Ensure tpm_inf_recv() does not overflow the provided buffer when
the TPM reports more data than the caller expects.

Signed-off-by: Shahriyar Jalayeri <shahriyar@posteo.de>
---
 drivers/char/tpm/tpm_infineon.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/drivers/char/tpm/tpm_infineon.c b/drivers/char/tpm/tpm_infineon.c
index 7638b65b8..eb6dd55ff 100644
--- a/drivers/char/tpm/tpm_infineon.c
+++ b/drivers/char/tpm/tpm_infineon.c
@@ -250,6 +250,12 @@ static int tpm_inf_recv(struct tpm_chip *chip, u8 * buf, size_t count)
 	number_of_wtx = 0;
 
 recv_begin:
+	if (count < 4) {
+		dev_err(&chip->dev,
+			"count less than the header size!\n");
+		return -EIO;
+	}
+
 	/* start receiving header */
 	for (i = 0; i < 4; i++) {
 		ret = wait(chip, STAT_RDA);
@@ -268,6 +274,12 @@ static int tpm_inf_recv(struct tpm_chip *chip, u8 * buf, size_t count)
 		/* size of the data received */
 		size = ((buf[2] << 8) | buf[3]);
 
+		if (size > count) {
+			dev_err(&chip->dev,
+				"Buffer too small for incoming data!\n");
+			return -EIO;
+		}
+
 		for (i = 0; i < size; i++) {
 			wait(chip, STAT_RDA);
 			buf[i] = tpm_data_in(RDFIFO);
-- 
2.43.0

On Fri, Oct 03, 2025 at 09:25:47AM +0000, Shahriyar Jalayeri wrote:
> Ensure tpm_inf_recv() does not overflow the provided buffer when
> the TPM reports more data than the caller expects.
> 
> Signed-off-by: Shahriyar Jalayeri <shahriyar@posteo.de>

missing:

Fixes: ebb81fdb3dd0 ("[PATCH] tpm: Support for Infineon TPM")

> ---
>  drivers/char/tpm/tpm_infineon.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/drivers/char/tpm/tpm_infineon.c b/drivers/char/tpm/tpm_infineon.c
> index 7638b65b8..eb6dd55ff 100644
> --- a/drivers/char/tpm/tpm_infineon.c
> +++ b/drivers/char/tpm/tpm_infineon.c
> @@ -250,6 +250,12 @@ static int tpm_inf_recv(struct tpm_chip *chip, u8 * buf, size_t count)
>  	number_of_wtx = 0;
>  
>  recv_begin:
> +	if (count < 4) {
> +		dev_err(&chip->dev,
> +			"count less than the header size!\n");
> +		return -EIO;
> +	}

Please remove dev_err() 

> +
>  	/* start receiving header */
>  	for (i = 0; i < 4; i++) {
>  		ret = wait(chip, STAT_RDA);
> @@ -268,6 +274,12 @@ static int tpm_inf_recv(struct tpm_chip *chip, u8 * buf, size_t count)
>  		/* size of the data received */
>  		size = ((buf[2] << 8) | buf[3]);
>  
> +		if (size > count) {
> +			dev_err(&chip->dev,
> +				"Buffer too small for incoming data!\n");
> +			return -EIO;
> +		}


Ditto
>
>  		for (i = 0; i < size; i++) {
>  			wait(chip, STAT_RDA);
>  			buf[i] = tpm_data_in(RDFIFO);
> -- 
> 2.43.0

BR, Jarkko

Dear Shahriyar,


Thank you for your patch.

Am 03.10.25 um 11:25 schrieb Shahriyar Jalayeri:
> Ensure tpm_inf_recv() does not overflow the provided buffer when
> the TPM reports more data than the caller expects.

Is it possible to enforce this situation to ensure your patch works?

> Signed-off-by: Shahriyar Jalayeri <shahriyar@posteo.de>
> ---
>   drivers/char/tpm/tpm_infineon.c | 12 ++++++++++++
>   1 file changed, 12 insertions(+)
> 
> diff --git a/drivers/char/tpm/tpm_infineon.c b/drivers/char/tpm/tpm_infineon.c
> index 7638b65b8..eb6dd55ff 100644
> --- a/drivers/char/tpm/tpm_infineon.c
> +++ b/drivers/char/tpm/tpm_infineon.c
> @@ -250,6 +250,12 @@ static int tpm_inf_recv(struct tpm_chip *chip, u8 * buf, size_t count)
>   	number_of_wtx = 0;
>   
>   recv_begin:
> +	if (count < 4) {
> +		dev_err(&chip->dev,
> +			"count less than the header size!\n");

Mention both values count and 4?

> +		return -EIO;
> +	}
> +

This is not described in the commit message.

>   	/* start receiving header */
>   	for (i = 0; i < 4; i++) {
>   		ret = wait(chip, STAT_RDA);
> @@ -268,6 +274,12 @@ static int tpm_inf_recv(struct tpm_chip *chip, u8 * buf, size_t count)
>   		/* size of the data received */
>   		size = ((buf[2] << 8) | buf[3]);
>   
> +		if (size > count) {
> +			dev_err(&chip->dev,
> +				"Buffer too small for incoming data!\n");

I’d log both values, and also specify that the operation is aborted.

> +			return -EIO;
> +		}
> +
>   		for (i = 0; i < size; i++) {
>   			wait(chip, STAT_RDA);
>   			buf[i] = tpm_data_in(RDFIFO);


Kind regards,

Paul

Dear Paul,

> 
> Is it possible to enforce this situation to ensure your patch works?
> 

Triggering the issue requires interposing the hardware or a MCU to 
emulate the TPM and send malformed TPM replies, I might be able to test 
this using Qemu, but that takes some time.

For the rest of your suggestions, I'll send another patch.

BR,
/shj