Re: [PATCH v5 07/13] soc: mediatek: cmdq: add write_s function

[Matthias Brugger <matthias.bgg@gmail.com>]

On 24/05/2020 19:26, Dennis-YC Hsieh wrote:
> Hi Mattias,
> 
> Thanks for your comment.
> 
> On Sat, 2020-05-16 at 20:14 +0200, Matthias Brugger wrote:
>>
>> On 08/03/2020 11:52, Dennis YC Hsieh wrote:
>>> add write_s function in cmdq helper functions which
>>> writes value contains in internal register to address
>>> with large dma access support.
>>>
>>> Signed-off-by: Dennis YC Hsieh <dennis-yc.hsieh@mediatek.com>
>>> Reviewed-by: CK Hu <ck.hu@mediatek.com>
>>> ---
>>>  drivers/soc/mediatek/mtk-cmdq-helper.c   | 34 +++++++++++++++++++++++-
>>>  include/linux/mailbox/mtk-cmdq-mailbox.h |  2 ++
>>>  include/linux/soc/mediatek/mtk-cmdq.h    | 20 ++++++++++++++
>>>  3 files changed, 55 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/soc/mediatek/mtk-cmdq-helper.c b/drivers/soc/mediatek/mtk-cmdq-helper.c
>>> index 33153d17c9d9..90f1ff2b4b00 100644
>>> --- a/drivers/soc/mediatek/mtk-cmdq-helper.c
>>> +++ b/drivers/soc/mediatek/mtk-cmdq-helper.c
>>> @@ -18,6 +18,10 @@ struct cmdq_instruction {
>>>  	union {
>>>  		u32 value;
>>>  		u32 mask;
>>> +		struct {
>>> +			u16 arg_c;
>>> +			u16 src_reg;
>>> +		};
>>>  	};
>>>  	union {
>>>  		u16 offset;
>>> @@ -29,7 +33,7 @@ struct cmdq_instruction {
>>>  		struct {
>>>  			u8 sop:5;
>>>  			u8 arg_c_t:1;
>>> -			u8 arg_b_t:1;
>>> +			u8 src_t:1;
>>
>> fixing patch 6/13 please. seems the struct should be added in this patch.
> 
> ok, will move to this patch.
> 
>>
>>>  			u8 dst_t:1;
>>>  		};
>>>  	};
>>> @@ -222,6 +226,34 @@ int cmdq_pkt_write_mask(struct cmdq_pkt *pkt, u8 subsys,
>>>  }
>>>  EXPORT_SYMBOL(cmdq_pkt_write_mask);
>>>  
>>> +int cmdq_pkt_write_s(struct cmdq_pkt *pkt, u16 high_addr_reg_idx,
>>> +		     u16 addr_low, u16 src_reg_idx, u32 mask)
>>> +{
>>> +	struct cmdq_instruction inst = { {0} };
>>> +	int err;
>>> +
>>> +	if (mask != U32_MAX) {
>>> +		inst.op = CMDQ_CODE_MASK;
>>> +		inst.mask = ~mask;
>>> +		err = cmdq_pkt_append_command(pkt, inst);
>>> +		if (err < 0)
>>> +			return err;
>>> +
>>> +		inst.mask = 0;
>>> +		inst.op = CMDQ_CODE_WRITE_S_MASK;
>>> +	} else {
>>> +		inst.op = CMDQ_CODE_WRITE_S;
>>> +	}
>>> +
>>> +	inst.src_t = CMDQ_REG_TYPE;
>>
>> Not defined.
>> Please make sure that every patch compiles on it's own and does not add a
>> regression. This is very helpful if we have to bisect the kernel in the future.
> 
> May I know which part not defined? The src_t defined on top of this
> patch and CMDQ_REG_TYPE defined in last patc (see 06/13).

correct, sorry for the noise.

> 
>>
>>> +	inst.sop = high_addr_reg_idx;
>>> +	inst.offset = addr_low;
>>> +	inst.src_reg = src_reg_idx;
>>> +
>>> +	return cmdq_pkt_append_command(pkt, inst);
>>> +}
>>> +EXPORT_SYMBOL(cmdq_pkt_write_s);
>>> +
>>>  int cmdq_pkt_wfe(struct cmdq_pkt *pkt, u16 event)
>>>  {
>>>  	struct cmdq_instruction inst = { {0} };
>>> diff --git a/include/linux/mailbox/mtk-cmdq-mailbox.h b/include/linux/mailbox/mtk-cmdq-mailbox.h
>>> index 121c3bb6d3de..8ef87e1bd03b 100644
>>> --- a/include/linux/mailbox/mtk-cmdq-mailbox.h
>>> +++ b/include/linux/mailbox/mtk-cmdq-mailbox.h
>>> @@ -59,6 +59,8 @@ enum cmdq_code {
>>>  	CMDQ_CODE_JUMP = 0x10,
>>>  	CMDQ_CODE_WFE = 0x20,
>>>  	CMDQ_CODE_EOC = 0x40,
>>> +	CMDQ_CODE_WRITE_S = 0x90,
>>> +	CMDQ_CODE_WRITE_S_MASK = 0x91,
>>>  	CMDQ_CODE_LOGIC = 0xa0,
>>>  };
>>>  
>>> diff --git a/include/linux/soc/mediatek/mtk-cmdq.h b/include/linux/soc/mediatek/mtk-cmdq.h
>>> index 83340211e1d3..c72d826d8934 100644
>>> --- a/include/linux/soc/mediatek/mtk-cmdq.h
>>> +++ b/include/linux/soc/mediatek/mtk-cmdq.h
>>> @@ -12,6 +12,8 @@
>>>  #include <linux/timer.h>
>>>  
>>>  #define CMDQ_NO_TIMEOUT		0xffffffffu
>>> +#define CMDQ_ADDR_HIGH(addr)	((u32)(((addr) >> 16) & GENMASK(31, 0)))
>>> +#define CMDQ_ADDR_LOW(addr)	((u16)(addr) | BIT(1))
>>>  
>>>  struct cmdq_pkt;
>>>  
>>> @@ -102,6 +104,24 @@ int cmdq_pkt_write(struct cmdq_pkt *pkt, u8 subsys, u16 offset, u32 value);
>>>  int cmdq_pkt_write_mask(struct cmdq_pkt *pkt, u8 subsys,
>>>  			u16 offset, u32 value, u32 mask);
>>>  
>>> +/**
>>> + * cmdq_pkt_write_s() - append write_s command to the CMDQ packet
>>> + * @pkt:	the CMDQ packet
>>> + * @high_addr_reg_idx:	internal regisger ID which contains high address of pa
>>
>> s/regisger/register
> 
> will fix
> 
>>
>>> + * @addr_low:	low address of pa
>>> + * @src_reg_idx:	the CMDQ internal register ID which cache source value
>>> + * @mask:	the specified target address mask, use U32_MAX if no need
>>> + *
>>> + * Return: 0 for success; else the error code is returned
>>> + *
>>> + * Support write value to physical address without subsys. Use CMDQ_ADDR_HIGH()
>>> + * to get high addrees and call cmdq_pkt_assign() to assign value into internal
>>
>> s/addrees/address
> 
> will fix
> 
>>
>>> + * reg. Also use CMDQ_ADDR_LOW() to get low address for addr_low parameterwhen
>>
>> s/parameterwhen/parameter when
> 
> will fix
> 
>>
>>> + * call to this function.
>>> + */
>>> +int cmdq_pkt_write_s(struct cmdq_pkt *pkt, u16 high_addr_reg_idx,
>>> +		     u16 addr_low, u16 src_reg_idx, u32 mask);
>>> +
>>
>> In general I wonder if we shouldn't provide two functions, one that writes a
>> mask and on for the else case.
> 
> ok, I'll separate this function to cmdq_pkt_write_s and
> cmdq_pkt_write_s_mask. Let the client choose which case is more
> suitable.

Sound good, thanks.


> 
> 
>>
>> Regards,
>> Matthias
>>
>>>  /**
>>>   * cmdq_pkt_wfe() - append wait for event command to the CMDQ packet
>>>   * @pkt:	the CMDQ packet
>>>
> 
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel