Re: [patch V4 part 5 13/31] x86/irq: Convey vector as argument and not in ptregs

[Lai Jiangshan]

Hello

On Mon, May 11, 2020 at 10:35 PM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>
> Lai,
>
> Lai Jiangshan <jiangshanlai+lkml@xxxxxxxxx> writes:
> > On Tue, May 5, 2020 at 10:23 PM Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
> >> +SYM_CODE_START(irq_entries_start)
> >> +    vector=FIRST_EXTERNAL_VECTOR
> >> +    .rept (FIRST_SYSTEM_VECTOR - FIRST_EXTERNAL_VECTOR)
> >> +       UNWIND_HINT_IRET_REGS
> >> +       .byte   0x6a, vector
> >> +       jmp     common_interrupt
> >> +       .align  8
> >> +    vector=vector+1
> >> +    .endr
> >> +SYM_CODE_END(irq_entries_start)
> >
> > Using ".byte   0x6a, vector" is somewhat ugly.
> >
> > I hope it should be " pushq $(s8_to_s64(vector))", which can also
> > help to reduce bunches of comments about ".byte   0x6a, vector".
> >
> > However, I don't know how to implement s8_to_s64() here.
>
> Neither do I.
>
> > But at least the following code works (generates the same two-byte
> > machine code as ".byte 0x6a, vector" does):
> >
> >         .if vector < 128
> >         pushq $(vector)
> >         .else
> >         pushq $(0xffffffffffffff00+vector)
> >         .endif
>
> Only slightly less ugly and needs as much commentry as the above.

Agree.

Just FYI, I tried this later, it can work.

#define S8_TO_S64(vector) ((vector>>7)*0xffffffffffffff00+vector)

Thanks
Lai