From: Alexander Aring <alex.aring@gmail.com>
To: Marcel Holtmann <marcel@holtmann.org>
Cc: linux-bluetooth <linux-bluetooth@vger.kernel.org>,
kernel@pengutronix.de,
Jukka Rissanen <jukka.rissanen@linux.intel.com>
Subject: Re: [RFC bluetooth-next 1/2] bluetooth: 6lowpan: avoid endless loop
Date: Sun, 25 Oct 2015 21:27:21 +0100 [thread overview]
Message-ID: <20151025202716.GA6435@omega> (raw)
In-Reply-To: <D10520FE-A8A8-49D9-B6BA-E6682CBCFB42@holtmann.org>
Hi Marcel,
On Mon, Oct 26, 2015 at 05:11:37AM +0900, Marcel Holtmann wrote:
> Hi Alex,
>
> > When -EAGAIN as return value for receive handling will do a retry of
> > parsing I can trigger a endless loop when iphc decompression e.g.
> > returns an errno because some missing function "-ENOTSUPP" or something
> > else. Somebody from outside can trigger an endless loop when sending a
> > an IPHC header which triggers this behaviour.
> >
> > NOTE: This really depends only if -EAGAIN means "try again to call the
> > receive handler with the skb". Sometimes we also drop (and kfree) the
> > skb, I think something is broken there... depends on the error branch.
> > When receiving failed simple free skb and return errno (which is not
> > -EAGAIN).
>
> I am lost on this comment, you need to explain this more and might actually want to put a comment in the code on this.
>
I am sorry, what I meant there is "How exactly does the '.recv' of
'l2cap_ops'" handle an -EAGAIN errno. If this means the callback will be
called again with the same parameters (What I am thinking of, when I
return -EAGAIN) then we have an endless loop which can be triggered from
the outside by sending an invalid/not supported IPHC 6LoWPAN header.
Because the function "chan_recv_cb" (which is ".recv") will call
recv_pkt (which return -EAGAIN on failure) and this function will call
ret = iphc_decompress(local_skb, dev, chan), which parse the IPHC Header
(can occur errno on invalid frames), then the complete calling chain
will return -EAGAIN at ".recv" callback of l2cap_ops". The IPHC header
comming from the "outside", everybody can manipulate this data.
Is it more clear now what I try to explained here?
First look on ".recv" callback handling in "net/bluetooth/l2cap_core.c"
a return of -EAGAIN, will not try to call the ".recv" callback with the
same parameters again.
Anyway, we should not use -EAGAIN here, or? Simple return the errno from
function which returns it.
- Alex
next prev parent reply other threads:[~2015-10-25 20:27 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-10-24 14:50 [RFC bluetooth-next 0/2] bluetooth: 6lowpan: suggestions for Jukka Alexander Aring
2015-10-24 14:50 ` [RFC bluetooth-next 1/2] bluetooth: 6lowpan: avoid endless loop Alexander Aring
2015-10-25 20:11 ` Marcel Holtmann
2015-10-25 20:27 ` Alexander Aring [this message]
2015-10-27 9:21 ` Jukka Rissanen
2015-10-24 14:50 ` [RFC bluetooth-next 2/2] bluetooth: 6lowpan: rework receive handling Alexander Aring
2015-10-28 8:57 ` Jukka Rissanen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20151025202716.GA6435@omega \
--to=alex.aring@gmail.com \
--cc=jukka.rissanen@linux.intel.com \
--cc=kernel@pengutronix.de \
--cc=linux-bluetooth@vger.kernel.org \
--cc=marcel@holtmann.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).