Re: [PATCH v3 00/23] Unrestricted media entity ID range support

[Javier Martinez Canillas <javier@osg.samsung.com>]

Hello Laurent,

On 12/27/2015 02:11 PM, Laurent Pinchart wrote:
> Hi Mauro,
> 
> On Wednesday 23 December 2015 10:32:42 Mauro Carvalho Chehab wrote:
>> Em Wed, 16 Dec 2015 16:03:01 +0200 Sakari Ailus escreveu:
>>> On Wed, Dec 16, 2015 at 03:32:15PM +0200, Sakari Ailus wrote:
>>>> This is the third version of the unrestricted media entity ID range
>>>> support set. I've taken Mauro's comments into account and fixed a number
>>>> of bugs as well (omap3isp memory leak and omap4iss stream start).
>>>
>>> Javier: Mauro told me you might have OMAP4 hardware. Would you be able to
>>> test the OMAP4 ISS with these patches?
>>>
>>> Thanks.
>>
>> Sakari,
>>
>> Testing with OMAP4 is not possible. The driver is broken: it doesn't
>> support DT, and the required pdata definition is missing.
> 
> What do you mean by missing ? struct iss_platform_data is defined in 
> include/media/omap4iss.h.
>

That's true but at the very least the omap4iss hwmod is broken in mainline
as you mentioned in the linux-media thread that Mauro shared below.

I think what Mauro meant is that there isn't an omap4 board supported in
mainline that makes use of the iss platform data structures. So testing the
driver in a popular omap4 board such as the pandaboard isn't possible without
adding plumbing board code. And even in that case, the driver fails to probe
due a missing iss_fck clock as Mauro mentioned in his boot log below as well.

I know is not a requirement for a driver to have mainline users but without
DT bindings, using this driver will not be possible sooner rather than later
once the mach-omap2 pdata-quirks.c workaround is removed from mainline since
the omap4 board files were deleted almost 3 years ago (3.11 according to git).
 
>> Both Javier and I tried to fix it in the last couple days, in order to test
>> it with a PandaBoard. We came with the enclosed patch, but it is still
>> incomplete. Based on what's written on this e-mail:
>> 	 https://www.mail-archive.com/linux-media@vger.kernel.org/msg89247.html
>>
>> It seems that this is an already known issue.
>>
>> So, I'm considering this driver as BROKEN. Not much sense on doing any
>> tests on it, while this doesn't get fixed.
>>
>> Regards,
>> Mauro
>>
>> PS.: With the enclosed patch, I got this error:
>> 	[    0.267639] platform omap4iss: failed to claim resource 2
>>
>> But, even if I comment out the platform code that returns this error,
>> there are still other missing things:
>> 	[    7.131622] omap4iss omap4iss: Unable to get iss_fck clock info
>> 	[    7.137878] omap4iss omap4iss: Unable to get clocks
>>
>> ---
>>
>> ARM: add a pdata quirks for OMAP4 panda camera
>>
>> This is a hack to make it to believe that the pandaboard
>> has a camera.
>>
>>
>> diff --git a/arch/arm/mach-omap2/pdata-quirks.c
>> b/arch/arm/mach-omap2/pdata-quirks.c index 1dfe34654c43..998bb6936dc0
>> 100644
>> --- a/arch/arm/mach-omap2/pdata-quirks.c
>> +++ b/arch/arm/mach-omap2/pdata-quirks.c
>> @@ -36,6 +36,8 @@
>>  #include "soc.h"
>>  #include "hsmmc.h"
>>
>> +#include "../../../drivers/staging/media/omap4iss/iss.h"
>> +
>>  struct pdata_init {
>>  	const char *compatible;
>>  	void (*fn)(void);
>> @@ -408,6 +410,124 @@ static void __init t410_abort_init(void)
>>  }
>>  #endif
>>
>> +#ifdef CONFIG_ARCH_OMAP4
>> +
>> +static struct resource panda_iss_resource[] = {
>> +	{
>> +		.start = 0x52000000,
>> +		.end = 0x52000000 + 0x100,
>> +		.name = "top",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52001000,
>> +		.end = 0x52001000 + 0x170,
>> +		.name = "csi2_a_regs1",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52001170,
>> +		.end = 0x52001170 + 0x020,
>> +		.name = "camerarx_core1",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52001400,
>> +		.end = 0x52001400 + 0x170,
>> +		.name = "csi2_b_regs1",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52001570,
>> +		.end = 0x52001570 + 0x020,
>> +		.name = "camerarx_core2",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52002000,
>> +		.end = 0x52002000 + 0x200,
>> +		.name = "bte",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52010000,
>> +		.end = 0x52010000 + 0x0a0,
>> +		.name = "isp_sys1",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52010400,
>> +		.end = 0x52010400 + 0x400,
>> +		.name = "isp_resizer",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52010800,
>> +		.end = 0x52010800 + 0x800,
>> +		.name = "isp_ipipe",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52011000,
>> +		.end = 0x52011000 + 0x200,
>> +		.name = "isp_isif",
>> +		.flags = IORESOURCE_MEM,
>> +	}, {
>> +		.start = 0x52011200,
>> +		.end = 0x52011200 + 0x080,
>> +		.name = "isp_ipipeif",
>> +		.flags = IORESOURCE_MEM,
>> +	}
>> +};
>> +
>> +static struct i2c_board_info panda_camera_i2c_device = {
>> +	I2C_BOARD_INFO("smia", 0x10),
>> +};
>> +
>> +static struct iss_subdev_i2c_board_info panda_camera_subdevs[] = {
>> +	{
>> +		.board_info = &panda_camera_i2c_device,
>> +		.i2c_adapter_id = 3,
>> +	},
>> +};
>> +
>> +static struct iss_v4l2_subdevs_group iss_subdevs[] = {
>> +	{
>> +		.subdevs = panda_camera_subdevs,
>> +		.interface = ISS_INTERFACE_CSI2A_PHY1,
>> +		.bus = {
>> +			.csi2 = {
>> +				.lanecfg = {
>> +					.clk = {
>> +						.pol = 0,
>> +						.pos = 2,
>> +					},
>> +					.data[0] = {
>> +						.pol = 0,
>> +						.pos = 1,
>> +					},
>> +					.data[1] = {
>> +						.pol = 0,
>> +						.pos = 3,
>> +					},
>> +				},
>> +			} },
>> +	},
>> +	{ /* sentinel */ },
>> +};
>> +
>> +static struct iss_platform_data iss_pdata = {
>> +	.subdevs = iss_subdevs,
>> +};
>> +
>> +static struct platform_device omap4iss_device = {
>> +	.name           = "omap4iss",
>> +	.id             = -1,
>> +	.dev = {
>> +		.platform_data = &iss_pdata,
>> +	},
>> +	.num_resources  = ARRAY_SIZE(panda_iss_resource),
>> +	.resource       = panda_iss_resource,
>> +};
>> +
>> +static void __init omap4_panda_legacy_init(void)
>> +{
>> +	platform_device_register(&omap4iss_device);
>> +}
>> +
>> +#endif /* CONFIG_ARCH_OMAP4 */
>> +
>>  #if defined(CONFIG_ARCH_OMAP4) || defined(CONFIG_SOC_OMAP5)
>>  static struct iommu_platform_data omap4_iommu_pdata = {
>>  	.reset_name = "mmu_cache",
>> @@ -539,6 +659,9 @@ static struct pdata_init pdata_quirks[] __initdata = {
>>  #ifdef CONFIG_SOC_TI81XX
>>  	{ "hp,t410", t410_abort_init, },
>>  #endif
>> +#ifdef CONFIG_ARCH_OMAP4
>> +	{ "ti,omap4-panda", omap4_panda_legacy_init, },
>> +#endif
>>  #ifdef CONFIG_SOC_OMAP5
>>  	{ "ti,omap5-uevm", omap5_uevm_legacy_init, },
>>  #endif
>>
>> diff --git a/drivers/staging/media/omap4iss/iss.c
>> b/drivers/staging/media/omap4iss/iss.c index 30b473cfb020..b528cacda17b
>> 100644
>> --- a/drivers/staging/media/omap4iss/iss.c
>> +++ b/drivers/staging/media/omap4iss/iss.c
>> @@ -1412,6 +1412,9 @@ static int iss_probe(struct platform_device *pdev)
>>  	unsigned int i;
>>  	int ret;
>>
>> +
>> +printk("%s: pdata=%p\n", __func__, pdata);
>> +
>>  	if (!pdata)
>>  		return -EINVAL;
>>
>> @@ -1437,24 +1440,33 @@ static int iss_probe(struct platform_device *pdev)
>>  	iss->syscon = syscon_regmap_lookup_by_compatible("syscon");
>>  	if (IS_ERR(iss->syscon)) {
>>  		ret = PTR_ERR(iss->syscon);
>> +		dev_err(iss->dev, "Unable to find syscon");
>>  		goto error;
>>  	}
>>
>>  	/* Clocks */
>>  	ret = iss_map_mem_resource(pdev, iss, OMAP4_ISS_MEM_TOP);
>> -	if (ret < 0)
>> +	if (ret < 0) {
>> +		dev_err(iss->dev, "Unable to map memory resource\n");
>>  		goto error;
>> +	}
>>
>>  	ret = iss_get_clocks(iss);
>> -	if (ret < 0)
>> +	if (ret < 0) {
>> +		dev_err(iss->dev, "Unable to get clocks\n");
>>  		goto error;
>> +	}
>>
>> -	if (!omap4iss_get(iss))
>> +	if (!omap4iss_get(iss)) {
>> +		dev_err(iss->dev, "Failed to acquire ISS resource\n");
>>  		goto error;
>> +	}
>>
>>  	ret = iss_reset(iss);
>> -	if (ret < 0)
>> +	if (ret < 0) {
>> +		dev_err(iss->dev, "Unable to reset ISS\n");
>>  		goto error_iss;
>> +	}
>>
>>  	iss->revision = iss_reg_read(iss, OMAP4_ISS_MEM_TOP, ISS_HL_REVISION);
>>  	dev_info(iss->dev, "Revision %08x found\n", iss->revision);
> 

Best regards,
-- 
Javier Martinez Canillas
Open Source Group
Samsung Research America