Re: [PATCH] LoongArch: Correct the definition of is_branch_ins()

[Jinyang He]

Hi, Tiezhu,


On 2022-12-14 16:30, Tiezhu Yang wrote:
> The current definition of is_branch_ins() is not correct,

But the branch instruction opcode only use the high 6 bits, [1]. That's 
meaningless because it is the same as the original result. If we care 
too much about the details of instruction format, we will lose a lot of 
tricks on LoongArch instruction coding.

[1] 
https://loongson.github.io/LoongArch-Documentation/LoongArch-Vol1-EN.html#table-of-instruction-encoding


> it was
> first introduced in commit 49aef111e2da ("LoongArch: Add prologue
> unwinder support"), in fact, there exist three branch instruction
> formats rather than only reg1i21_format.
>
> Signed-off-by: Tiezhu Yang <yangtiezhu@xxxxxxxxxxx>
> ---
>   arch/loongarch/include/asm/inst.h | 27 +++++++++++++++++++++++++--
>   1 file changed, 25 insertions(+), 2 deletions(-)
>
> diff --git a/arch/loongarch/include/asm/inst.h b/arch/loongarch/include/asm/inst.h
> index c00e151..42be39be 100644
> --- a/arch/loongarch/include/asm/inst.h
> +++ b/arch/loongarch/include/asm/inst.h
> @@ -329,8 +329,31 @@ static inline bool is_pc_ins(union loongarch_instruction *ip)
>   
>   static inline bool is_branch_ins(union loongarch_instruction *ip)
>   {
> -	return ip->reg1i21_format.opcode >= beqz_op &&
> -		ip->reg1i21_format.opcode <= bgeu_op;
> +	switch (ip->reg0i26_format.opcode) {
> +	case b_op:
> +	case bl_op:
> +		return true;
> +	}
> +
> +	switch (ip->reg1i21_format.opcode) {
> +	case beqz_op:
> +	case bnez_op:
> +	case bceqz_op:
> +		return true;
> +	}
> +
> +	switch (ip->reg2i16_format.opcode) {
> +	case jirl_op:
> +	case beq_op:
> +	case bne_op:
> +	case blt_op:
> +	case bge_op:
> +	case bltu_op:
> +	case bgeu_op:
> +		return true;
> +	}
> +
> +	return false;
>   }
>   
>   static inline bool is_ra_save_ins(union loongarch_instruction *ip)