Re: 2.0.30 serial.c, ppp.c and pppd-2.2 questions
Rogier Wolff (R.E.Wolff@bitwizard.nl)
Thu, 24 Jul 1997 09:09:43 +0200 (MET DST)
Theodore Y. Ts'o wrote:
>
> Date: Wed, 23 Jul 1997 08:07:23 +0200 (MET DST)
> From: R.E.Wolff@BitWizard.nl (Rogier Wolff)
>
> Yes, I agree. At 38k4 you need 130 ms to fill the flip buffer. And if
> you see this kind of delays there is something "wrong". An easy
> solution like increasing the buffer size may delay the fenomenon to
> the point that it is hard to reproduce, but we haven't really fixed
> anything. In some cases not receiving any chars for 130 ms can be
> annoying too....
>
> And at 115kbaud you need 44ms to fill the flip buffer. So you'd have to
> miss 4 clock ticks before the flip buffer would be overflowing. I can
> believe we're missing 1, or maybe even 2, but 4?
I was naming the 130 ms numbers because that's what I'm seeing.
On a specialix card running at 38k4, one recieve interrupt per 6
chars, I'm getting those flip buffer overflows. Especially when
the machine starts swapping.
If you have any hints on how to debug this, I'd appreciate it.
Roger.