Re: (NEWBIE) Networking Problems

From: jpd (read_the_sig_at_do.not.spam.it)
Date: 04/14/04 

Date: Wed, 14 Apr 2004 05:39:21 +0000 (UTC)

On 2004-04-14, glen herrmannsfeldt <gah@ugcs.caltech.edu> wrote:
[snip]
>
> Couldn't it enable interrupts with interrupt vectors pointing
> to different places that would indicate which one it was, and
> then convince the card to interrupt, such as sending a packet
> and emptying the transmit buffer?

*crunch* *crunch* ehrm. Oh, I think I get what you want to do.
I still don't see the point in it, though. And usually you need
interrupts to efficiently *receive* packets, not transmit them.

But still, that doesn't help you with the case where you don't
know where the card is interrupting. You can't just send a packet
and hope somebody sends one back, monitoring all IRQs, then try
and correlate which one it is. Before you know it you'll end up
abusing the clock or keyboard or disk interrupt or so.

> Maybe drivers aren't allowed to do that during boot time, though.

At boot time you either have a preconfigured kernel or you tell the
kernel using the boot configurator where the apropriate irqs and
addresses are. 

-- 
  j p d (at) d s b (dot) t u d e l f t (dot) n l .

Relevant Pages

  • Re: IF_HANDOFF vs. IFQ_HANDOFF
    ... There should be a nearly-complete interrupt ... > an interrupt after every packet. ... interrupt mark on fragments in the middle of a packet. ... before sending frames in sk_start if number of available Tx descriptors ...
    (freebsd-net)
  • Re: RFC: issues concerning the next NAPI interface
    ... If NAPI disables interrupts and keeps them disabled while there are more packets arriving or more transmits being completed, why do hardware interrupt mitigation / coalescing features of the network silicon help? ... dominates the packet processing costs. ... A driver's poll() would set an internal flag and record the current jiffies value when finding workdone=0 rather than doing an immediate napi_complete. ...
    (Linux-Kernel)
  • Re: NIPS solutions
    ... > machines are used? ... each packet that is arriving at the ... > network interface triggers an hardware interrupt that is instantly ... the soft interrupt packet queue. ...
    (Focus-IDS)
  • Re: [PATCH 7/14] random: Remove SA_SAMPLE_RANDOM from network drivers
    ... The PHY of the network card has some tiny delays internal ... and if a free packet buffer is available, ... If the chip has hw interrupt mitigation enabled, ...
    (Linux-Kernel)
  • Re: very busy syslog server
    ... For the syslog daemon, I have chosen rsyslogd, and the backend is mysql. ... broadcast/multicast datagrams dropped due to no socket ->>> 123677 dropped due to full socket buffers 0 not for hashed pcb ... than one packet per interrupt. ...
    (freebsd-performance)