Subject: Re: change cgd(4) to use bufpool ?
To: Darrin B. Jewell <>
From: Roland C. Dowdeswell <>
List: tech-kern
Date: 07/18/2004 18:05:31
On 1090179095 seconds since the Beginning of the UNIX epoch
"Darrin B. Jewell" wrote:

>Is there a good reason that the cgd(4) device uses its
>own buffer structure and pool?  The following patch allows
>it to use the regular bufpool by taking advantage of the
>bp->b_private field to store the old buffer.

Hmmm, no not really.  Although, we really should have some way of
ensuring that we can obtain at least one buf per cgd(4) at some
point in the future to prevent possible deadlocks.

In cgdiodone, now one calls getcgd_softc() which can fail.  It
might be nice to put a DIAGNOSTIC in there to check it, although
not necessary.

    Roland Dowdeswell                      http://www.Imrryr.ORG/~elric/