Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220915164730.515767-1-e.shatokhin@yadro.com> <20220915164730.515767-2-e.shatokhin@yadro.com>
In-Reply-To: <20220915164730.515767-2-e.shatokhin@yadro.com>
From:   Jassi Brar <jassisinghbrar@gmail.com>
Date:   Fri, 16 Sep 2022 10:57:44 -0500
Message-ID: <CABb+yY06VZB8mk7S=N1LO=AcwyOH2nfuGo-eyxRdRx782jDB_Q@mail.gmail.com>
Subject: Re: [PATCH 1/2] mailbox: Propagate errors from .send_data() callback
 to mbox_send_message()
To:     Evgenii Shatokhin <e.shatokhin@yadro.com>
Cc:     linux-kernel@vger.kernel.org, Ilya Kuznetsov <ilya@yadro.com>,
        linux@yadro.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Sep 15, 2022 at 11:49 AM Evgenii Shatokhin
<e.shatokhin@yadro.com> wrote:
>
> msg_submit() calls .send_data() function from the mailbox controller driver
> to place the first of the queued messages to the mailbox. Depending on the
> actual driver used, this operation could fail.
>
> In this case, if mbox_send_message() is called in blocking mode, it will
> always return -ETIME rather than the actual error reported by the
> underlying driver. This could be confusing, so let us propagate the
> error from msg_submit() to mbox_send_message().
>
In blocking mode, the client gets -ETIME because the api waited long
enough for the transfer to be successful.
It is the job of the underlying controller driver to be in a
consistent state and recover from errors. It may print its own error
but the client shouldn't have to differentiate between failure causes.

thanks