Re: [PATCH] dmaengine: qcom: bam_dma: Add LOCK and UNLOCK flag bit support

From: mdalam@codeaurora.org
To: Vinod Koul <vkoul@kernel.org>
Cc: corbet@lwn.net, agross@kernel.org, bjorn.andersson@linaro.org,
	dan.j.williams@intel.com, dmaengine@vger.kernel.org,
	linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-arm-msm@vger.kernel.org, sricharan@codeaurora.org
Subject: Re: [PATCH] dmaengine: qcom: bam_dma: Add LOCK and UNLOCK flag bit support
Date: Mon, 21 Dec 2020 23:03:42 +0530	[thread overview]
Message-ID: <efcc74bbdf36b4ddbf764eb6b4ed99f2@codeaurora.org> (raw)
In-Reply-To: <20201221092355.GA3323@vkoul-mobl>

On 2020-12-21 14:53, Vinod Koul wrote:
> Hello,
> 
> On 17-12-20, 20:07, Md Sadre Alam wrote:
>> This change will add support for LOCK & UNLOCK flag bit support
>> on CMD descriptor.
>> 
>> If DMA_PREP_LOCK flag passed in prep_slave_sg then requester of this
>> transaction wanted to lock the DMA controller for this transaction so
>> BAM driver should set LOCK bit for the HW descriptor.
>> 
>> If DMA_PREP_UNLOCK flag passed in prep_slave_sg then requester of this
>> transaction wanted to unlock the DMA controller.so BAM driver should 
>> set
>> UNLOCK bit for the HW descriptor.
> 
> Can you explain why would we need to first lock and then unlock..? How
> would this be used in real world.
> 
> I have read a bit of documentation but is unclear to me. Also should
> this be exposed as an API to users, sounds like internal to driver..?
> 

IPQ5018 SoC having only one Crypto Hardware Engine. This Crypto Hardware 
Engine
will be shared between A53 core & ubi32 core. There is two separate 
driver dedicated
to A53 core and ubi32 core. So to use Crypto Hardware Engine parallelly 
for encryption/description
we need bam locking mechanism. if one driver will submit the request for 
encryption/description
to Crypto then first it has to set LOCK flag bit on command descriptor 
so that other pipes will
get locked.

The Pipe Locking/Unlocking will be only on command-descriptor. Upon 
encountering a command descriptor
with LOCK bit set, The BAM will lock all other pipes not related to the 
current pipe group, and keep
handling the current pipe only until it sees the UNLOCK set then it will 
release all locked pipes.
locked pipe will not fetch new descriptors even if it got event/events 
adding more descriptors for
this pipe (locked pipe).

No need to expose as an API to user because its internal to driver, so 
while preparing command descriptor
just we have to update the LOCK/UNLOCK flag.

> 
>> 
>> Signed-off-by: Md Sadre Alam <mdalam@codeaurora.org>
>> ---
>>  Documentation/driver-api/dmaengine/provider.rst | 9 +++++++++
>>  drivers/dma/qcom/bam_dma.c                      | 9 +++++++++
>>  include/linux/dmaengine.h                       | 5 +++++
>>  3 files changed, 23 insertions(+)
>> 
>> diff --git a/Documentation/driver-api/dmaengine/provider.rst 
>> b/Documentation/driver-api/dmaengine/provider.rst
>> index ddb0a81..d7516e2 100644
>> --- a/Documentation/driver-api/dmaengine/provider.rst
>> +++ b/Documentation/driver-api/dmaengine/provider.rst
>> @@ -599,6 +599,15 @@ DMA_CTRL_REUSE
>>    - This flag is only supported if the channel reports the 
>> DMA_LOAD_EOT
>>      capability.
>> 
>> +- DMA_PREP_LOCK
>> +
>> +  - If set , the client driver tells DMA controller I am locking you 
>> for
>> +    this transcation.
>> +
>> +- DMA_PREP_UNLOCK
>> +
>> +  - If set, the client driver will tells DMA controller I am 
>> releasing the lock
>> +
>>  General Design Notes
>>  ====================
>> 
>> diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
>> index 4eeb8bb..cdbe395 100644
>> --- a/drivers/dma/qcom/bam_dma.c
>> +++ b/drivers/dma/qcom/bam_dma.c
>> @@ -58,6 +58,8 @@ struct bam_desc_hw {
>>  #define DESC_FLAG_EOB BIT(13)
>>  #define DESC_FLAG_NWD BIT(12)
>>  #define DESC_FLAG_CMD BIT(11)
>> +#define DESC_FLAG_LOCK BIT(10)
>> +#define DESC_FLAG_UNLOCK BIT(9)
>> 
>>  struct bam_async_desc {
>>  	struct virt_dma_desc vd;
>> @@ -644,6 +646,13 @@ static struct dma_async_tx_descriptor 
>> *bam_prep_slave_sg(struct dma_chan *chan,
>> 
>>  	/* fill in temporary descriptors */
>>  	desc = async_desc->desc;
>> +	if (flags & DMA_PREP_CMD) {
>> +		if (flags & DMA_PREP_LOCK)
>> +			desc->flags |= cpu_to_le16(DESC_FLAG_LOCK);
>> +		if (flags & DMA_PREP_UNLOCK)
>> +			desc->flags |= cpu_to_le16(DESC_FLAG_UNLOCK);
>> +	}
>> +
>>  	for_each_sg(sgl, sg, sg_len, i) {
>>  		unsigned int remainder = sg_dma_len(sg);
>>  		unsigned int curr_offset = 0;
>> diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
>> index dd357a7..79ccadb4 100644
>> --- a/include/linux/dmaengine.h
>> +++ b/include/linux/dmaengine.h
>> @@ -190,6 +190,9 @@ struct dma_interleaved_template {
>>   *  transaction is marked with DMA_PREP_REPEAT will cause the new 
>> transaction
>>   *  to never be processed and stay in the issued queue forever. The 
>> flag is
>>   *  ignored if the previous transaction is not a repeated 
>> transaction.
>> + * @DMA_PREP_LOCK: tell the driver that DMA HW engine going to be 
>> locked for this
>> + *  transaction , until not seen DMA_PREP_UNLOCK flag set.
>> + * @DMA_PREP_UNLOCK: tell the driver to unlock the DMA HW engine.
>>   */
>>  enum dma_ctrl_flags {
>>  	DMA_PREP_INTERRUPT = (1 << 0),
>> @@ -202,6 +205,8 @@ enum dma_ctrl_flags {
>>  	DMA_PREP_CMD = (1 << 7),
>>  	DMA_PREP_REPEAT = (1 << 8),
>>  	DMA_PREP_LOAD_EOT = (1 << 9),
>> +	DMA_PREP_LOCK = (1 << 10),
>> +	DMA_PREP_UNLOCK = (1 << 11),
>>  };
>> 
>>  /**
>> --
>> 2.7.4