Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 02 Aug 2018 20:50:16 +0000
From:      bugzilla-noreply@freebsd.org
To:        bugs@FreeBSD.org
Subject:   [Bug 230304] Malloc while lock is held in crypto code
Message-ID:  <bug-230304-227-Hy2liL4r9X@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-230304-227@https.bugs.freebsd.org/bugzilla/>
References:  <bug-230304-227@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=3D230304

Conrad Meyer <cem@freebsd.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|New                         |Open
           Assignee|bugs@FreeBSD.org            |cem@freebsd.org

--- Comment #1 from Conrad Meyer <cem@freebsd.org> ---
The issue is that crypto_newsession() holds the normal mutex ("Crypto
driver table" / CRYPTO_DRIVER_LOCK()) over the CRYPTODEV_NEWSESSION()
call, and the cryptosoft driver allocates extra memory with M_WAITOK.

The short term solution is to change crypto driver allocations back to
M_NOWAIT unconditionally and be sure to check all error paths for bugs
and memory leaks.

But I don't think that's a reasonable long term approach.  I think we
probably want to pass a flags argument from crypto_newsession all the
way down into individual driver newsessions to fix this right.  That
will take a decent amount of churn.  crypto_newsession() also needs to
be fixed to avoid deadlock when the caller allows M_WAITOK by dropping
the lock over the call.

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-230304-227-Hy2liL4r9X>