Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 06 Jul 2000 02:36:33 -0700
From:      Mike Smith <msmith@freebsd.org>
To:        Gerhard Sittig <Gerhard.Sittig@gmx.net>
Cc:        freebsd-stable@FreeBSD.ORG
Subject:   Re: Dump takes ages on SCSI. (several times faster on IDE) 
Message-ID:  <200007060936.CAA01043@mass.osd.bsdi.com>
In-Reply-To: Your message of "Tue, 04 Jul 2000 19:16:30 %2B0200." <20000704191630.U5945@speedy.gsinet> 

next in thread | previous in thread | raw e-mail | index | archive | help
> On Tue, Jul 04, 2000 at 05:22 -0700, Mike Smith wrote:
> > > 
> > > The Problem with aha1542 (ISA busmaster capable controller)
> > > in PCI boards is that it is located "at the end" of a long
> > > chain to memory.  Chances are, busmastering will not work at
> > > all behind the PCI to ISA bridge since all the PCI devices
> > > have higher priority and the ISA gadget never gets any access
> > > granted or at least always too late.  You're better off with
> > > a PCI controller.
> > 
> > This is Just Not True.
> 
> Can you please tell me more about it?  I'm not sure of the
> priority thing I mention above, but the fact that ISA busmaster
> devices and PCI boards somewhat collide or counter operate
> (sorry, lacking better words once more) is what I got from
> serious papers here.

Even this isn't the case.  The critical issue is simply that the ISA 
busmaster has to operate correctly in the presence of other masters.  The 
ISA bridge is responsible for translating the master operations from the 
ISA bus to the PCI bus.

In your previous statement, you allude to "priority" wrt. PCI busmaster 
DMA - there is no such thing.  As far as a busmaster device like the 1542 
is concerned, there is no such thing as "too late" for DMA to/from memory.

> How direct is the access method an ISA
> controller has to the memory when sitting behind a bridge?  I
> would be glad to believe a 1542B/C in a PCI machine is A Good
> Thing(TM).  :)

It's not a "good thing" - it's a slow controller behind two bridges, and 
it's not going to perform _well_, but modulo broken hardware it _will_ 
work correctly.

You're quite correct when you say that you're better off with a PCI 
controller, however.

-- 
... every activity meets with opposition, everyone who acts has his
rivals and unfortunately opponents also.  But not because people want
to be opponents, rather because the tasks and relationships force
people to take different points of view.  [Dr. Fritz Todt]




To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200007060936.CAA01043>