RE: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
Santiago, Kevin, -Original Message- From: Kevin Hilman [mailto:khil...@deeprootsystems.com] Sent: Monday, October 26, 2009 5:35 PM To: santiago.nu...@ridgerun.com Cc: Narnakaje, Snehaprabha; davinci-linux-open- sou...@linux.davincidsp.com; todd.fisc...@ridgerun.com; linux- me...@vger.kernel.org Subject: Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board Santiago Nunez-Corrales snu...@ridgerun.com writes: Kevin Hilman wrote: santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach- davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. Kevin Kevin, I've been working on this particular implementation. This board-dm365-evm.c is specific to the board, therefore I don't still get the point of not having those values wired to the board file, but I know it'd be nice to have the CPLD configuration triggered upon TVP7002 detection. I see two options: Having them in the board file is appropriate, what I object to is the selection by Kconfig. Run-time detection is always preferred when possible. We have this CPLD init API - evm_init_cpld() called from the dm365_evm_init() function. The CPLD init API was trying to initialize the CPLD, based on the default configuration. I believe David Brownell had this placeholder for have_imager() and have_tvp7002() APIs since, we have different CPLD settings for the imager, tvp7002 and tvp5146 for the video input source. 1. Do the callback function inside pdata and initialize it at driver load time (tvp7002_probe). Set tvp5146 as default and override when driver loads (and restore when unloads). This is the preferred option to me. We can decide on the default video input source to be TVP5146. However we do not need a new callback function. We already have the VPFE .setup_input callback API dm365evm_setup_video_input() for the same purpose. The VPFE .setup_input API is called when each of the decoders (sub-devices) registered with VPFE capture driver. It is also called when the application decides on an input source and switches between the input sources. So, TVP5146 remains as the default video input source, only until VPFE probe is called, which registers the all decoders (sub-devices) defined in the VPFE platform data. This also means that the last decoder in the VPFE platform data remains active after boot-up. One can change the order in the VPFE platform data if a particular input source needs to remain active after boot-up. Note that application can always switch the input-source at run time. I quickly tried removing the have_imager() construct in the evm_init_cpld() and here is the output - Starting kernel ... Uncompressing Linux... done, booting the kernel. Linux version 2.6.32-rc2-davinci1-dirty ... CPU: Testing write buffer coherency: ok DaVinci: 8 gpio irqs NET: Registered protocol family 16 EVM: tvp5146 SD video input bio: create slab bio-0 at 0 SCSI subsystem initialized ... vpfe_init vpfe-capture: vpss clock vpss_master enabled vpfe-capture vpfe-capture: v4l2 device registered vpfe-capture vpfe-capture: video device registered EVM: switch to tvp5146 SD video input tvp514x 1-005d: tvp514x 1-005d decoder driver registered !! vpfe-capture vpfe-capture: v4l2 sub device tvp5146 registered EVM: switch to tvp7002 HD video input tvp7002 1-005c: tvp7002 1-005c decoder driver registered !! vpfe-capture vpfe-capture: v4l2 sub device tvp7002
Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
Sneha, So, if I got it right, have_tvp7002 is unnecessary since that initialization is done via the CPLD interface. Therefore, all I need to do is actually remove the logic for the device from board-dm365-evm.h. Am I right? If so, should I also remove the logic for tvp5146? Regards, Narnakaje, Snehaprabha wrote: Santiago, Kevin, -Original Message- From: Kevin Hilman [mailto:khil...@deeprootsystems.com] Sent: Monday, October 26, 2009 5:35 PM To: santiago.nu...@ridgerun.com Cc: Narnakaje, Snehaprabha; davinci-linux-open- sou...@linux.davincidsp.com; todd.fisc...@ridgerun.com; linux- me...@vger.kernel.org Subject: Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board Santiago Nunez-Corrales snu...@ridgerun.com writes: Kevin Hilman wrote: santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach- davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. Kevin Kevin, I've been working on this particular implementation. This board-dm365-evm.c is specific to the board, therefore I don't still get the point of not having those values wired to the board file, but I know it'd be nice to have the CPLD configuration triggered upon TVP7002 detection. I see two options: Having them in the board file is appropriate, what I object to is the selection by Kconfig. Run-time detection is always preferred when possible. We have this CPLD init API - evm_init_cpld() called from the dm365_evm_init() function. The CPLD init API was trying to initialize the CPLD, based on the default configuration. I believe David Brownell had this placeholder for have_imager() and have_tvp7002() APIs since, we have different CPLD settings for the imager, tvp7002 and tvp5146 for the video input source. 1. Do the callback function inside pdata and initialize it at driver load time (tvp7002_probe). Set tvp5146 as default and override when driver loads (and restore when unloads). This is the preferred option to me. We can decide on the default video input source to be TVP5146. However we do not need a new callback function. We already have the VPFE .setup_input callback API dm365evm_setup_video_input() for the same purpose. The VPFE .setup_input API is called when each of the decoders (sub-devices) registered with VPFE capture driver. It is also called when the application decides on an input source and switches between the input sources. So, TVP5146 remains as the default video input source, only until VPFE probe is called, which registers the all decoders (sub-devices) defined in the VPFE platform data. This also means that the last decoder in the VPFE platform data remains active after boot-up. One can change the order in the VPFE platform data if a particular input source needs to remain active after boot-up. Note that application can always switch the input-source at run time. I quickly tried removing the have_imager() construct in the evm_init_cpld() and here is the output - Starting kernel ... Uncompressing Linux... done, booting the kernel. Linux version 2.6.32-rc2-davinci1-dirty ... CPU: Testing write buffer coherency: ok DaVinci: 8 gpio irqs NET: Registered protocol family 16 EVM: tvp5146 SD video input bio: create slab bio-0 at 0 SCSI subsystem initialized ... vpfe_init vpfe-capture: vpss clock vpss_master enabled vpfe-capture vpfe-capture: v4l2 device registered vpfe-capture vpfe-capture: video device
RE: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
-Original Message- From: Santiago Nunez-Corrales [mailto:snu...@ridgerun.com] Sent: Tuesday, October 27, 2009 11:54 AM To: Narnakaje, Snehaprabha Cc: Kevin Hilman; davinci-linux-open-sou...@linux.davincidsp.com; todd.fisc...@ridgerun.com; linux-media@vger.kernel.org Subject: Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board Sneha, So, if I got it right, have_tvp7002 is unnecessary since that initialization is done via the CPLD interface. Therefore, all I need to do is actually remove the logic for the device from board-dm365-evm.h. Am I right? If so, should I also remove the logic for tvp5146? Santiago, Yes, have_tvp7002 is not necessary. In fact evm_init_cpld() is also not necessary from video input source selection perspective. At this point, you can remove have_tvp7002() and keep the default as TVP5146. The evm_init_cpld() requires some cleanup for have_imager() as well. I am expecting Murali to provide patches for this sometime soon. This brings up another concern. Your board setup related patches are dependent on one of the old patches from Neil Sikka - to add TVP5146 capture support. That patch is not in the mainline yet and requires re-submission by Murali. I suggest you complete the TVP7002 sub-device driver (tvp7002.c and tvp7002.h), submit patches and follow up with the approval in the V4L2 community. Patches corresponding to VPFE and DM365 board setup can be handled by Murali, working with you. Thanks Sneha Regards, Narnakaje, Snehaprabha wrote: Santiago, Kevin, -Original Message- From: Kevin Hilman [mailto:khil...@deeprootsystems.com] Sent: Monday, October 26, 2009 5:35 PM To: santiago.nu...@ridgerun.com Cc: Narnakaje, Snehaprabha; davinci-linux-open- sou...@linux.davincidsp.com; todd.fisc...@ridgerun.com; linux- me...@vger.kernel.org Subject: Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board Santiago Nunez-Corrales snu...@ridgerun.com writes: Kevin Hilman wrote: santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach- davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. Kevin Kevin, I've been working on this particular implementation. This board-dm365-evm.c is specific to the board, therefore I don't still get the point of not having those values wired to the board file, but I know it'd be nice to have the CPLD configuration triggered upon TVP7002 detection. I see two options: Having them in the board file is appropriate, what I object to is the selection by Kconfig. Run-time detection is always preferred when possible. We have this CPLD init API - evm_init_cpld() called from the dm365_evm_init() function. The CPLD init API was trying to initialize the CPLD, based on the default configuration. I believe David Brownell had this placeholder for have_imager() and have_tvp7002() APIs since, we have different CPLD settings for the imager, tvp7002 and tvp5146 for the video input source. 1. Do the callback function inside pdata and initialize it at driver load time (tvp7002_probe). Set tvp5146 as default and override when driver loads (and restore when unloads). This is the preferred option to me. We can decide on the default video input source to be TVP5146. However we do not need a new callback function. We already have the VPFE .setup_input callback API dm365evm_setup_video_input() for the same purpose. The VPFE .setup_input API is called when each of the decoders (sub-devices
Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
Understood. I will send my refactoring according to community revisions I have up to now and wait for integration with Murali. I have some questions before fully using his DV interfaces. Regards, Narnakaje, Snehaprabha wrote: -Original Message- From: Santiago Nunez-Corrales [mailto:snu...@ridgerun.com] Sent: Tuesday, October 27, 2009 11:54 AM To: Narnakaje, Snehaprabha Cc: Kevin Hilman; davinci-linux-open-sou...@linux.davincidsp.com; todd.fisc...@ridgerun.com; linux-media@vger.kernel.org Subject: Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board Sneha, So, if I got it right, have_tvp7002 is unnecessary since that initialization is done via the CPLD interface. Therefore, all I need to do is actually remove the logic for the device from board-dm365-evm.h. Am I right? If so, should I also remove the logic for tvp5146? Santiago, Yes, have_tvp7002 is not necessary. In fact evm_init_cpld() is also not necessary from video input source selection perspective. At this point, you can remove have_tvp7002() and keep the default as TVP5146. The evm_init_cpld() requires some cleanup for have_imager() as well. I am expecting Murali to provide patches for this sometime soon. This brings up another concern. Your board setup related patches are dependent on one of the old patches from Neil Sikka - to add TVP5146 capture support. That patch is not in the mainline yet and requires re-submission by Murali. I suggest you complete the TVP7002 sub-device driver (tvp7002.c and tvp7002.h), submit patches and follow up with the approval in the V4L2 community. Patches corresponding to VPFE and DM365 board setup can be handled by Murali, working with you. Thanks Sneha Regards, Narnakaje, Snehaprabha wrote: Santiago, Kevin, -Original Message- From: Kevin Hilman [mailto:khil...@deeprootsystems.com] Sent: Monday, October 26, 2009 5:35 PM To: santiago.nu...@ridgerun.com Cc: Narnakaje, Snehaprabha; davinci-linux-open- sou...@linux.davincidsp.com; todd.fisc...@ridgerun.com; linux- me...@vger.kernel.org Subject: Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board Santiago Nunez-Corrales snu...@ridgerun.com writes: Kevin Hilman wrote: santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach- davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. Kevin Kevin, I've been working on this particular implementation. This board-dm365-evm.c is specific to the board, therefore I don't still get the point of not having those values wired to the board file, but I know it'd be nice to have the CPLD configuration triggered upon TVP7002 detection. I see two options: Having them in the board file is appropriate, what I object to is the selection by Kconfig. Run-time detection is always preferred when possible. We have this CPLD init API - evm_init_cpld() called from the dm365_evm_init() function. The CPLD init API was trying to initialize the CPLD, based on the default configuration. I believe David Brownell had this placeholder for have_imager() and have_tvp7002() APIs since, we have different CPLD settings for the imager, tvp7002 and tvp5146 for the video input source. 1. Do the callback function inside pdata and initialize it at driver load time (tvp7002_probe). Set tvp5146 as default and override when driver loads (and restore when unloads). This is the preferred option to me. We can decide on the default
Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
Santiago Nunez-Corrales snu...@ridgerun.com writes: Kevin Hilman wrote: santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach-davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. Kevin Kevin, I've been working on this particular implementation. This board-dm365-evm.c is specific to the board, therefore I don't still get the point of not having those values wired to the board file, but I know it'd be nice to have the CPLD configuration triggered upon TVP7002 detection. I see two options: Having them in the board file is appropriate, what I object to is the selection by Kconfig. Run-time detection is always preferred when possible. 1. Do the callback function inside pdata and initialize it at driver load time (tvp7002_probe). Set tvp5146 as default and override when driver loads (and restore when unloads). This is the preferred option to me. 2. Add an entry to sysfs such that it can be user-configurable whether to activate one of the other regardless of whether tvp5156 or tvp7002 are actually there (the only result would be fail to access the device). Why do you need sysfs options for switching? Wouldn't building as modules and loading/unloading the needed modules serve the same purpose? Remeber that the 'probe' isn't going to be called until the platform_driver is registered, and that will (usually) happen at module load time. Sneha, do you have any suggestions on this one? Kevin -- To unsubscribe from this list: send the line unsubscribe linux-media in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
Kevin Hilman wrote: santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach-davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. Kevin Kevin, I've been working on this particular implementation. This board-dm365-evm.c is specific to the board, therefore I don't still get the point of not having those values wired to the board file, but I know it'd be nice to have the CPLD configuration triggered upon TVP7002 detection. I see two options: 1. Do the callback function inside pdata and initialize it at driver load time (tvp7002_probe). Set tvp5146 as default and override when driver loads (and restore when unloads). 2. Add an entry to sysfs such that it can be user-configurable whether to activate one of the other regardless of whether tvp5156 or tvp7002 are actually there (the only result would be fail to access the device). Sneha, do you have any suggestions on this one? Regards, -- Santiago Nunez-Corrales, Eng. RidgeRun Engineering, LLC Guayabos, Curridabat San Jose, Costa Rica +(506) 2271 1487 +(506) 8313 0536 http://www.ridgerun.com -- To unsubscribe from this list: send the line unsubscribe linux-media in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
Sekhar, Nori, Sekhar wrote: On Fri, Oct 16, 2009 at 00:17:46, Kevin Hilman wrote: santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach-davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. But this wouldn't work when TVP7002 is built as a module. Correct? The current patch does not take care of the module case as well. Patch 6/6 of this series does seem to make the TVP7002 driver available as module. Well, that was the intention given the inherent convenience of loading/unloading the TVP7002 driver for applications. Now, given that scenario, I know the #ifdef option is not elegant, but it is simple and accomplishes the purpose with the module approach. Any other suggestions/ideas? Thanks, Sekhar Regards, -- Santiago Nunez-Corrales, Eng. RidgeRun Engineering, LLC Guayabos, Curridabat San Jose, Costa Rica +(506) 2271 1487 +(506) 8313 0536 http://www.ridgerun.com -- To unsubscribe from this list: send the line unsubscribe linux-media in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
[PATCH 2/6 v5] Support for TVP7002 in dm365 board
From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach-davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif } @@ -96,8 +104,26 @@ static inline int have_tvp7002(void) #define CPLD_CCD_DIR3 CPLD_OFFSET(0x3f,0) #define CPLD_CCD_IO3 CPLD_OFFSET(0x3f,1) +#define CPLD_VIDEO_INPUT_MUX_MASK 0x7 +#define CPLD_VIDEO_INPUT_MUX_TVP7002 0x1 +#define CPLD_VIDEO_INPUT_MUX_IMAGER0x2 +#define CPLD_VIDEO_INPUT_MUX_TVP5146 0x5 + static void __iomem *cpld; +static struct tvp514x_platform_data tvp5146_pdata = { + .clk_polarity = 0, + .hs_polarity = 1, + .vs_polarity = 1 +}; + +/* tvp7002 platform data, used during reset and probe operations */ +static struct tvp7002_config tvp7002_pdata = { + .clk_polarity = 0, + .hs_polarity = 0, + .vs_polarity = 0, + .fid_polarity = 0, +}; /* NOTE: this is geared for the standard config, with a socketed * 2 GByte Micron NAND (MT29F16G08FAA) using 128KB sectors. If you @@ -207,6 +233,140 @@ static int cpld_mmc_get_ro(int module) return !!(__raw_readb(cpld + CPLD_CARDSTAT) BIT(module ? 5 : 1)); } +#define TVP514X_STD_ALL(V4L2_STD_NTSC | V4L2_STD_PAL) +/* Inputs available at the TVP5146 */ +static struct v4l2_input tvp5146_inputs[] = { + { + .index = 0, + .name = Composite, + .type = V4L2_INPUT_TYPE_CAMERA, + .std = TVP514X_STD_ALL, + }, + { + .index = 1, + .name = S-Video, + .type = V4L2_INPUT_TYPE_CAMERA, + .std = TVP514X_STD_ALL, + }, +}; + +#define TVP7002_STD_ALL(V4L2_STD_525P_60 | V4L2_STD_625P_50 |\ + V4L2_STD_NTSC | V4L2_STD_PAL |\ + V4L2_STD_720P_50 | V4L2_STD_720P_60 |\ + V4L2_STD_1080I_50 | V4L2_STD_1080I_60 |\ + V4L2_STD_1080P_50 | V4L2_STD_1080P_60) + +/* Inputs available at the TVP7002 */ +static struct v4l2_input tvp7002_inputs[] = { + { + .index = 0, + .name = Component, + .type = V4L2_INPUT_TYPE_CAMERA, + .std = TVP7002_STD_ALL, + }, +}; + +/* + * this is the route info for connecting each input to decoder + * ouput that goes to vpfe. There is a one to one correspondence + * with tvp5146_inputs + */ +static struct vpfe_route tvp5146_routes[] = { + { + .input = INPUT_CVBS_VI2B, + .output = OUTPUT_10BIT_422_EMBEDDED_SYNC, + }, +{ + .input = INPUT_SVIDEO_VI2C_VI1C, + .output = OUTPUT_10BIT_422_EMBEDDED_SYNC, + }, +}; + +static struct vpfe_subdev_info vpfe_sub_devs[] = { +{ + .module_name = tvp5146, + .grp_id = VPFE_SUBDEV_TVP5146, + .num_inputs = ARRAY_SIZE(tvp5146_inputs), + .inputs = tvp5146_inputs, + .routes = tvp5146_routes, + .can_route = 1, + .ccdc_if_params = { + .if_type = VPFE_BT656, + .hdpol = VPFE_PINPOL_POSITIVE, + .vdpol = VPFE_PINPOL_POSITIVE, + }, + .board_info = { + I2C_BOARD_INFO(tvp5146, 0x5d), + .platform_data = tvp5146_pdata, + }, + }, + { + .module_name = tvp7002, + .grp_id = VPFE_SUBDEV_TVP7002, + .num_inputs = ARRAY_SIZE(tvp7002_inputs), + .inputs = tvp7002_inputs, + .ccdc_if_params = { + .if_type = VPFE_BT1120, + .hdpol = VPFE_PINPOL_POSITIVE, + .vdpol = VPFE_PINPOL_POSITIVE, + }, + .board_info = { + I2C_BOARD_INFO(tvp7002, 0x5c), + .platform_data = tvp7002_pdata, + }, + }, + { + .module_name =
Re: [PATCH 2/6 v5] Support for TVP7002 in dm365 board
santiago.nu...@ridgerun.com writes: From: Santiago Nunez-Corrales santiago.nu...@ridgerun.com This patch provides support for TVP7002 in architecture definitions within DM365. Signed-off-by: Santiago Nunez-Corrales santiago.nu...@ridgerun.com --- arch/arm/mach-davinci/board-dm365-evm.c | 170 ++- 1 files changed, 166 insertions(+), 4 deletions(-) diff --git a/arch/arm/mach-davinci/board-dm365-evm.c b/arch/arm/mach-davinci/board-dm365-evm.c index a1d5e7d..6c544d3 100644 --- a/arch/arm/mach-davinci/board-dm365-evm.c +++ b/arch/arm/mach-davinci/board-dm365-evm.c @@ -38,6 +38,11 @@ #include mach/common.h #include mach/mmc.h #include mach/nand.h +#include mach/gpio.h +#include linux/videodev2.h +#include media/tvp514x.h +#include media/tvp7002.h +#include media/davinci/videohd.h static inline int have_imager(void) @@ -48,8 +53,11 @@ static inline int have_imager(void) static inline int have_tvp7002(void) { - /* REVISIT when it's supported, trigger via Kconfig */ +#ifdef CONFIG_VIDEO_TVP7002 + return 1; +#else return 0; +#endif I've said this before, but I'll say it again. I don't like the #ifdef-on-Kconfig-option here. Can you add a probe hook to the platform_data so that when the tvp7002 is found it can call pdata-probe() which could then set a flag for use by have_tvp7002(). This will have he same effect without the ifdef since if the driver is not compiled in, its probe can never be triggered. Kevin -- To unsubscribe from this list: send the line unsubscribe linux-media in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html