Re: [RFC PATCH 4/6] accel/tcg: Inline cpu_mmu_lookup function

[Richard Henderson]

On 2/15/24 09:28, Max Chou wrote:
> Signed-off-by: Max Chou <max.chou@sifive.com>
> ---
>   accel/tcg/user-exec.c | 7 +++++--
>   1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c
> index 68b252cb8e8..c5453810eee 100644
> --- a/accel/tcg/user-exec.c
> +++ b/accel/tcg/user-exec.c
> @@ -942,8 +942,11 @@ void page_reset_target_data(target_ulong start, 
> target_ulong last) { }
>   
>   /* The system-mode versions of these helpers are in cputlb.c.  */
>   
> -static void *cpu_mmu_lookup(CPUState *cpu, vaddr addr,
> -                            MemOp mop, uintptr_t ra, MMUAccessType type)
> +static inline QEMU_ALWAYS_INLINE void *cpu_mmu_lookup(CPUState *cpu,
> +                                                      vaddr addr,
> +                                                      MemOp mop,
> +                                                      uintptr_t ra,
> +                                                      MMUAccessType type)
>   {
>       int a_bits = get_alignment_bits(mop);
>       void *ret;

This is a large function.  Why does it need to be inlined?
For this and the next two patches I require evidence, because I don't believe you are 
attacking the problem correctly.


r~