pwd->h[3] = pwt->h[3];
+    } else {
+        pwd->h[7] = pws->h[6];
+        pwd->h[6] = pwt->h[6];
+        pwd->h[5] = pws->h[4];
+        pwd->h[4] = pwt->h[4];
+        pwd->h[3] = pws->h[2];
+        pwd->h[2] = pwt->h[2];
+        pwd->h[1] = pws->h[0];
+        pwd->h[0] = pwt->h[0];
+    }
 }
 
 void helper_msa_ilvev_w(CPUMIPSState *env,
@@ -4220,17 +4220,17 @@ void helper_msa_ilvev_w(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->w[2] = pws->w[3];
-    pwd->w[3] = pwt->w[3];
-    pwd->w[0] = pws->w[1];
-    pwd->w[1] = pwt->w[1];
-#else
-    pwd->w[3] = pws->w[2];
-    pwd->w[2] = pwt->w[2];
-    pwd->w[1] = pws->w[0];
-    pwd->w[0] = pwt->w[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->w[2] = pws->w[3];
+        pwd->w[3] = pwt->w[3];
+        pwd->w[0] = pws->w[1];
+        pwd->w[1] = pwt->w[1];
+    } else {
+        pwd->w[3] = pws->w[2];
+        pwd->w[2] = pwt->w[2];
+        pwd->w[1] = pws->w[0];
+        pwd->w[0] = pwt->w[0];
+    }
 }
 
 void helper_msa_ilvev_d(CPUMIPSState *env,
@@ -4252,41 +4252,41 @@ void helper_msa_ilvod_b(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->b[7]  = pwt->b[6];
-    pwd->b[6]  = pws->b[6];
-    pwd->b[5]  = pwt->b[4];
-    pwd->b[4]  = pws->b[4];
-    pwd->b[3]  = pwt->b[2];
-    pwd->b[2]  = pws->b[2];
-    pwd->b[1]  = pwt->b[0];
-    pwd->b[0]  = pws->b[0];
-    pwd->b[15] = pwt->b[14];
-    pwd->b[14] = pws->b[14];
-    pwd->b[13] = pwt->b[12];
-    pwd->b[12] = pws->b[12];
-    pwd->b[11] = pwt->b[10];
-    pwd->b[10] = pws->b[10];
-    pwd->b[9]  = pwt->b[8];
-    pwd->b[8]  = pws->b[8];
-#else
-    pwd->b[0]  = pwt->b[1];
-    pwd->b[1]  = pws->b[1];
-    pwd->b[2]  = pwt->b[3];
-    pwd->b[3]  = pws->b[3];
-    pwd->b[4]  = pwt->b[5];
-    pwd->b[5]  = pws->b[5];
-    pwd->b[6]  = pwt->b[7];
-    pwd->b[7]  = pws->b[7];
-    pwd->b[8]  = pwt->b[9];
-    pwd->b[9]  = pws->b[9];
-    pwd->b[10] = pwt->b[11];
-    pwd->b[11] = pws->b[11];
-    pwd->b[12] = pwt->b[13];
-    pwd->b[13] = pws->b[13];
-    pwd->b[14] = pwt->b[15];
-    pwd->b[15] = pws->b[15];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->b[7]  = pwt->b[6];
+        pwd->b[6]  = pws->b[6];
+        pwd->b[5]  = pwt->b[4];
+        pwd->b[4]  = pws->b[4];
+        pwd->b[3]  = pwt->b[2];
+        pwd->b[2]  = pws->b[2];
+        pwd->b[1]  = pwt->b[0];
+        pwd->b[0]  = pws->b[0];
+        pwd->b[15] = pwt->b[14];
+        pwd->b[14] = pws->b[14];
+        pwd->b[13] = pwt->b[12];
+        pwd->b[12] = pws->b[12];
+        pwd->b[11] = pwt->b[10];
+        pwd->b[10] = pws->b[10];
+        pwd->b[9]  = pwt->b[8];
+        pwd->b[8]  = pws->b[8];
+    } else {
+        pwd->b[0]  = pwt->b[1];
+        pwd->b[1]  = pws->b[1];
+        pwd->b[2]  = pwt->b[3];
+        pwd->b[3]  = pws->b[3];
+        pwd->b[4]  = pwt->b[5];
+        pwd->b[5]  = pws->b[5];
+        pwd->b[6]  = pwt->b[7];
+        pwd->b[7]  = pws->b[7];
+        pwd->b[8]  = pwt->b[9];
+        pwd->b[9]  = pws->b[9];
+        pwd->b[10] = pwt->b[11];
+        pwd->b[11] = pws->b[11];
+        pwd->b[12] = pwt->b[13];
+        pwd->b[13] = pws->b[13];
+        pwd->b[14] = pwt->b[15];
+        pwd->b[15] = pws->b[15];
+    }
 }
 
 void helper_msa_ilvod_h(CPUMIPSState *env,
@@ -4296,25 +4296,25 @@ void helper_msa_ilvod_h(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->h[3] = pwt->h[2];
-    pwd->h[2] = pws->h[2];
-    pwd->h[1] = pwt->h[0];
-    pwd->h[0] = pws->h[0];
-    pwd->h[7] = pwt->h[6];
-    pwd->h[6] = pws->h[6];
-    pwd->h[5] = pwt->h[4];
-    pwd->h[4] = pws->h[4];
-#else
-    pwd->h[0] = pwt->h[1];
-    pwd->h[1] = pws->h[1];
-    pwd->h[2] = pwt->h[3];
-    pwd->h[3] = pws->h[3];
-    pwd->h[4] = pwt->h[5];
-    pwd->h[5] = pws->h[5];
-    pwd->h[6] = pwt->h[7];
-    pwd->h[7] = pws->h[7];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->h[3] = pwt->h[2];
+        pwd->h[2] = pws->h[2];
+        pwd->h[1] = pwt->h[0];
+        pwd->h[0] = pws->h[0];
+        pwd->h[7] = pwt->h[6];
+        pwd->h[6] = pws->h[6];
+        pwd->h[5] = pwt->h[4];
+        pwd->h[4] = pws->h[4];
+    } else {
+        pwd->h[0] = pwt->h[1];
+        pwd->h[1] = pws->h[1];
+        pwd->h[2] = pwt->h[3];
+        pwd->h[3] = pws->h[3];
+        pwd->h[4] = pwt->h[5];
+        pwd->h[5] = pws->h[5];
+        pwd->h[6] = pwt->h[7];
+        pwd->h[7] = pws->h[7];
+    }
 }
 
 void helper_msa_ilvod_w(CPUMIPSState *env,
@@ -4324,17 +4324,17 @@ void helper_msa_ilvod_w(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->w[1] = pwt->w[0];
-    pwd->w[0] = pws->w[0];
-    pwd->w[3] = pwt->w[2];
-    pwd->w[2] = pws->w[2];
-#else
-    pwd->w[0] = pwt->w[1];
-    pwd->w[1] = pws->w[1];
-    pwd->w[2] = pwt->w[3];
-    pwd->w[3] = pws->w[3];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->w[1] = pwt->w[0];
+        pwd->w[0] = pws->w[0];
+        pwd->w[3] = pwt->w[2];
+        pwd->w[2] = pws->w[2];
+    } else {
+        pwd->w[0] = pwt->w[1];
+        pwd->w[1] = pws->w[1];
+        pwd->w[2] = pwt->w[3];
+        pwd->w[3] = pws->w[3];
+    }
 }
 
 void helper_msa_ilvod_d(CPUMIPSState *env,
@@ -4356,41 +4356,41 @@ void helper_msa_ilvl_b(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->b[7]  = pwt->b[15];
-    pwd->b[6]  = pws->b[15];
-    pwd->b[5]  = pwt->b[14];
-    pwd->b[4]  = pws->b[14];
-    pwd->b[3]  = pwt->b[13];
-    pwd->b[2]  = pws->b[13];
-    pwd->b[1]  = pwt->b[12];
-    pwd->b[0]  = pws->b[12];
-    pwd->b[15] = pwt->b[11];
-    pwd->b[14] = pws->b[11];
-    pwd->b[13] = pwt->b[10];
-    pwd->b[12] = pws->b[10];
-    pwd->b[11] = pwt->b[9];
-    pwd->b[10] = pws->b[9];
-    pwd->b[9]  = pwt->b[8];
-    pwd->b[8]  = pws->b[8];
-#else
-    pwd->b[0]  = pwt->b[8];
-    pwd->b[1]  = pws->b[8];
-    pwd->b[2]  = pwt->b[9];
-    pwd->b[3]  = pws->b[9];
-    pwd->b[4]  = pwt->b[10];
-    pwd->b[5]  = pws->b[10];
-    pwd->b[6]  = pwt->b[11];
-    pwd->b[7]  = pws->b[11];
-    pwd->b[8]  = pwt->b[12];
-    pwd->b[9]  = pws->b[12];
-    pwd->b[10] = pwt->b[13];
-    pwd->b[11] = pws->b[13];
-    pwd->b[12] = pwt->b[14];
-    pwd->b[13] = pws->b[14];
-    pwd->b[14] = pwt->b[15];
-    pwd->b[15] = pws->b[15];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->b[7]  = pwt->b[15];
+        pwd->b[6]  = pws->b[15];
+        pwd->b[5]  = pwt->b[14];
+        pwd->b[4]  = pws->b[14];
+        pwd->b[3]  = pwt->b[13];
+        pwd->b[2]  = pws->b[13];
+        pwd->b[1]  = pwt->b[12];
+        pwd->b[0]  = pws->b[12];
+        pwd->b[15] = pwt->b[11];
+        pwd->b[14] = pws->b[11];
+        pwd->b[13] = pwt->b[10];
+        pwd->b[12] = pws->b[10];
+        pwd->b[11] = pwt->b[9];
+        pwd->b[10] = pws->b[9];
+        pwd->b[9]  = pwt->b[8];
+        pwd->b[8]  = pws->b[8];
+    } else {
+        pwd->b[0]  = pwt->b[8];
+        pwd->b[1]  = pws->b[8];
+        pwd->b[2]  = pwt->b[9];
+        pwd->b[3]  = pws->b[9];
+        pwd->b[4]  = pwt->b[10];
+        pwd->b[5]  = pws->b[10];
+        pwd->b[6]  = pwt->b[11];
+        pwd->b[7]  = pws->b[11];
+        pwd->b[8]  = pwt->b[12];
+        pwd->b[9]  = pws->b[12];
+        pwd->b[10] = pwt->b[13];
+        pwd->b[11] = pws->b[13];
+        pwd->b[12] = pwt->b[14];
+        pwd->b[13] = pws->b[14];
+        pwd->b[14] = pwt->b[15];
+        pwd->b[15] = pws->b[15];
+    }
 }
 
 void helper_msa_ilvl_h(CPUMIPSState *env,
@@ -4400,25 +4400,25 @@ void helper_msa_ilvl_h(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->h[3] = pwt->h[7];
-    pwd->h[2] = pws->h[7];
-    pwd->h[1] = pwt->h[6];
-    pwd->h[0] = pws->h[6];
-    pwd->h[7] = pwt->h[5];
-    pwd->h[6] = pws->h[5];
-    pwd->h[5] = pwt->h[4];
-    pwd->h[4] = pws->h[4];
-#else
-    pwd->h[0] = pwt->h[4];
-    pwd->h[1] = pws->h[4];
-    pwd->h[2] = pwt->h[5];
-    pwd->h[3] = pws->h[5];
-    pwd->h[4] = pwt->h[6];
-    pwd->h[5] = pws->h[6];
-    pwd->h[6] = pwt->h[7];
-    pwd->h[7] = pws->h[7];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->h[3] = pwt->h[7];
+        pwd->h[2] = pws->h[7];
+        pwd->h[1] = pwt->h[6];
+        pwd->h[0] = pws->h[6];
+        pwd->h[7] = pwt->h[5];
+        pwd->h[6] = pws->h[5];
+        pwd->h[5] = pwt->h[4];
+        pwd->h[4] = pws->h[4];
+    } else {
+        pwd->h[0] = pwt->h[4];
+        pwd->h[1] = pws->h[4];
+        pwd->h[2] = pwt->h[5];
+        pwd->h[3] = pws->h[5];
+        pwd->h[4] = pwt->h[6];
+        pwd->h[5] = pws->h[6];
+        pwd->h[6] = pwt->h[7];
+        pwd->h[7] = pws->h[7];
+    }
 }
 
 void helper_msa_ilvl_w(CPUMIPSState *env,
@@ -4428,17 +4428,17 @@ void helper_msa_ilvl_w(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->w[1] = pwt->w[3];
-    pwd->w[0] = pws->w[3];
-    pwd->w[3] = pwt->w[2];
-    pwd->w[2] = pws->w[2];
-#else
-    pwd->w[0] = pwt->w[2];
-    pwd->w[1] = pws->w[2];
-    pwd->w[2] = pwt->w[3];
-    pwd->w[3] = pws->w[3];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->w[1] = pwt->w[3];
+        pwd->w[0] = pws->w[3];
+        pwd->w[3] = pwt->w[2];
+        pwd->w[2] = pws->w[2];
+    } else {
+        pwd->w[0] = pwt->w[2];
+        pwd->w[1] = pws->w[2];
+        pwd->w[2] = pwt->w[3];
+        pwd->w[3] = pws->w[3];
+    }
 }
 
 void helper_msa_ilvl_d(CPUMIPSState *env,
@@ -4460,41 +4460,41 @@ void helper_msa_ilvr_b(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->b[8]  = pws->b[0];
-    pwd->b[9]  = pwt->b[0];
-    pwd->b[10] = pws->b[1];
-    pwd->b[11] = pwt->b[1];
-    pwd->b[12] = pws->b[2];
-    pwd->b[13] = pwt->b[2];
-    pwd->b[14] = pws->b[3];
-    pwd->b[15] = pwt->b[3];
-    pwd->b[0]  = pws->b[4];
-    pwd->b[1]  = pwt->b[4];
-    pwd->b[2]  = pws->b[5];
-    pwd->b[3]  = pwt->b[5];
-    pwd->b[4]  = pws->b[6];
-    pwd->b[5]  = pwt->b[6];
-    pwd->b[6]  = pws->b[7];
-    pwd->b[7]  = pwt->b[7];
-#else
-    pwd->b[15] = pws->b[7];
-    pwd->b[14] = pwt->b[7];
-    pwd->b[13] = pws->b[6];
-    pwd->b[12] = pwt->b[6];
-    pwd->b[11] = pws->b[5];
-    pwd->b[10] = pwt->b[5];
-    pwd->b[9]  = pws->b[4];
-    pwd->b[8]  = pwt->b[4];
-    pwd->b[7]  = pws->b[3];
-    pwd->b[6]  = pwt->b[3];
-    pwd->b[5]  = pws->b[2];
-    pwd->b[4]  = pwt->b[2];
-    pwd->b[3]  = pws->b[1];
-    pwd->b[2]  = pwt->b[1];
-    pwd->b[1]  = pws->b[0];
-    pwd->b[0]  = pwt->b[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->b[8]  = pws->b[0];
+        pwd->b[9]  = pwt->b[0];
+        pwd->b[10] = pws->b[1];
+        pwd->b[11] = pwt->b[1];
+        pwd->b[12] = pws->b[2];
+        pwd->b[13] = pwt->b[2];
+        pwd->b[14] = pws->b[3];
+        pwd->b[15] = pwt->b[3];
+        pwd->b[0]  = pws->b[4];
+        pwd->b[1]  = pwt->b[4];
+        pwd->b[2]  = pws->b[5];
+        pwd->b[3]  = pwt->b[5];
+        pwd->b[4]  = pws->b[6];
+        pwd->b[5]  = pwt->b[6];
+        pwd->b[6]  = pws->b[7];
+        pwd->b[7]  = pwt->b[7];
+    } else {
+        pwd->b[15] = pws->b[7];
+        pwd->b[14] = pwt->b[7];
+        pwd->b[13] = pws->b[6];
+        pwd->b[12] = pwt->b[6];
+        pwd->b[11] = pws->b[5];
+        pwd->b[10] = pwt->b[5];
+        pwd->b[9]  = pws->b[4];
+        pwd->b[8]  = pwt->b[4];
+        pwd->b[7]  = pws->b[3];
+        pwd->b[6]  = pwt->b[3];
+        pwd->b[5]  = pws->b[2];
+        pwd->b[4]  = pwt->b[2];
+        pwd->b[3]  = pws->b[1];
+        pwd->b[2]  = pwt->b[1];
+        pwd->b[1]  = pws->b[0];
+        pwd->b[0]  = pwt->b[0];
+    }
 }
 
 void helper_msa_ilvr_h(CPUMIPSState *env,
@@ -4504,25 +4504,25 @@ void helper_msa_ilvr_h(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->h[4] = pws->h[0];
-    pwd->h[5] = pwt->h[0];
-    pwd->h[6] = pws->h[1];
-    pwd->h[7] = pwt->h[1];
-    pwd->h[0] = pws->h[2];
-    pwd->h[1] = pwt->h[2];
-    pwd->h[2] = pws->h[3];
-    pwd->h[3] = pwt->h[3];
-#else
-    pwd->h[7] = pws->h[3];
-    pwd->h[6] = pwt->h[3];
-    pwd->h[5] = pws->h[2];
-    pwd->h[4] = pwt->h[2];
-    pwd->h[3] = pws->h[1];
-    pwd->h[2] = pwt->h[1];
-    pwd->h[1] = pws->h[0];
-    pwd->h[0] = pwt->h[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->h[4] = pws->h[0];
+        pwd->h[5] = pwt->h[0];
+        pwd->h[6] = pws->h[1];
+        pwd->h[7] = pwt->h[1];
+        pwd->h[0] = pws->h[2];
+        pwd->h[1] = pwt->h[2];
+        pwd->h[2] = pws->h[3];
+        pwd->h[3] = pwt->h[3];
+    } else {
+        pwd->h[7] = pws->h[3];
+        pwd->h[6] = pwt->h[3];
+        pwd->h[5] = pws->h[2];
+        pwd->h[4] = pwt->h[2];
+        pwd->h[3] = pws->h[1];
+        pwd->h[2] = pwt->h[1];
+        pwd->h[1] = pws->h[0];
+        pwd->h[0] = pwt->h[0];
+    }
 }
 
 void helper_msa_ilvr_w(CPUMIPSState *env,
@@ -4532,17 +4532,17 @@ void helper_msa_ilvr_w(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->w[2] = pws->w[0];
-    pwd->w[3] = pwt->w[0];
-    pwd->w[0] = pws->w[1];
-    pwd->w[1] = pwt->w[1];
-#else
-    pwd->w[3] = pws->w[1];
-    pwd->w[2] = pwt->w[1];
-    pwd->w[1] = pws->w[0];
-    pwd->w[0] = pwt->w[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->w[2] = pws->w[0];
+        pwd->w[3] = pwt->w[0];
+        pwd->w[0] = pws->w[1];
+        pwd->w[1] = pwt->w[1];
+    } else {
+        pwd->w[3] = pws->w[1];
+        pwd->w[2] = pwt->w[1];
+        pwd->w[1] = pws->w[0];
+        pwd->w[0] = pwt->w[0];
+    }
 }
 
 void helper_msa_ilvr_d(CPUMIPSState *env,
@@ -4663,41 +4663,41 @@ void helper_msa_pckev_b(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->b[8]  = pws->b[9];
-    pwd->b[10] = pws->b[13];
-    pwd->b[12] = pws->b[1];
-    pwd->b[14] = pws->b[5];
-    pwd->b[0]  = pwt->b[9];
-    pwd->b[2]  = pwt->b[13];
-    pwd->b[4]  = pwt->b[1];
-    pwd->b[6]  = pwt->b[5];
-    pwd->b[9]  = pws->b[11];
-    pwd->b[13] = pws->b[3];
-    pwd->b[1]  = pwt->b[11];
-    pwd->b[5]  = pwt->b[3];
-    pwd->b[11] = pws->b[15];
-    pwd->b[3]  = pwt->b[15];
-    pwd->b[15] = pws->b[7];
-    pwd->b[7]  = pwt->b[7];
-#else
-    pwd->b[15] = pws->b[14];
-    pwd->b[13] = pws->b[10];
-    pwd->b[11] = pws->b[6];
-    pwd->b[9]  = pws->b[2];
-    pwd->b[7]  = pwt->b[14];
-    pwd->b[5]  = pwt->b[10];
-    pwd->b[3]  = pwt->b[6];
-    pwd->b[1]  = pwt->b[2];
-    pwd->b[14] = pws->b[12];
-    pwd->b[10] = pws->b[4];
-    pwd->b[6]  = pwt->b[12];
-    pwd->b[2]  = pwt->b[4];
-    pwd->b[12] = pws->b[8];
-    pwd->b[4]  = pwt->b[8];
-    pwd->b[8]  = pws->b[0];
-    pwd->b[0]  = pwt->b[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->b[8]  = pws->b[9];
+        pwd->b[10] = pws->b[13];
+        pwd->b[12] = pws->b[1];
+        pwd->b[14] = pws->b[5];
+        pwd->b[0]  = pwt->b[9];
+        pwd->b[2]  = pwt->b[13];
+        pwd->b[4]  = pwt->b[1];
+        pwd->b[6]  = pwt->b[5];
+        pwd->b[9]  = pws->b[11];
+        pwd->b[13] = pws->b[3];
+        pwd->b[1]  = pwt->b[11];
+        pwd->b[5]  = pwt->b[3];
+        pwd->b[11] = pws->b[15];
+        pwd->b[3]  = pwt->b[15];
+        pwd->b[15] = pws->b[7];
+        pwd->b[7]  = pwt->b[7];
+    } else {
+        pwd->b[15] = pws->b[14];
+        pwd->b[13] = pws->b[10];
+        pwd->b[11] = pws->b[6];
+        pwd->b[9]  = pws->b[2];
+        pwd->b[7]  = pwt->b[14];
+        pwd->b[5]  = pwt->b[10];
+        pwd->b[3]  = pwt->b[6];
+        pwd->b[1]  = pwt->b[2];
+        pwd->b[14] = pws->b[12];
+        pwd->b[10] = pws->b[4];
+        pwd->b[6]  = pwt->b[12];
+        pwd->b[2]  = pwt->b[4];
+        pwd->b[12] = pws->b[8];
+        pwd->b[4]  = pwt->b[8];
+        pwd->b[8]  = pws->b[0];
+        pwd->b[0]  = pwt->b[0];
+    }
 }
 
 void helper_msa_pckev_h(CPUMIPSState *env,
@@ -4707,25 +4707,25 @@ void helper_msa_pckev_h(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->h[4] = pws->h[5];
-    pwd->h[6] = pws->h[1];
-    pwd->h[0] = pwt->h[5];
-    pwd->h[2] = pwt->h[1];
-    pwd->h[5] = pws->h[7];
-    pwd->h[1] = pwt->h[7];
-    pwd->h[7] = pws->h[3];
-    pwd->h[3] = pwt->h[3];
-#else
-    pwd->h[7] = pws->h[6];
-    pwd->h[5] = pws->h[2];
-    pwd->h[3] = pwt->h[6];
-    pwd->h[1] = pwt->h[2];
-    pwd->h[6] = pws->h[4];
-    pwd->h[2] = pwt->h[4];
-    pwd->h[4] = pws->h[0];
-    pwd->h[0] = pwt->h[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->h[4] = pws->h[5];
+        pwd->h[6] = pws->h[1];
+        pwd->h[0] = pwt->h[5];
+        pwd->h[2] = pwt->h[1];
+        pwd->h[5] = pws->h[7];
+        pwd->h[1] = pwt->h[7];
+        pwd->h[7] = pws->h[3];
+        pwd->h[3] = pwt->h[3];
+    } else {
+        pwd->h[7] = pws->h[6];
+        pwd->h[5] = pws->h[2];
+        pwd->h[3] = pwt->h[6];
+        pwd->h[1] = pwt->h[2];
+        pwd->h[6] = pws->h[4];
+        pwd->h[2] = pwt->h[4];
+        pwd->h[4] = pws->h[0];
+        pwd->h[0] = pwt->h[0];
+    }
 }
 
 void helper_msa_pckev_w(CPUMIPSState *env,
@@ -4735,17 +4735,17 @@ void helper_msa_pckev_w(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->w[2] = pws->w[3];
-    pwd->w[0] = pwt->w[3];
-    pwd->w[3] = pws->w[1];
-    pwd->w[1] = pwt->w[1];
-#else
-    pwd->w[3] = pws->w[2];
-    pwd->w[1] = pwt->w[2];
-    pwd->w[2] = pws->w[0];
-    pwd->w[0] = pwt->w[0];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->w[2] = pws->w[3];
+        pwd->w[0] = pwt->w[3];
+        pwd->w[3] = pws->w[1];
+        pwd->w[1] = pwt->w[1];
+    } else {
+        pwd->w[3] = pws->w[2];
+        pwd->w[1] = pwt->w[2];
+        pwd->w[2] = pws->w[0];
+        pwd->w[0] = pwt->w[0];
+    }
 }
 
 void helper_msa_pckev_d(CPUMIPSState *env,
@@ -4767,42 +4767,41 @@ void helper_msa_pckod_b(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->b[7]  = pwt->b[6];
-    pwd->b[5]  = pwt->b[2];
-    pwd->b[3]  = pwt->b[14];
-    pwd->b[1]  = pwt->b[10];
-    pwd->b[15] = pws->b[6];
-    pwd->b[13] = pws->b[2];
-    pwd->b[11] = pws->b[14];
-    pwd->b[9]  = pws->b[10];
-    pwd->b[6]  = pwt->b[4];
-    pwd->b[2]  = pwt->b[12];
-    pwd->b[14] = pws->b[4];
-    pwd->b[10] = pws->b[12];
-    pwd->b[4]  = pwt->b[0];
-    pwd->b[12] = pws->b[0];
-    pwd->b[0]  = pwt->b[8];
-    pwd->b[8]  = pws->b[8];
-#else
-    pwd->b[0]  = pwt->b[1];
-    pwd->b[2]  = pwt->b[5];
-    pwd->b[4]  = pwt->b[9];
-    pwd->b[6]  = pwt->b[13];
-    pwd->b[8]  = pws->b[1];
-    pwd->b[10] = pws->b[5];
-    pwd->b[12] = pws->b[9];
-    pwd->b[14] = pws->b[13];
-    pwd->b[1]  = pwt->b[3];
-    pwd->b[5]  = pwt->b[11];
-    pwd->b[9]  = pws->b[3];
-    pwd->b[13] = pws->b[11];
-    pwd->b[3]  = pwt->b[7];
-    pwd->b[11] = pws->b[7];
-    pwd->b[7]  = pwt->b[15];
-    pwd->b[15] = pws->b[15];
-#endif
-
+    if (HOST_BIG_ENDIAN) {
+        pwd->b[7]  = pwt->b[6];
+        pwd->b[5]  = pwt->b[2];
+        pwd->b[3]  = pwt->b[14];
+        pwd->b[1]  = pwt->b[10];
+        pwd->b[15] = pws->b[6];
+        pwd->b[13] = pws->b[2];
+        pwd->b[11] = pws->b[14];
+        pwd->b[9]  = pws->b[10];
+        pwd->b[6]  = pwt->b[4];
+        pwd->b[2]  = pwt->b[12];
+        pwd->b[14] = pws->b[4];
+        pwd->b[10] = pws->b[12];
+        pwd->b[4]  = pwt->b[0];
+        pwd->b[12] = pws->b[0];
+        pwd->b[0]  = pwt->b[8];
+        pwd->b[8]  = pws->b[8];
+    } else {
+        pwd->b[0]  = pwt->b[1];
+        pwd->b[2]  = pwt->b[5];
+        pwd->b[4]  = pwt->b[9];
+        pwd->b[6]  = pwt->b[13];
+        pwd->b[8]  = pws->b[1];
+        pwd->b[10] = pws->b[5];
+        pwd->b[12] = pws->b[9];
+        pwd->b[14] = pws->b[13];
+        pwd->b[1]  = pwt->b[3];
+        pwd->b[5]  = pwt->b[11];
+        pwd->b[9]  = pws->b[3];
+        pwd->b[13] = pws->b[11];
+        pwd->b[3]  = pwt->b[7];
+        pwd->b[11] = pws->b[7];
+        pwd->b[7]  = pwt->b[15];
+        pwd->b[15] = pws->b[15];
+    }
 }
 
 void helper_msa_pckod_h(CPUMIPSState *env,
@@ -4812,25 +4811,25 @@ void helper_msa_pckod_h(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->h[3] = pwt->h[2];
-    pwd->h[1] = pwt->h[6];
-    pwd->h[7] = pws->h[2];
-    pwd->h[5] = pws->h[6];
-    pwd->h[2] = pwt->h[0];
-    pwd->h[6] = pws->h[0];
-    pwd->h[0] = pwt->h[4];
-    pwd->h[4] = pws->h[4];
-#else
-    pwd->h[0] = pwt->h[1];
-    pwd->h[2] = pwt->h[5];
-    pwd->h[4] = pws->h[1];
-    pwd->h[6] = pws->h[5];
-    pwd->h[1] = pwt->h[3];
-    pwd->h[5] = pws->h[3];
-    pwd->h[3] = pwt->h[7];
-    pwd->h[7] = pws->h[7];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->h[3] = pwt->h[2];
+        pwd->h[1] = pwt->h[6];
+        pwd->h[7] = pws->h[2];
+        pwd->h[5] = pws->h[6];
+        pwd->h[2] = pwt->h[0];
+        pwd->h[6] = pws->h[0];
+        pwd->h[0] = pwt->h[4];
+        pwd->h[4] = pws->h[4];
+    } else {
+        pwd->h[0] = pwt->h[1];
+        pwd->h[2] = pwt->h[5];
+        pwd->h[4] = pws->h[1];
+        pwd->h[6] = pws->h[5];
+        pwd->h[1] = pwt->h[3];
+        pwd->h[5] = pws->h[3];
+        pwd->h[3] = pwt->h[7];
+        pwd->h[7] = pws->h[7];
+    }
 }
 
 void helper_msa_pckod_w(CPUMIPSState *env,
@@ -4840,17 +4839,17 @@ void helper_msa_pckod_w(CPUMIPSState *env,
     wr_t *pws = &(env->active_fpu.fpr[ws].wr);
     wr_t *pwt = &(env->active_fpu.fpr[wt].wr);
 
-#if HOST_BIG_ENDIAN
-    pwd->w[1] = pwt->w[0];
-    pwd->w[3] = pws->w[0];
-    pwd->w[0] = pwt->w[2];
-    pwd->w[2] = pws->w[2];
-#else
-    pwd->w[0] = pwt->w[1];
-    pwd->w[2] = pws->w[1];
-    pwd->w[1] = pwt->w[3];
-    pwd->w[3] = pws->w[3];
-#endif
+    if (HOST_BIG_ENDIAN) {
+        pwd->w[1] = pwt->w[0];
+        pwd->w[3] = pws->w[0];
+        pwd->w[0] = pwt->w[2];
+        pwd->w[2] = pws->w[2];
+    } else {
+        pwd->w[0] = pwt->w[1];
+        pwd->w[2] = pws->w[1];
+        pwd->w[1] = pwt->w[3];
+        pwd->w[3] = pws->w[3];
+    }
 }
 
 void helper_msa_pckod_d(CPUMIPSState *env,
@@ -5928,13 +5927,13 @@ void helper_msa_copy_s_b(CPUMIPSState *env, uint32_t rd,
                          uint32_t ws, uint32_t n)
 {
     n %= 16;
-#if HOST_BIG_ENDIAN
-    if (n < 8) {
-        n = 8 - n - 1;
-    } else {
-        n = 24 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 8) {
+            n = 8 - n - 1;
+        } else {
+            n = 24 - n - 1;
+        }
     }
-#endif
     env->active_tc.gpr[rd] = (int8_t)env->active_fpu.fpr[ws].wr.b[n];
 }
 
@@ -5942,13 +5941,13 @@ void helper_msa_copy_s_h(CPUMIPSState *env, uint32_t rd,
                          uint32_t ws, uint32_t n)
 {
     n %= 8;
-#if HOST_BIG_ENDIAN
-    if (n < 4) {
-        n = 4 - n - 1;
-    } else {
-        n = 12 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 4) {
+            n = 4 - n - 1;
+        } else {
+            n = 12 - n - 1;
+        }
     }
-#endif
     env->active_tc.gpr[rd] = (int16_t)env->active_fpu.fpr[ws].wr.h[n];
 }
 
@@ -5956,13 +5955,13 @@ void helper_msa_copy_s_w(CPUMIPSState *env, uint32_t rd,
                          uint32_t ws, uint32_t n)
 {
     n %= 4;
-#if HOST_BIG_ENDIAN
-    if (n < 2) {
-        n = 2 - n - 1;
-    } else {
-        n = 6 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 2) {
+            n = 2 - n - 1;
+        } else {
+            n = 6 - n - 1;
+        }
     }
-#endif
     env->active_tc.gpr[rd] = (int32_t)env->active_fpu.fpr[ws].wr.w[n];
 }
 
@@ -5977,13 +5976,13 @@ void helper_msa_copy_u_b(CPUMIPSState *env, uint32_t rd,
                          uint32_t ws, uint32_t n)
 {
     n %= 16;
-#if HOST_BIG_ENDIAN
-    if (n < 8) {
-        n = 8 - n - 1;
-    } else {
-        n = 24 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 8) {
+            n = 8 - n - 1;
+        } else {
+            n = 24 - n - 1;
+        }
     }
-#endif
     env->active_tc.gpr[rd] = (uint8_t)env->active_fpu.fpr[ws].wr.b[n];
 }
 
@@ -5991,13 +5990,13 @@ void helper_msa_copy_u_h(CPUMIPSState *env, uint32_t rd,
                          uint32_t ws, uint32_t n)
 {
     n %= 8;
-#if HOST_BIG_ENDIAN
-    if (n < 4) {
-        n = 4 - n - 1;
-    } else {
-        n = 12 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 4) {
+            n = 4 - n - 1;
+        } else {
+            n = 12 - n - 1;
+        }
     }
-#endif
     env->active_tc.gpr[rd] = (uint16_t)env->active_fpu.fpr[ws].wr.h[n];
 }
 
@@ -6005,13 +6004,13 @@ void helper_msa_copy_u_w(CPUMIPSState *env, uint32_t rd,
                          uint32_t ws, uint32_t n)
 {
     n %= 4;
-#if HOST_BIG_ENDIAN
-    if (n < 2) {
-        n = 2 - n - 1;
-    } else {
-        n = 6 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 2) {
+            n = 2 - n - 1;
+        } else {
+            n = 6 - n - 1;
+        }
     }
-#endif
     env->active_tc.gpr[rd] = (uint32_t)env->active_fpu.fpr[ws].wr.w[n];
 }
 
@@ -6021,13 +6020,13 @@ void helper_msa_insert_b(CPUMIPSState *env, uint32_t wd,
     wr_t *pwd = &(env->active_fpu.fpr[wd].wr);
     target_ulong rs = env->active_tc.gpr[rs_num];
     n %= 16;
-#if HOST_BIG_ENDIAN
-    if (n < 8) {
-        n = 8 - n - 1;
-    } else {
-        n = 24 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 8) {
+            n = 8 - n - 1;
+        } else {
+            n = 24 - n - 1;
+        }
     }
-#endif
     pwd->b[n] = (int8_t)rs;
 }
 
@@ -6037,13 +6036,13 @@ void helper_msa_insert_h(CPUMIPSState *env, uint32_t wd,
     wr_t *pwd = &(env->active_fpu.fpr[wd].wr);
     target_ulong rs = env->active_tc.gpr[rs_num];
     n %= 8;
-#if HOST_BIG_ENDIAN
-    if (n < 4) {
-        n = 4 - n - 1;
-    } else {
-        n = 12 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 4) {
+            n = 4 - n - 1;
+        } else {
+            n = 12 - n - 1;
+        }
     }
-#endif
     pwd->h[n] = (int16_t)rs;
 }
 
@@ -6053,13 +6052,13 @@ void helper_msa_insert_w(CPUMIPSState *env, uint32_t wd,
     wr_t *pwd = &(env->active_fpu.fpr[wd].wr);
     target_ulong rs = env->active_tc.gpr[rs_num];
     n %= 4;
-#if HOST_BIG_ENDIAN
-    if (n < 2) {
-        n = 2 - n - 1;
-    } else {
-        n = 6 - n - 1;
+    if (HOST_BIG_ENDIAN) {
+        if (n < 2) {
+            n = 2 - n - 1;
+        } else {
+            n = 6 - n - 1;
+        }
     }
-#endif
     pwd->w[n] = (int32_t)rs;
 }
 
Signature · 1 line
2.51.0
#17 lore raw

PATCH 12/16 target/ppc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 target/ppc/arch_dump.c              |  9 ++-------
 target/ppc/int_helper.c             | 28 ++++++++++++++--------------
 target/ppc/kvm.c                    | 25 +++++++++----------------
 target/ppc/translate/vmx-impl.c.inc | 14 +++++++-------
 target/ppc/translate/vsx-impl.c.inc |  6 +++---
 tcg/ppc/tcg-target.c.inc            | 24 ++++++++++++------------
 6 files changed, 47 insertions(+), 59 deletions(-)
target/ppc/arch_dump.c +2 -7
index 80ac6c3e320..5cb8dbe9a6a 100644
@@ -158,21 +158,16 @@ static void ppc_write_elf_vmxregset(NoteFuncArg *arg, PowerPCCPU *cpu, int id)
     struct PPCElfVmxregset *vmxregset;
     Note *note = &arg->note;
     DumpState *s = arg->state;
+    const int host_data_order = HOST_BIG_ENDIAN ? ELFDATA2MSB : ELFDATA2LSB;
+    const bool needs_byteswap = s->dump_info.d_endian == host_data_order;
 
     note->hdr.n_type = cpu_to_dump32(s, NT_PPC_VMX);
     vmxregset = &note->contents.vmxregset;
     memset(vmxregset, 0, sizeof(*vmxregset));
 
     for (i = 0; i < 32; i++) {
-        bool needs_byteswap;
         ppc_avr_t *avr = cpu_avr_ptr(&cpu->env, i);
 
-#if HOST_BIG_ENDIAN
-        needs_byteswap = s->dump_info.d_endian == ELFDATA2LSB;
-#else
-        needs_byteswap = s->dump_info.d_endian == ELFDATA2MSB;
-#endif
-
         if (needs_byteswap) {
             vmxregset->avr[i].u64[0] = bswap64(avr->u64[1]);
             vmxregset->avr[i].u64[1] = bswap64(avr->u64[0]);
target/ppc/int_helper.c +14 -14
index ef4b2e75d60..0c6f5b2e519 100644
@@ -1678,13 +1678,13 @@ void helper_vslo(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
 {
     int sh = (b->VsrB(0xf) >> 3) & 0xf;
 
-#if HOST_BIG_ENDIAN
-    memmove(&r->u8[0], &a->u8[sh], 16 - sh);
-    memset(&r->u8[16 - sh], 0, sh);
-#else
-    memmove(&r->u8[sh], &a->u8[0], 16 - sh);
-    memset(&r->u8[0], 0, sh);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        memmove(&r->u8[0], &a->u8[sh], 16 - sh);
+        memset(&r->u8[16 - sh], 0, sh);
+    } else {
+        memmove(&r->u8[sh], &a->u8[0], 16 - sh);
+        memset(&r->u8[0], 0, sh);
+    }
 }
 
 #if HOST_BIG_ENDIAN
@@ -1898,13 +1898,13 @@ void helper_vsro(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
 {
     int sh = (b->VsrB(0xf) >> 3) & 0xf;
 
-#if HOST_BIG_ENDIAN
-    memmove(&r->u8[sh], &a->u8[0], 16 - sh);
-    memset(&r->u8[0], 0, sh);
-#else
-    memmove(&r->u8[0], &a->u8[sh], 16 - sh);
-    memset(&r->u8[16 - sh], 0, sh);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        memmove(&r->u8[sh], &a->u8[0], 16 - sh);
+        memset(&r->u8[0], 0, sh);
+    } else {
+        memmove(&r->u8[0], &a->u8[sh], 16 - sh);
+        memset(&r->u8[16 - sh], 0, sh);
+    }
 }
 
 void helper_vsumsws(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
target/ppc/kvm.c +9 -16
index 2521ff65c6c..c00d29ce2c8 100644
@@ -651,13 +651,13 @@ static int kvm_put_fp(CPUState *cs)
             uint64_t *fpr = cpu_fpr_ptr(env, i);
             uint64_t *vsrl = cpu_vsrl_ptr(env, i);
 
-#if HOST_BIG_ENDIAN
-            vsr[0] = float64_val(*fpr);
-            vsr[1] = *vsrl;
-#else
-            vsr[0] = *vsrl;
-            vsr[1] = float64_val(*fpr);
-#endif
+            if (HOST_BIG_ENDIAN) {
+                vsr[0] = float64_val(*fpr);
+                vsr[1] = *vsrl;
+            } else {
+                vsr[0] = *vsrl;
+                vsr[1] = float64_val(*fpr);
+            }
             reg.addr = (uintptr_t) &vsr;
             reg.id = vsx ? KVM_REG_PPC_VSR(i) : KVM_REG_PPC_FPR(i);
 
@@ -728,17 +728,10 @@ static int kvm_get_fp(CPUState *cs)
                                         strerror(errno));
                 return ret;
             } else {
-#if HOST_BIG_ENDIAN
-                *fpr = vsr[0];
+                *fpr = vsr[!HOST_BIG_ENDIAN];
                 if (vsx) {
-                    *vsrl = vsr[1];
+                    *vsrl = vsr[HOST_BIG_ENDIAN];
                 }
-#else
-                *fpr = vsr[1];
-                if (vsx) {
-                    *vsrl = vsr[0];
-                }
-#endif
             }
         }
     }
target/ppc/translate/vmx-impl.c.inc +7 -7
index 92d6e8c6032..ca9cf1823d4 100644
@@ -134,9 +134,9 @@ static void gen_mtvscr(DisasContext *ctx)
 
     val = tcg_temp_new_i32();
     bofs = avr_full_offset(rB(ctx->opcode));
-#if HOST_BIG_ENDIAN
-    bofs += 3 * 4;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        bofs += 3 * 4;
+    }
 
     tcg_gen_ld_i32(val, tcg_env, bofs);
     gen_helper_mtvscr(tcg_env, val);
@@ -1528,10 +1528,10 @@ static void gen_vsplt(DisasContext *ctx, int vece)
 
     /* Experimental testing shows that hardware masks the immediate.  */
     bofs += (uimm << vece) & 15;
-#if !HOST_BIG_ENDIAN
-    bofs ^= 15;
-    bofs &= ~((1 << vece) - 1);
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        bofs ^= 15;
+        bofs &= ~((1 << vece) - 1);
+    }
 
     tcg_gen_gvec_dup_mem(vece, dofs, bofs, 16, 16);
 }
target/ppc/translate/vsx-impl.c.inc +3 -3
index 00ad57c6282..8e5c75961f4 100644
@@ -1642,9 +1642,9 @@ static bool trans_XXSPLTW(DisasContext *ctx, arg_XX2_uim *a)
     tofs = vsr_full_offset(a->xt);
     bofs = vsr_full_offset(a->xb);
     bofs += a->uim << MO_32;
-#if !HOST_BIG_ENDIAN
-    bofs ^= 8 | 4;
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        bofs ^= 8 | 4;
+    }
 
     tcg_gen_gvec_dup_mem(MO_32, tofs, bofs, 16, 16);
     return true;
tcg/ppc/tcg-target.c.inc +12 -12
index b8b23d44d5e..61aa77f5454 100644
@@ -3951,9 +3951,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
             tcg_out_mem_long(s, 0, LVEBX, out, base, offset);
         }
         elt = extract32(offset, 0, 4);
-#if !HOST_BIG_ENDIAN
-        elt ^= 15;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 15;
+        }
         tcg_out32(s, VSPLTB | VRT(out) | VRB(out) | (elt << 16));
         break;
     case MO_16:
@@ -3964,9 +3964,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
             tcg_out_mem_long(s, 0, LVEHX, out, base, offset);
         }
         elt = extract32(offset, 1, 3);
-#if !HOST_BIG_ENDIAN
-        elt ^= 7;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 7;
+        }
         tcg_out32(s, VSPLTH | VRT(out) | VRB(out) | (elt << 16));
         break;
     case MO_32:
@@ -3977,9 +3977,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
         tcg_debug_assert((offset & 3) == 0);
         tcg_out_mem_long(s, 0, LVEWX, out, base, offset);
         elt = extract32(offset, 2, 2);
-#if !HOST_BIG_ENDIAN
-        elt ^= 3;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 3;
+        }
         tcg_out32(s, VSPLTW | VRT(out) | VRB(out) | (elt << 16));
         break;
     case MO_64:
@@ -3991,9 +3991,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
         tcg_out_mem_long(s, 0, LVX, out, base, offset & -16);
         tcg_out_vsldoi(s, TCG_VEC_TMP1, out, out, 8);
         elt = extract32(offset, 3, 1);
-#if !HOST_BIG_ENDIAN
-        elt = !elt;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt = !elt;
+        }
         if (elt) {
             tcg_out_vsldoi(s, out, out, TCG_VEC_TMP1, 8);
         } else {
Signature · 1 line
2.51.0
Reviewed-by on v1 12/16
#18 lore raw

Re: [PATCH 12/16] target/ppc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 21 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
  target/ppc/arch_dump.c              |  9 ++-------
  target/ppc/int_helper.c             | 28 ++++++++++++++--------------
  target/ppc/kvm.c                    | 25 +++++++++----------------
  target/ppc/translate/vmx-impl.c.inc | 14 +++++++-------
  target/ppc/translate/vsx-impl.c.inc |  6 +++---
  tcg/ppc/tcg-target.c.inc            | 24 ++++++++++++------------
  6 files changed, 47 insertions(+), 59 deletions(-)
target/ppc/arch_dump.c +0 -0
index 80ac6c3e320..5cb8dbe9a6a 100644

target/ppc/arch_dump.c static void ppc_write_elf_vmxregset(NoteFuncArg *arg, PowerPCCPU *cpu, int id) · line 161

Quoted 4 lines
target/ppc/arch_dump.c +1 -0
@@ -158,21 +158,16 @@ static void ppc_write_elf_vmxregset(NoteFuncArg *arg, PowerPCCPU *cpu, int id)
      struct PPCElfVmxregset *vmxregset;
      Note *note = &arg->note;
      DumpState *s = arg->state;
+    const int host_data_order = HOST_BIG_ENDIAN ? ELFDATA2MSB : ELFDATA2LSB;

I think the order of return values of ternary op needs to be swapped.

                            HOST_BIG_ENDIAN ? ELFDATA2LSB : ELFDATA2MSB;

Otherwise, looks fine. With that fixed:

Quoted 215 lines
target/ppc/arch_dump.c +1 -7
+    const bool needs_byteswap = s->dump_info.d_endian == host_data_order;
 
      note->hdr.n_type = cpu_to_dump32(s, NT_PPC_VMX);
      vmxregset = &note->contents.vmxregset;
      memset(vmxregset, 0, sizeof(*vmxregset));
 
      for (i = 0; i < 32; i++) {
-        bool needs_byteswap;
          ppc_avr_t *avr = cpu_avr_ptr(&cpu->env, i);
 
-#if HOST_BIG_ENDIAN
-        needs_byteswap = s->dump_info.d_endian == ELFDATA2LSB;
-#else
-        needs_byteswap = s->dump_info.d_endian == ELFDATA2MSB;
-#endif
-
          if (needs_byteswap) {
              vmxregset->avr[i].u64[0] = bswap64(avr->u64[1]);
              vmxregset->avr[i].u64[1] = bswap64(avr->u64[0]);
target/ppc/int_helper.c +14 -14
index ef4b2e75d60..0c6f5b2e519 100644
@@ -1678,13 +1678,13 @@ void helper_vslo(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
  {
      int sh = (b->VsrB(0xf) >> 3) & 0xf;
 
-#if HOST_BIG_ENDIAN
-    memmove(&r->u8[0], &a->u8[sh], 16 - sh);
-    memset(&r->u8[16 - sh], 0, sh);
-#else
-    memmove(&r->u8[sh], &a->u8[0], 16 - sh);
-    memset(&r->u8[0], 0, sh);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        memmove(&r->u8[0], &a->u8[sh], 16 - sh);
+        memset(&r->u8[16 - sh], 0, sh);
+    } else {
+        memmove(&r->u8[sh], &a->u8[0], 16 - sh);
+        memset(&r->u8[0], 0, sh);
+    }
  }
 
  #if HOST_BIG_ENDIAN
@@ -1898,13 +1898,13 @@ void helper_vsro(ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
  {
      int sh = (b->VsrB(0xf) >> 3) & 0xf;
 
-#if HOST_BIG_ENDIAN
-    memmove(&r->u8[sh], &a->u8[0], 16 - sh);
-    memset(&r->u8[0], 0, sh);
-#else
-    memmove(&r->u8[0], &a->u8[sh], 16 - sh);
-    memset(&r->u8[16 - sh], 0, sh);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        memmove(&r->u8[sh], &a->u8[0], 16 - sh);
+        memset(&r->u8[0], 0, sh);
+    } else {
+        memmove(&r->u8[0], &a->u8[sh], 16 - sh);
+        memset(&r->u8[16 - sh], 0, sh);
+    }
  }
 
  void helper_vsumsws(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
target/ppc/kvm.c +9 -16
index 2521ff65c6c..c00d29ce2c8 100644
@@ -651,13 +651,13 @@ static int kvm_put_fp(CPUState *cs)
              uint64_t *fpr = cpu_fpr_ptr(env, i);
              uint64_t *vsrl = cpu_vsrl_ptr(env, i);
 
-#if HOST_BIG_ENDIAN
-            vsr[0] = float64_val(*fpr);
-            vsr[1] = *vsrl;
-#else
-            vsr[0] = *vsrl;
-            vsr[1] = float64_val(*fpr);
-#endif
+            if (HOST_BIG_ENDIAN) {
+                vsr[0] = float64_val(*fpr);
+                vsr[1] = *vsrl;
+            } else {
+                vsr[0] = *vsrl;
+                vsr[1] = float64_val(*fpr);
+            }
              reg.addr = (uintptr_t) &vsr;
              reg.id = vsx ? KVM_REG_PPC_VSR(i) : KVM_REG_PPC_FPR(i);
 
@@ -728,17 +728,10 @@ static int kvm_get_fp(CPUState *cs)
                                          strerror(errno));
                  return ret;
              } else {
-#if HOST_BIG_ENDIAN
-                *fpr = vsr[0];
+                *fpr = vsr[!HOST_BIG_ENDIAN];
                  if (vsx) {
-                    *vsrl = vsr[1];
+                    *vsrl = vsr[HOST_BIG_ENDIAN];
                  }
-#else
-                *fpr = vsr[1];
-                if (vsx) {
-                    *vsrl = vsr[0];
-                }
-#endif
              }
          }
      }
target/ppc/translate/vmx-impl.c.inc +7 -7
index 92d6e8c6032..ca9cf1823d4 100644
@@ -134,9 +134,9 @@ static void gen_mtvscr(DisasContext *ctx)
 
      val = tcg_temp_new_i32();
      bofs = avr_full_offset(rB(ctx->opcode));
-#if HOST_BIG_ENDIAN
-    bofs += 3 * 4;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        bofs += 3 * 4;
+    }
 
      tcg_gen_ld_i32(val, tcg_env, bofs);
      gen_helper_mtvscr(tcg_env, val);
@@ -1528,10 +1528,10 @@ static void gen_vsplt(DisasContext *ctx, int vece)
 
      /* Experimental testing shows that hardware masks the immediate.  */
      bofs += (uimm << vece) & 15;
-#if !HOST_BIG_ENDIAN
-    bofs ^= 15;
-    bofs &= ~((1 << vece) - 1);
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        bofs ^= 15;
+        bofs &= ~((1 << vece) - 1);
+    }
 
      tcg_gen_gvec_dup_mem(vece, dofs, bofs, 16, 16);
  }
target/ppc/translate/vsx-impl.c.inc +3 -3
index 00ad57c6282..8e5c75961f4 100644
@@ -1642,9 +1642,9 @@ static bool trans_XXSPLTW(DisasContext *ctx, arg_XX2_uim *a)
      tofs = vsr_full_offset(a->xt);
      bofs = vsr_full_offset(a->xb);
      bofs += a->uim << MO_32;
-#if !HOST_BIG_ENDIAN
-    bofs ^= 8 | 4;
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        bofs ^= 8 | 4;
+    }
 
      tcg_gen_gvec_dup_mem(MO_32, tofs, bofs, 16, 16);
      return true;
tcg/ppc/tcg-target.c.inc +12 -12
index b8b23d44d5e..61aa77f5454 100644
@@ -3951,9 +3951,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
              tcg_out_mem_long(s, 0, LVEBX, out, base, offset);
          }
          elt = extract32(offset, 0, 4);
-#if !HOST_BIG_ENDIAN
-        elt ^= 15;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 15;
+        }
          tcg_out32(s, VSPLTB | VRT(out) | VRB(out) | (elt << 16));
          break;
      case MO_16:
@@ -3964,9 +3964,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
              tcg_out_mem_long(s, 0, LVEHX, out, base, offset);
          }
          elt = extract32(offset, 1, 3);
-#if !HOST_BIG_ENDIAN
-        elt ^= 7;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 7;
+        }
          tcg_out32(s, VSPLTH | VRT(out) | VRB(out) | (elt << 16));
          break;
      case MO_32:
@@ -3977,9 +3977,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
          tcg_debug_assert((offset & 3) == 0);
          tcg_out_mem_long(s, 0, LVEWX, out, base, offset);
          elt = extract32(offset, 2, 2);
-#if !HOST_BIG_ENDIAN
-        elt ^= 3;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 3;
+        }
          tcg_out32(s, VSPLTW | VRT(out) | VRB(out) | (elt << 16));
          break;
      case MO_64:
@@ -3991,9 +3991,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece,
          tcg_out_mem_long(s, 0, LVX, out, base, offset & -16);
          tcg_out_vsldoi(s, TCG_VEC_TMP1, out, out, 8);
          elt = extract32(offset, 3, 1);
-#if !HOST_BIG_ENDIAN
-        elt = !elt;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt = !elt;
+        }
          if (elt) {
              tcg_out_vsldoi(s, out, out, TCG_VEC_TMP1, 8);
          } else {
#19 lore raw

Re: [PATCH 12/16] target/ppc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 33 lines · Harsh Prateek Bora
Quoted 21 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
  target/ppc/arch_dump.c              |  9 ++------- target/ppc/int_helper.c             | 28 ++++++++++++++-------------- target/ppc/kvm.c                    | 25 +++++++++---------------- target/ppc/translate/vmx-impl.c.inc | 14 +++++++------- target/ppc/translate/vsx-impl.c.inc |  6 +++--- tcg/ppc/tcg-target.c.inc            | 24 ++++++++++++------------

  6 files changed, 47 insertions(+), 59 deletions(-)
target/ppc/arch_dump.c +0 -0
index 80ac6c3e320..5cb8dbe9a6a 100644

target/ppc/arch_dump.c static void ppc_write_elf_vmxregset(NoteFuncArg · line 163

Quoted 6 lines
target/ppc/arch_dump.c +1 -0
@@ -158,21 +158,16 @@ static void ppc_write_elf_vmxregset(NoteFuncArg 
 *arg, PowerPCCPU *cpu, int id)
       struct PPCElfVmxregset *vmxregset;
       Note *note = &arg->note;
       DumpState *s = arg->state;
+    const int host_data_order = HOST_BIG_ENDIAN ? ELFDATA2MSB : 
 ELFDATA2LSB;

I think the order of return values of ternary op needs to be swapped.

Quoted 8 lines

                            HOST_BIG_ENDIAN ? ELFDATA2LSB : ELFDATA2MSB;

Otherwise, looks fine. With that fixed:

Quoted 2 lines
target/ppc/arch_dump.c +1 -0
+    const bool needs_byteswap = s->dump_info.d_endian == 
 host_data_order;

Or, a better fix would be to change == to != ?

Quoted 209 lines
Quoted 208 lines
target/ppc/arch_dump.c +0 -7
       note->hdr.n_type = cpu_to_dump32(s, NT_PPC_VMX);
       vmxregset = &note->contents.vmxregset;
       memset(vmxregset, 0, sizeof(*vmxregset));
       for (i = 0; i < 32; i++) {
-        bool needs_byteswap;
           ppc_avr_t *avr = cpu_avr_ptr(&cpu->env, i);
-#if HOST_BIG_ENDIAN
-        needs_byteswap = s->dump_info.d_endian == ELFDATA2LSB;
-#else
-        needs_byteswap = s->dump_info.d_endian == ELFDATA2MSB;
-#endif
-
           if (needs_byteswap) {
               vmxregset->avr[i].u64[0] = bswap64(avr->u64[1]);
               vmxregset->avr[i].u64[1] = bswap64(avr->u64[0]);
target/ppc/int_helper.c +14 -14
index ef4b2e75d60..0c6f5b2e519 100644
@@ -1678,13 +1678,13 @@ void helper_vslo(ppc_avr_t *r, ppc_avr_t *a, 
 ppc_avr_t *b)
   {
       int sh = (b->VsrB(0xf) >> 3) & 0xf;
-#if HOST_BIG_ENDIAN
-    memmove(&r->u8[0], &a->u8[sh], 16 - sh);
-    memset(&r->u8[16 - sh], 0, sh);
-#else
-    memmove(&r->u8[sh], &a->u8[0], 16 - sh);
-    memset(&r->u8[0], 0, sh);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        memmove(&r->u8[0], &a->u8[sh], 16 - sh);
+        memset(&r->u8[16 - sh], 0, sh);
+    } else {
+        memmove(&r->u8[sh], &a->u8[0], 16 - sh);
+        memset(&r->u8[0], 0, sh);
+    }
   }
   #if HOST_BIG_ENDIAN
@@ -1898,13 +1898,13 @@ void helper_vsro(ppc_avr_t *r, ppc_avr_t *a, 
 ppc_avr_t *b)
   {
       int sh = (b->VsrB(0xf) >> 3) & 0xf;
-#if HOST_BIG_ENDIAN
-    memmove(&r->u8[sh], &a->u8[0], 16 - sh);
-    memset(&r->u8[0], 0, sh);
-#else
-    memmove(&r->u8[0], &a->u8[sh], 16 - sh);
-    memset(&r->u8[16 - sh], 0, sh);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        memmove(&r->u8[sh], &a->u8[0], 16 - sh);
+        memset(&r->u8[0], 0, sh);
+    } else {
+        memmove(&r->u8[0], &a->u8[sh], 16 - sh);
+        memset(&r->u8[16 - sh], 0, sh);
+    }
   }
   void helper_vsumsws(CPUPPCState *env, ppc_avr_t *r, ppc_avr_t *a, ppc_avr_t *b)
target/ppc/kvm.c +9 -16
index 2521ff65c6c..c00d29ce2c8 100644
@@ -651,13 +651,13 @@ static int kvm_put_fp(CPUState *cs)
               uint64_t *fpr = cpu_fpr_ptr(env, i);
               uint64_t *vsrl = cpu_vsrl_ptr(env, i);
-#if HOST_BIG_ENDIAN
-            vsr[0] = float64_val(*fpr);
-            vsr[1] = *vsrl;
-#else
-            vsr[0] = *vsrl;
-            vsr[1] = float64_val(*fpr);
-#endif
+            if (HOST_BIG_ENDIAN) {
+                vsr[0] = float64_val(*fpr);
+                vsr[1] = *vsrl;
+            } else {
+                vsr[0] = *vsrl;
+                vsr[1] = float64_val(*fpr);
+            }
               reg.addr = (uintptr_t) &vsr;
               reg.id = vsx ? KVM_REG_PPC_VSR(i) : KVM_REG_PPC_FPR(i);
@@ -728,17 +728,10 @@ static int kvm_get_fp(CPUState *cs)
                                           strerror(errno));
                   return ret;
               } else {
-#if HOST_BIG_ENDIAN
-                *fpr = vsr[0];
+                *fpr = vsr[!HOST_BIG_ENDIAN];
                   if (vsx) {
-                    *vsrl = vsr[1];
+                    *vsrl = vsr[HOST_BIG_ENDIAN];
                   }
-#else
-                *fpr = vsr[1];
-                if (vsx) {
-                    *vsrl = vsr[0];
-                }
-#endif
               }
           }
       }
target/ppc/translate/vmx-impl.c.inc +7 -7
index 92d6e8c6032..ca9cf1823d4 100644
@@ -134,9 +134,9 @@ static void gen_mtvscr(DisasContext *ctx)
       val = tcg_temp_new_i32();
       bofs = avr_full_offset(rB(ctx->opcode));
-#if HOST_BIG_ENDIAN
-    bofs += 3 * 4;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        bofs += 3 * 4;
+    }
       tcg_gen_ld_i32(val, tcg_env, bofs);
       gen_helper_mtvscr(tcg_env, val);
@@ -1528,10 +1528,10 @@ static void gen_vsplt(DisasContext *ctx, int 
 vece)
       /* Experimental testing shows that hardware masks the immediate.  */
       bofs += (uimm << vece) & 15;
-#if !HOST_BIG_ENDIAN
-    bofs ^= 15;
-    bofs &= ~((1 << vece) - 1);
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        bofs ^= 15;
+        bofs &= ~((1 << vece) - 1);
+    }
       tcg_gen_gvec_dup_mem(vece, dofs, bofs, 16, 16);
   }
target/ppc/translate/vsx-impl.c.inc +3 -3
index 00ad57c6282..8e5c75961f4 100644
@@ -1642,9 +1642,9 @@ static bool trans_XXSPLTW(DisasContext *ctx, 
 arg_XX2_uim *a)
       tofs = vsr_full_offset(a->xt);
       bofs = vsr_full_offset(a->xb);
       bofs += a->uim << MO_32;
-#if !HOST_BIG_ENDIAN
-    bofs ^= 8 | 4;
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        bofs ^= 8 | 4;
+    }
       tcg_gen_gvec_dup_mem(MO_32, tofs, bofs, 16, 16);
       return true;
tcg/ppc/tcg-target.c.inc +3 -3
index b8b23d44d5e..61aa77f5454 100644
@@ -3951,9 +3951,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, 
 TCGType type, unsigned vece,
               tcg_out_mem_long(s, 0, LVEBX, out, base, offset);
           }
           elt = extract32(offset, 0, 4);
-#if !HOST_BIG_ENDIAN
-        elt ^= 15;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 15;
+        }
           tcg_out32(s, VSPLTB | VRT(out) | VRB(out) | (elt << 16));
           break;

      case MO_16:

tcg/ppc/tcg-target.c.inc +3 -3
@@ -3964,9 +3964,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, 
 TCGType type, unsigned vece,
               tcg_out_mem_long(s, 0, LVEHX, out, base, offset);
           }
           elt = extract32(offset, 1, 3);
-#if !HOST_BIG_ENDIAN
-        elt ^= 7;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 7;
+        }
           tcg_out32(s, VSPLTH | VRT(out) | VRB(out) | (elt << 16));
           break;

      case MO_32:

tcg/ppc/tcg-target.c.inc +3 -3
@@ -3977,9 +3977,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, 
 TCGType type, unsigned vece,
           tcg_debug_assert((offset & 3) == 0);
           tcg_out_mem_long(s, 0, LVEWX, out, base, offset);
           elt = extract32(offset, 2, 2);
-#if !HOST_BIG_ENDIAN
-        elt ^= 3;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt ^= 3;
+        }
           tcg_out32(s, VSPLTW | VRT(out) | VRB(out) | (elt << 16));
           break;

      case MO_64:

tcg/ppc/tcg-target.c.inc +3 -3
@@ -3991,9 +3991,9 @@ static bool tcg_out_dupm_vec(TCGContext *s, 
 TCGType type, unsigned vece,
           tcg_out_mem_long(s, 0, LVX, out, base, offset & -16);
           tcg_out_vsldoi(s, TCG_VEC_TMP1, out, out, 8);
           elt = extract32(offset, 3, 1);
-#if !HOST_BIG_ENDIAN
-        elt = !elt;
-#endif
+        if (!HOST_BIG_ENDIAN) {
+            elt = !elt;
+        }
           if (elt) {
               tcg_out_vsldoi(s, out, out, TCG_VEC_TMP1, 8);

          } else {

#20 lore raw

PATCH 13/16 target/riscv: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 target/riscv/vector_helper.c            | 32 ++++++++++++-------------
 target/riscv/insn_trans/trans_rvv.c.inc | 16 ++++++-------
 2 files changed, 24 insertions(+), 24 deletions(-)
target/riscv/vector_helper.c +16 -16
index 41ea2231067..2de3358ee86 100644
@@ -235,26 +235,26 @@ vext_continuous_ldst_host(CPURISCVState *env, vext_ldst_elem_fn_host *ldst_host,
                         void *vd, uint32_t evl, uint32_t reg_start, void *host,
                         uint32_t esz, bool is_load)
 {
-#if HOST_BIG_ENDIAN
-    for (; reg_start < evl; reg_start++, host += esz) {
-        ldst_host(vd, reg_start, host);
-    }
-#else
-    if (esz == 1) {
-        uint32_t byte_offset = reg_start * esz;
-        uint32_t size = (evl - reg_start) * esz;
-
-        if (is_load) {
-            memcpy(vd + byte_offset, host, size);
-        } else {
-            memcpy(host, vd + byte_offset, size);
-        }
-    } else {
+    if (HOST_BIG_ENDIAN) {
         for (; reg_start < evl; reg_start++, host += esz) {
             ldst_host(vd, reg_start, host);
         }
+    } else {
+        if (esz == 1) {
+            uint32_t byte_offset = reg_start * esz;
+            uint32_t size = (evl - reg_start) * esz;
+
+            if (is_load) {
+                memcpy(vd + byte_offset, host, size);
+            } else {
+                memcpy(host, vd + byte_offset, size);
+            }
+        } else {
+            for (; reg_start < evl; reg_start++, host += esz) {
+                ldst_host(vd, reg_start, host);
+            }
+        }
     }
-#endif
 }
 
 static void vext_set_tail_elems_1s(target_ulong vl, void *vd,
target/riscv/insn_trans/trans_rvv.c.inc +8 -8
index f4b5460340e..2a487179f63 100644
@@ -3351,19 +3351,19 @@ static void load_element(TCGv_i64 dest, TCGv_ptr base,
 /* offset of the idx element with base register r */
 static uint32_t endian_ofs(DisasContext *s, int r, int idx)
 {
-#if HOST_BIG_ENDIAN
-    return vreg_ofs(s, r) + ((idx ^ (7 >> s->sew)) << s->sew);
-#else
-    return vreg_ofs(s, r) + (idx << s->sew);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        return vreg_ofs(s, r) + ((idx ^ (7 >> s->sew)) << s->sew);
+    } else {
+        return vreg_ofs(s, r) + (idx << s->sew);
+    }
 }
 
 /* adjust the index according to the endian */
 static void endian_adjust(TCGv_i32 ofs, int sew)
 {
-#if HOST_BIG_ENDIAN
-    tcg_gen_xori_i32(ofs, ofs, 7 >> sew);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        tcg_gen_xori_i32(ofs, ofs, 7 >> sew);
+    }
 }
 
 /* Load idx >= VLMAX ? 0 : vreg[idx] */
Signature · 1 line
2.51.0
Reviewed-by on v1 13/16
#21 lore raw

Re: [PATCH 13/16] target/riscv: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 8 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

Alistair

Quoted 88 lines
---
 target/riscv/vector_helper.c            | 32 ++++++++++++-------------
 target/riscv/insn_trans/trans_rvv.c.inc | 16 ++++++-------
 2 files changed, 24 insertions(+), 24 deletions(-)
target/riscv/vector_helper.c +16 -16
index 41ea2231067..2de3358ee86 100644
@@ -235,26 +235,26 @@ vext_continuous_ldst_host(CPURISCVState *env, vext_ldst_elem_fn_host *ldst_host,
                         void *vd, uint32_t evl, uint32_t reg_start, void *host,
                         uint32_t esz, bool is_load)
 {
-#if HOST_BIG_ENDIAN
-    for (; reg_start < evl; reg_start++, host += esz) {
-        ldst_host(vd, reg_start, host);
-    }
-#else
-    if (esz == 1) {
-        uint32_t byte_offset = reg_start * esz;
-        uint32_t size = (evl - reg_start) * esz;
-
-        if (is_load) {
-            memcpy(vd + byte_offset, host, size);
-        } else {
-            memcpy(host, vd + byte_offset, size);
-        }
-    } else {
+    if (HOST_BIG_ENDIAN) {
         for (; reg_start < evl; reg_start++, host += esz) {
             ldst_host(vd, reg_start, host);
         }
+    } else {
+        if (esz == 1) {
+            uint32_t byte_offset = reg_start * esz;
+            uint32_t size = (evl - reg_start) * esz;
+
+            if (is_load) {
+                memcpy(vd + byte_offset, host, size);
+            } else {
+                memcpy(host, vd + byte_offset, size);
+            }
+        } else {
+            for (; reg_start < evl; reg_start++, host += esz) {
+                ldst_host(vd, reg_start, host);
+            }
+        }
     }
-#endif
 }
 
 static void vext_set_tail_elems_1s(target_ulong vl, void *vd,
target/riscv/insn_trans/trans_rvv.c.inc +8 -9
index f4b5460340e..2a487179f63 100644
@@ -3351,19 +3351,19 @@ static void load_element(TCGv_i64 dest, TCGv_ptr base,
 /* offset of the idx element with base register r */
 static uint32_t endian_ofs(DisasContext *s, int r, int idx)
 {
-#if HOST_BIG_ENDIAN
-    return vreg_ofs(s, r) + ((idx ^ (7 >> s->sew)) << s->sew);
-#else
-    return vreg_ofs(s, r) + (idx << s->sew);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        return vreg_ofs(s, r) + ((idx ^ (7 >> s->sew)) << s->sew);
+    } else {
+        return vreg_ofs(s, r) + (idx << s->sew);
+    }
 }
 
 /* adjust the index according to the endian */
 static void endian_adjust(TCGv_i32 ofs, int sew)
 {
-#if HOST_BIG_ENDIAN
-    tcg_gen_xori_i32(ofs, ofs, 7 >> sew);
-#endif
+    if (HOST_BIG_ENDIAN) {
+        tcg_gen_xori_i32(ofs, ofs, 7 >> sew);
+    }
 }
 
 /* Load idx >= VLMAX ? 0 : vreg[idx] */
--

2.51.0

#22 lore raw

PATCH 14/16 target/s390x: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 target/s390x/tcg/translate.c        | 6 +++---
 target/s390x/tcg/translate_vx.c.inc | 6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)
target/s390x/tcg/translate.c +3 -3
index c7e8574438c..ec9e5a07516 100644
@@ -258,9 +258,9 @@ static inline int vec_reg_offset(uint8_t reg, uint8_t enr, MemOp es)
      * 16 byte operations to handle it in a special way.
      */
     g_assert(es <= MO_64);
-#if !HOST_BIG_ENDIAN
-    offs ^= (8 - bytes);
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        offs ^= (8 - bytes);
+    }
     return offs + vec_full_reg_offset(reg);
 }
 
target/s390x/tcg/translate_vx.c.inc +3 -3
index e073e5ad3aa..f3b4b48ab7b 100644
@@ -175,9 +175,9 @@ static void get_vec_element_ptr_i64(TCGv_ptr ptr, uint8_t reg, TCGv_i64 enr,
 
     /* convert it to an element offset relative to tcg_env (vec_reg_offset() */
     tcg_gen_shli_i64(tmp, tmp, es);
-#if !HOST_BIG_ENDIAN
-    tcg_gen_xori_i64(tmp, tmp, 8 - NUM_VEC_ELEMENT_BYTES(es));
-#endif
+    if (!HOST_BIG_ENDIAN) {
+        tcg_gen_xori_i64(tmp, tmp, 8 - NUM_VEC_ELEMENT_BYTES(es));
+    }
     tcg_gen_addi_i64(tmp, tmp, vec_full_reg_offset(reg));
 
     /* generate the final ptr by adding tcg_env */
Signature · 1 line
2.51.0
Reviewed-by on v1 14/16
#23 lore raw

Re: [PATCH 14/16] target/s390x: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

Quoted 8 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---

Signature · 3 lines
Cheers

David / dhildenb
#24 lore raw

PATCH 15/16 target/sparc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 target/sparc/vis_helper.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
target/sparc/vis_helper.c +9 -9
index 371f5445a1f..72926dfb7b0 100644
@@ -462,15 +462,15 @@ uint64_t helper_bshuffle(uint64_t gsr, uint64_t src1, uint64_t src2)
     uint32_t i, mask, host;
 
     /* Set up S such that we can index across all of the bytes.  */
-#if HOST_BIG_ENDIAN
-    s.ll[0] = src1;
-    s.ll[1] = src2;
-    host = 0;
-#else
-    s.ll[1] = src1;
-    s.ll[0] = src2;
-    host = 15;
-#endif
+    if (HOST_BIG_ENDIAN) {
+        s.ll[0] = src1;
+        s.ll[1] = src2;
+        host = 0;
+    } else {
+        s.ll[1] = src1;
+        s.ll[0] = src2;
+        host = 15;
+    }
     mask = gsr >> 32;
 
     for (i = 0; i < 8; ++i) {
Signature · 1 line
2.51.0
#25 lore raw

PATCH 16/16 util/bitmap: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

In reply to Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

---
 util/bitmap.c | 19 +++++++------------
 1 file changed, 7 insertions(+), 12 deletions(-)
util/bitmap.c +7 -12
index 8d12e90a5a4..2346159116e 100644
@@ -421,19 +421,14 @@ static void bitmap_to_from_le(unsigned long *dst,
 {
     long len = BITS_TO_LONGS(nbits);
 
-#if HOST_BIG_ENDIAN
-    long index;
-
-    for (index = 0; index < len; index++) {
-# if HOST_LONG_BITS == 64
-        dst[index] = bswap64(src[index]);
-# else
-        dst[index] = bswap32(src[index]);
-# endif
+    if (HOST_BIG_ENDIAN) {
+        for (long index = 0; index < len; index++) {
+            dst[index] = (HOST_LONG_BITS == 64) ? bswap64(src[index])
+                                                : bswap32(src[index]);
+        }
+    } else {
+        memcpy(dst, src, len * sizeof(unsigned long));
     }
-#else
-    memcpy(dst, src, len * sizeof(unsigned long));
-#endif
 }
 
 void bitmap_from_le(unsigned long *dst, const unsigned long *src,
Signature · 1 line
2.51.0
maintainer
#26 lore raw
In reply to Philippe Mathieu-Daudé
Quoted 5 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

No need to repost (and I didn't review), but please change throughout the commit message to mention a *compile-time* if() check. The code for the wrong endianness will not make it past compilation, and mentioning runtime checks left me wondering if you had mistaken HOST for TARGET.

Thanks,

Paolo

Quoted 54 lines

Philippe Mathieu-Daudé (16):
linux-user/arm: Checkpatch style cleanups
linux-user/arm: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check
ui: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check
net: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check
disas: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check hw/core/loader: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check hw/display: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check hw/virtio: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/alpha: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/arm: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/mips: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/ppc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/riscv: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/s390x: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check target/sparc: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check util/bitmap: Replace HOST_BIG_ENDIAN #ifdef with runtime if() check

  include/hw/virtio/virtio-access.h       |   6 +-
  include/hw/virtio/virtio-gpu-bswap.h    |   6 +-
  target/arm/cpu.h                        |   8 +-
  target/arm/tcg/translate-a64.h          |   5 +-
  disas/disas-host.c                      |   6 +-
  hw/core/loader.c                        |   5 +-
  hw/display/artist.c                     |  12 +-
  hw/display/vga.c                        |  20 +-
  hw/display/virtio-gpu-gl.c              |   8 +-
  hw/display/virtio-gpu-rutabaga.c        |   8 +-
  hw/virtio/vhost.c                       |   7 +-
  linux-user/arm/nwfpe/double_cpdo.c      |  12 +-
  linux-user/arm/nwfpe/fpa11_cpdt.c       |  77 +-
  net/net.c                               |  14 +-
  target/alpha/translate.c                |  11 +-
  target/arm/tcg/sve_helper.c             |  10 +-
  target/arm/tcg/translate-sve.c          |  22 +-
  target/arm/tcg/translate-vfp.c          |   8 +-
  target/arm/tcg/translate.c              |   6 +-
  target/mips/tcg/msa_helper.c            | 889 ++++++++++++------------
  target/ppc/arch_dump.c                  |   9 +-
  target/ppc/int_helper.c                 |  28 +-
  target/ppc/kvm.c                        |  25 +-
  target/riscv/vector_helper.c            |  32 +-
  target/s390x/tcg/translate.c            |   6 +-
  target/sparc/vis_helper.c               |  18 +-
  ui/vdagent.c                            |  16 +-
  ui/vnc.c                                |   6 +-
  util/bitmap.c                           |  19 +-
  target/ppc/translate/vmx-impl.c.inc     |  14 +-
  target/ppc/translate/vsx-impl.c.inc     |   6 +-
  target/riscv/insn_trans/trans_rvv.c.inc |  16 +-
  target/s390x/tcg/translate_vx.c.inc     |   6 +-
  tcg/ppc/tcg-target.c.inc                |  24 +-
  34 files changed, 656 insertions(+), 709 deletions(-)
#27 lore raw
Quoted 8 lines · Paolo Bonzini
Quoted 5 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

No functional change intended.

No need to repost (and I didn't review), but please change throughout the commit message to mention a *compile-time* if() check.  The code for the wrong endianness will not make it past compilation, and mentioning runtime checks left me wondering if you had mistaken HOST for TARGET.

I want HOST, and hope the changes are correct: I meant to express the code is elided by the compiler, indeed not at *runtime* 🤦 I'll reword the commit description. Thanks for catching this.

Quoted 4 lines

Thanks,

Paolo

#28 lore raw
Quoted 6 lines · Philippe Mathieu-Daudé
Quoted 5 lines · Paolo Bonzini
Quoted 4 lines · Philippe Mathieu-Daudé

Replace compile-time #ifdef with a runtime check to ensure all code paths are built and tested. This reduces build-time configuration complexity and improves maintainability.

Quoted 7 lines
Quoted 3 lines
Quoted 1 line

No functional change intended.

No need to repost (and I didn't review), but please change throughout the commit message to mention a *compile-time* if() check.  The code for the wrong endianness will not make it past compilation, and mentioning runtime checks left me wondering if you had mistaken HOST for TARGET.

I want HOST, and hope the changes are correct: I meant to express the code is elided by the compiler, indeed not at *runtime* 🤦 I'll reword the commit description. Thanks for catching this.

What about:

Replace preprocessor-time #ifdef with a compile-time check to ensure all code paths are built and tested. This reduces build-time configuration complexity and simplifies code maintainability.

?