lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Fri, 17 Feb 2017 08:26:55 +0100
From:   Peter Huewe <>
To:     Christophe Ricard <>,
        Peter Huewe <>,
        Jarkko Sakkinen <>
CC:     Jason Gunthorpe <>,,,
        Christophe Ricard <>,,
        Alexander Steffen <>
Subject: Re: [PATCH 5/5] tpm_tis_spi: Add small delay after last transfer

Am 17. Februar 2017 06:09:42 MEZ schrieb Christophe Ricard <>:
>Are you sure it is not better to introduce this delay directly in the 
>rpi spi driver ?
>Other than that i don't see any issue with it.

Yes - it would be perfect to fix the issue in the upstream rpi spi master driver.
However this only happens sporadically in some strange cases (i.e. 300-500khz, single byte transfer after having more bytes in the same #cs frame) (unrelated to the tpm)

We are still looking into it, why /where it happens and how to reproduce it reliably and then file a bug/fix it.

For now however the priority is to make the tpm_tis_spi driver work reliably also on the rpi, and this is what this workaround does.

We can simply remove it once the rpi spi master is fixed.

>On 16/02/2017 08:09, Peter Huewe wrote:
>> Testing the implementation with a Raspberry Pi 2 showed that under
>> circumstances its SPI master erroneously releases the CS line before
>> transfer is complete, i.e. before the end of the last clock. In this
>> the TPM ignores the transfer and misses for example the GO command.
>> driver is unable to detect this communication problem and will wait
>for a
>> command response that is never going to arrive, timing out
>> As a workaround, the small delay ensures that the CS line is held
>> enough, even with a faulty SPI master. Other SPI masters are not
>> except for a negligible performance penalty.
>> Cc: <>
>> Fixes: 0edbfea537d1 ("tpm/tpm_tis_spi: Add support for spi phy")
>> Signed-off-by: Alexander Steffen <>
>> Signed-off-by: Peter Huewe <>
>> ---
>>   drivers/char/tpm/tpm_tis_spi.c | 1 +
>>   1 file changed, 1 insertion(+)
>> diff --git a/drivers/char/tpm/tpm_tis_spi.c
>> index b50c5b072df3..685c51bf5d7e 100644
>> --- a/drivers/char/tpm/tpm_tis_spi.c
>> +++ b/drivers/char/tpm/tpm_tis_spi.c
>> @@ -110,6 +110,7 @@ static int tpm_tis_spi_transfer(struct
>tpm_tis_data *data, u32 addr, u8 len,
>>   		spi_xfer.cs_change = 0;
>>   		spi_xfer.len = transfer_len;
>> +		spi_xfer.delay_usecs = 5;
>>   		if (direction) {
>>   			spi_xfer.tx_buf = NULL;

Sent from my mobile

Powered by blists - more mailing lists