DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 7ECE3208E4
Content-Type: text/plain;
        charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 12.1 \(3445.101.1\))
Subject: Re: [PATCH 1/1 v2] Bluetooth: make the balance of judgement condition
 to fix a false report
From:   Johan Hedberg <johan.hedberg@gmail.com>
In-Reply-To: <1704c39f-c005-d0cc-9bab-c21e537a07be@gmail.com>
Date:   Tue, 27 Nov 2018 23:19:14 +0200
Cc:     linux-bluetooth@vger.kernel.org
Content-Transfer-Encoding: quoted-printable
Message-Id: <566C2A3E-8924-455C-93B1-B514A4CB0511@gmail.com>
References: <5A23C871-B30F-44C4-898B-C138EFDB4893@holtmann.org>
 <1542245517-150977-1-git-send-email-zumeng.chen@gmail.com>
 <00BD5234-6645-4E40-9B3C-859C147EEB7C@gmail.com>
 <3D63F92F-1BA5-415D-AED4-A38FB2E29C00@gmail.com>
 <1704c39f-c005-d0cc-9bab-c21e537a07be@gmail.com>
To:     Zumeng Chen <zumeng.chen@gmail.com>
Sender: linux-bluetooth-owner@vger.kernel.org
Precedence: bulk

Hi Zumeng,


> On 27 Nov 2018, at 14.22, Zumeng Chen <zumeng.chen@gmail.com> wrote:
> There is a related kernel as shown in the below list
>=20
> https://bugzilla.kernel.org/show_bug.cgi?id=3D198699
>=20
> I didn't think my patch is to fix this one because I think we need to =
sort out
> the whole logic of CMD_COMPLETE. But my patch is good to fix the issue
> described by subject.

I=E2=80=99m still failing to see any major problem with command complete =
handling. This seems more related to sending HCI commands that do not =
complete in Command Complete (such as Inquiry or connection creation) =
through the hci_request framework together with hci_req_run_skb(). This =
could either happen by directly using the hci_req_sync() API, such as in =
the case of the legacy inquiry ioctl, or by some code using =
hci_cmd_sync() to send a command that does not complete in Command =
Complete. In these cases you end up with a Command Status with status =3D=3D=
 0, which the hci_request code interprets as completion of the request, =
but is unable to fetch any return parameters from it (which for these =
cases is fine).

Since what I've describe above is normal usage of the hci_req_sync() and =
hci_cmd_sync() APIs I still don=E2=80=99t see any other issue except =
that an error log was thrown rather than than (at most) a debug log. So =
I=E2=80=99d still go for the patch that I submitted earlier today.

Johan=