Re: [PATCH net-next] net: lan966x: Improve the CPU TX bitrate.

['Horatiu Vultur']

The 03/08/2022 22:46, David Laight wrote:
> 
> From: Horatiu Vultur
> > Sent: 08 March 2022 22:30
> >
> > The 03/08/2022 22:36, Andrew Lunn wrote:
> > >
> > > >  static int lan966x_port_inj_ready(struct lan966x *lan966x, u8 grp)
> > > >  {
> > > > -     u32 val;
> > > > +     unsigned long time = jiffies + usecs_to_jiffies(READL_TIMEOUT_US);
> > > > +     int ret = 0;
> > > >
> > > > -     return readx_poll_timeout_atomic(lan966x_port_inj_status, lan966x, val,
> > > > -                                      QS_INJ_STATUS_FIFO_RDY_GET(val) & BIT(grp),
> > > > -                                      READL_SLEEP_US, READL_TIMEOUT_US);
> > > > +     while (!(lan_rd(lan966x, QS_INJ_STATUS) &
> > > > +              QS_INJ_STATUS_FIFO_RDY_SET(BIT(grp)))) {
> > > > +             if (time_after(jiffies, time)) {
> > > > +                     ret = -ETIMEDOUT;
> > > > +                     break;
> > > > +             }
> > >
> > > Did you try setting READL_SLEEP_US to 0? readx_poll_timeout_atomic()
> > > explicitly supports that.
> >
> > I have tried but it didn't improve. It was the same as before.
> 
> How many times round the loop is it going ?

In the tests that I have done, I have never seen entering in the loop.

> 
> It might be that by the time readx_poll_timeout_atomic()
> gets around to reading the register the fifo is actually ready.
> 
> The there is the delay between detecting 'ready' and writing
> the next data.
> That delay might be cumulative and affect performance.

There is also a small delay between writting the word and checking if
more words can be written.

> 
> OTOH spinning waiting for fifo space is just plain horrid.

Yes, I agree with you.

> Reminds me of 3C509 drivers :-)

> 
>         David
> 
> -
> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
> Registration No: 1397386 (Wales)

-- 
/Horatiu