diff mbox series

[v3,3/7] i2c: aspeed: Fix unhandled Tx done with NAK

Message ID 20210519074934.20712-4-quan@os.amperecomputing.com
State Superseded
Headers show
Series Add Aspeed SSIF BMC driver | expand

Commit Message

Quan Nguyen May 19, 2021, 7:49 a.m. UTC
It is observed that in normal condition, when the last byte sent by
slave, the Tx Done with NAK irq will raise.
But it is also observed that sometimes master issues next transaction
too quick while the slave irq handler is not yet invoked and Tx Done
with NAK irq of last byte of previous READ PROCESSED was not ack'ed.
This Tx Done with NAK irq is raised together with the Slave Match and
Rx Done irq of the next coming transaction from master.
Unfortunately, the current slave irq handler handles the Slave Match and
Rx Done only in higher priority and ignore the Tx Done with NAK, causing
the complain as below:
"aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected
0x00000086, but was 0x00000084"

This commit handles this case by emitting a Slave Stop event for the
Tx Done with NAK before processing Slave Match and Rx Done for the
coming transaction from master.

Signed-off-by: Quan Nguyen <quan@os.amperecomputing.com>
---
v3:
  + First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 5 +++++
 1 file changed, 5 insertions(+)

Comments

Joel Stanley May 19, 2021, 11:28 p.m. UTC | #1
Ryan, can you please review this change?

On Wed, 19 May 2021 at 07:50, Quan Nguyen <quan@os.amperecomputing.com> wrote:
>
> It is observed that in normal condition, when the last byte sent by
> slave, the Tx Done with NAK irq will raise.
> But it is also observed that sometimes master issues next transaction
> too quick while the slave irq handler is not yet invoked and Tx Done
> with NAK irq of last byte of previous READ PROCESSED was not ack'ed.
> This Tx Done with NAK irq is raised together with the Slave Match and
> Rx Done irq of the next coming transaction from master.
> Unfortunately, the current slave irq handler handles the Slave Match and
> Rx Done only in higher priority and ignore the Tx Done with NAK, causing
> the complain as below:
> "aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected
> 0x00000086, but was 0x00000084"
>
> This commit handles this case by emitting a Slave Stop event for the
> Tx Done with NAK before processing Slave Match and Rx Done for the
> coming transaction from master.

It sounds like this patch is independent of the rest of the series,
and can go in on it's own. Please send it separately to the i2c
maintainers and add a suitable Fixes line, such as:

  Fixes: f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C driver")

>
> Signed-off-by: Quan Nguyen <quan@os.amperecomputing.com>
> ---
> v3:
>   + First introduce in v3 [Quan]
>
>  drivers/i2c/busses/i2c-aspeed.c | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/drivers/i2c/busses/i2c-aspeed.c b/drivers/i2c/busses/i2c-aspeed.c
> index 724bf30600d6..3fb37c3f23d4 100644
> --- a/drivers/i2c/busses/i2c-aspeed.c
> +++ b/drivers/i2c/busses/i2c-aspeed.c
> @@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status)
>
>         /* Slave was requested, restart state machine. */
>         if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {

Can you explain why you need to do this handing inside the SLAVE_MATCH case?

Could you instead move the TX_NAK handling to be above the SLAVE_MATCH case?

> +               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
> +                   bus->slave_state == ASPEED_I2C_SLAVE_READ_PROCESSED) {

Either way, this needs a comment to explain what we're working around.

> +                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
> +                       i2c_slave_event(slave, I2C_SLAVE_STOP, &value);
> +               }
>                 irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
>                 bus->slave_state = ASPEED_I2C_SLAVE_START;
>         }
> --
> 2.28.0
>
Ryan Chen May 20, 2021, 11:28 a.m. UTC | #2
> -----Original Message-----
> From: Joel Stanley <joel@jms.id.au>
> Sent: Thursday, May 20, 2021 7:29 AM
> To: Quan Nguyen <quan@os.amperecomputing.com>; Ryan Chen
> <ryan_chen@aspeedtech.com>
> Cc: Corey Minyard <minyard@acm.org>; Rob Herring <robh+dt@kernel.org>;
> Andrew Jeffery <andrew@aj.id.au>; Brendan Higgins
> <brendanhiggins@google.com>; Benjamin Herrenschmidt
> <benh@kernel.crashing.org>; Wolfram Sang <wsa@kernel.org>; Philipp Zabel
> <p.zabel@pengutronix.de>; openipmi-developer@lists.sourceforge.net;
> devicetree <devicetree@vger.kernel.org>; Linux ARM
> <linux-arm-kernel@lists.infradead.org>; linux-aspeed
> <linux-aspeed@lists.ozlabs.org>; Linux Kernel Mailing List
> <linux-kernel@vger.kernel.org>; linux-i2c@vger.kernel.org; Open Source
> Submission <patches@amperecomputing.com>; Phong Vo
> <phong@os.amperecomputing.com>; Thang Q . Nguyen
> <thang@os.amperecomputing.com>; OpenBMC Maillist
> <openbmc@lists.ozlabs.org>
> Subject: Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK
> 
> Ryan, can you please review this change?
> 
> On Wed, 19 May 2021 at 07:50, Quan Nguyen
> <quan@os.amperecomputing.com> wrote:
> >
> > It is observed that in normal condition, when the last byte sent by
> > slave, the Tx Done with NAK irq will raise.
> > But it is also observed that sometimes master issues next transaction
> > too quick while the slave irq handler is not yet invoked and Tx Done
> > with NAK irq of last byte of previous READ PROCESSED was not ack'ed.
> > This Tx Done with NAK irq is raised together with the Slave Match and
> > Rx Done irq of the next coming transaction from master.
> > Unfortunately, the current slave irq handler handles the Slave Match
> > and Rx Done only in higher priority and ignore the Tx Done with NAK,
> > causing the complain as below:
> > "aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected
> > 0x00000086, but was 0x00000084"
> >
> > This commit handles this case by emitting a Slave Stop event for the
> > Tx Done with NAK before processing Slave Match and Rx Done for the
> > coming transaction from master.
> 
> It sounds like this patch is independent of the rest of the series, and can go in
> on it's own. Please send it separately to the i2c maintainers and add a suitable
> Fixes line, such as:
> 
>   Fixes: f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C
> driver")
> 
> >
> > Signed-off-by: Quan Nguyen <quan@os.amperecomputing.com>
> > ---
> > v3:
> >   + First introduce in v3 [Quan]
> >
> >  drivers/i2c/busses/i2c-aspeed.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/drivers/i2c/busses/i2c-aspeed.c
> > b/drivers/i2c/busses/i2c-aspeed.c index 724bf30600d6..3fb37c3f23d4
> > 100644
> > --- a/drivers/i2c/busses/i2c-aspeed.c
> > +++ b/drivers/i2c/busses/i2c-aspeed.c
> > @@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct
> > aspeed_i2c_bus *bus, u32 irq_status)
> >
> >         /* Slave was requested, restart state machine. */
> >         if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {
> 
> Can you explain why you need to do this handing inside the SLAVE_MATCH
> case?
> 
> Could you instead move the TX_NAK handling to be above the SLAVE_MATCH
> case?
> 
> > +               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
> > +                   bus->slave_state ==
> > + ASPEED_I2C_SLAVE_READ_PROCESSED) {
> 
> Either way, this needs a comment to explain what we're working around.
> 
> > +                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
> > +                       i2c_slave_event(slave, I2C_SLAVE_STOP,
> &value);

According the patch assume slave receive TX_NAK will be go to SLAVE_STOP state?

> > +               }
> >                 irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
> >                 bus->slave_state = ASPEED_I2C_SLAVE_START;
> >         }
> > --
> > 2.28.0
> >
Quan Nguyen May 20, 2021, 1:48 p.m. UTC | #3
On 20/05/2021 06:28, Joel Stanley wrote:
> Ryan, can you please review this change?
> 
> On Wed, 19 May 2021 at 07:50, Quan Nguyen <quan@os.amperecomputing.com> wrote:
>>
>> It is observed that in normal condition, when the last byte sent by
>> slave, the Tx Done with NAK irq will raise.
>> But it is also observed that sometimes master issues next transaction
>> too quick while the slave irq handler is not yet invoked and Tx Done
>> with NAK irq of last byte of previous READ PROCESSED was not ack'ed.
>> This Tx Done with NAK irq is raised together with the Slave Match and
>> Rx Done irq of the next coming transaction from master.
>> Unfortunately, the current slave irq handler handles the Slave Match and
>> Rx Done only in higher priority and ignore the Tx Done with NAK, causing
>> the complain as below:
>> "aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected
>> 0x00000086, but was 0x00000084"
>>
>> This commit handles this case by emitting a Slave Stop event for the
>> Tx Done with NAK before processing Slave Match and Rx Done for the
>> coming transaction from master.
> 
> It sounds like this patch is independent of the rest of the series,
> and can go in on it's own. Please send it separately to the i2c
> maintainers and add a suitable Fixes line, such as:
> 
>    Fixes: f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C driver")
> 
Will separate this patch into independent series in next version.

>>
>> Signed-off-by: Quan Nguyen <quan@os.amperecomputing.com>
>> ---
>> v3:
>>    + First introduce in v3 [Quan]
>>
>>   drivers/i2c/busses/i2c-aspeed.c | 5 +++++
>>   1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/i2c/busses/i2c-aspeed.c b/drivers/i2c/busses/i2c-aspeed.c
>> index 724bf30600d6..3fb37c3f23d4 100644
>> --- a/drivers/i2c/busses/i2c-aspeed.c
>> +++ b/drivers/i2c/busses/i2c-aspeed.c
>> @@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status)
>>
>>          /* Slave was requested, restart state machine. */
>>          if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {
> 
> Can you explain why you need to do this handing inside the SLAVE_MATCH case?

> Could you instead move the TX_NAK handling to be above the SLAVE_MATCH case?
>
>> +               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
>> +                   bus->slave_state == ASPEED_I2C_SLAVE_READ_PROCESSED) {
> 
> Either way, this needs a comment to explain what we're working around.
>
Let me explain with the two examples below in normal case and the case 
where this patch is for:

In normal case:
The first transaction is Slave send (Master read):
    20(addr) 03(singlepart read) 03 1c 2e d5

Then the second Master write follow as below:
    20(addr) 02(singlepart write) 02 18 08 59

The irq will raise in sequence below:

  irq      data  from-state      to-state
00000084  20    INACTIVE        WRITE_RECEIVED
00000004  03    WRITE_RECEIVED  WRITE_RECEIVED <= RX_DONE
00000084  03    WRITE_RECEIVED  READ_PROCESSED
00000001  1c    READ_PROCESSED  READ_PROCESSED <= TX_ACK
00000001  2e    READ_PROCESSED  READ_PROCESSED
00000001  d5    READ_PROCESSED  READ_PROCESSED
00000002  xx    READ_PROCESSED  INACTIVE       <= TX_NAK

00000084  20    INACTIVE        WRITE_RECEIVED <= SLAVE_MATCH & RX_DONE
00000004  02    WRITE_RECEIVED  WRITE_RECEIVED
00000084  02    WRITE_RECEIVED  WRITE_RECEIVED
00000004  18    WRITE_RECEIVED  WRITE_RECEIVED
00000004  08    WRITE_RECEIVED  WRITE_RECEIVED
00000004  59    WRITE_RECEIVED  WRITE_RECEIVED
00000010  xx    WRITE_RECEIVED  INACTIVE

But sometimes:
The first transaction is Slave send (Master read):
    20(addr) 03(singlepart read) 03 1c 42 cc a5

Then the second Master write follow as below:
    20(addr) 02(singlepart write) 03 18 42 0c 63

The irq will raise in sequence below:

  irq      data  from-state      to-state
00000084  20    INACTIVE        WRITE_RECEIVED
00000004  03    WRITE_RECEIVED  WRITE_RECEIVED
00000084  03    WRITE_RECEIVED  READ_PROCESSED
00000001  1c    READ_PROCESSED  READ_PROCESSED
00000001  42    READ_PROCESSED  READ_PROCESSED
00000001  0c    READ_PROCESSED  READ_PROCESSED
00000001  63    READ_PROCESSED  READ_PROCESSED

00000086  20    READ_PROCESSED  WRITE_RECEIVED <= both 3 irqs raised
00000004  02    WRITE_RECEIVED  WRITE_RECEIVED
00000084  03    WRITE_RECEIVED  WRITE_RECEIVED
00000004  18    WRITE_RECEIVED  WRITE_RECEIVED
00000004  42    WRITE_RECEIVED  WRITE_RECEIVED
00000004  0c    WRITE_RECEIVED  WRITE_RECEIVED
00000004  63    WRITE_RECEIVED  WRITE_RECEIVED
00000010  xx    WRITE_RECEIVED  INACTIVE

This patch is to address this case where TX_NAK, SLAVE_MATCH and RX_DONE 
are raised together.

>> +                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
>> +                       i2c_slave_event(slave, I2C_SLAVE_STOP, &value);
>> +               }
>>                  irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
>>                  bus->slave_state = ASPEED_I2C_SLAVE_START;
>>          }
>> --
>> 2.28.0
>>
Quan Nguyen May 20, 2021, 2:15 p.m. UTC | #4
On 20/05/2021 18:28, Ryan Chen wrote:
>> -----Original Message-----
>> From: Joel Stanley <joel@jms.id.au>
>> Sent: Thursday, May 20, 2021 7:29 AM
>> To: Quan Nguyen <quan@os.amperecomputing.com>; Ryan Chen
>> <ryan_chen@aspeedtech.com>
>> Cc: Corey Minyard <minyard@acm.org>; Rob Herring <robh+dt@kernel.org>;
>> Andrew Jeffery <andrew@aj.id.au>; Brendan Higgins
>> <brendanhiggins@google.com>; Benjamin Herrenschmidt
>> <benh@kernel.crashing.org>; Wolfram Sang <wsa@kernel.org>; Philipp Zabel
>> <p.zabel@pengutronix.de>; openipmi-developer@lists.sourceforge.net;
>> devicetree <devicetree@vger.kernel.org>; Linux ARM
>> <linux-arm-kernel@lists.infradead.org>; linux-aspeed
>> <linux-aspeed@lists.ozlabs.org>; Linux Kernel Mailing List
>> <linux-kernel@vger.kernel.org>; linux-i2c@vger.kernel.org; Open Source
>> Submission <patches@amperecomputing.com>; Phong Vo
>> <phong@os.amperecomputing.com>; Thang Q . Nguyen
>> <thang@os.amperecomputing.com>; OpenBMC Maillist
>> <openbmc@lists.ozlabs.org>
>> Subject: Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK
>>
>> Ryan, can you please review this change?
>>
>> On Wed, 19 May 2021 at 07:50, Quan Nguyen
>> <quan@os.amperecomputing.com> wrote:
>>>
>>> It is observed that in normal condition, when the last byte sent by
>>> slave, the Tx Done with NAK irq will raise.
>>> But it is also observed that sometimes master issues next transaction
>>> too quick while the slave irq handler is not yet invoked and Tx Done
>>> with NAK irq of last byte of previous READ PROCESSED was not ack'ed.
>>> This Tx Done with NAK irq is raised together with the Slave Match and
>>> Rx Done irq of the next coming transaction from master.
>>> Unfortunately, the current slave irq handler handles the Slave Match
>>> and Rx Done only in higher priority and ignore the Tx Done with NAK,
>>> causing the complain as below:
>>> "aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected
>>> 0x00000086, but was 0x00000084"
>>>
>>> This commit handles this case by emitting a Slave Stop event for the
>>> Tx Done with NAK before processing Slave Match and Rx Done for the
>>> coming transaction from master.
>>
>> It sounds like this patch is independent of the rest of the series, and can go in
>> on it's own. Please send it separately to the i2c maintainers and add a suitable
>> Fixes line, such as:
>>
>>    Fixes: f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C
>> driver")
>>
>>>
>>> Signed-off-by: Quan Nguyen <quan@os.amperecomputing.com>
>>> ---
>>> v3:
>>>    + First introduce in v3 [Quan]
>>>
>>>   drivers/i2c/busses/i2c-aspeed.c | 5 +++++
>>>   1 file changed, 5 insertions(+)
>>>
>>> diff --git a/drivers/i2c/busses/i2c-aspeed.c
>>> b/drivers/i2c/busses/i2c-aspeed.c index 724bf30600d6..3fb37c3f23d4
>>> 100644
>>> --- a/drivers/i2c/busses/i2c-aspeed.c
>>> +++ b/drivers/i2c/busses/i2c-aspeed.c
>>> @@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct
>>> aspeed_i2c_bus *bus, u32 irq_status)
>>>
>>>          /* Slave was requested, restart state machine. */
>>>          if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {
>>
>> Can you explain why you need to do this handing inside the SLAVE_MATCH
>> case?
>>
>> Could you instead move the TX_NAK handling to be above the SLAVE_MATCH
>> case?
>>
>>> +               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
>>> +                   bus->slave_state ==
>>> + ASPEED_I2C_SLAVE_READ_PROCESSED) {
>>
>> Either way, this needs a comment to explain what we're working around.
>>
>>> +                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
>>> +                       i2c_slave_event(slave, I2C_SLAVE_STOP,
>> &value);
> 
> According the patch assume slave receive TX_NAK will be go to SLAVE_STOP state?
> 
Hi Ryan,

As per my explain in other email, we need to emit one SLAVE_STOP event 
to complete the previous transaction before start processing for the 
next transaction.

- Quan

>>> +               }
>>>                  irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
>>>                  bus->slave_state = ASPEED_I2C_SLAVE_START;
>>>          }
>>> --
>>> 2.28.0
>>>
diff mbox series

Patch

diff --git a/drivers/i2c/busses/i2c-aspeed.c b/drivers/i2c/busses/i2c-aspeed.c
index 724bf30600d6..3fb37c3f23d4 100644
--- a/drivers/i2c/busses/i2c-aspeed.c
+++ b/drivers/i2c/busses/i2c-aspeed.c
@@ -254,6 +254,11 @@  static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status)
 
 	/* Slave was requested, restart state machine. */
 	if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {
+		if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
+		    bus->slave_state == ASPEED_I2C_SLAVE_READ_PROCESSED) {
+			irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+			i2c_slave_event(slave, I2C_SLAVE_STOP, &value);
+		}
 		irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
 		bus->slave_state = ASPEED_I2C_SLAVE_START;
 	}