[PATCH v2 1/2] i2c: core: ACPI: Properly set status byte to 0 for multi-byte writes

[PATCH v2 1/2] i2c: core: ACPI: Properly set status byte to 0 for multi-byte writes

[Hans de Goede]

acpi_gsb_i2c_write_bytes() returns i2c_transfer()'s return value, which
is the number of transfers executed on success, so 1.

The ACPI code expects us to store 0 in gsb->status for success, not 1.

Specifically this breaks the following code in the Thinkpad 8 DSDT:

            ECWR = I2CW = ECWR /* \_SB_.I2C1.BAT0.ECWR */
            If ((ECST == Zero))
            {
                ECRD = I2CR /* \_SB_.I2C1.I2CR */
            }

Before this commit we set ECST to 1, causing the read to never happen
breaking battery monitoring on the Thinkpad 8.

This commit makes acpi_gsb_i2c_write_bytes() return 0 when i2c_transfer()
returns 1, so the single write transfer completed successfully, and
makes it return -EIO on for other (unexpected) return values >= 0.

Cc: stable@vger.kernel.org
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
Changes in v2:
-Modify the value which acpi_gsb_i2c_write_bytes() returns instead of
 checking + modifying the return value in its caller
---
 drivers/i2c/i2c-core-acpi.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/i2c/i2c-core-acpi.c b/drivers/i2c/i2c-core-acpi.c
index 7c3b4740b94b..b8f303dea305 100644
--- a/drivers/i2c/i2c-core-acpi.c
+++ b/drivers/i2c/i2c-core-acpi.c
@@ -482,11 +482,16 @@ static int acpi_gsb_i2c_write_bytes(struct i2c_client *client,
 	msgs[0].buf = buffer;
 
 	ret = i2c_transfer(client->adapter, msgs, ARRAY_SIZE(msgs));
-	if (ret < 0)
-		dev_err(&client->adapter->dev, "i2c write failed\n");
 
 	kfree(buffer);
-	return ret;
+
+	if (ret < 0) {
+		dev_err(&client->adapter->dev, "i2c write failed: %d\n", ret);
+		return ret;
+	}
+
+	/* 1 transfer must have completed successfully */
+	return (ret == 1) ? 0 : -EIO;
 }
 
 static acpi_status
-- 
2.18.0

Re: [PATCH v2 1/2] i2c: core: ACPI: Properly set status byte to 0 for multi-byte writes

[Peter Rosin]

On 2018-08-12 12:53, Hans de Goede wrote:
> acpi_gsb_i2c_write_bytes() returns i2c_transfer()'s return value, which
> is the number of transfers executed on success, so 1.
> 
> The ACPI code expects us to store 0 in gsb->status for success, not 1.
> 
> Specifically this breaks the following code in the Thinkpad 8 DSDT:
> 
>             ECWR = I2CW = ECWR /* \_SB_.I2C1.BAT0.ECWR */
>             If ((ECST == Zero))
>             {
>                 ECRD = I2CR /* \_SB_.I2C1.I2CR */
>             }
> 
> Before this commit we set ECST to 1, causing the read to never happen
> breaking battery monitoring on the Thinkpad 8.
> 
> This commit makes acpi_gsb_i2c_write_bytes() return 0 when i2c_transfer()
> returns 1, so the single write transfer completed successfully, and
> makes it return -EIO on for other (unexpected) return values >= 0.

I'm wondering if this might be fallout from one of
35cd67a0caf7 ("i2c: viperboard: return message count on master_xfer success")
de9a8634f1cb ("i2c: pmcmsp: return message count on master_xfer success")

But I have no idea what i2c driver these Thinkpads are using, so it is not
unlikely that I'm way off...

Cheers,
Peter

> Cc: stable@vger.kernel.org
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
> ---
> Changes in v2:
> -Modify the value which acpi_gsb_i2c_write_bytes() returns instead of
>  checking + modifying the return value in its caller

Re: [PATCH v2 1/2] i2c: core: ACPI: Properly set status byte to 0 for multi-byte writes

[Hans de Goede]

Hi,

On 12-08-18 22:39, Peter Rosin wrote:
> On 2018-08-12 12:53, Hans de Goede wrote:
>> acpi_gsb_i2c_write_bytes() returns i2c_transfer()'s return value, which
>> is the number of transfers executed on success, so 1.
>>
>> The ACPI code expects us to store 0 in gsb->status for success, not 1.
>>
>> Specifically this breaks the following code in the Thinkpad 8 DSDT:
>>
>>              ECWR = I2CW = ECWR /* \_SB_.I2C1.BAT0.ECWR */
>>              If ((ECST == Zero))
>>              {
>>                  ECRD = I2CR /* \_SB_.I2C1.I2CR */
>>              }
>>
>> Before this commit we set ECST to 1, causing the read to never happen
>> breaking battery monitoring on the Thinkpad 8.
>>
>> This commit makes acpi_gsb_i2c_write_bytes() return 0 when i2c_transfer()
>> returns 1, so the single write transfer completed successfully, and
>> makes it return -EIO on for other (unexpected) return values >= 0.
> 
> I'm wondering if this might be fallout from one of
> 35cd67a0caf7 ("i2c: viperboard: return message count on master_xfer success")
> de9a8634f1cb ("i2c: pmcmsp: return message count on master_xfer success")

No, these are using the designware driver, so these commits are not
the cause, this simply is a long standing bug.

Regards,

Hans

Re: [PATCH v2 1/2] i2c: core: ACPI: Properly set status byte to 0 for multi-byte writes

[Mika Westerberg]

On Sun, Aug 12, 2018 at 12:53:20PM +0200, Hans de Goede wrote:
> acpi_gsb_i2c_write_bytes() returns i2c_transfer()'s return value, which
> is the number of transfers executed on success, so 1.
> 
> The ACPI code expects us to store 0 in gsb->status for success, not 1.
> 
> Specifically this breaks the following code in the Thinkpad 8 DSDT:
> 
>             ECWR = I2CW = ECWR /* \_SB_.I2C1.BAT0.ECWR */
>             If ((ECST == Zero))
>             {
>                 ECRD = I2CR /* \_SB_.I2C1.I2CR */
>             }
> 
> Before this commit we set ECST to 1, causing the read to never happen
> breaking battery monitoring on the Thinkpad 8.
> 
> This commit makes acpi_gsb_i2c_write_bytes() return 0 when i2c_transfer()
> returns 1, so the single write transfer completed successfully, and
> makes it return -EIO on for other (unexpected) return values >= 0.
> 
> Cc: stable@vger.kernel.org
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>

Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

Re: [PATCH v2 1/2] i2c: core: ACPI: Properly set status byte to 0 for multi-byte writes

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 1007 bytes --]

On Sun, Aug 12, 2018 at 12:53:20PM +0200, Hans de Goede wrote:
> acpi_gsb_i2c_write_bytes() returns i2c_transfer()'s return value, which
> is the number of transfers executed on success, so 1.
> 
> The ACPI code expects us to store 0 in gsb->status for success, not 1.
> 
> Specifically this breaks the following code in the Thinkpad 8 DSDT:
> 
>             ECWR = I2CW = ECWR /* \_SB_.I2C1.BAT0.ECWR */
>             If ((ECST == Zero))
>             {
>                 ECRD = I2CR /* \_SB_.I2C1.I2CR */
>             }
> 
> Before this commit we set ECST to 1, causing the read to never happen
> breaking battery monitoring on the Thinkpad 8.
> 
> This commit makes acpi_gsb_i2c_write_bytes() return 0 when i2c_transfer()
> returns 1, so the single write transfer completed successfully, and
> makes it return -EIO on for other (unexpected) return values >= 0.
> 
> Cc: stable@vger.kernel.org
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>

Applied to for-next, thanks!


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]