Re: [PATCH 2/6] i2c: qup: Add V2 tags support

[Sricharan R <sricharan-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>]

Hi Ivan,

On 03/25/2015 05:54 PM, Ivan T. Ivanov wrote:
>
> Hi Sricharan,
>
> On Fri, 2015-03-13 at 23:19 +0530, Sricharan R wrote:
>> From: Andy Gross <agross-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
>>
>> QUP from version 2.1.1 onwards, supports a new format of
>> i2c command tags. Tag codes instructs the controller to
>> perform a operation like read/write. This new tagging version
>> supports bam dma and transfers of more than 256 bytes without 'stop'
>> in between. Adding the support for the same.
>>
>> For each block a data_write/read tag and data_len tag is added to
>> the output fifo. For the final block of data write_stop/read_stop
>> tag is used.
>>
>> Signed-off-by: Andy Gross <agross-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
>> Signed-off-by: Sricharan R <sricharan-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
>> ---
>>   drivers/i2c/busses/i2c-qup.c | 342 ++++++++++++++++++++++++++++++++++++++-----
>>
>
> <snip>
>
>> +#define I2C_QUP_CLK_MAX_FREQ           3400000
>
> unused?
>
Ok, will remove this.

>> +
>>   /* Status, Error flags */
>>   #define I2C_STATUS_WR_BUFFER_FULL      BIT(0)
>>   #define I2C_STATUS_BUS_ACTIVE          BIT(8)
>> @@ -99,6 +117,11 @@
>>
>>   #define QUP_READ_LIMIT                 256
>>
>> +struct qup_i2c_config {
>> +       int tag_ver;
>> +       int max_freq;
>> +};
>> +
>
> Do you really need this one. It is referenced only during probe,
> but information contained in is already available by other means.
>
Ok, agree that this can be removed and the information can be fixed
for V1 version and same for the rest of the other versions. That would
need to use compatible check though.
>
>>   struct qup_i2c_dev {
>>          struct device*dev;
>>          void __iomem*base;
>> @@ -112,9 +135,20 @@ struct qup_i2c_dev {
>>          int     in_fifo_sz;
>>          int     out_blk_sz;
>>          int     in_blk_sz;
>> -
>> +       int     blocks;
>> +       u8      *block_tag_len;
>> +       int     *block_data_len;
>
> Maybe these could be organized in struct?
>
ok, will group it.
> <snip>
>
>>
>> +static void qup_i2c_create_tag_v2(struct qup_i2c_dev *qup,
>> +                                               struct i2c_msg *msg)
>> +{
>> +       u16 addr = (msg->addr << 1) | ((msg->flags & I2C_M_RD) == I2C_M_RD);
>> +       int len = 0, prev_len = 0;
>> +       int blocks = 0;
>> +       int rem;
>> +       int block_len = 0;
>> +       int data_len;
>> +
>> +       qup->block_pos = 0;
>> +       qup->pos = 0;
>> +       qup->blocks = (msg->len + QUP_READ_LIMIT - 1) / (QUP_READ_LIMIT);
>
> Braces around QUP_READ_LIMIT are not needed.
>
ok.
>> +       rem = msg->len % QUP_READ_LIMIT;
>> +
>> +       /* 2 tag bytes for each block + 2 extra bytes for first block */
>> +       qup->tags = kzalloc((qup->blocks << 1) + 2, GFP_KERNEL);
>
> Don't play tricks with shifts, just use multiplication.
>
why ?
>> +       qup->block_tag_len = kzalloc(qup->blocks, GFP_KERNEL);
>> +       qup->block_data_len = kzalloc(sizeof(int) * qup->blocks, GFP_KERNEL);
>
> Better use sizeof(*qup->block_data_len).
>
ok.
>> +
>> +       while (blocks < qup->blocks) {
>> +               /* 0 is used to specify a READ_LIMIT of 256 bytes */
>> +               data_len = (blocks < (qup->blocks - 1)) ? 0 : rem;
>> +
>> +               /* Send START and ADDR bytes only for the first block */
>> +               if (!blocks) {
>> +                       qup->tags[len++] = QUP_TAG_V2_START;
>> +
>> +                       if (qup->is_hs) {
>> +                               qup->tags[len++] = QUP_TAG_V2_HS;
>> +                               qup->tags[len++] = QUP_TAG_V2_START;
>
> Is this second QUP_TAG_V2_START intentional?
>
  Yes, for HS mode 2 start tags are required.
>> +                       }
>> +
>> +                       qup->tags[len++] = addr & 0xff;
>> +                       if (msg->flags & I2C_M_TEN)
>> +                               qup->tags[len++] = addr >> 8;
>> +               }
>> +
>> +               /* Send _STOP commands for the last block */
>> +               if (blocks == (qup->blocks - 1)) {
>> +                       if (msg->flags & I2C_M_RD)
>> +                               qup->tags[len++] = QUP_TAG_V2_DATARD_STOP;
>> +                       else
>> +                               qup->tags[len++] = QUP_TAG_V2_DATAWR_STOP;
>> +               } else {
>> +                       if (msg->flags & I2C_M_RD)
>> +                               qup->tags[len++] = QUP_TAG_V2_DATARD;
>> +                       else
>> +                               qup->tags[len++] = QUP_TAG_V2_DATAWR;
>> +               }
>> +
>> +               qup->tags[len++] = data_len;
>> +               block_len = len - prev_len;
>> +               prev_len = len;
>> +               qup->block_tag_len[blocks] = block_len;
>> +
>> +               if (!data_len)
>> +                       qup->block_data_len[blocks] = QUP_READ_LIMIT;
>> +               else
>> +                       qup->block_data_len[blocks] = data_len;
>> +
>> +               qup->tags_pos = 0;
>> +               blocks++;
>> +       }
>> +
>> +       qup->tx_tag_len = len;
>> +
>> +       if (msg->flags & I2C_M_RD)
>> +               qup->rx_tag_len = (qup->blocks << 1);
>
> here again.
>
hmm, why not shift ?
>> +       else
>> +               qup->rx_tag_len = 0;
>> +}
>> +
>> +static u32 qup_i2c_xfer_data(struct qup_i2c_dev *qup, int len,
>> +                                                               u8 *buf, int last)
>> +{
>
> I think that xfer is too vague in this case, prefer write or send.
>
ok. Will change it to send.
>> +       static u32 val, idx;
>
> static? please fix.
  That was intentional. Using to pack tag and data in to one word across
  two calls. So preserving val, idx across calls.
>
>> +       u32  t, rem, pos = 0;
>> +
>> +       rem = len - pos + idx;
>> +
>> +       while (rem) {
>> +               if (qup_i2c_wait_ready(qup, QUP_OUT_FULL, 0, 4)) {
>> +                       dev_err(qup->dev, "timeout for fifo out full");
>> +                       break;
>> +               }
>> +
>> +               t = (rem >= 4) ? 4 : rem;
>> +
>> +               while (idx < t)
>> +                       val |= buf[pos++] << (idx++ << 3);
>
> here again, multiplication or shift?
>
  I felt shifts should be readable. No ?
>> +
>> +               if (t == 4) {
>> +                       writel(val, qup->base + QUP_OUT_FIFO_BASE);
>> +                       idx = val = 0;
>> +               }
>> +
>> +               rem = len - pos;
>> +       }
>> +
>
> What will happen if they are less than 4 bytes to send?
>
It gets written over in the next call and gets written here itself if it 
is the last block of data.
>> +       if (last) {
>> +               writel(val, qup->base + QUP_OUT_FIFO_BASE);
>> +               idx = val = 0;
>> +       }
>> +
>> +       return 0;
>> +}
>> +
>>
>
> <snip>
>
>>
>> -static void qup_i2c_issue_read(struct qup_i2c_dev *qup, struct i2c_msg *msg)
>> +static void qup_i2c_issue_read_v1(struct qup_i2c_dev *qup, struct
>> +                                               i2c_msg *msg)
>
> Please, move struct on the same line as i2c_msg.
>
ok.

>>   {
>>          u32 addr, len, val;
>>
>> @@ -395,24 +576,33 @@ static void qup_i2c_issue_read(struct qup_i2c_dev *qup, struct i2c_msg *msg)
>>          /* 0 is used to specify a length 256 (QUP_READ_LIMIT) */
>>          len = (msg->len == QUP_READ_LIMIT) ? 0 : msg->len;
>>
>> -       val = ((QUP_TAG_REC | len) << QUP_MSW_SHIFT) | QUP_TAG_START | addr;
>> +       val = ((QUP_TAG_REC | len) << QUP_MSW_SHIFT) | QUP_TAG_START |
>> +                       addr;
>
> Unrelated change?
  hmm. Must have come for adjusting 80 chars. Will remove this here.
>
>>          writel(val, qup->base + QUP_OUT_FIFO_BASE);
>>   }
>>
>>
>
> <snip>
>
>>
>> +static void qup_i2c_read_fifo_v2(struct qup_i2c_dev *qup, struct
>> +                                       i2c_msg *msg)
>> +{
>> +       u32 val;
>> +       int idx;
>> +       int pos = 0;
>
>> +       int total = qup->block_data_len[qup->block_pos] + 2;
>
> Could we  have at least comment what is this +2?
>
  Ok. The +2 is for the read tags that are extra bytes read for each
  block. Will add comment for that.

> <snip>
>
>> +
>> +static void qup_i2c_read_fifo(struct qup_i2c_dev *qup, struct i2c_msg
>> +                                                                       *msg)
>
> Bad indent?
>
  ok. Will fix it.
> <snip>
>
>
>> +static const struct qup_i2c_config configs[] = {
>> +       { 0, 400000, },
>> +       { 1, 3400000, },
>> +};
>> +
>> +static const struct of_device_id qup_i2c_dt_match[] = {
>> +       { .compatible = "qcom,i2c-qup-v1.1.1", .data = &configs[0] },
>> +       { .compatible = "qcom,i2c-qup-v2.1.1", .data = &configs[1] },
>> +       { .compatible = "qcom,i2c-qup-v2.2.1", .data = &configs[1] },
>> +       {}
>> +};
>> +MODULE_DEVICE_TABLE(of, qup_i2c_dt_match);
>> +
>>   static void qup_i2c_disable_clocks(struct qup_i2c_dev *qup)
>>   {
>>          u32 config;
>> @@ -579,6 +837,8 @@ static int qup_i2c_probe(struct platform_device *pdev)
>>          int ret, fs_div, hs_div;
>>          int src_clk_freq;
>>          u32 clk_freq = 100000;
>> +       const struct qup_i2c_config *config;
>> +       const struct of_device_id *of_id;
>>
>>          qup = devm_kzalloc(&pdev->dev, sizeof(*qup), GFP_KERNEL);
>>          if (!qup)
>> @@ -590,8 +850,15 @@ static int qup_i2c_probe(struct platform_device *pdev)
>>
>>          of_property_read_u32(node, "clock-frequency", &clk_freq);
>>
>> -       /* We support frequencies up to FAST Mode (400KHz) */
>> -       if (!clk_freq || clk_freq > 400000) {
>> +       of_id = of_match_node(qup_i2c_dt_match, node);
>> +       if (!of_id)
>> +               return -EINVAL;
>
> this could not happen.
>
  Correct. Will remove it.
>> +
>> +       config = of_id->data;
>> +       qup->use_v2_tags = !!config->tag_ver;
>> +
>> +       /* We support frequencies up to HIGH SPEED Mode (3400KHz) */
>> +       if (!clk_freq || clk_freq > config->max_freq) {
>
> Looks like if controller is v > 1 it supports I2C_QUP_CLK_MAX_FREQ.
> I don't see need for struct qup_i2c_config.
>
Ok, this can removed if compatible check is used to differentiate 
between V1 and rest of the versions.

Regards,
  Sricharan


-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a 
member of Code Aurora Forum, hosted by The Linux Foundation