| Message ID | 1610478161-4877-2-git-send-email-skomatineni@nvidia.com |
|---|---|
| State | Accepted |
| Headers | show |
| Series | [v3] i2c: tegra: Create i2c_writesl_vi() to use with VI I2C for filling TX FIFO | expand |
12.01.2021 22:02, Sowjanya Komatineni пишет: > VI I2C controller has known hardware bug where immediate multiple > writes to TX_FIFO register gets stuck. > > Recommended software work around is to read I2C register after > each write to TX_FIFO register to flush out the data. > > This patch implements this work around for VI I2C controller. You could have kept Thierry's ack since it was given to the fix in general. No need to make v4. > Signed-off-by: Sowjanya Komatineni <skomatineni@nvidia.com> > --- > drivers/i2c/busses/i2c-tegra.c | 22 +++++++++++++++++++++- > 1 file changed, 21 insertions(+), 1 deletion(-) > > diff --git a/drivers/i2c/busses/i2c-tegra.c b/drivers/i2c/busses/i2c-tegra.c > index 6f08c0c..4a27782 100644 > --- a/drivers/i2c/busses/i2c-tegra.c > +++ b/drivers/i2c/busses/i2c-tegra.c > @@ -326,6 +326,8 @@ static void i2c_writel(struct tegra_i2c_dev *i2c_dev, u32 val, unsigned int reg) > /* read back register to make sure that register writes completed */ > if (reg != I2C_TX_FIFO) > readl_relaxed(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, reg)); > + else if (i2c_dev->is_vi) > + readl_relaxed(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, I2C_INT_STATUS)); > } > > static u32 i2c_readl(struct tegra_i2c_dev *i2c_dev, unsigned int reg) > @@ -339,6 +341,21 @@ static void i2c_writesl(struct tegra_i2c_dev *i2c_dev, void *data, > writesl(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, reg), data, len); > } > > +static void i2c_writesl_vi(struct tegra_i2c_dev *i2c_dev, void *data, > + unsigned int reg, unsigned int len) > +{ > + u32 *data32 = data; > + > + /* > + * VI I2C controller has known hardware bug where writes get stuck > + * when immediate multiple writes happen to TX_FIFO register. > + * Recommended software work around is to read I2C register after > + * each write to TX_FIFO register to flush out the data. > + */ > + while (len--) > + i2c_writel(i2c_dev, *data32++, reg); > +} > + > static void i2c_readsl(struct tegra_i2c_dev *i2c_dev, void *data, > unsigned int reg, unsigned int len) > { > @@ -811,7 +828,10 @@ static int tegra_i2c_fill_tx_fifo(struct tegra_i2c_dev *i2c_dev) > i2c_dev->msg_buf_remaining = buf_remaining; > i2c_dev->msg_buf = buf + words_to_transfer * BYTES_PER_FIFO_WORD; > > - i2c_writesl(i2c_dev, buf, I2C_TX_FIFO, words_to_transfer); > + if (i2c_dev->is_vi) > + i2c_writesl_vi(i2c_dev, buf, I2C_TX_FIFO, words_to_transfer); > + else > + i2c_writesl(i2c_dev, buf, I2C_TX_FIFO, words_to_transfer); > > buf += words_to_transfer * BYTES_PER_FIFO_WORD; > } > Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
On Tue, Jan 12, 2021 at 11:02:41AM -0800, Sowjanya Komatineni wrote: > VI I2C controller has known hardware bug where immediate multiple > writes to TX_FIFO register gets stuck. > > Recommended software work around is to read I2C register after > each write to TX_FIFO register to flush out the data. > > This patch implements this work around for VI I2C controller. > > Signed-off-by: Sowjanya Komatineni <skomatineni@nvidia.com> > --- > drivers/i2c/busses/i2c-tegra.c | 22 +++++++++++++++++++++- > 1 file changed, 21 insertions(+), 1 deletion(-) Acked-by: Thierry Reding <treding@nvidia.com>
On Tue, Jan 12, 2021 at 11:02:41AM -0800, Sowjanya Komatineni wrote: > VI I2C controller has known hardware bug where immediate multiple > writes to TX_FIFO register gets stuck. > > Recommended software work around is to read I2C register after > each write to TX_FIFO register to flush out the data. > > This patch implements this work around for VI I2C controller. > > Signed-off-by: Sowjanya Komatineni <skomatineni@nvidia.com> Applied to for-current, thanks!
diff --git a/drivers/i2c/busses/i2c-tegra.c b/drivers/i2c/busses/i2c-tegra.c index 6f08c0c..4a27782 100644 --- a/drivers/i2c/busses/i2c-tegra.c +++ b/drivers/i2c/busses/i2c-tegra.c @@ -326,6 +326,8 @@ static void i2c_writel(struct tegra_i2c_dev *i2c_dev, u32 val, unsigned int reg) /* read back register to make sure that register writes completed */ if (reg != I2C_TX_FIFO) readl_relaxed(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, reg)); + else if (i2c_dev->is_vi) + readl_relaxed(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, I2C_INT_STATUS)); } static u32 i2c_readl(struct tegra_i2c_dev *i2c_dev, unsigned int reg) @@ -339,6 +341,21 @@ static void i2c_writesl(struct tegra_i2c_dev *i2c_dev, void *data, writesl(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, reg), data, len); } +static void i2c_writesl_vi(struct tegra_i2c_dev *i2c_dev, void *data, + unsigned int reg, unsigned int len) +{ + u32 *data32 = data; + + /* + * VI I2C controller has known hardware bug where writes get stuck + * when immediate multiple writes happen to TX_FIFO register. + * Recommended software work around is to read I2C register after + * each write to TX_FIFO register to flush out the data. + */ + while (len--) + i2c_writel(i2c_dev, *data32++, reg); +} + static void i2c_readsl(struct tegra_i2c_dev *i2c_dev, void *data, unsigned int reg, unsigned int len) { @@ -811,7 +828,10 @@ static int tegra_i2c_fill_tx_fifo(struct tegra_i2c_dev *i2c_dev) i2c_dev->msg_buf_remaining = buf_remaining; i2c_dev->msg_buf = buf + words_to_transfer * BYTES_PER_FIFO_WORD; - i2c_writesl(i2c_dev, buf, I2C_TX_FIFO, words_to_transfer); + if (i2c_dev->is_vi) + i2c_writesl_vi(i2c_dev, buf, I2C_TX_FIFO, words_to_transfer); + else + i2c_writesl(i2c_dev, buf, I2C_TX_FIFO, words_to_transfer); buf += words_to_transfer * BYTES_PER_FIFO_WORD; }
VI I2C controller has known hardware bug where immediate multiple writes to TX_FIFO register gets stuck. Recommended software work around is to read I2C register after each write to TX_FIFO register to flush out the data. This patch implements this work around for VI I2C controller. Signed-off-by: Sowjanya Komatineni <skomatineni@nvidia.com> --- drivers/i2c/busses/i2c-tegra.c | 22 +++++++++++++++++++++- 1 file changed, 21 insertions(+), 1 deletion(-)