[PATCH] remoteproc: mediatek: enable cache for mt8186 SCP

[PATCH] remoteproc: mediatek: enable cache for mt8186 SCP

[Allen-KH Cheng]

1. Set SCP cache size before loading SCP FW. (8KB+8KB)
2. Adjust ipi_buf_offset from 0x7bdb0 to 0x3BDB0 for enableing cache

SCP side
 - IPI Buffer: 0x3BDB0 <-> 0x3C000
 - Cache: 0x3C000 <-> 0x40000

Signed-off-by: Allen-KH Cheng <allen-kh.cheng@mediatek.com>
---
 drivers/remoteproc/mtk_scp.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/remoteproc/mtk_scp.c b/drivers/remoteproc/mtk_scp.c
index 38609153bf64..24065b6b4da8 100644
--- a/drivers/remoteproc/mtk_scp.c
+++ b/drivers/remoteproc/mtk_scp.c
@@ -401,6 +401,14 @@ static int mt8186_scp_before_load(struct mtk_scp *scp)
 	writel(0x0, scp->reg_base + MT8186_SCP_L1_SRAM_PD_P1);
 	writel(0x0, scp->reg_base + MT8186_SCP_L1_SRAM_PD_p2);
 
+	/*
+	 * Set I-cache and D-cache size before loading SCP FW.
+	 * SCP SRAM logical address may change when cache size setting differs.
+	 */
+	writel(MT8183_SCP_CACHE_CON_WAYEN | MT8183_SCP_CACHESIZE_8KB,
+	       scp->reg_base + MT8183_SCP_CACHE_CON);
+	writel(MT8183_SCP_CACHESIZE_8KB, scp->reg_base + MT8183_SCP_DCACHE_CON);
+
 	return 0;
 }
 
@@ -905,7 +913,7 @@ static const struct mtk_scp_of_data mt8186_of_data = {
 	.scp_da_to_va = mt8183_scp_da_to_va,
 	.host_to_scp_reg = MT8183_HOST_TO_SCP,
 	.host_to_scp_int_bit = MT8183_HOST_IPC_INT_BIT,
-	.ipi_buf_offset = 0x7bdb0,
+	.ipi_buf_offset = 0x3bdb0,
 };
 
 static const struct mtk_scp_of_data mt8192_of_data = {
-- 
2.18.0

Re: [PATCH] remoteproc: mediatek: enable cache for mt8186 SCP

[Mathieu Poirier]

Hi Allen,

On Tue, Mar 22, 2022 at 08:28:45PM +0800, Allen-KH Cheng wrote:
> 1. Set SCP cache size before loading SCP FW. (8KB+8KB)
> 2. Adjust ipi_buf_offset from 0x7bdb0 to 0x3BDB0 for enableing cache
> 
> SCP side
>  - IPI Buffer: 0x3BDB0 <-> 0x3C000
>  - Cache: 0x3C000 <-> 0x40000
>

I would also like to find in this changelog "why" this patch is needed and more
importantly, how it remains compatible with existing implementation.  I am
mostly worried about implemenations where the application processor is using
caches while the SCP is not.

I will also need a couple of "Tested-by" tags before moving forward with this
patch.

Thanks,
Mathieu

> Signed-off-by: Allen-KH Cheng <allen-kh.cheng@mediatek.com>
> ---
>  drivers/remoteproc/mtk_scp.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/remoteproc/mtk_scp.c b/drivers/remoteproc/mtk_scp.c
> index 38609153bf64..24065b6b4da8 100644
> --- a/drivers/remoteproc/mtk_scp.c
> +++ b/drivers/remoteproc/mtk_scp.c
> @@ -401,6 +401,14 @@ static int mt8186_scp_before_load(struct mtk_scp *scp)
>  	writel(0x0, scp->reg_base + MT8186_SCP_L1_SRAM_PD_P1);
>  	writel(0x0, scp->reg_base + MT8186_SCP_L1_SRAM_PD_p2);
>  
> +	/*
> +	 * Set I-cache and D-cache size before loading SCP FW.
> +	 * SCP SRAM logical address may change when cache size setting differs.
> +	 */
> +	writel(MT8183_SCP_CACHE_CON_WAYEN | MT8183_SCP_CACHESIZE_8KB,
> +	       scp->reg_base + MT8183_SCP_CACHE_CON);
> +	writel(MT8183_SCP_CACHESIZE_8KB, scp->reg_base + MT8183_SCP_DCACHE_CON);
> +
>  	return 0;
>  }
>  
> @@ -905,7 +913,7 @@ static const struct mtk_scp_of_data mt8186_of_data = {
>  	.scp_da_to_va = mt8183_scp_da_to_va,
>  	.host_to_scp_reg = MT8183_HOST_TO_SCP,
>  	.host_to_scp_int_bit = MT8183_HOST_IPC_INT_BIT,
> -	.ipi_buf_offset = 0x7bdb0,
> +	.ipi_buf_offset = 0x3bdb0,
>  };
>  
>  static const struct mtk_scp_of_data mt8192_of_data = {
> -- 
> 2.18.0
>