Re: [PATCH v1 07/12] soc: mediatek: cmdq: add write_s function

[Dennis-YC Hsieh]

Hi CK,

On Fri, 2019-11-22 at 16:56 +0800, CK Hu wrote:
> Hi, Dennis:
> 
> On Thu, 2019-11-21 at 17:12 +0800, Dennis YC Hsieh wrote:
> > add write_s function in cmdq helper functions which
> > support large dma access.
> > 
> > Signed-off-by: Dennis YC Hsieh <dennis-yc.hsieh@xxxxxxxxxxxx>
> > ---
> >  drivers/soc/mediatek/mtk-cmdq-helper.c   |   34 ++++++++++++++++++++++++++++++
> >  include/linux/mailbox/mtk-cmdq-mailbox.h |    2 ++
> >  include/linux/soc/mediatek/mtk-cmdq.h    |   13 ++++++++++++
> >  3 files changed, 49 insertions(+)
> > 
> > diff --git a/drivers/soc/mediatek/mtk-cmdq-helper.c b/drivers/soc/mediatek/mtk-cmdq-helper.c
> > index d419e99..1b074a9 100644
> > --- a/drivers/soc/mediatek/mtk-cmdq-helper.c
> > +++ b/drivers/soc/mediatek/mtk-cmdq-helper.c
> > @@ -15,6 +15,9 @@
> >  #define CMDQ_EOC_CMD		((u64)((CMDQ_CODE_EOC << CMDQ_OP_CODE_SHIFT)) \
> >  				<< 32 | CMDQ_EOC_IRQ_EN)
> >  #define CMDQ_REG_TYPE		1
> > +#define CMDQ_ADDR_HIGH(addr)	((u32)(((addr) >> 16) & GENMASK(31, 0)))
> > +#define CMDQ_ADDR_LOW_BIT	BIT(1)
> > +#define CMDQ_ADDR_LOW(addr)	((u16)(addr) | CMDQ_ADDR_LOW_BIT)
> >  
> >  struct cmdq_instruction {
> >  	union {
> > @@ -224,6 +227,37 @@ 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, dma_addr_t addr,
> > +		     u32 value, u32 mask)
> > +{
> > +	struct cmdq_instruction inst = { {0} };
> > +	int err;
> > +	const u16 dst_reg_idx = CMDQ_SPR_TEMP;
> > +
> > +	err = cmdq_pkt_assign(pkt, dst_reg_idx, CMDQ_ADDR_HIGH(addr));
> > +	if (err < 0)
> > +		return 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.op = CMDQ_CODE_WRITE_S_MASK;
> > +	} else {
> > +		inst.op = CMDQ_CODE_WRITE_S;
> > +	}
> > +
> > +	inst.sop = dst_reg_idx;
> > +	inst.offset = CMDQ_ADDR_LOW(addr);
> > +	inst.value = value;
> > +
> > +	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 121c3bb..8ef87e1 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 8334021..8dbd046 100644
> > --- a/include/linux/soc/mediatek/mtk-cmdq.h
> > +++ b/include/linux/soc/mediatek/mtk-cmdq.h
> > @@ -12,6 +12,7 @@
> >  #include <linux/timer.h>
> >  
> >  #define CMDQ_NO_TIMEOUT		0xffffffffu
> > +#define CMDQ_SPR_TEMP		0
> >  
> >  struct cmdq_pkt;
> >  
> > @@ -103,6 +104,18 @@ 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 with mask to the CMDQ packet
> > + * @pkt:	the CMDQ packet
> > + * @addr:	the physical address of register or dma
> > + * @value:	the specified target value
> > + * @mask:	the specified target mask
> > + *
> > + * Return: 0 for success; else the error code is returned
> > + */
> > +int cmdq_pkt_write_s(struct cmdq_pkt *pkt, dma_addr_t addr,
> > +		     u32 value, u32 mask);
> 
> You have an API cmdq_pkt_read_s() which read data into gce internal
> register, so I expect that cmdq_pkt_write_s() is an API which write data
> from gce internal register, the expected prototype is
> 
> int cmdq_pkt_write_s(struct cmdq_pkt *pkt, phys_addr_t addr, u16
> reg_idx);
> 
> Your version would confuse the user because you hide the internal
> register parameter. If you want to provide this service, I would like
> you to change the function name so that user would not be confused and
> easily to understand what you want to do in this function.
> 
> Another choice is: cmdq_pkt_write_s() is implemented in my definition,
> and user could call cmdq_pkt_assign() and cmdq_pkt_write_s() to achieve
> this function.
> 
> Regards,
> CK
> 

Thanks for your comment.

Ok, we have to provide write constant value service to client, so I will
change the function name to cmdq_pkt_write_s_value() in this patch.

And since it is better to provide consistent API so I will design
another function with interface as your suggestion:
int cmdq_pkt_write_s(struct cmdq_pkt *pkt, phys_addr_t addr, u16
reg_idx);

In another patch I provide cmdq_pkt_mem_move(). I will move part of
implementation to cmdq_pkt_write_s(), so that cmdq_pkt_mem_move() can be
combination of cmdq_pkt_read_s() and cmdq_pkt_write_s().

How do you think?


Regards,
Dennis

> > +
> > +/**
> >   * cmdq_pkt_wfe() - append wait for event command to the CMDQ packet
> >   * @pkt:	the CMDQ packet
> >   * @event:	the desired event type to "wait and CLEAR"
> 
>