Re: [PATCH] usb: host: ohci-at91: suspend/resume ports after/before OHCI accesses

From: <Claudiu.Beznea@microchip.com>
To: <stern@rowland.harvard.edu>
Cc: <gregkh@linuxfoundation.org>, <Nicolas.Ferre@microchip.com>,
	<alexandre.belloni@bootlin.com>,
	<Ludovic.Desroches@microchip.com>,
	<Cristian.Birsan@microchip.com>, <linux-usb@vger.kernel.org>,
	<linux-arm-kernel@lists.infradead.org>,
	<linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] usb: host: ohci-at91: suspend/resume ports after/before OHCI accesses
Date: Thu, 24 Jun 2021 06:40:25 +0000	[thread overview]
Message-ID: <f03ccb09-4b5e-4db7-2cf0-375d53234099@microchip.com> (raw)
In-Reply-To: <20210623164148.GC499969@rowland.harvard.edu>

On 23.06.2021 19:41, Alan Stern wrote:
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
> 
> On Wed, Jun 23, 2021 at 02:33:14PM +0000, Claudiu.Beznea@microchip.com wrote:
>> On 23.06.2021 17:19, Alan Stern wrote:
>>> EXTERNAL EMAIL: Do not click links or open attachments unless you know the content is safe
>>>
>>> On Wed, Jun 23, 2021 at 02:09:16PM +0000, Claudiu.Beznea@microchip.com wrote:
>>>> On 23.06.2021 16:59, Alan Stern wrote:
>>>>> One thing you might consider changing: The name of the
>>>>> ohci_at91_port_suspend routine is misleading.  It doesn't really
>>>>> handle suspending the port; instead it handles the clocks that drive
>>>>> the entire OHCI controller.  Right?
>>>>
>>>> It does both as far as I can tell at the moment.
>>>
>>> But the name suggests that it only handles suspending a port.  That's
>>> misleading.
>>>
>>> And the way it is used in the SetPortFeature(USB_PORT_FEAT_SUSPEND)
>>> case in ohci_at91_hub_control is just plain wrong.  It won't merely
>>> suspend a single port; it will disable the entire OHCI controller.
>>
>> Agree with all the above!
> 
> Are there any systems beside the SAMA7G5 and others you tested which
> might be affected by this patch?  Do they all work pretty much the
> same way?  (I want to make sure no others will be adversely affected
> by this change.)

I tested it on SAMA7G5, SAMA5D2 and SAM9X60. I tested the suspend/resume
to/from mem. On SAMA5D2 and SAM9X60 there is no clock provided by
transceiver A to OHCI. I encountered no issues on tested systems. These IPs
are also present on SAMA5D3 and SAMA5D4 systems which I haven't tested as I
expect to behave as SAMA5D2 (as the clocking scheme is the same with
SAMA5D2). I can also try it on a SAMA5D3 (I don't have a SAMA5D4 with me at
the moment), tough, just to be sure nothing is broken there too.

Thank you,
Claudiu Beznea

> 
> Alan Stern
> 