linux-kernel - Re: [PATCH v1] i2c: tegra: Fix i2c_writesl() to use writel() instead of writesl()

lists.openwall.net		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]

Message-ID: <20201020074846.GA1877013@ulmo>
Date:   Tue, 20 Oct 2020 09:48:46 +0200
From:   Thierry Reding <thierry.reding@...il.com>
To:     Sowjanya Komatineni <skomatineni@...dia.com>
Cc:     jonathanh@...dia.com, digetx@...il.com,
        linux-tegra@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-i2c@...r.kernel.org
Subject: Re: [PATCH v1] i2c: tegra: Fix i2c_writesl() to use writel() instead
 of writesl()

On Mon, Oct 19, 2020 at 09:03:54PM -0700, Sowjanya Komatineni wrote:
> VI I2C don't have DMA support and uses PIO mode all the time.
> 
> Current driver uses writesl() to fill TX FIFO based on available
> empty slots and with this seeing strange silent hang during any I2C
> register access after filling TX FIFO with 8 words.
> 
> Using writel() followed by i2c_readl() in a loop to write all words
> to TX FIFO instead of using writesl() helps for large transfers in
> PIO mode.
> 
> So, this patch updates i2c_writesl() API to use writel() in a loop
> instead of writesl().
> 
> Signed-off-by: Sowjanya Komatineni <skomatineni@...dia.com>
> ---
>  drivers/i2c/busses/i2c-tegra.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-tegra.c b/drivers/i2c/busses/i2c-tegra.c
> index 6f08c0c..274bf3a 100644
> --- a/drivers/i2c/busses/i2c-tegra.c
> +++ b/drivers/i2c/busses/i2c-tegra.c
> @@ -333,10 +333,13 @@ static u32 i2c_readl(struct tegra_i2c_dev *i2c_dev, unsigned int reg)
>  	return readl_relaxed(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, reg));
>  }
>  
> -static void i2c_writesl(struct tegra_i2c_dev *i2c_dev, void *data,
> +static void i2c_writesl(struct tegra_i2c_dev *i2c_dev, u32 *data,
>  			unsigned int reg, unsigned int len)
>  {
> -	writesl(i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, reg), data, len);
> +	while (len--) {
> +		writel(*data++, i2c_dev->base + tegra_i2c_reg_addr(i2c_dev, reg));
> +		i2c_readl(i2c_dev, I2C_INT_STATUS);
> +	}
>  }
>  
>  static void i2c_readsl(struct tegra_i2c_dev *i2c_dev, void *data,
> @@ -811,7 +814,7 @@ static int tegra_i2c_fill_tx_fifo(struct tegra_i2c_dev *i2c_dev)
>  		i2c_dev->msg_buf_remaining = buf_remaining;
>  		i2c_dev->msg_buf = buf + words_to_transfer * BYTES_PER_FIFO_WORD;
>  
> -		i2c_writesl(i2c_dev, buf, I2C_TX_FIFO, words_to_transfer);
> +		i2c_writesl(i2c_dev, (u32 *)buf, I2C_TX_FIFO, words_to_transfer);

I've thought a bit more about this and I wonder if we're simply reading
out the wrong value for tx_fifo_avail and therefore end up overflowing
the TX FIFO. Have you checked what the value is for tx_fifo_avail when
this silent hang occurs? Given that this is specific to the VI I2C I'm
wondering if this is perhaps a hardware bug where we read the wrong TX
FIFO available count.

Thierry

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)