2019-11-10 16:34:18

by Amir Mizinski

[permalink] [raw]
Subject: [PATCH v1 3/5] char: tpm: rewrite "tpm_tis_req_canceled()"

From: Amir Mizinski <[email protected]>

using this function while read/write data resulted in aborted operation.
after investigating according to TCG TPM Profile (PTP) Specifications,
i found cancel should happen only if TPM_STS.commandReady bit is lit
and couldn't find a case when the current condition is valid.
also only cmdReady bit need to be compared instead of the full lower status register byte.

Signed-off-by: Amir Mizinski <[email protected]>
---
drivers/char/tpm/tpm_tis_core.c | 12 +-----------
1 file changed, 1 insertion(+), 11 deletions(-)

diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/tpm_tis_core.c
index ce7f8a1..9016f06 100644
--- a/drivers/char/tpm/tpm_tis_core.c
+++ b/drivers/char/tpm/tpm_tis_core.c
@@ -627,17 +627,7 @@ static int probe_itpm(struct tpm_chip *chip)

static bool tpm_tis_req_canceled(struct tpm_chip *chip, u8 status)
{
- struct tpm_tis_data *priv = dev_get_drvdata(&chip->dev);
-
- switch (priv->manufacturer_id) {
- case TPM_VID_WINBOND:
- return ((status == TPM_STS_VALID) ||
- (status == (TPM_STS_VALID | TPM_STS_COMMAND_READY)));
- case TPM_VID_STM:
- return (status == (TPM_STS_VALID | TPM_STS_COMMAND_READY));
- default:
- return (status == TPM_STS_COMMAND_READY);
- }
+ return ((status & TPM_STS_COMMAND_READY) == TPM_STS_COMMAND_READY);
}

static irqreturn_t tis_int_handler(int dummy, void *dev_id)
--
2.7.4


2019-11-10 18:04:08

by Jerry Snitselaar

[permalink] [raw]
Subject: Re: [PATCH v1 3/5] char: tpm: rewrite "tpm_tis_req_canceled()"

On Sun Nov 10 19, [email protected] wrote:
>From: Amir Mizinski <[email protected]>
>
>using this function while read/write data resulted in aborted operation.
>after investigating according to TCG TPM Profile (PTP) Specifications,
>i found cancel should happen only if TPM_STS.commandReady bit is lit
>and couldn't find a case when the current condition is valid.
>also only cmdReady bit need to be compared instead of the full lower status register byte.
>
>Signed-off-by: Amir Mizinski <[email protected]>
>---
> drivers/char/tpm/tpm_tis_core.c | 12 +-----------
> 1 file changed, 1 insertion(+), 11 deletions(-)
>
>diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/tpm_tis_core.c
>index ce7f8a1..9016f06 100644
>--- a/drivers/char/tpm/tpm_tis_core.c
>+++ b/drivers/char/tpm/tpm_tis_core.c
>@@ -627,17 +627,7 @@ static int probe_itpm(struct tpm_chip *chip)
>
> static bool tpm_tis_req_canceled(struct tpm_chip *chip, u8 status)
> {
>- struct tpm_tis_data *priv = dev_get_drvdata(&chip->dev);
>-
>- switch (priv->manufacturer_id) {
>- case TPM_VID_WINBOND:
>- return ((status == TPM_STS_VALID) ||
>- (status == (TPM_STS_VALID | TPM_STS_COMMAND_READY)));
>- case TPM_VID_STM:
>- return (status == (TPM_STS_VALID | TPM_STS_COMMAND_READY));

Stefan were these cases you found that were deviating from the spec? Wondering
if dropping these will cause issues for these devices.

>- default:
>- return (status == TPM_STS_COMMAND_READY);
>- }
>+ return ((status & TPM_STS_COMMAND_READY) == TPM_STS_COMMAND_READY);
> }
>
> static irqreturn_t tis_int_handler(int dummy, void *dev_id)
>--
>2.7.4
>

2019-11-12 01:27:04

by Stefan Berger

[permalink] [raw]
Subject: Re: [PATCH v1 3/5] char: tpm: rewrite "tpm_tis_req_canceled()"

On 11/10/19 1:00 PM, Jerry Snitselaar wrote:
> On Sun Nov 10 19, [email protected] wrote:
>> From: Amir Mizinski <[email protected]>
>>
>> using this function while read/write data resulted in aborted operation.
>> after investigating according to TCG TPM Profile (PTP) Specifications,
>> i found cancel should happen only if TPM_STS.commandReady bit is lit
>> and couldn't find a case when the current condition is valid.
>> also only cmdReady bit need to be compared instead of the full lower
>> status register byte.
>>
>> Signed-off-by: Amir Mizinski <[email protected]>
>> ---
>> drivers/char/tpm/tpm_tis_core.c | 12 +-----------
>> 1 file changed, 1 insertion(+), 11 deletions(-)
>>
>> diff --git a/drivers/char/tpm/tpm_tis_core.c
>> b/drivers/char/tpm/tpm_tis_core.c
>> index ce7f8a1..9016f06 100644
>> --- a/drivers/char/tpm/tpm_tis_core.c
>> +++ b/drivers/char/tpm/tpm_tis_core.c
>> @@ -627,17 +627,7 @@ static int probe_itpm(struct tpm_chip *chip)
>>
>> static bool tpm_tis_req_canceled(struct tpm_chip *chip, u8 status)
>> {
>> -??? struct tpm_tis_data *priv = dev_get_drvdata(&chip->dev);
>> -
>> -??? switch (priv->manufacturer_id) {
>> -??? case TPM_VID_WINBOND:
>> -??????? return ((status == TPM_STS_VALID) ||
>> -??????????? (status == (TPM_STS_VALID | TPM_STS_COMMAND_READY)));
>> -??? case TPM_VID_STM:
>> -??????? return (status == (TPM_STS_VALID | TPM_STS_COMMAND_READY));
>
> Stefan were these cases you found that were deviating from the spec?
> Wondering
> if dropping these will cause issues for these devices.


I believe these devices needed special handling of the status register
as they didn't behave as the 'other' devices, so I would expect issues.

?? Stefan