linux-kernel - Re: [PATCH v4] i2c: imx: support DMA defer probing

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: <153e8e36-7b0e-4379-9cc3-6dacb5d705be@pengutronix.de>
Date: Wed, 27 Nov 2024 09:38:23 +0100
From: Ahmad Fatoum <a.fatoum@...gutronix.de>
To: carlos.song@....com, mkl@...gutronix.de, frank.li@....com,
 o.rempel@...gutronix.de, kernel@...gutronix.de, andi.shyti@...nel.org,
 shawnguo@...nel.org, s.hauer@...gutronix.de, festevam@...il.com
Cc: imx@...ts.linux.dev, linux-i2c@...r.kernel.org,
 linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4] i2c: imx: support DMA defer probing

Hello Carlos,

On 27.11.24 09:38, carlos.song@....com wrote:
> From: Carlos Song <carlos.song@....com>
> 
> Return -EPROBE_DEFER when dma_request_slave_channel() because DMA driver
> have not ready yet.
> 
> Move i2c_imx_dma_request() before registering I2C adapter to avoid
> infinite loop of .probe() calls to the same driver, see "e8c220fac415
> Revert "i2c: imx: improve the error handling in i2c_imx_dma_request()""
> and "Documentation/driver-api/driver-model/driver.rst".
> 
> Use CPU mode to avoid stuck registering i2c adapter when DMA resources
> are unavailable.

Please try to address open questions before sending new versions of
the patch set. Otherwise, it's difficult to follow the conversation.

Did you see my question[1] on your v2:

| Wouldn't this break probe for all i2c-imx users who have CONFIG_IMX_SDMA
| disabled?
|
| Also I am wondering on what kernel version and what configuration
| (CONFIG_I2C_IMX=?, CONFIG_IMX_SDMA=?) you have that made you run into
| this situation.
|
| I'd have expected that with fw_devlink enabled, the I2C controller wouldn't
| be probed before the DMA provider is available.

[1]: https://lore.kernel.org/all/19a43db4-db5c-4638-9778-d94fb571a206@pengutronix.de/

Thanks,
Ahma

> 
> Signed-off-by: Carlos Song <carlos.song@....com>
> Signed-off-by: Clark Wang <xiaoning.wang@....com>
> ---
> Change for V4:
> - Output "Only use PIO mode" log in debug level when no DMA configure.
> Change for V3:
> - According to Marc's comment, remove error print when defer probe.
>   Add info log when DMA has not been enabled and add error log when
>   DMA error, both won't stuck the i2c adapter register, just for reminding,
>   i2c adapter is working only in PIO mode.
> Change for V2:
> - According to Frank's comments, wrap at 75 char and Simplify fix code
>   at i2c_imx_dma_request().
> - Use strict patch check, fix alignment warning at i2c_imx_dma_request()
> ---
>  drivers/i2c/busses/i2c-imx.c | 31 +++++++++++++++++++++++--------
>  1 file changed, 23 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-imx.c b/drivers/i2c/busses/i2c-imx.c
> index 5ed4cb61e262..b11d66d56c55 100644
> --- a/drivers/i2c/busses/i2c-imx.c
> +++ b/drivers/i2c/busses/i2c-imx.c
> @@ -397,17 +397,16 @@ static void i2c_imx_reset_regs(struct imx_i2c_struct *i2c_imx)
>  }
>  
>  /* Functions for DMA support */
> -static void i2c_imx_dma_request(struct imx_i2c_struct *i2c_imx,
> -						dma_addr_t phy_addr)
> +static int i2c_imx_dma_request(struct imx_i2c_struct *i2c_imx, dma_addr_t phy_addr)
>  {
>  	struct imx_i2c_dma *dma;
>  	struct dma_slave_config dma_sconfig;
> -	struct device *dev = &i2c_imx->adapter.dev;
> +	struct device *dev = i2c_imx->adapter.dev.parent;
>  	int ret;
>  
>  	dma = devm_kzalloc(dev, sizeof(*dma), GFP_KERNEL);
>  	if (!dma)
> -		return;
> +		return -ENOMEM;
>  
>  	dma->chan_tx = dma_request_chan(dev, "tx");
>  	if (IS_ERR(dma->chan_tx)) {
> @@ -452,7 +451,7 @@ static void i2c_imx_dma_request(struct imx_i2c_struct *i2c_imx,
>  	dev_info(dev, "using %s (tx) and %s (rx) for DMA transfers\n",
>  		dma_chan_name(dma->chan_tx), dma_chan_name(dma->chan_rx));
>  
> -	return;
> +	return 0;
>  
>  fail_rx:
>  	dma_release_channel(dma->chan_rx);
> @@ -460,6 +459,8 @@ static void i2c_imx_dma_request(struct imx_i2c_struct *i2c_imx,
>  	dma_release_channel(dma->chan_tx);
>  fail_al:
>  	devm_kfree(dev, dma);
> +
> +	return ret;
>  }
>  
>  static void i2c_imx_dma_callback(void *arg)
> @@ -1803,6 +1804,23 @@ static int i2c_imx_probe(struct platform_device *pdev)
>  	if (ret == -EPROBE_DEFER)
>  		goto clk_notifier_unregister;
>  
> +	/*
> +	 * Init DMA config if supported, -ENODEV means DMA not enabled at
> +	 * this platform, that is not a real error, so just remind "only
> +	 * PIO mode is used". If DMA is enabled, but meet error when request
> +	 * DMA channel, error should be showed in probe error log. PIO mode
> +	 * should be available regardless of DMA.
> +	 */
> +	ret = i2c_imx_dma_request(i2c_imx, phy_addr);
> +	if (ret) {
> +		if (ret == -EPROBE_DEFER)
> +			goto clk_notifier_unregister;
> +		else if (ret == -ENODEV)
> +			dev_dbg(&pdev->dev, "Only use PIO mode\n");
> +		else
> +			dev_err_probe(&pdev->dev, ret, "Failed to setup DMA, only use PIO mode\n");
> +	}
> +
>  	/* Add I2C adapter */
>  	ret = i2c_add_numbered_adapter(&i2c_imx->adapter);
>  	if (ret < 0)
> @@ -1817,9 +1835,6 @@ static int i2c_imx_probe(struct platform_device *pdev)
>  		i2c_imx->adapter.name);
>  	dev_info(&i2c_imx->adapter.dev, "IMX I2C adapter registered\n");
>  
> -	/* Init DMA config if supported */
> -	i2c_imx_dma_request(i2c_imx, phy_addr);
> -
>  	return 0;   /* Return OK */
>  
>  clk_notifier_unregister:


-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |