linux-kernel - Re: [PATCH 1/2] dmaengine: sprd: delete enable opreation in probe

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date: Tue, 19 Dec 2023 15:00:24 +0800
From: Baolin Wang <baolin.wang@...ux.alibaba.com>
To: liu kaiwei <liukaiwei086@...il.com>, Vinod Koul <vkoul@...nel.org>
Cc: Kaiwei Liu <kaiwei.liu@...soc.com>, Orson Zhai <orsonzhai@...il.com>,
 Chunyan Zhang <zhang.lyra@...il.com>, dmaengine@...r.kernel.org,
 linux-kernel@...r.kernel.org, Wenming Wu <wenming.wu@...soc.com>
Subject: Re: [PATCH 1/2] dmaengine: sprd: delete enable opreation in probe



On 12/19/2023 1:21 PM, liu kaiwei wrote:
> On Mon, Dec 11, 2023 at 7:41 PM Vinod Koul <vkoul@...nel.org> wrote:
>>
>> On 06-12-23, 17:32, liu kaiwei wrote:
>>> On Fri, Nov 24, 2023 at 9:11 PM Vinod Koul <vkoul@...nel.org> wrote:
>>>>
>>>> On 02-11-23, 20:16, Kaiwei Liu wrote:
>>>>> From: "kaiwei.liu" <kaiwei.liu@...soc.com>
>>>>
>>>> Typo is subject line
>>>>
>>>>>
>>>>> In the probe of dma, it will allocate device memory and do some
>>>>> initalization settings. All operations are only at the software
>>>>> level and don't need the DMA hardware power on. It doesn't need
>>>>> to resume the device and set the device active as well. here
>>>>> delete unnecessary operation.
>>>>
>>>> Don't you need to read or write to the device? Without enable that wont
>>>> work right?
>>>>
>>>
>>> Yes, it doesn't need to read or write to the device in the probe of DMA.
>>> We will enable the DMA when allocating the DMA channel.
>>
>> So you will probe even if device is not present! I think it makes sense
>> to access device registers in probe!
> 
> There is another reason why we delete enable/disable and not to access
> device in probe. The current driver is applicable to two DMA devices
> in different
> power domain. For some scenes, one of the domain is power off and when you
> probe,  enable the dma with the domain power off may cause crash.
> 
> For example, one case is for audio co-processor and DMA serves for it,
> DMA's power domain is off during initialization since audio is not used
> at that time, so we cannot read/write DMA's register for this kind of cases.
> 
> @Baolin Wang
> Hi baolin，what's your opinion?

Please add your explanation into the commit message.

Moreover, I think Vinod's concern is reasonable, so you can not move the 
pm_runtime_enable() after registering the devices, which means users can 
access the device without powering on.

To solve your problem, I think you can move the pm_runtime_enable() 
before dma_async_device_register(), then if users want to use DMA in 
probe stage, the dma_chan_get()--->sprd_dma_alloc_chan_resources() will 
help to power on it.