[2/3] hw/ppc/sam460ex: Replace tswap32() by stl_endian_p()

Commit Message

Philippe Mathieu-Daudé Sept. 30, 2024, 12:53 p.m. UTC

Replace the target-specific tswap32() call by stl_endian_p()
which does the same but takes the endianness as argument, thus
is target-agnostic.
Get the vCPU endianness calling ppc_cpu_is_big_endian().

Signed-off-by: Philippe Mathieu-Daudé <[e-mail]>
---
 hw/ppc/sam460ex.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

Comments

BALATON Zoltan Sept. 30, 2024, 1:49 p.m. UTC | #1

On Mon, 30 Sep 2024, Philippe Mathieu-Daudé wrote:
> Replace the target-specific tswap32() call by stl_endian_p()
> which does the same but takes the endianness as argument, thus
> is target-agnostic.
> Get the vCPU endianness calling ppc_cpu_is_big_endian().
>
> Signed-off-by: Philippe Mathieu-Daudé <[e-mail]>
> ---
> hw/ppc/sam460ex.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/hw/ppc/sam460ex.c b/hw/ppc/sam460ex.c
> index 8dc75fb9f0..6257ddbec6 100644
> --- a/hw/ppc/sam460ex.c
> +++ b/hw/ppc/sam460ex.c
> @@ -248,10 +248,11 @@ static void mmubooke_create_initial_mapping(CPUPPCState *env,
> static void main_cpu_reset(void *opaque)
> {
>     PowerPCCPU *cpu = opaque;
> +    CPUState *cs = CPU(cpu);
>     CPUPPCState *env = &cpu->env;
>     struct boot_info *bi = env->load_info;
>
> -    cpu_reset(CPU(cpu));
> +    cpu_reset(cs);
>
>     /* either we have a kernel to boot or we jump to U-Boot */
>     if (bi->entry != UBOOT_ENTRY) {
> @@ -261,7 +262,7 @@ static void main_cpu_reset(void *opaque)
>
>         /* Create a mapping for the kernel.  */
>         mmubooke_create_initial_mapping(env, 0, 0);
> -        env->gpr[6] = tswap32(EPAPR_MAGIC);
> +        stl_endian_p(ppc_cpu_is_big_endian(cs), &env->gpr[6], EPAPR_MAGIC);

I think this slightly changes behaviour even if getting the same result. 
The ppc_cpu_is_big_endian() checks the CPU bit for current mode while 
previously tswap32 only checks for host endianness vs. big endian which is 
the default mode for PPC which is what the CPU is in during boot where 
this data is used. So even with checking the bit it'd be the same but the 
check is not needed. I think you could/should just have hard coded big 
endian here to preserve the current behaviour.

(There were some discussion a while back if this EPAPR_MAGIC is correct or 
needed at all but preserving current behaviour is a good enough for now.)

Regards,
BALATON Zoltan

>         env->gpr[7] = (16 * MiB) - 8; /* bi->ima_size; */
>
>     } else {
>

Patch

diff --git a/hw/ppc/sam460ex.c b/hw/ppc/sam460ex.c
index 8dc75fb9f0..6257ddbec6 100644
--- a/hw/ppc/sam460ex.c
+++ b/hw/ppc/sam460ex.c
@@ -248,10 +248,11 @@  static void mmubooke_create_initial_mapping(CPUPPCState *env,
 static void main_cpu_reset(void *opaque)
 {
     PowerPCCPU *cpu = opaque;
+    CPUState *cs = CPU(cpu);
     CPUPPCState *env = &cpu->env;
     struct boot_info *bi = env->load_info;
 
-    cpu_reset(CPU(cpu));
+    cpu_reset(cs);
 
     /* either we have a kernel to boot or we jump to U-Boot */
     if (bi->entry != UBOOT_ENTRY) {
@@ -261,7 +262,7 @@  static void main_cpu_reset(void *opaque)
 
         /* Create a mapping for the kernel.  */
         mmubooke_create_initial_mapping(env, 0, 0);
-        env->gpr[6] = tswap32(EPAPR_MAGIC);
+        stl_endian_p(ppc_cpu_is_big_endian(cs), &env->gpr[6], EPAPR_MAGIC);
         env->gpr[7] = (16 * MiB) - 8; /* bi->ima_size; */
 
     } else {