kern/72041: Deadlock when disk is destroyed while user process closes

[Poul-Henning Kamp]

The following reply was made to PR kern/72041; it has been noted by GNATS.

From: "Poul-Henning Kamp" <phk at phk.freebsd.dk>
To: Brian Eng <brian at midstream.com>
Cc: freebsd-gnats-submit at freebsd.org
Subject: Re: kern/72041: Deadlock when disk is destroyed while user process closes 
Date: Thu, 23 Sep 2004 20:48:58 +0200

 In message <200409231827.i8NIR3TK071354 at www.freebsd.org>, Brian Eng writes:
 
 >One perspective on this is that cambio inverted the layers; normally,
 >geom code calls cam code, but in the 'camcontrol rescan' case, cam
 >code calls geom code, resulting in locks being taken in opposite
 >order.  Perhaps disk_destroy could just queue to g_event and not
 >wait for completion.
 
 A lot of the trouble in this area comes from the fact that the disk_*()
 api has to isolate giant free code from giant infected code.
 
 There are no easy fixes to currently remaining problems, they will take
 serious work to fix.
 
 I know there is some work on CAM going on behind the scenes and hopefully
 they will take this sort of thing into account.
 
 -- 
 Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
 phk at FreeBSD.ORG         | TCP/IP since RFC 956
 FreeBSD committer       | BSD since 4.3-tahoe    
 Never attribute to malice what can adequately be explained by incompetence.