Re: [PATCH v1 1/4] LoongArch: Move three functions from kprobes.c to inst.h

[Tiezhu Yang]

On 04/12/2023 06:39 PM, Xi Ruoyao wrote:
> On Wed, 2023-04-12 at 18:04 +0800, Tiezhu Yang wrote:
> > +static inline bool insns_not_supported(union loongarch_instruction insn)
>
> This function seems long enough (to me) not to use "static inline".
> Note that most part of this function belongs to a cold path, and IMO
> it's bad to inline a cold path into every caller.

OK, I will move the three functions from kprobes.c to inst.c.

> > +{
> > +       switch (insn.reg2i14_format.opcode) {
> > +       case llw_op:
> > +       case lld_op:
> > +       case scw_op:
> > +       case scd_op:
> > +               pr_notice("kprobe: ll and sc instructions are not supported\n");
> > +               return true;
> > +       }
> > +
> > +       switch (insn.reg1i21_format.opcode) {
> > +       case bceqz_op:
> > +               pr_notice("kprobe: bceqz and bcnez instructions are not supported\n");
> > +               return true;
> > +       }
> > +
> > +       return false;
> > +}
> > +
> > +static inline bool insns_need_simulation(union loongarch_instruction insn)
> > +{
> > +       if (is_pc_ins(&insn))
> > +               return true;
> > +
> > +       if (is_branch_ins(&insn))
> > +               return true;
> > +
> > +       return false;
>
> I'd write "return is_pc_ins(&insn) || is_branch_ins(&insn);" here, but
> there is no behavioral difference anyway.

I prefer leave it as it is.

Thanks,
Tiezhu