Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1556311 > unrolled thread

[PATCH v2] i2c: piix4: Avoid race conditions with IMC

Back to article view | Back to linux.kernel

Contents

  [PATCH v2] i2c: piix4: Avoid race conditions with IMC Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com> - 2017-01-11 10:20 +0100
    Re: [PATCH v2] i2c: piix4: Avoid race conditions with IMC Andy Shevchenko <andy.shevchenko@gmail.com> - 2017-01-11 17:40 +0100
      Re: [PATCH v2] i2c: piix4: Avoid race conditions with IMC Wolfram Sang <wsa@the-dreams.de> - 2017-01-12 20:40 +0100
        Re: [PATCH v2] i2c: piix4: Avoid race conditions with IMC Andy Shevchenko <andy.shevchenko@gmail.com> - 2017-01-12 20:40 +0100
          Re: [PATCH v2] i2c: piix4: Avoid race conditions with IMC Wolfram Sang <wsa@the-dreams.de> - 2017-01-12 21:00 +0100
            Re: [PATCH v2] i2c: piix4: Avoid race conditions with IMC Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com> - 2017-01-12 21:10 +0100
              Re: [PATCH v2] i2c: piix4: Avoid race conditions with IMC Wolfram Sang <wsa@the-dreams.de> - 2017-01-12 21:30 +0100
              Re: [PATCH v2] i2c: piix4: Avoid race conditions with IMC Andy Shevchenko <andy.shevchenko@gmail.com> - 2017-01-12 21:50 +0100

#1556311 — [PATCH v2] i2c: piix4: Avoid race conditions with IMC

On AMD's SB800 and upwards, the SMBus is shared with the Integrated
Micro Controller (IMC).

The platform provides a hardware semaphore to avoid race conditions
among them. (Check page 288 of the SB800-Series Southbridges Register
Reference Guide http://support.amd.com/TechDocs/45482.pdf)

Without this patch, many access to the SMBus end with an invalid
transaction or even with the bus stalled.

Credit-to: Alexandre Desnoyers <alex@qtec.com>
Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
---

v2: Suggestions by Andy Shevchenko <andy.shevchenko@gmail.com>:
 -Rename timeout to retries
 -Use do {} while(--retries) pattern
 -Group new variables


 drivers/i2c/busses/i2c-piix4.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/drivers/i2c/busses/i2c-piix4.c b/drivers/i2c/busses/i2c-piix4.c
index f0563f7ce01b..81d06be0a72d 100644
--- a/drivers/i2c/busses/i2c-piix4.c
+++ b/drivers/i2c/busses/i2c-piix4.c
@@ -585,10 +585,29 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
 		 u8 command, int size, union i2c_smbus_data *data)
 {
 	struct i2c_piix4_adapdata *adapdata = i2c_get_adapdata(adap);
+	unsigned short piix4_smba = adapdata->smba;
+	int retries = MAX_TIMEOUT;
+	int smbslvcnt;
 	u8 smba_en_lo;
 	u8 port;
 	int retval;
 
+	/* Request the SMBUS semaphore, avoid conflicts with the IMC */
+	smbslvcnt  = inb_p(SMBSLVCNT);
+	do {
+		outb_p(smbslvcnt | 0x10, SMBSLVCNT);
+
+		/* Check the semaphore status */
+		smbslvcnt  = inb_p(SMBSLVCNT);
+		if (smbslvcnt & 0x10)
+			break;
+
+		usleep_range(1000, 2000);
+	} while (--retries);
+	/* SMBus is still owned by the IMC, we give up */
+	if (!retries)
+		return -EBUSY;
+
 	mutex_lock(&piix4_mutex_sb800);
 
 	outb_p(piix4_port_sel_sb800, SB800_PIIX4_SMB_IDX);
@@ -606,6 +625,9 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
 
 	mutex_unlock(&piix4_mutex_sb800);
 
+	/* Release the semaphore */
+	outb_p(smbslvcnt | 0x20, SMBSLVCNT);
+
 	return retval;
 }
 
-- 
2.11.0

[toc] | [next] | [standalone]

#1556698

On Wed, Jan 11, 2017 at 11:11 AM, Ricardo Ribalda Delgado
<ricardo.ribalda@gmail.com> wrote:
> On AMD's SB800 and upwards, the SMBus is shared with the Integrated
> Micro Controller (IMC).
>
> The platform provides a hardware semaphore to avoid race conditions
> among them. (Check page 288 of the SB800-Series Southbridges Register
> Reference Guide http://support.amd.com/TechDocs/45482.pdf)
>
> Without this patch, many access to the SMBus end with an invalid
> transaction or even with the bus stalled.
>

After addressing below comment
FWIW: Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>:

> Credit-to: Alexandre Desnoyers <alex@qtec.com>

It would be nice to put in plain test what you tell in the discussion
here instead of non-standard (see submitting-patches.rst) tag.

> Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
> ---
>
> v2: Suggestions by Andy Shevchenko <andy.shevchenko@gmail.com>:
>  -Rename timeout to retries
>  -Use do {} while(--retries) pattern
>  -Group new variables
>
>
>  drivers/i2c/busses/i2c-piix4.c | 22 ++++++++++++++++++++++
>  1 file changed, 22 insertions(+)
>
> diff --git a/drivers/i2c/busses/i2c-piix4.c b/drivers/i2c/busses/i2c-piix4.c
> index f0563f7ce01b..81d06be0a72d 100644
> --- a/drivers/i2c/busses/i2c-piix4.c
> +++ b/drivers/i2c/busses/i2c-piix4.c
> @@ -585,10 +585,29 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
>                  u8 command, int size, union i2c_smbus_data *data)
>  {
>         struct i2c_piix4_adapdata *adapdata = i2c_get_adapdata(adap);
> +       unsigned short piix4_smba = adapdata->smba;
> +       int retries = MAX_TIMEOUT;
> +       int smbslvcnt;
>         u8 smba_en_lo;
>         u8 port;
>         int retval;
>
> +       /* Request the SMBUS semaphore, avoid conflicts with the IMC */
> +       smbslvcnt  = inb_p(SMBSLVCNT);
> +       do {
> +               outb_p(smbslvcnt | 0x10, SMBSLVCNT);
> +
> +               /* Check the semaphore status */
> +               smbslvcnt  = inb_p(SMBSLVCNT);
> +               if (smbslvcnt & 0x10)
> +                       break;
> +
> +               usleep_range(1000, 2000);
> +       } while (--retries);
> +       /* SMBus is still owned by the IMC, we give up */
> +       if (!retries)
> +               return -EBUSY;
> +
>         mutex_lock(&piix4_mutex_sb800);
>
>         outb_p(piix4_port_sel_sb800, SB800_PIIX4_SMB_IDX);
> @@ -606,6 +625,9 @@ static s32 piix4_access_sb800(struct i2c_adapter *adap, u16 addr,
>
>         mutex_unlock(&piix4_mutex_sb800);
>
> +       /* Release the semaphore */
> +       outb_p(smbslvcnt | 0x20, SMBSLVCNT);
> +
>         return retval;
>  }
>
> --
> 2.11.0
>



-- 
With Best Regards,
Andy Shevchenko

[toc] | [prev] | [next] | [standalone]

#1557747

> After addressing below comment
> FWIW: Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>:

@andy: Please don't prefix tags. Patchwork doesn't seem to find them like this :/

> > Credit-to: Alexandre Desnoyers <alex@qtec.com>
> 
> It would be nice to put in plain test what you tell in the discussion
> here instead of non-standard (see submitting-patches.rst) tag.

@ricardo: I agree. Is this Reported-by maybe?

[toc] | [prev] | [next] | [standalone]

#1557750

On Thu, Jan 12, 2017 at 9:29 PM, Wolfram Sang <wsa@the-dreams.de> wrote:
>
>> After addressing below comment
>> FWIW:

Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>:

> @andy: Please don't prefix tags. Patchwork doesn't seem to find them like this :/

Oops, sorry, didn't notice that. I usually use a new line for them.

>> > Credit-to: Alexandre Desnoyers <alex@qtec.com>
>>
>> It would be nice to put in plain test what you tell in the discussion
>> here instead of non-standard (see submitting-patches.rst) tag.
>
> @ricardo: I agree. Is this Reported-by maybe?

Yeah, since it was mentioned IIRC that guy is hardware engineer I
don't think Suggested-by would work here better.

-- 
With Best Regards,
Andy Shevchenko

[toc] | [prev] | [next] | [standalone]

#1557757

> > @andy: Please don't prefix tags. Patchwork doesn't seem to find them like this :/
> 
> Oops, sorry, didn't notice that. I usually use a new line for them.

Thanks for fixing it :)

> > @ricardo: I agree. Is this Reported-by maybe?
> 
> Yeah, since it was mentioned IIRC that guy is hardware engineer I
> don't think Suggested-by would work here better.

So, I changed that and applied to for-current so it can be in my next
pull-request.

Thanks!

[toc] | [prev] | [next] | [standalone]

#1557764

Hi Wolfram and Andy,


I have already sent a v3 with the latest suggested changes.

On Thu, Jan 12, 2017 at 8:54 PM, Wolfram Sang <wsa@the-dreams.de> wrote:
>> > @andy: Please don't prefix tags. Patchwork doesn't seem to find them like this :/
>>
>> Oops, sorry, didn't notice that. I usually use a new line for them.
>
> Thanks for fixing it :)
>
>> > @ricardo: I agree. Is this Reported-by maybe?

I still believe that we miss a Credit-to: or Co-author: tag ;). Most
times we do not give enough credit to the hardware engineers. I would
have taken me a long time through dark AMD doc to figure this out. But
for the meantime Reported-by will do the job.

Thanks all of you for your fast review!


-- 
Ricardo Ribalda

[toc] | [prev] | [next] | [standalone]

#1557778

[Multipart message — attachments visible in raw view] — view raw

> I still believe that we miss a Credit-to: or Co-author: tag ;). Most

Co-author is just a second signed-off (in my book at least).

> times we do not give enough credit to the hardware engineers.

True. Yet again, I'd favor simply multiple SoB here.

> Thanks all of you for your fast review!

You are welcome!

[toc] | [prev] | [next] | [standalone]

#1557795

On Thu, Jan 12, 2017 at 10:02 PM, Ricardo Ribalda Delgado
<ricardo.ribalda@gmail.com> wrote:
>>> > @ricardo: I agree. Is this Reported-by maybe?
>
> I still believe that we miss a Credit-to: or Co-author: tag ;)

We have more than enough of tags.

-- 
With Best Regards,
Andy Shevchenko

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web