In loongarch_cpu_class_by_name(char *cpu_model) function,
the argument cpu_model already has the suffix '-loongarch-cpu',
so we should remove the LOONGARCH_CPU_TYPE_NAME(cpu_model) macro.
And add the assertion that 'cpu_model' resolves to a class of the
appropriate type.

Signed-off-by: Xiaojuan Yang <yangxiaojuan@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/loongarch/cpu.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/target/loongarch/cpu.c b/target/loongarch/cpu.c
index e21715592a..ed26f9beed 100644
--- a/target/loongarch/cpu.c
+++ b/target/loongarch/cpu.c
@@ -571,11 +571,12 @@ static void loongarch_cpu_init(Object *obj)
 static ObjectClass *loongarch_cpu_class_by_name(const char *cpu_model)
 {
     ObjectClass *oc;
-    char *typename;
 
-    typename = g_strdup_printf(LOONGARCH_CPU_TYPE_NAME("%s"), cpu_model);
-    oc = object_class_by_name(typename);
-    g_free(typename);
+    oc = object_class_by_name(cpu_model);
+    if (!oc || !object_class_dynamic_cast(oc, TYPE_LOONGARCH_CPU) ||
+        object_class_is_abstract(oc)) {
+        return NULL;
+    }
     return oc;
 }
 
-- 
2.31.1

On 7/15/22 11:37, Xiaojuan Yang wrote:
> In loongarch_cpu_class_by_name(char *cpu_model) function,
> the argument cpu_model already has the suffix '-loongarch-cpu',
> so we should remove the LOONGARCH_CPU_TYPE_NAME(cpu_model) macro.
> And add the assertion that 'cpu_model' resolves to a class of the
> appropriate type.
> 
> Signed-off-by: Xiaojuan Yang <yangxiaojuan@loongson.cn>
> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

This patch causes tests to fail, e.g.

   TEST    float_convd on loongarch64

qemu-loongarch64: unable to find CPU model 'la464'

make[1]: *** [/home/rth/qemu/src/tests/tcg/multiarch/Makefile.target:29: run-float_convd] 
Error 1


What caused you assume that all cpu models are already suffixed?


r~

On 7/19/22 12:16, Richard Henderson wrote:
> On 7/15/22 11:37, Xiaojuan Yang wrote:
>> In loongarch_cpu_class_by_name(char *cpu_model) function,
>> the argument cpu_model already has the suffix '-loongarch-cpu',
>> so we should remove the LOONGARCH_CPU_TYPE_NAME(cpu_model) macro.
>> And add the assertion that 'cpu_model' resolves to a class of the
>> appropriate type.
>>
>> Signed-off-by: Xiaojuan Yang <yangxiaojuan@loongson.cn>
>> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
> 
> This patch causes tests to fail, e.g.
> 
>    TEST    float_convd on loongarch64
> 
> qemu-loongarch64: unable to find CPU model 'la464'
> 
> make[1]: *** [/home/rth/qemu/src/tests/tcg/multiarch/Makefile.target:29: run-float_convd] 
> Error 1
> 
> 
> What caused you assume that all cpu models are already suffixed?

Mm.  I suppose the use over in hw/loongarch/loongson3.c.
I will make this function match target/alpha/cpu.c, which checks cpu_model as-is, and then 
tries again with the suffix.


r~

On Tue, 19 Jul 2022 12:22:14 +0530
Richard Henderson <richard.henderson@linaro.org> wrote:

> On 7/19/22 12:16, Richard Henderson wrote:
> > On 7/15/22 11:37, Xiaojuan Yang wrote:  
> >> In loongarch_cpu_class_by_name(char *cpu_model) function,
> >> the argument cpu_model already has the suffix '-loongarch-cpu',
> >> so we should remove the LOONGARCH_CPU_TYPE_NAME(cpu_model) macro.
> >> And add the assertion that 'cpu_model' resolves to a class of the
> >> appropriate type.
> >>
> >> Signed-off-by: Xiaojuan Yang <yangxiaojuan@loongson.cn>
> >> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>  
> > 
> > This patch causes tests to fail, e.g.
> > 
> >    TEST    float_convd on loongarch64
> > 
> > qemu-loongarch64: unable to find CPU model 'la464'
> > 
> > make[1]: *** [/home/rth/qemu/src/tests/tcg/multiarch/Makefile.target:29: run-float_convd] 
> > Error 1
> > 
> > 
> > What caused you assume that all cpu models are already suffixed?  
> 
> Mm.  I suppose the use over in hw/loongarch/loongson3.c.
> I will make this function match target/alpha/cpu.c, which checks cpu_model as-is, and then 
> tries again with the suffix.

if we think about adding hotplug/qmp/-device support for CPUs in future,
those interface operate with type-names directly without any pre-processing whatsoever.

so I'd very much prefer just as-is (i.e. direct type-name matching)
(all other model parsing usually comes from legacy code)

> 
> 
> r~
>