linux-kernel - Re: [PATCH v2 10/22] media: rockchip: rga: prepare cmdbuf on streamon

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: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <0dbdad63-df04-4220-96b6-b6af7e6e5a5f@pengutronix.de>
Date: Wed, 7 Jan 2026 13:00:28 +0100
From: Sven Püschel <s.pueschel@...gutronix.de>
To: Nicolas Dufresne <nicolas@...fresne.ca>,
 Jacob Chen <jacob-chen@...wrt.com>,
 Ezequiel Garcia <ezequiel@...guardiasur.com.ar>,
 Mauro Carvalho Chehab <mchehab@...nel.org>, Heiko Stuebner
 <heiko@...ech.de>, Rob Herring <robh@...nel.org>,
 Krzysztof Kozlowski <krzk+dt@...nel.org>, Conor Dooley <conor+dt@...nel.org>
Cc: linux-media@...r.kernel.org, linux-rockchip@...ts.infradead.org,
 linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
 devicetree@...r.kernel.org, kernel@...gutronix.de
Subject: Re: [PATCH v2 10/22] media: rockchip: rga: prepare cmdbuf on streamon

Hi,

On 12/24/25 4:29 PM, Nicolas Dufresne wrote:
> Hi,
>
> Le mercredi 03 décembre 2025 à 16:52 +0100, Sven Püschel a écrit :
>> Prepare the command buffer on streamon to reuse it's contents instead of
>> completely writing it for every frame. Due to the stream settings being
>> fixed after a streamon we only need to replace the source and destination
>> addresses for each frame. This reduces the amount of CPU and memory
>> operations done in each frame.
> My speculation is that flushing to device is usually more expensive then this,

Yeah, it was an idea to maybe get a bit more performance. I didn't see a 
performance improvement after implementing it.

But I thought that it could be beneficial when the CPU is fully loaded 
(instead of my small test with an idle CPU). Therefore I've left it in.

Sincerely
     Sven

> but I didn't check closely if you have enabled the caches. That being said, this
> is good change. It highlight the inflexible nature of V4L2 imho, which in this
> case can be used to optimize.
>
> Reviewed-by: Nicolas Dufresne <nicolas.dufresne@...labora.com>
>
>> Signed-off-by: Sven Püschel <s.pueschel@...gutronix.de>
>> ---
>>   drivers/media/platform/rockchip/rga/rga-hw.c | 13 +++++++++----
>>   drivers/media/platform/rockchip/rga/rga.c    | 13 ++++++++++++-
>>   drivers/media/platform/rockchip/rga/rga.h    |  1 +
>>   3 files changed, 22 insertions(+), 5 deletions(-)
>>
>> diff --git a/drivers/media/platform/rockchip/rga/rga-hw.c b/drivers/media/platform/rockchip/rga/rga-hw.c
>> index 56a2558539bfb..8cdfe089fd636 100644
>> --- a/drivers/media/platform/rockchip/rga/rga-hw.c
>> +++ b/drivers/media/platform/rockchip/rga/rga-hw.c
>> @@ -408,8 +408,6 @@ static void rga_cmd_set(struct rga_ctx *ctx,
>>   {
>>   	struct rockchip_rga *rga = ctx->rga;
>>   
>> -	memset(ctx->cmdbuf_virt, 0, RGA_CMDBUF_SIZE * 4);
>> -
>>   	rga_cmd_set_src_addr(ctx, src->dma_desc_pa);
>>   	/*
>>   	 * Due to hardware bug,
>> @@ -418,11 +416,9 @@ static void rga_cmd_set(struct rga_ctx *ctx,
>>   	rga_cmd_set_src1_addr(ctx, dst->dma_desc_pa);
>>   
>>   	rga_cmd_set_dst_addr(ctx, dst->dma_desc_pa);
>> -	rga_cmd_set_mode(ctx);
>>   
>>   	rga_cmd_set_src_info(ctx, &src->offset);
>>   	rga_cmd_set_dst_info(ctx, &dst->offset);
>> -	rga_cmd_set_trans_info(ctx);
>>   
>>   	rga_write(rga, RGA_CMD_BASE, ctx->cmdbuf_phy);
>>   
>> @@ -431,6 +427,14 @@ static void rga_cmd_set(struct rga_ctx *ctx,
>>   				   PAGE_SIZE, DMA_BIDIRECTIONAL);
>>   }
>>   
>> +static void rga_hw_setup_cmdbuf(struct rga_ctx *ctx)
>> +{
>> +	memset(ctx->cmdbuf_virt, 0, RGA_CMDBUF_SIZE * 4);
>> +
>> +	rga_cmd_set_mode(ctx);
>> +	rga_cmd_set_trans_info(ctx);
>> +}
>> +
>>   static void rga_hw_start(struct rockchip_rga *rga,
>>   			 struct rga_vb_buffer *src,  struct rga_vb_buffer *dst)
>>   {
>> @@ -622,6 +626,7 @@ const struct rga_hw rga2_hw = {
>>   	.max_height = MAX_HEIGHT,
>>   	.stride_alignment = 4,
>>   
>> +	.setup_cmdbuf = rga_hw_setup_cmdbuf,
>>   	.start = rga_hw_start,
>>   	.handle_irq = rga_handle_irq,
>>   	.get_version = rga_get_version,
>> diff --git a/drivers/media/platform/rockchip/rga/rga.c b/drivers/media/platform/rockchip/rga/rga.c
>> index 592c977a07cf3..f02ae02de26ca 100644
>> --- a/drivers/media/platform/rockchip/rga/rga.c
>> +++ b/drivers/media/platform/rockchip/rga/rga.c
>> @@ -523,6 +523,17 @@ static int vidioc_s_selection(struct file *file, void *priv,
>>   	return ret;
>>   }
>>   
>> +static int vidioc_streamon(struct file *file, void *priv,
>> +			   enum v4l2_buf_type type)
>> +{
>> +	struct rga_ctx *ctx = file_to_rga_ctx(file);
>> +	const struct rga_hw *hw = ctx->rga->hw;
>> +
>> +	hw->setup_cmdbuf(ctx);
>> +
>> +	return v4l2_m2m_streamon(file, ctx->fh.m2m_ctx, type);
>> +}
>> +
>>   static const struct v4l2_ioctl_ops rga_ioctl_ops = {
>>   	.vidioc_querycap = vidioc_querycap,
>>   
>> @@ -547,7 +558,7 @@ static const struct v4l2_ioctl_ops rga_ioctl_ops = {
>>   	.vidioc_subscribe_event = v4l2_ctrl_subscribe_event,
>>   	.vidioc_unsubscribe_event = v4l2_event_unsubscribe,
>>   
>> -	.vidioc_streamon = v4l2_m2m_ioctl_streamon,
>> +	.vidioc_streamon = vidioc_streamon,
>>   	.vidioc_streamoff = v4l2_m2m_ioctl_streamoff,
>>   
>>   	.vidioc_g_selection = vidioc_g_selection,
>> diff --git a/drivers/media/platform/rockchip/rga/rga.h b/drivers/media/platform/rockchip/rga/rga.h
>> index 0aef348dddb95..93162b118d069 100644
>> --- a/drivers/media/platform/rockchip/rga/rga.h
>> +++ b/drivers/media/platform/rockchip/rga/rga.h
>> @@ -154,6 +154,7 @@ struct rga_hw {
>>   	u32 max_width, max_height;
>>   	u8 stride_alignment;
>>   
>> +	void (*setup_cmdbuf)(struct rga_ctx *ctx);
>>   	void (*start)(struct rockchip_rga *rga,
>>   		      struct rga_vb_buffer *src, struct rga_vb_buffer *dst);
>>   	bool (*handle_irq)(struct rockchip_rga *rga);