[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <30b1aa51-2fac-16b8-8fba-e82a88e3219d@nxp.com>
Date: Wed, 1 Feb 2023 10:41:15 +0200
From: Iuliana Prodan <iuliana.prodan@....com>
To: Peng Fan <peng.fan@....nxp.com>,
"Iuliana Prodan (OSS)" <iuliana.prodan@....nxp.com>,
Bjorn Andersson <andersson@...nel.org>,
Mathieu Poirier <mathieu.poirier@...aro.org>,
Shawn Guo <shawnguo@...nel.org>,
Sascha Hauer <s.hauer@...gutronix.de>,
"S.J. Wang" <shengjiu.wang@....com>,
Fabio Estevam <festevam@...il.com>,
Daniel Baluta <daniel.baluta@....com>,
Paul Olaru <paul.olaru@....com>,
Laurentiu Mihalcea <laurentiu.mihalcea@....com>
Cc: linux-imx <linux-imx@....com>, linux-remoteproc@...r.kernel.org,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
Pengutronix Kernel Team <kernel@...gutronix.de>
Subject: Re: [PATCH v2] remoteproc: imx_dsp_rproc: add custom memory copy
implementation for i.MX DSP Cores
Hi Peng,
On 2/1/2023 2:27 AM, Peng Fan wrote:
>
>
> On 2/1/2023 1:04 AM, Iuliana Prodan (OSS) wrote:
>> From: Iuliana Prodan <iuliana.prodan@....com>
>>
>> The IRAM is part of the HiFi DSP.
>> According to hardware specification only 32-bits write are allowed
>> otherwise we get a Kernel panic.
>>
>> Therefore add a custom memory copy and memset functions to deal with
>> the above restriction.
>
> Which platform has this limitation? This driver has been landed for
> quite some time, is there any specific condition to trigger the issue?
>
> Regards,
> Peng.
>
Any platform with HiFi DSP.
As I explained in the previous version, until now, it was used in a
limited scenario and the firmware was correctly built to respect the
write restriction - having the IRAM sections size a multiple of 4bytes.
Now, I was trying a simple hello_world sample from Zephyr, compiled with
GCC and I crashed the Kernel trying to load it on the hifi4 DSP.
Thanks,
Iulia
Powered by blists - more mailing lists