Received-SPF: pass (google.com: domain of linux-kernel+bounces-127204-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <e172edd3-824b-4ee4-b650-be1a8b9286fd@ti.com>
Date: Mon, 1 Apr 2024 18:50:37 -0500
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH 13/13] mailbox: omap: Remove kernel FIFO message queuing
To: Hari Nagalla <hnagalla@ti.com>, Jassi Brar <jassisinghbrar@gmail.com>,
        Nick Saulnier <nsaulnier@ti.com>,
        Bjorn Andersson <andersson@kernel.org>,
        Mathieu Poirier <mathieu.poirier@linaro.org>
CC: <linux-remoteproc@vger.kernel.org>, <linux-kernel@vger.kernel.org>
References: <20240325172045.113047-1-afd@ti.com>
 <20240325172045.113047-14-afd@ti.com>
 <761aa56f-55c4-e0d4-9f75-eef8035aa25b@ti.com>
Content-Language: en-US
From: Andrew Davis <afd@ti.com>
In-Reply-To: <761aa56f-55c4-e0d4-9f75-eef8035aa25b@ti.com>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit

On 4/1/24 6:39 PM, Hari Nagalla wrote:
> On 3/25/24 12:20, Andrew Davis wrote:
>> The kernel FIFO queue has a couple issues. The biggest issue is that
>> it causes extra latency in a path that can be used in real-time tasks,
>> such as communication with real-time remote processors.
>>
>> The whole FIFO idea itself looks to be a leftover from before the
>> unified mailbox framework. The current mailbox framework expects
>> mbox_chan_received_data() to be called with data immediately as it
>> arrives. Remove the FIFO and pass the messages to the mailbox
>> framework directly.
> Yes, this would definitely speed up the message receive path. With RT linux, the irq runs in thread context, so that is Ok. But with non-RT the whole receive path runs in interrupt context. So, i think it would be appropriate to use a threaded_irq()?

I was thinking the same at first, but seems some mailbox drivers use threaded, others
use non-threaded context. Since all we do in the IRQ context anymore is call
mbox_chan_received_data(), which is supposed to be IRQ safe, then it should be fine
either way. So for now I just kept this using the regular IRQ context as before.

If that does turn out to be an issue then let's switch to threaded.

Andrew