Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 31 Jan 2017 08:03:05 +0000
From:      bugzilla-noreply@freebsd.org
To:        freebsd-net@FreeBSD.org
Subject:   [Bug 204340] [panic] nfsd, em, msix, fatal trap 9
Message-ID:  <bug-204340-2472-vPOEobzSvJ@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-204340-2472@https.bugs.freebsd.org/bugzilla/>
References:  <bug-204340-2472@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D204340

--- Comment #22 from Andriy Gapon <avg@FreeBSD.org> ---
(In reply to Rick Macklem from comment #20)

Rick, thank you very much for the explanation!  I knew that nfsd processes =
were
special as they 'lend their stacks to kernel' or something like that.  But I
failed to realise that that put restrictions in the signals as well.

I should also explain that kill -9 was not used to shutdown nfsd or as a
replacement for the normal nfsd management.  It was used just to demonstrate
the problem.

I think that originally the problem happened when gdb was used on an nfsd
process.

I understand that the nfsd processes are special.  But the situation seems =
to
be a bit fragile.  The current design is old and proven.  But perhaps we co=
uld
switch to using kernel processes or maybe we could mark the nfsd processes =
with
a special flag somehow as to prevent them being killed SIGKILL or stopped w=
ith
SIGSTOP (i.e. prevent normal signal delivery for all signals).

Lastly, just to clarify, should we avoid using debuggers / SIGSTOP with nfs=
d?

--=20
You are receiving this mail because:
You are on the CC list for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-204340-2472-vPOEobzSvJ>