Re: [RESEND, PATCH v13 09/12] soc: mediatek: cmdq: define the instruction struct

From: Bibby Hsieh
Date: Tue Aug 27 2019 - 00:12:11 EST

Next message: Chen-Yu Tsai: "Re: [linux-sunxi] [PATCH v6 1/3] ASoC: sun4i-i2s: incorrect regmap for A83T"
Previous message: Bibby Hsieh: "Re: [RESEND, PATCH v13 10/12] soc: mediatek: cmdq: add polling function"
Next in thread: Matthias Brugger: "Re: [RESEND, PATCH v13 09/12] soc: mediatek: cmdq: define the instruction struct"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Fri, 2019-08-23 at 15:50 +0200, Matthias Brugger wrote:
>
> On 20/08/2019 10:49, Bibby Hsieh wrote:
> > Define an instruction structure for gce driver to append command.
> > This structure can make the client's code more readability.
> >
> > Signed-off-by: Bibby Hsieh <bibby.hsieh@xxxxxxxxxxxx>
> > Reviewed-by: CK Hu <ck.hu@xxxxxxxxxxxx>
> > ---
> > drivers/soc/mediatek/mtk-cmdq-helper.c | 106 +++++++++++++++--------
> > include/linux/mailbox/mtk-cmdq-mailbox.h | 2 +
> > 2 files changed, 74 insertions(+), 34 deletions(-)
> >
> > diff --git a/drivers/soc/mediatek/mtk-cmdq-helper.c b/drivers/soc/mediatek/mtk-cmdq-helper.c
> > index 7aa0517ff2f3..e3d5b0be8e79 100644
> > --- a/drivers/soc/mediatek/mtk-cmdq-helper.c
> > +++ b/drivers/soc/mediatek/mtk-cmdq-helper.c
> > @@ -9,12 +9,24 @@
> > #include <linux/mailbox_controller.h>
> > #include <linux/soc/mediatek/mtk-cmdq.h>
> >
> > -#define CMDQ_ARG_A_WRITE_MASK 0xffff
> > #define CMDQ_WRITE_ENABLE_MASK BIT(0)
> > #define CMDQ_EOC_IRQ_EN BIT(0)
> > #define CMDQ_EOC_CMD ((u64)((CMDQ_CODE_EOC << CMDQ_OP_CODE_SHIFT)) \
> > << 32 | CMDQ_EOC_IRQ_EN)
> >
> > +struct cmdq_instruction {
> > + union {
> > + u32 value;
> > + u32 mask;
> > + };
> > + union {
> > + u16 offset;
> > + u16 event;
> > + };
> > + u8 subsys;
> > + u8 op;
> > +};
> > +
> > static void cmdq_client_timeout(struct timer_list *t)
> > {
> > struct cmdq_client *client = from_timer(client, t, timer);
> > @@ -110,10 +122,8 @@ void cmdq_pkt_destroy(struct cmdq_pkt *pkt)
> > }
> > EXPORT_SYMBOL(cmdq_pkt_destroy);
> >
> > -static int cmdq_pkt_append_command(struct cmdq_pkt *pkt, enum cmdq_code code,
> > - u32 arg_a, u32 arg_b)
> > +static struct cmdq_instruction *cmdq_pkt_append_command(struct cmdq_pkt *pkt)
> > {
> > - u64 *cmd_ptr;
> >
> > if (unlikely(pkt->cmd_buf_size + CMDQ_INST_SIZE > pkt->buf_size)) {
> > /*
> > @@ -127,81 +137,109 @@ static int cmdq_pkt_append_command(struct cmdq_pkt *pkt, enum cmdq_code code,
> > pkt->cmd_buf_size += CMDQ_INST_SIZE;
> > WARN_ONCE(1, "%s: buffer size %u is too small !\n",
> > __func__, (u32)pkt->buf_size);
> > - return -ENOMEM;
> > + return NULL;
> > }
> > - cmd_ptr = pkt->va_base + pkt->cmd_buf_size;
> > - (*cmd_ptr) = (u64)((code << CMDQ_OP_CODE_SHIFT) | arg_a) << 32 | arg_b;
> > +
> > + *(u64 *)(pkt->va_base + pkt->cmd_buf_size) = 0;> pkt->cmd_buf_size += CMDQ_INST_SIZE;
> >
> > - return 0;
> > + return pkt->va_base + pkt->cmd_buf_size - CMDQ_INST_SIZE;
> > }
> >
> > int cmdq_pkt_write(struct cmdq_pkt *pkt, u8 subsys, u16 offset, u32 value)
> > {
> > - u32 arg_a = (offset & CMDQ_ARG_A_WRITE_MASK) |
> > - (subsys << CMDQ_SUBSYS_SHIFT);
> > + struct cmdq_instruction *inst;
> > +
> > + inst = cmdq_pkt_append_command(pkt);
> > + if (!inst)
> > + return -ENOMEM;
> > +
> > + inst->op = CMDQ_CODE_WRITE;
> > + inst->value = value;
> > + inst->offset = offset;
> > + inst->subsys = subsys;
> >
>
> I can see that using cmdq_instruction will make the code more readable, but I
> dislike the approach that cmdq_pkt_append_command returns a pointer where we
> write the instruction to. Better we pass inst to cmdq_pkt_append_command() and
> write it there to cmd_ptr.
>
> I think this way we can get rid of explicitly setting the memory to zero:
> *(u64 *)(pkt->va_base + pkt->cmd_buf_size) = 0;
>
> And if we pass the inst to the append_command we don't have to change the return
> value handling of cmdq_pkt_append_command(), which makes the patch easier to
> understand.

Ok, I will change and resend it.

>
> > - return cmdq_pkt_append_command(pkt, CMDQ_CODE_WRITE, arg_a, value);
> > + return 0;
> > }
> > EXPORT_SYMBOL(cmdq_pkt_write);
> >
> > int cmdq_pkt_write_mask(struct cmdq_pkt *pkt, u8 subsys,
> > u16 offset, u32 value, u32 mask)
> > {
> > - u32 offset_mask = offset;
> > - int err = 0;
> > + struct cmdq_instruction *inst;
> > + u16 offset_mask = offset;
> >
> > if (mask != 0xffffffff) {
> > - err = cmdq_pkt_append_command(pkt, CMDQ_CODE_MASK, 0, ~mask);
> > + inst = cmdq_pkt_append_command(pkt);
> > + if (!inst)
> > + return -ENOMEM;
> > +
> > + inst->op = CMDQ_CODE_MASK;
> > + inst->mask = ~mask;
> > offset_mask |= CMDQ_WRITE_ENABLE_MASK;
> > }
> > - err |= cmdq_pkt_write(pkt, value, subsys, offset_mask);
> >
> > - return err;
> > + return cmdq_pkt_write(pkt, subsys, offset_mask, value);
> > }
> > EXPORT_SYMBOL(cmdq_pkt_write_mask);
> >
> > int cmdq_pkt_wfe(struct cmdq_pkt *pkt, u16 event)
> > {
> > - u32 arg_b;
> > + struct cmdq_instruction *inst;
> >
> > if (event >= CMDQ_MAX_EVENT)
> > return -EINVAL;
> >
> > - /*
> > - * WFE arg_b
> > - * bit 0-11: wait value
> > - * bit 15: 1 - wait, 0 - no wait
> > - * bit 16-27: update value
> > - * bit 31: 1 - update, 0 - no update
> > - */
>
> I have no strong opinion of CMDQ_WFE_OPTION but if you want to introduce it,
> then please copy the comment over to include/linux/mailbox/mtk-cmdq-mailbox.h

Ok. let's move the descriptions to header.
>
> Just one question, why did you call it _OPTION? It's not really expressive for me.

Actually, _OPTION is come from our hardware design name...

>
> > - arg_b = CMDQ_WFE_UPDATE | CMDQ_WFE_WAIT | CMDQ_WFE_WAIT_VALUE;
> > + inst = cmdq_pkt_append_command(pkt);
> > + if (!inst)
> > + return -ENOMEM;
> > +
> > + inst->op = CMDQ_CODE_WFE;
> > + inst->value = CMDQ_WFE_OPTION;
> > + inst->event = event;
> >
> > - return cmdq_pkt_append_command(pkt, CMDQ_CODE_WFE, event, arg_b);
> > + return 0;
> > }
> > EXPORT_SYMBOL(cmdq_pkt_wfe);
> >
> > int cmdq_pkt_clear_event(struct cmdq_pkt *pkt, u16 event)
> > {
> > + struct cmdq_instruction *inst;
> > +
> > if (event >= CMDQ_MAX_EVENT)
> > return -EINVAL;
> >
> > - return cmdq_pkt_append_command(pkt, CMDQ_CODE_WFE, event,
> > - CMDQ_WFE_UPDATE);
> > + inst = cmdq_pkt_append_command(pkt);
> > + if (!inst)
> > + return -ENOMEM;
> > +
> > + inst->op = CMDQ_CODE_WFE;
> > + inst->value = CMDQ_WFE_UPDATE;
> > + inst->event = event;
> > +
> > + return 0;
> > }
> > EXPORT_SYMBOL(cmdq_pkt_clear_event);
> >
> > static int cmdq_pkt_finalize(struct cmdq_pkt *pkt)
> > {
> > - int err;
> > + struct cmdq_instruction *inst;
> > +
> > + inst = cmdq_pkt_append_command(pkt);
> > + if (!inst)
> > + return -ENOMEM;
> >
> > - /* insert EOC and generate IRQ for each command iteration */
>
> Please don't delete the comment.
>
> > - err = cmdq_pkt_append_command(pkt, CMDQ_CODE_EOC, 0, CMDQ_EOC_IRQ_EN);
> > + inst->op = CMDQ_CODE_EOC;
> > + inst->value = CMDQ_EOC_IRQ_EN;
> >
> > - /* JUMP to end */
>
> Same here.
>
> Regards,
> Matthias
>
> > - err |= cmdq_pkt_append_command(pkt, CMDQ_CODE_JUMP, 0, CMDQ_JUMP_PASS);
> > + inst = cmdq_pkt_append_command(pkt);
> > + if (!inst)
> > + return -ENOMEM;
> > +
> > + inst->op = CMDQ_CODE_JUMP;
> > + inst->value = CMDQ_JUMP_PASS;
> >
> > - return err;
> > + return 0;
> > }
> >
> > static void cmdq_pkt_flush_async_cb(struct cmdq_cb_data data)
> > diff --git a/include/linux/mailbox/mtk-cmdq-mailbox.h b/include/linux/mailbox/mtk-cmdq-mailbox.h
> > index 911475da7a53..c8adedefaf42 100644
> > --- a/include/linux/mailbox/mtk-cmdq-mailbox.h
> > +++ b/include/linux/mailbox/mtk-cmdq-mailbox.h
> > @@ -19,6 +19,8 @@
> > #define CMDQ_WFE_UPDATE BIT(31)
> > #define CMDQ_WFE_WAIT BIT(15)
> > #define CMDQ_WFE_WAIT_VALUE 0x1
> > +#define CMDQ_WFE_OPTION (CMDQ_WFE_UPDATE | CMDQ_WFE_WAIT | \
> > + CMDQ_WFE_WAIT_VALUE)
> > /** cmdq event maximum */
> > #define CMDQ_MAX_EVENT 0x3ff
> >
> >

--
Bibby

Next message: Chen-Yu Tsai: "Re: [linux-sunxi] [PATCH v6 1/3] ASoC: sun4i-i2s: incorrect regmap for A83T"
Previous message: Bibby Hsieh: "Re: [RESEND, PATCH v13 10/12] soc: mediatek: cmdq: add polling function"
Next in thread: Matthias Brugger: "Re: [RESEND, PATCH v13 09/12] soc: mediatek: cmdq: define the instruction struct"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]