All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jean Delvare <jdelvare@suse.de>
To: Heiner Kallweit <hkallweit1@gmail.com>
Cc: linux-i2c@vger.kernel.org
Subject: Re: [PATCH] i2c: i801: Fix handling SMBHSTCNT_PEC_EN
Date: Wed, 21 Jul 2021 14:46:20 +0200	[thread overview]
Message-ID: <20210721144620.00671c3d@endymion> (raw)
In-Reply-To: <15db81d0-ddbd-b590-3996-51e588c5b10a@gmail.com>

Hi Heiner,

On Mon, 21 Jun 2021 23:08:40 +0200, Heiner Kallweit wrote:
> Bit SMBHSTCNT_PEC_EN is used only if software calculates the CRC and
> uses register SMBPEC. This is not supported by the driver, it supports
> hw-calculation of CRC only (using bit SMBAUXSTS_CRCE). The chip spec
> states the following, therefore never set bit SMBHSTCNT_PEC_EN.
> 
> Chapter SMBus CRC Generation and Checking
> If the AAC bit is set in the Auxiliary Control register, the PCH
> automatically calculates and drives CRC at the end of the transmitted
> packet for write cycles, and will check the CRC for read cycles. It will
> not transmit the contents of the PEC register for CRC. The PEC bit must
> not be set in the Host Control register. If this bit is set, unspecified
> behavior will result.
> 
> This patch is based solely on the specification and compile-tested only,
> because I have no PEC-capable devices.
> 
> Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
> ---
> This patch may be a candidate for stable (with a little bit of fuzz)
> once somebody with a PEC-capable device has tested it.

I don't think so. One of the conditions for stable branch acceptance is
that the patch must fix a real problem that bothers people. Here we
have an issue that is present for 20 years and nobody ever reported.

I'm all for fixing it upstream, but not in stable kernels.

As for testing, I also don't have a PEC-cable device at hand. However,
we may still be able to test this change:
* If you have a device at 0x69 on the i801 SMBus of any of your system,
  that would be a clock device, which almost always support PEC.
* If you have EEPROMs on your i801 SMBus, you may be lucky and find a
  sequence of bytes where the PEC computation leads to exactly the
  value of the following byte. I remember doing that years ago, sadly I
  can no longer find the script I wrote at that time. Be careful when
  accessing SPD EEPROMs, you want to read from them, not write to them
  ;-) Incidentally i2c-tools was just improved to allow arbitrary SMBus
  block read commands so i2cget can be used for easy testing from
  user-space.

> ---
>  drivers/i2c/busses/i2c-i801.c | 23 +++++++++--------------
>  1 file changed, 9 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-i801.c b/drivers/i2c/busses/i2c-i801.c
> index 0153a21f4..161176fcd 100644
> --- a/drivers/i2c/busses/i2c-i801.c
> +++ b/drivers/i2c/busses/i2c-i801.c
> @@ -516,12 +516,9 @@ static int i801_transaction(struct i801_priv *priv, int xact)
>  
>  static int i801_block_transaction_by_block(struct i801_priv *priv,
>  					   union i2c_smbus_data *data,
> -					   char read_write, int command,
> -					   int hwpec)
> +					   char read_write, int command)
>  {
> -	int i, len;
> -	int status;
> -	int xact = hwpec ? SMBHSTCNT_PEC_EN : 0;
> +	int i, len, status, xact = 0;

This can be simplified further. xact no longer needs to be initialized
here, you can set its value directly below instead of using bit masking.

>  
>  	switch (command) {
>  	case I2C_SMBUS_BLOCK_PROC_CALL:
> @@ -678,8 +675,7 @@ static irqreturn_t i801_isr(int irq, void *dev_id)
>   */
>  static int i801_block_transaction_byte_by_byte(struct i801_priv *priv,
>  					       union i2c_smbus_data *data,
> -					       char read_write, int command,
> -					       int hwpec)
> +					       char read_write, int command)
>  {
>  	int i, len;
>  	int smbcmd;
> @@ -777,9 +773,8 @@ static int i801_set_block_buffer_mode(struct i801_priv *priv)
>  }
>  
>  /* Block transaction function */
> -static int i801_block_transaction(struct i801_priv *priv,
> -				  union i2c_smbus_data *data, char read_write,
> -				  int command, int hwpec)
> +static int i801_block_transaction(struct i801_priv *priv, union i2c_smbus_data *data,
> +				  char read_write, int command)
>  {
>  	int result = 0;
>  	unsigned char hostc;
> @@ -815,11 +810,11 @@ static int i801_block_transaction(struct i801_priv *priv,
>  	 && i801_set_block_buffer_mode(priv) == 0)
>  		result = i801_block_transaction_by_block(priv, data,
>  							 read_write,
> -							 command, hwpec);
> +							 command);
>  	else
>  		result = i801_block_transaction_byte_by_byte(priv, data,
>  							     read_write,
> -							     command, hwpec);
> +							     command);
>  
>  	if (command == I2C_SMBUS_I2C_BLOCK_DATA
>  	 && read_write == I2C_SMBUS_WRITE) {
> @@ -936,8 +931,7 @@ static s32 i801_access(struct i2c_adapter *adap, u16 addr,
>  		       SMBAUXCTL(priv));
>  
>  	if (block)
> -		ret = i801_block_transaction(priv, data, read_write, size,
> -					     hwpec);
> +		ret = i801_block_transaction(priv, data, read_write, size);
>  	else
>  		ret = i801_transaction(priv, xact);
>  
> @@ -1625,6 +1619,7 @@ static void i801_setup_hstcfg(struct i801_priv *priv)
>  	unsigned char hstcfg = priv->original_hstcfg;
>  
>  	hstcfg &= ~SMBHSTCFG_I2C_EN;	/* SMBus timing */
> +	hstcfg &= ~SMBHSTCNT_PEC_EN;

A comment like /* Disable software PEC */ could be added here.

>  	hstcfg |= SMBHSTCFG_HST_EN;
>  	pci_write_config_byte(priv->pci_dev, SMBHSTCFG, hstcfg);
>  


-- 
Jean Delvare
SUSE L3 Support

  parent reply	other threads:[~2021-07-21 12:46 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-06-21 21:08 [PATCH] i2c: i801: Fix handling SMBHSTCNT_PEC_EN Heiner Kallweit
2021-07-18 13:51 ` Heiner Kallweit
2021-07-21  8:55   ` Jean Delvare
2021-07-21 12:46 ` Jean Delvare [this message]
2021-07-22  8:34   ` Jean Delvare
2021-07-22 20:02     ` Heiner Kallweit
2021-07-27  9:38       ` Jean Delvare

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210721144620.00671c3d@endymion \
    --to=jdelvare@suse.de \
    --cc=hkallweit1@gmail.com \
    --cc=linux-i2c@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.