Re: kern/38889

[Masao Uebayashi <uebayasi%tombi.co.jp@localhost>]

The following reply was made to PR kern/38889; it has been noted by GNATS.

From: Masao Uebayashi <uebayasi%tombi.co.jp@localhost>
To: YAMAMOTO Takashi <yamt%mwd.biglobe.ne.jp@localhost>
Cc: netbsd-bugs%NetBSD.org@localhost, gnats-bugs%NetBSD.org@localhost
Subject: Re: kern/38889
Date: Fri, 25 May 2012 08:57:48 +0900

 On Thu, May 24, 2012 at 03:00:02AM +0000, YAMAMOTO Takashi wrote:
 > hi,
 > 
 > > I wonder if we can completely disallow mmap'ing block devices?  I suspect
 > > it has never worked nor needed by anyone - otherwise why such a serious
 > > problem has been open for a long time!
 > 
 > i don't know.  but it might be useful for applications which want to
 > do mmap+send for zero copy.  eg. an iscsi target
 
 Understood, but that could be properly designed/implamented later.  I
 think it's better to just disallow mmap(bdev) on netbsd-6.
 
 (I also think that genfs is a wrong place to handle mmap(bdev).  IIUC, bdev's
 data has nothing to do with its hosting mount/filesystem.  Handling VREG and
 VBLK in one place sounds very odd to me.)
 
 > 
 > please keep gnats-bugs%NetBSD.org@localhost in Cc: so that it's filed in 
 > gnats.
 > 
 > YAMAMOTO Takashi