diff mbox series

[v9,19/19] target/riscv/tcg: do not support profiles for 'max' CPU

Message ID 20231102224445.527355-20-dbarboza@ventanamicro.com (mailing list archive)
State New, archived
Headers show
Series rv64i and rva22u64 CPUs, RVA22U64 profile support | expand

Commit Message

Daniel Henrique Barboza Nov. 2, 2023, 10:44 p.m. UTC
There's no gain in allowing the 'max' CPU to support profiles, since it
already contains everything that QEMU can support. And we'll open the
door for 'unorthodox' stuff like users disabling profiles of the 'max'
CPU.

Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
---
 target/riscv/tcg/tcg-cpu.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

Comments

Andrew Jones Nov. 3, 2023, 9:01 a.m. UTC | #1
On Thu, Nov 02, 2023 at 07:44:45PM -0300, Daniel Henrique Barboza wrote:
> There's no gain in allowing the 'max' CPU to support profiles, since it
> already contains everything that QEMU can support. And we'll open the
> door for 'unorthodox' stuff like users disabling profiles of the 'max'
> CPU.

I don't see a lot of value in this patch, but maybe I'm just too cruel to
users that don't know what they're doing. I even see a negative value to
this patch because I can conceive of writing a script where I generally
want to use rv64i with my explicit list of profiles/extensions, but then
I may want to temporarily "boost" my CPU to 'max' for some reason. If
I write my script like

 CPU=rv64i
 EXTENSIONS=profile=on,extension=on
 qemu -cpu $CPU,$EXTENSIONS ...

then I can't just do

 CPU=max ./my-script

to boost my CPU, since max will error out when it sees profiles being
enabled (even though that should be no-op for it). Instead, I need to
do

 CPU=max EXTENSIONS= ./my-script

which isn't horrible, but a bit annoying.

So, personally, I would drop this patch.

Thanks,
drew

> 
> Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
> ---
>  target/riscv/tcg/tcg-cpu.c | 16 +++++++++++-----
>  1 file changed, 11 insertions(+), 5 deletions(-)
> 
> diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c
> index 553fb337e7..9a964a426e 100644
> --- a/target/riscv/tcg/tcg-cpu.c
> +++ b/target/riscv/tcg/tcg-cpu.c
> @@ -825,6 +825,11 @@ static bool riscv_cpu_is_vendor(Object *cpu_obj)
>      return object_dynamic_cast(cpu_obj, TYPE_RISCV_VENDOR_CPU) != NULL;
>  }
>  
> +static bool riscv_cpu_has_max_extensions(Object *cpu_obj)
> +{
> +    return object_dynamic_cast(cpu_obj, TYPE_RISCV_CPU_MAX) != NULL;
> +}
> +
>  /*
>   * We'll get here via the following path:
>   *
> @@ -1003,6 +1008,12 @@ static void cpu_set_profile(Object *obj, Visitor *v, const char *name,
>          return;
>      }
>  
> +    if (riscv_cpu_has_max_extensions(obj)) {
> +        error_setg(errp, "Profile %s is not available for the 'max' CPU",
> +                   profile->name);
> +        return;
> +    }
> +
>      if (cpu->env.misa_mxl != MXL_RV64) {
>          error_setg(errp, "Profile %s only available for 64 bit CPUs",
>                     profile->name);
> @@ -1251,11 +1262,6 @@ static void riscv_init_max_cpu_extensions(Object *obj)
>      }
>  }
>  
> -static bool riscv_cpu_has_max_extensions(Object *cpu_obj)
> -{
> -    return object_dynamic_cast(cpu_obj, TYPE_RISCV_CPU_MAX) != NULL;
> -}
> -
>  static void tcg_cpu_instance_init(CPUState *cs)
>  {
>      RISCVCPU *cpu = RISCV_CPU(cs);
> -- 
> 2.41.0
>
Daniel Henrique Barboza Nov. 3, 2023, 11:13 a.m. UTC | #2
On 11/3/23 06:01, Andrew Jones wrote:
> On Thu, Nov 02, 2023 at 07:44:45PM -0300, Daniel Henrique Barboza wrote:
>> There's no gain in allowing the 'max' CPU to support profiles, since it
>> already contains everything that QEMU can support. And we'll open the
>> door for 'unorthodox' stuff like users disabling profiles of the 'max'
>> CPU.
> 
> I don't see a lot of value in this patch, but maybe I'm just too cruel to
> users that don't know what they're doing. I even see a negative value to
> this patch because I can conceive of writing a script where I generally
> want to use rv64i with my explicit list of profiles/extensions, but then
> I may want to temporarily "boost" my CPU to 'max' for some reason. If
> I write my script like
> 
>   CPU=rv64i
>   EXTENSIONS=profile=on,extension=on
>   qemu -cpu $CPU,$EXTENSIONS ...
> 
> then I can't just do
> 
>   CPU=max ./my-script
> 
> to boost my CPU, since max will error out when it sees profiles being
> enabled (even though that should be no-op for it). Instead, I need to
> do
> 
>   CPU=max EXTENSIONS= ./my-script
> 
> which isn't horrible, but a bit annoying.
> 
> So, personally, I would drop this patch.


Fair enough. I wasn't creative enough with scripting to see the value of 'max' and
profiles. Let's drop it.

Thanks,

Daniel

> 
> Thanks,
> drew
> 
>>
>> Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
>> ---
>>   target/riscv/tcg/tcg-cpu.c | 16 +++++++++++-----
>>   1 file changed, 11 insertions(+), 5 deletions(-)
>>
>> diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c
>> index 553fb337e7..9a964a426e 100644
>> --- a/target/riscv/tcg/tcg-cpu.c
>> +++ b/target/riscv/tcg/tcg-cpu.c
>> @@ -825,6 +825,11 @@ static bool riscv_cpu_is_vendor(Object *cpu_obj)
>>       return object_dynamic_cast(cpu_obj, TYPE_RISCV_VENDOR_CPU) != NULL;
>>   }
>>   
>> +static bool riscv_cpu_has_max_extensions(Object *cpu_obj)
>> +{
>> +    return object_dynamic_cast(cpu_obj, TYPE_RISCV_CPU_MAX) != NULL;
>> +}
>> +
>>   /*
>>    * We'll get here via the following path:
>>    *
>> @@ -1003,6 +1008,12 @@ static void cpu_set_profile(Object *obj, Visitor *v, const char *name,
>>           return;
>>       }
>>   
>> +    if (riscv_cpu_has_max_extensions(obj)) {
>> +        error_setg(errp, "Profile %s is not available for the 'max' CPU",
>> +                   profile->name);
>> +        return;
>> +    }
>> +
>>       if (cpu->env.misa_mxl != MXL_RV64) {
>>           error_setg(errp, "Profile %s only available for 64 bit CPUs",
>>                      profile->name);
>> @@ -1251,11 +1262,6 @@ static void riscv_init_max_cpu_extensions(Object *obj)
>>       }
>>   }
>>   
>> -static bool riscv_cpu_has_max_extensions(Object *cpu_obj)
>> -{
>> -    return object_dynamic_cast(cpu_obj, TYPE_RISCV_CPU_MAX) != NULL;
>> -}
>> -
>>   static void tcg_cpu_instance_init(CPUState *cs)
>>   {
>>       RISCVCPU *cpu = RISCV_CPU(cs);
>> -- 
>> 2.41.0
>>
diff mbox series

Patch

diff --git a/target/riscv/tcg/tcg-cpu.c b/target/riscv/tcg/tcg-cpu.c
index 553fb337e7..9a964a426e 100644
--- a/target/riscv/tcg/tcg-cpu.c
+++ b/target/riscv/tcg/tcg-cpu.c
@@ -825,6 +825,11 @@  static bool riscv_cpu_is_vendor(Object *cpu_obj)
     return object_dynamic_cast(cpu_obj, TYPE_RISCV_VENDOR_CPU) != NULL;
 }
 
+static bool riscv_cpu_has_max_extensions(Object *cpu_obj)
+{
+    return object_dynamic_cast(cpu_obj, TYPE_RISCV_CPU_MAX) != NULL;
+}
+
 /*
  * We'll get here via the following path:
  *
@@ -1003,6 +1008,12 @@  static void cpu_set_profile(Object *obj, Visitor *v, const char *name,
         return;
     }
 
+    if (riscv_cpu_has_max_extensions(obj)) {
+        error_setg(errp, "Profile %s is not available for the 'max' CPU",
+                   profile->name);
+        return;
+    }
+
     if (cpu->env.misa_mxl != MXL_RV64) {
         error_setg(errp, "Profile %s only available for 64 bit CPUs",
                    profile->name);
@@ -1251,11 +1262,6 @@  static void riscv_init_max_cpu_extensions(Object *obj)
     }
 }
 
-static bool riscv_cpu_has_max_extensions(Object *cpu_obj)
-{
-    return object_dynamic_cast(cpu_obj, TYPE_RISCV_CPU_MAX) != NULL;
-}
-
 static void tcg_cpu_instance_init(CPUState *cs)
 {
     RISCVCPU *cpu = RISCV_CPU(cs);