From owner-freebsd-hackers  Wed Apr  2 16:23:52 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id QAA09496
          for hackers-outgoing; Wed, 2 Apr 1997 16:23:52 -0800 (PST)
Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50])
          by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id QAA09491
          for <freebsd-hackers@freebsd.org>; Wed, 2 Apr 1997 16:23:48 -0800 (PST)
Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id RAA15015; Wed, 2 Apr 1997 17:06:49 -0700
From: Terry Lambert <terry@lambert.org>
Message-Id: <199704030006.RAA15015@phaeton.artisoft.com>
Subject: Re: VOP_PUTPAGES
To: pmchen@eecs.umich.edu (Peter M. Chen)
Date: Wed, 2 Apr 1997 17:06:48 -0700 (MST)
Cc: freebsd-hackers@freebsd.org
In-Reply-To: <199704022320.SAA19709@life.eecs.umich.edu> from "Peter M. Chen" at Apr 2, 97 06:20:42 pm
X-Mailer: ELM [version 2.4 PL24]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

> I feel stupid for asking this, but what does VOP_PUTPAGES call?  I tried
> tracing through the maze of macros but ended up at vp->v_op[0], which I
> think is an error function.
> 
> VOP_PUTPAGES (kern/vnode_if.h) calls
>     VCALL(vp, VOFFSET(vop_putpages), &a)

yes.

>     which becomes
>     VOCALL(vp->v_op, VOFFSET(vop_putpages),&a)

yes.

>     which becomes
>     ( *(vp->v_op[VOFFSET(vop_putpages)])) (&a)

yes.  And VOFFSET(vop_putpages) is at offset 43 in vfs_op_descs[].

>     which becomes
>     ( *(vp->v_op[((&vop_putpages_desc)->vdesc_offset) ])) (&a)

yes.

>     which becomes
>     ( *(vp->v_op[0])) (&a)

BZZZZZZZZZZTTTT.  For FS's where the op is supported, it's the offset
of the per FS specific *_putpage.  For most FS's, it's not supported,
but getpages is -- see ffs_getpages() routine in the source file
/sys/ufs/ufs/ufs_readwrite.c

For those where it is not, a "failed" call will be called.

I don't believe the code is ever really called anywhere -- instead,
you should grep for "putpages" in /sys/vm/*.c, and look around there
(mostly "default_pager.c", device_pager.c", "swap_pager.c", and
"vn_pager.c").  These are called explicity instead of the VOP's.

One of the most useful places to look would be in the mmap() code,
since that is where the VOP_PUTPAGES() would be most easily
implemented (the most interesting code in that case is the vn_pager.c).


In general, the getpages() is only used for paging in pages from a file
being used as a swap backing store for an executable -- in other words,
*clean* pages.

The putpages() routine is for writing dirty pages out... and is not
currently used.

For more details, you should see the posting in -current about John
Dyson's modifications to add the op's into the list of VOPS supported
as "standard", about 6 moths to a year ago.  There is every intent
to actually use the putpages VOP -- eventually.


					Regards,
					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.