Re: [PATCH v4 3/5] target/riscv: Move misa_mxl_max to class

[Akihiko Odaki]

On 2023/10/18 22:01, Daniel Henrique Barboza wrote:
> On 10/17/23 15:53, Akihiko Odaki wrote:
> > misa_mxl_max is common for all instances of a RISC-V CPU class so they
> > are better put into class.
> >
> > Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
> > ---
>
> I'll repeat what I said in the v1: this patch is adding an extra class 
> parameter,
> an extra param required to each class_init, and an extra CPUClass cast 
> every time
> we want to read misa_mxl_max, all of that because we want to assign 
> gdb_core_xml_file
> earlier.
>
> If my previous suggestion of assigning gdb_core directly into 
> riscv_cpu_class_init()
> doesn't work necause we need misa_mxl_max to do it, a good alternative 
> is setting
> gdb_core_xml_file in riscv_cpu_post_init(), which is executed after all 
> cpu_init()
> functions where we already have env->misa_mxl_max set properly.

We want to assign gdb_core_xml_file *earlier* so assigning it after 
cpu_init() is not OK. In general it should be considered unsafe to 
initialize a class variable after class_init(); otherwise other 
subsystems cannot know when it becomes available.