Re: [PATCH] i2c: designware: fix idx_write_cnt in read loop

[David Zheng]

On Mon, May 22, 2023 at 05:58:26PM +0300, Jarkko Nikula wrote:
> Hi
> 
> On 5/18/23 21:06, David Zheng wrote:
> > With IC_INTR_RX_FULL slave interrupt handler reads data in a loop until
> > RX FIFO is empty. When testing with the slave-eeprom, each transaction
> > has 2 bytes for address/index and 1 byte for value, the address byte
> > can be written as data byte due to dropping STOP condition.
> > 
> > In the test below, the master continuously writes to the slave, first 2
> > bytes are index, 3rd byte is value and follow by a STOP condition.
> > 
> >   i2c_write: i2c-3 #0 a=04b f=0000 l=3 [00-D1-D1]
> >   i2c_write: i2c-3 #0 a=04b f=0000 l=3 [00-D2-D2]
> >   i2c_write: i2c-3 #0 a=04b f=0000 l=3 [00-D3-D3]
> > 
> > Upon receiving STOP condition slave eeprom would reset `idx_write_cnt` so
> > next 2 bytes can be treated as buffer index for upcoming transaction.
> > Supposedly the slave eeprom buffer would be written as
> > 
> >   EEPROM[0x00D1] = 0xD1
> >   EEPROM[0x00D2] = 0xD2
> >   EEPROM[0x00D3] = 0xD3
> > 
> > When CPU load is high the slave irq handler may not read fast enough,
> > the interrupt status can be seen as 0x204 with both DW_IC_INTR_STOP_DET
> > (0x200) and DW_IC_INTR_RX_FULL (0x4) bits. The slave device may see
> > the transactions below.
> > 
> >   0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
> >   0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
> >   0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
> >   0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1794 : INTR_STAT=0x204
> >   0x1 STATUS SLAVE_ACTIVITY=0x0 : RAW_INTR_STAT=0x1790 : INTR_STAT=0x200
> >   0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
> >   0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
> >   0x1 STATUS SLAVE_ACTIVITY=0x1 : RAW_INTR_STAT=0x1594 : INTR_STAT=0x4
> > 
> > After `D1` is received, read loop continues to read `00` which is the
> > first bype of next index. Since STOP condition is ignored by the loop,
> > eeprom buffer index increased to `D2` and `00` is written as value.
> > 
> > So the slave eeprom buffer becomes
> > 
> >   EEPROM[0x00D1] = 0xD1
> >   EEPROM[0x00D2] = 0x00
> >   EEPROM[0x00D3] = 0xD3
> > 
> > The fix is to use `FIRST_DATA_BYTE` (bit 11) in `IC_DATA_CMD` to split
> > the transactions. The first index byte in this case would have bit 11
> > set. Check this indication to inject I2C_SLAVE_WRITE_REQUESTED event
> > which will reset `idx_write_cnt` in slave eeprom.
> > 
> > Signed-off-by: David Zheng <david.zheng@xxxxxxxxx>
> > ---
> >   drivers/i2c/busses/i2c-designware-core.h  | 2 ++
> >   drivers/i2c/busses/i2c-designware-slave.c | 6 ++++--
> >   2 files changed, 6 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/i2c/busses/i2c-designware-core.h b/drivers/i2c/busses/i2c-designware-core.h
> > index c5d87aae39c6..8b85147bd518 100644
> > --- a/drivers/i2c/busses/i2c-designware-core.h
> > +++ b/drivers/i2c/busses/i2c-designware-core.h
> > @@ -123,6 +123,8 @@
> >   #define DW_IC_COMP_PARAM_1_SPEED_MODE_HIGH	(BIT(2) | BIT(3))
> >   #define DW_IC_COMP_PARAM_1_SPEED_MODE_MASK	GENMASK(3, 2)
> > +#define DW_IC_DATA_CMD_FIRST_DATA_BYTE		BIT(11)
> > +
> >   /*
> >    * Sofware status flags
> >    */
> > diff --git a/drivers/i2c/busses/i2c-designware-slave.c b/drivers/i2c/busses/i2c-designware-slave.c
> > index cec25054bb24..9549cbcf50aa 100644
> > --- a/drivers/i2c/busses/i2c-designware-slave.c
> > +++ b/drivers/i2c/busses/i2c-designware-slave.c
> > @@ -170,12 +170,14 @@ static irqreturn_t i2c_dw_isr_slave(int this_irq, void *dev_id)
> >   		if (!(dev->status & STATUS_WRITE_IN_PROGRESS)) {
> >   			dev->status |= STATUS_WRITE_IN_PROGRESS;
> >   			dev->status &= ~STATUS_READ_IN_PROGRESS;
> > -			i2c_slave_event(dev->slave, I2C_SLAVE_WRITE_REQUESTED,
> > -					&val);
> >   		}
> >   		do {
> >   			regmap_read(dev->map, DW_IC_DATA_CMD, &tmp);
> > +			if (tmp & DW_IC_DATA_CMD_FIRST_DATA_BYTE)
> > +				i2c_slave_event(dev->slave,
> > +						I2C_SLAVE_WRITE_REQUESTED,
> > +						&val);
> >   			val = tmp;
> >   			i2c_slave_event(dev->slave, I2C_SLAVE_WRITE_RECEIVED,
> >   					&val);
> I fear this might cause regression on some use case on HW that doesn't have
> the FIRST_DATA_BYTE bit in IC_DATA_CMD. That is available on newer Synopsys
> I2C IPs only. For example my test HW doesn't have it.
> 
> This means the I2C_SLAVE_WRITE_REQUESTED is never delivered on these HWs
> that don't implement the FIRST_DATA_BYTE.
> 
> My quick tests using i2c-slave-eeprom didn't show regression but I'm sure
> there is a case that will regress because of that.

I2C_SLAVE_WRITE_REQUESTED can be sent in the if block. Restoring the
removed lines should cover the use case for HW does not have FIRST_DATA_BYTE.

There is no harm to send it again in read loop for FIRST_DATA_BYTE.
Will resubmit the patch with the change.

Thanks
David