Re: HEAD UP: non-MPSAFE network drivers to be disabled (was: 8.0 network stack MPsafety goals (fwd))

From: Peter Wemm <peter_at_wemm.org>
Date: Tue, 27 May 2008 15:30:28 -0700
On Tue, May 27, 2008 at 8:19 AM, Bruce M. Simpson <bms_at_freebsd.org> wrote:
> Ed Schouten wrote:
>>>>
>>>> With the word `should' I meant that it would still be possible to
>>>> implement multiple line disciplines with the mpsafetty code. But I
>>>> really think line disciplines should go.
>>>>
>>>
>>> No, don't bother.  Remove the entire "abstraction" of linedisciplines.
>>>
>>
>> Your wish is my command.
>>
>
> Like Julian and Robert said, please don't break Netgraph.
>
> It still needs to hook into tty at a very low level; and the components such
> as Bluetooth which can use ttys for low level stream processing, are built
> on Netgraph.
>
> Both SLIP and PPP can be built using Netgraph.
>
> Other than that, line disciplines can go away.

I think line disciplines should go away as such.  It would be nice to
have a place to hook into (even linux has this), but the line
discipline concept is 30 years stale.

IMHO, priority #1 is give ttys the first class treatment they need,
then priority #2 is to put in some low level hooks for ng_tty or
whatever.  It would be hard to come up with a more inconvenient
interface than we have now.

As an example.. look at the CAN_BYPASS_L_RINT stuff.  If we had a way
to pass a block of characters upstream instead of a single character
at a time, then we could build this magic into the input routines
instead of the device drivers.

The bypass is still useful because it saves considerable cycles for
raw tty modes, eg: userland ppp, zmodem upload/downloads etc etc.  But
the logic for it shouldn't have to be scattered all over the tree due
to poor line discipline APIs.

-- 
Peter Wemm - peter_at_wemm.org; peter_at_FreeBSD.org; peter_at_yahoo-inc.com
"All of this is for nothing if we don't go to the stars" - JMS/B5
"If Java had true garbage collection, most programs would delete
themselves upon execution." -- Robert Sewell
Received on Tue May 27 2008 - 20:30:29 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:31 UTC