Date: Sat, 13 Jun 2015 17:46:03 -0500 From: "Matthew D. Fuller" <fullermd@over-yonder.net> To: =?iso-8859-1?Q?Jos=E9_Garc=EDa?= Juanino <jjuanino@gmail.com> Cc: freebsd-geom@freebsd.org Subject: Re: geli panics my system after suspend-resume: g_eli_orphan_spoil_assert() called for cd0.eli Message-ID: <20150613224602.GB1733@over-yonder.net> In-Reply-To: <CAAVO5%2BLR9MfHvt0tR5G4FyXKRt_hzkR3ZKUxvEA8Z%2BjWuEFq3g@mail.gmail.com> References: <CAAVO5%2BLR9MfHvt0tR5G4FyXKRt_hzkR3ZKUxvEA8Z%2BjWuEFq3g@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
[ Redirect to -geom, bcc'd to -current for transition ]
On Sat, Jun 13, 2015 at 04:10:38PM +0200 I heard the voice of
José García Juanino, and lo! it spake thus:
> Hi FreeBSD current,
>
> I get a reproducible panic following these steps:
>
> 1- Mount a geli encrypted DVD:
[...]
> 4- Open again the lid a wait. The system resumes, but panics after a
> few seconds:
>
> panic: Function g_eli_orphan_spoil_assert() called for cd0.eli.
This is presumably what's hooked up in g_eli_create():
/*
* Spoiling cannot happen actually, because we keep provider open for
* writing all the time or provider is read-only.
*/
gp->spoiled = g_eli_orphan_spoil_assert;
The backtrace makes me think that coming out of resume will cause a
media change event on the cd0 device, which sounds a _little_ odd, but
may be necessary due to Stupid Reality Crap. That pushes a spoil
event down to g_eli. And g_eli seems to assume that can't happen to
a thing it's got around an open.
With the caveat that I'm no expert on ELI or GEOM, and any advice I
might give could just be the trigger for a nasal demon invasion of our
world, one thing you might try is taking that event as a hint to just
blow away the .eli device; treat it like an orphan. So changing that
callback to using the g_eli_orphan func instead. e.g., something like
Index: g_eli.c
===================================================================
--- g_eli.c (revision 284359)
+++ g_eli.c (working copy)
@@ -730,7 +730,7 @@ g_eli_create(struct gctl_req *req, struct g_class
* Spoiling cannot happen actually, because we keep provider open for
* writing all the time or provider is read-only.
*/
- gp->spoiled = g_eli_orphan_spoil_assert;
+ gp->spoiled = g_eli_orphan;
gp->orphan = g_eli_orphan;
gp->dumpconf = g_eli_dumpconf;
/*
(not even remotely tested; beware of leopard)
--
Matthew Fuller (MF4839) | fullermd@over-yonder.net
Systems/Network Administrator | http://www.over-yonder.net/~fullermd/
On the Internet, nobody can hear you scream.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150613224602.GB1733>