| Message ID | 20220510115859.19777-1-laurent.pinchart@ideasonboard.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | staging: media: imx: Prepare destaging of imx7-media-csi | expand |
Hi Laurent, Thanks for this small change :). On Tue, May 10, 2022 at 02:58:09PM +0300, Laurent Pinchart wrote: > Hello, > > This patch series prepares the imx7-media-csi for destaging by > decoupling it from the helpers shared with the i.MX6 IPUv3. > > The strategy Paul and I have followed is to import copies of helper code > and, refactor it within the imx7-media-csi driver, and repeat until no > more shared helpers are used. There is still room for refactoring and > simplification of the imx7-media-csi driver, but I believe it is now in > a state clean enough to be moved out of staging. I agree. > > The series also includes a few fixes or improvements in supported > formats that are now made possible thanks to this refactoring. See > patches 45/50 and 46/50 for details. > > The code size has grown as a result. This is partly offset by code in > the shared helpers that can be removed or simplified, but I haven't > starting working on that. The helpers are now used for the i.MX6 IPUv3 > only, so I will leave this exercise to anyone who would be interested in > destaging that driver as well. > > Some of the items in the TODO file related to the imx7-media-csi driver > have been addressed. The two remaining items are frame interval monitor > support and restricting the list of supported formats to the SoC > version. The former isn't a destaging blocker in my opinion, as the > feature can be added later if desired (and frame interval monitoring > should then be moved to the V4L2 core). I believe the latter could also > be addressed after destaging the driver, but in any case, this is a > discussion for a future destaging series (which may come as soon as this > one is accepted). > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > imx7/imx8mm media / csi patches" series. I went over all patches and I have 2 small remarks: 1. Shouldn't we change the connection between imx-media objects and imx7-csi also in kconfig? Since at the end of this series they are completely independent. Yeah, it can be done in a follow up patch on the unstaging, for me that's fine also. 2. Something that caught my eye on patch 2/50. But nothing functional. Once again many thanks for continuing investing in this code. for the all series (except patch 2/50): Acked-by: Rui Miguel Silva <rmfrfs@gmail.com> Cheers, Rui > > Laurent Pinchart (48): > staging: media: imx: imx7-media-csi: Initialize locks early on > staging: media: imx: imx7-media-csi: Split imx_media_dev from probe() > staging: media: imx: imx7-media-csi: Import notifier helpers > staging: media: imx: imx7-media-csi: Drop duplicate link creation > staging: media: imx: imx7-media-csi: Drop the imx_media notifier > staging: media: imx: imx7-media-csi: Don't populate vdev lists > staging: media: imx: imx7-media-csi: Drop unused frame_interval > staging: media: imx: imx7-media-csi: Move format init to probe time > staging: media: imx: imx7-media-csi: Import video device helpers > staging: media: imx: imx7-media-csi: Drop legacy video device support > staging: media: imx: imx7-media-csi: Drop unused controls support > staging: media: imx: imx7-media-csi: Reorganize imx7_csi structure > staging: media: imx: imx7-media-csi: Fold capture_priv into imx7_csi > staging: media: imx: imx7-media-csi: Ensure consistent function prefix > staging: media: imx: imx7-media-csi: Don't set subdev group id > staging: media: imx: imx7-media-csi: Import imx_media_dev_init() > helper > staging: media: imx: imx7-media-csi: Embed imx_media_dev in imx7_csi > staging: media: imx: imx7-media-csi: Drop imx_media_add_video_device > call > staging: media: imx: imx7-media-csi: Don't initialize unused fields > staging: media: imx: imx7-media-csi: Inline imx_media_pipeline_pad() > staging: media: imx: imx7-media-csi: Import > imx_media_pipeline_set_stream() > staging: media: imx: imx7-media-csi: Avoid unnecessary casts > staging: media: imx: imx7-media-csi: Inline pipeline start/stop > staging: media: imx: imx7-media-csi: Fold imx_media_dev into imx7_csi > staging: media: imx: imx7-media-csi: Decouple from imx_media_buffer > staging: media: imx: imx7-media-csi: Fold imx_media_video_dev into > imx7_csi > staging: media: imx: imx7-media-csi: Store imx7_csi in drv data > staging: media: imx: imx7-media-csi: Decouple from imx_media_dma_buf > staging: media: imx: imx7-media-csi: Decouple from shared macros > staging: media: imx: imx7-media-csi: Drop error message on alloc > failure > staging: media: imx: imx7-media-csi: Import format helpers > staging: media: imx: imx7-media-csi: Replace ipu_color_space with bool > yuv field > staging: media: imx: imx7-media-csi: Drop IC support from > imx7_csi_try_colorimetry() > staging: media: imx: imx7-media-csi: Drop IPU-only formats > staging: media: imx: imx7-media-csi: Drop unsupported YUV and RGB > formats > staging: media: imx: imx7-media-csi: Make default formats consistent > staging: media: imx: imx7-media-csi: Define macro for default mbus > code > staging: media: imx: imx7-media-csi: Simplify default mbus code in > try_fmt > staging: media: imx: imx7-media-csi: Drop YUV/RGB/BAYER format > selectors > staging: media: imx: imx7-media-csi: Drop unneeded imx7_csi_pixfmt > fields > staging: media: imx: imx7-media-csi: Inline imx7_csi_init_mbus_fmt() > staging: media: imx: imx7-media-csi: Simplify default format in > try_fmt > staging: media: imx: imx7-media-csi: Fix list of supported formats > staging: media: imx: imx7-media-csi: Add V4L2_PIX_FMT_Y14 support > staging: media: imx: imx7-media-csi: Drop unneeded pixel format > validation > staging: media: imx: imx7-media-csi: Inline > imx7_csi_enum_pixel_formats() > staging: media: imx: imx7-media-csi: Drop V4L2 events support > staging: media: imx: imx7-media-csi: Drop usage of shared helpers > > Paul Elder (2): > staging: media: imx: imx7-media-csi: Move misc init out of probe() > staging: media: imx: imx7-media-csi: Remove imx_media_of_add_csi > > drivers/staging/media/imx/imx7-media-csi.c | 1370 +++++++++++++++++--- > 1 file changed, 1172 insertions(+), 198 deletions(-) > > > base-commit: c5eb0a61238dd6faf37f58c9ce61c9980aaffd7a > -- > Regards, > > Laurent Pinchart >
Hi Rui, On Tue, May 10, 2022 at 03:04:16PM +0100, Rui Miguel Silva wrote: > Hi Laurent, > Thanks for this small change :). Thanks for your small review ;-) > On Tue, May 10, 2022 at 02:58:09PM +0300, Laurent Pinchart wrote: > > Hello, > > > > This patch series prepares the imx7-media-csi for destaging by > > decoupling it from the helpers shared with the i.MX6 IPUv3. > > > > The strategy Paul and I have followed is to import copies of helper code > > and, refactor it within the imx7-media-csi driver, and repeat until no > > more shared helpers are used. There is still room for refactoring and > > simplification of the imx7-media-csi driver, but I believe it is now in > > a state clean enough to be moved out of staging. > > I agree. > > > The series also includes a few fixes or improvements in supported > > formats that are now made possible thanks to this refactoring. See > > patches 45/50 and 46/50 for details. > > > > The code size has grown as a result. This is partly offset by code in > > the shared helpers that can be removed or simplified, but I haven't > > starting working on that. The helpers are now used for the i.MX6 IPUv3 > > only, so I will leave this exercise to anyone who would be interested in > > destaging that driver as well. > > > > Some of the items in the TODO file related to the imx7-media-csi driver > > have been addressed. The two remaining items are frame interval monitor > > support and restricting the list of supported formats to the SoC > > version. The former isn't a destaging blocker in my opinion, as the > > feature can be added later if desired (and frame interval monitoring > > should then be moved to the V4L2 core). I believe the latter could also > > be addressed after destaging the driver, but in any case, this is a > > discussion for a future destaging series (which may come as soon as this > > one is accepted). > > > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > > imx7/imx8mm media / csi patches" series. > > I went over all patches and I have 2 small remarks: > > 1. Shouldn't we change the connection between imx-media objects and > imx7-csi also in kconfig? Since at the end of this series they are > completely independent. Yeah, it can be done in a follow up > patch on the unstaging, for me that's fine also. I was thinking of doing it while destaging, as I found it a bit pointless to reorganize Kconfig in staging to then immediately move the driver out of staging, but I don't mind already handling this in staging if preferred. > 2. Something that caught my eye on patch 2/50. But nothing functional. I'll reply to that patch. > Once again many thanks for continuing investing in this code. > for the all series (except patch 2/50): > > Acked-by: Rui Miguel Silva <rmfrfs@gmail.com> Thank you ! > > Laurent Pinchart (48): > > staging: media: imx: imx7-media-csi: Initialize locks early on > > staging: media: imx: imx7-media-csi: Split imx_media_dev from probe() > > staging: media: imx: imx7-media-csi: Import notifier helpers > > staging: media: imx: imx7-media-csi: Drop duplicate link creation > > staging: media: imx: imx7-media-csi: Drop the imx_media notifier > > staging: media: imx: imx7-media-csi: Don't populate vdev lists > > staging: media: imx: imx7-media-csi: Drop unused frame_interval > > staging: media: imx: imx7-media-csi: Move format init to probe time > > staging: media: imx: imx7-media-csi: Import video device helpers > > staging: media: imx: imx7-media-csi: Drop legacy video device support > > staging: media: imx: imx7-media-csi: Drop unused controls support > > staging: media: imx: imx7-media-csi: Reorganize imx7_csi structure > > staging: media: imx: imx7-media-csi: Fold capture_priv into imx7_csi > > staging: media: imx: imx7-media-csi: Ensure consistent function prefix > > staging: media: imx: imx7-media-csi: Don't set subdev group id > > staging: media: imx: imx7-media-csi: Import imx_media_dev_init() > > helper > > staging: media: imx: imx7-media-csi: Embed imx_media_dev in imx7_csi > > staging: media: imx: imx7-media-csi: Drop imx_media_add_video_device > > call > > staging: media: imx: imx7-media-csi: Don't initialize unused fields > > staging: media: imx: imx7-media-csi: Inline imx_media_pipeline_pad() > > staging: media: imx: imx7-media-csi: Import > > imx_media_pipeline_set_stream() > > staging: media: imx: imx7-media-csi: Avoid unnecessary casts > > staging: media: imx: imx7-media-csi: Inline pipeline start/stop > > staging: media: imx: imx7-media-csi: Fold imx_media_dev into imx7_csi > > staging: media: imx: imx7-media-csi: Decouple from imx_media_buffer > > staging: media: imx: imx7-media-csi: Fold imx_media_video_dev into > > imx7_csi > > staging: media: imx: imx7-media-csi: Store imx7_csi in drv data > > staging: media: imx: imx7-media-csi: Decouple from imx_media_dma_buf > > staging: media: imx: imx7-media-csi: Decouple from shared macros > > staging: media: imx: imx7-media-csi: Drop error message on alloc > > failure > > staging: media: imx: imx7-media-csi: Import format helpers > > staging: media: imx: imx7-media-csi: Replace ipu_color_space with bool > > yuv field > > staging: media: imx: imx7-media-csi: Drop IC support from > > imx7_csi_try_colorimetry() > > staging: media: imx: imx7-media-csi: Drop IPU-only formats > > staging: media: imx: imx7-media-csi: Drop unsupported YUV and RGB > > formats > > staging: media: imx: imx7-media-csi: Make default formats consistent > > staging: media: imx: imx7-media-csi: Define macro for default mbus > > code > > staging: media: imx: imx7-media-csi: Simplify default mbus code in > > try_fmt > > staging: media: imx: imx7-media-csi: Drop YUV/RGB/BAYER format > > selectors > > staging: media: imx: imx7-media-csi: Drop unneeded imx7_csi_pixfmt > > fields > > staging: media: imx: imx7-media-csi: Inline imx7_csi_init_mbus_fmt() > > staging: media: imx: imx7-media-csi: Simplify default format in > > try_fmt > > staging: media: imx: imx7-media-csi: Fix list of supported formats > > staging: media: imx: imx7-media-csi: Add V4L2_PIX_FMT_Y14 support > > staging: media: imx: imx7-media-csi: Drop unneeded pixel format > > validation > > staging: media: imx: imx7-media-csi: Inline > > imx7_csi_enum_pixel_formats() > > staging: media: imx: imx7-media-csi: Drop V4L2 events support > > staging: media: imx: imx7-media-csi: Drop usage of shared helpers > > > > Paul Elder (2): > > staging: media: imx: imx7-media-csi: Move misc init out of probe() > > staging: media: imx: imx7-media-csi: Remove imx_media_of_add_csi > > > > drivers/staging/media/imx/imx7-media-csi.c | 1370 +++++++++++++++++--- > > 1 file changed, 1172 insertions(+), 198 deletions(-) > > > > > > base-commit: c5eb0a61238dd6faf37f58c9ce61c9980aaffd7a
Hello Laurent, thanks for pushing out this huge set. Am Dienstag, 10. Mai 2022, 13:58:09 CEST schrieb Laurent Pinchart: > Hello, > > This patch series prepares the imx7-media-csi for destaging by > decoupling it from the helpers shared with the i.MX6 IPUv3. > > The strategy Paul and I have followed is to import copies of helper code > and, refactor it within the imx7-media-csi driver, and repeat until no > more shared helpers are used. There is still room for refactoring and > simplification of the imx7-media-csi driver, but I believe it is now in > a state clean enough to be moved out of staging. > > The series also includes a few fixes or improvements in supported > formats that are now made possible thanks to this refactoring. See > patches 45/50 and 46/50 for details. > > The code size has grown as a result. This is partly offset by code in > the shared helpers that can be removed or simplified, but I haven't > starting working on that. The helpers are now used for the i.MX6 IPUv3 > only, so I will leave this exercise to anyone who would be interested in > destaging that driver as well. > > Some of the items in the TODO file related to the imx7-media-csi driver > have been addressed. The two remaining items are frame interval monitor > support and restricting the list of supported formats to the SoC > version. The former isn't a destaging blocker in my opinion, as the > feature can be added later if desired (and frame interval monitoring > should then be moved to the V4L2 core). I believe the latter could also > be addressed after destaging the driver, but in any case, this is a > discussion for a future destaging series (which may come as soon as this > one is accepted). > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > imx7/imx8mm media / csi patches" series. Thanks for putting me on CC. I'll try to get my system running your patches ASAP. Best regards, Alexander
Hello Laurent, Am Dienstag, 10. Mai 2022, 13:58:09 CEST schrieb Laurent Pinchart: > Hello, > > This patch series prepares the imx7-media-csi for destaging by > decoupling it from the helpers shared with the i.MX6 IPUv3. > > The strategy Paul and I have followed is to import copies of helper code > and, refactor it within the imx7-media-csi driver, and repeat until no > more shared helpers are used. There is still room for refactoring and > simplification of the imx7-media-csi driver, but I believe it is now in > a state clean enough to be moved out of staging. > > The series also includes a few fixes or improvements in supported > formats that are now made possible thanks to this refactoring. See > patches 45/50 and 46/50 for details. > > The code size has grown as a result. This is partly offset by code in > the shared helpers that can be removed or simplified, but I haven't > starting working on that. The helpers are now used for the i.MX6 IPUv3 > only, so I will leave this exercise to anyone who would be interested in > destaging that driver as well. > > Some of the items in the TODO file related to the imx7-media-csi driver > have been addressed. The two remaining items are frame interval monitor > support and restricting the list of supported formats to the SoC > version. The former isn't a destaging blocker in my opinion, as the > feature can be added later if desired (and frame interval monitoring > should then be moved to the V4L2 core). I believe the latter could also > be addressed after destaging the driver, but in any case, this is a > discussion for a future destaging series (which may come as soon as this > one is accepted). > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > imx7/imx8mm media / csi patches" series. Thanks again for this patchset. This does not only simplify my series, it pretty much makes it obsolete. Only one change is still valid (rounding up width depending on bpp), but I would skip that for now. Just using my DT is enough. Nice! If you respin a v2, I'll retest again for a proper Tested-by tag. Regards, Alexander
Hi Alexander, On Wed, May 11, 2022 at 03:36:14PM +0200, Alexander Stein wrote: > Hello Laurent, > > thanks for pushing out this huge set. > > Am Dienstag, 10. Mai 2022, 13:58:09 CEST schrieb Laurent Pinchart: > > Hello, > > > > This patch series prepares the imx7-media-csi for destaging by > > decoupling it from the helpers shared with the i.MX6 IPUv3. > > > > The strategy Paul and I have followed is to import copies of helper code > > and, refactor it within the imx7-media-csi driver, and repeat until no > > more shared helpers are used. There is still room for refactoring and > > simplification of the imx7-media-csi driver, but I believe it is now in > > a state clean enough to be moved out of staging. > > > > The series also includes a few fixes or improvements in supported > > formats that are now made possible thanks to this refactoring. See > > patches 45/50 and 46/50 for details. > > > > The code size has grown as a result. This is partly offset by code in > > the shared helpers that can be removed or simplified, but I haven't > > starting working on that. The helpers are now used for the i.MX6 IPUv3 > > only, so I will leave this exercise to anyone who would be interested in > > destaging that driver as well. > > > > Some of the items in the TODO file related to the imx7-media-csi driver > > have been addressed. The two remaining items are frame interval monitor > > support and restricting the list of supported formats to the SoC > > version. The former isn't a destaging blocker in my opinion, as the > > feature can be added later if desired (and frame interval monitoring > > should then be moved to the V4L2 core). I believe the latter could also > > be addressed after destaging the driver, but in any case, this is a > > discussion for a future destaging series (which may come as soon as this > > one is accepted). > > > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > > imx7/imx8mm media / csi patches" series. > > Thanks for putting me on CC. I'll try to get my system running your patches > ASAP. As only minor changes were needed, I ended up posting v1.1 of the corresponding patches instead of spamming everybody with a full v2. For your convenience, I've pushed the result to git://linuxtv.org/pinchartl/media.git imx7/destage I know it's a bit of a short notice, but I'd like to send a pull request in time for v5.20, when do you think you would be able to test the series ?
Hello Laurent, Am Donnerstag, 7. Juli 2022, 02:23:55 CEST schrieb Laurent Pinchart: > Hi Alexander, > > On Wed, May 11, 2022 at 03:36:14PM +0200, Alexander Stein wrote: > > Hello Laurent, > > > > thanks for pushing out this huge set. > > > > Am Dienstag, 10. Mai 2022, 13:58:09 CEST schrieb Laurent Pinchart: > > > Hello, > > > > > > This patch series prepares the imx7-media-csi for destaging by > > > decoupling it from the helpers shared with the i.MX6 IPUv3. > > > > > > The strategy Paul and I have followed is to import copies of helper code > > > and, refactor it within the imx7-media-csi driver, and repeat until no > > > more shared helpers are used. There is still room for refactoring and > > > simplification of the imx7-media-csi driver, but I believe it is now in > > > a state clean enough to be moved out of staging. > > > > > > The series also includes a few fixes or improvements in supported > > > formats that are now made possible thanks to this refactoring. See > > > patches 45/50 and 46/50 for details. > > > > > > The code size has grown as a result. This is partly offset by code in > > > the shared helpers that can be removed or simplified, but I haven't > > > starting working on that. The helpers are now used for the i.MX6 IPUv3 > > > only, so I will leave this exercise to anyone who would be interested in > > > destaging that driver as well. > > > > > > Some of the items in the TODO file related to the imx7-media-csi driver > > > have been addressed. The two remaining items are frame interval monitor > > > support and restricting the list of supported formats to the SoC > > > version. The former isn't a destaging blocker in my opinion, as the > > > feature can be added later if desired (and frame interval monitoring > > > should then be moved to the V4L2 core). I believe the latter could also > > > be addressed after destaging the driver, but in any case, this is a > > > discussion for a future destaging series (which may come as soon as this > > > one is accepted). > > > > > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > > > imx7/imx8mm media / csi patches" series. > > > > Thanks for putting me on CC. I'll try to get my system running your > > patches > > ASAP. > > As only minor changes were needed, I ended up posting v1.1 of the > corresponding patches instead of spamming everybody with a full v2. For > your convenience, I've pushed the result to > > git://linuxtv.org/pinchartl/media.git imx7/destage > > I know it's a bit of a short notice, but I'd like to send a pull request > in time for v5.20, when do you think you would be able to test the > series ? I happen to have my setup already available, so I could test your updated series pretty fast. Accessing imx327 still works for WIP config. So Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> Best regards, Alexander
Hi Alexander, On Thu, Jul 07, 2022 at 12:45:01PM +0200, Alexander Stein wrote: > Am Donnerstag, 7. Juli 2022, 02:23:55 CEST schrieb Laurent Pinchart: > > On Wed, May 11, 2022 at 03:36:14PM +0200, Alexander Stein wrote: > > > Hello Laurent, > > > > > > thanks for pushing out this huge set. > > > > > > Am Dienstag, 10. Mai 2022, 13:58:09 CEST schrieb Laurent Pinchart: > > > > Hello, > > > > > > > > This patch series prepares the imx7-media-csi for destaging by > > > > decoupling it from the helpers shared with the i.MX6 IPUv3. > > > > > > > > The strategy Paul and I have followed is to import copies of helper code > > > > and, refactor it within the imx7-media-csi driver, and repeat until no > > > > more shared helpers are used. There is still room for refactoring and > > > > simplification of the imx7-media-csi driver, but I believe it is now in > > > > a state clean enough to be moved out of staging. > > > > > > > > The series also includes a few fixes or improvements in supported > > > > formats that are now made possible thanks to this refactoring. See > > > > patches 45/50 and 46/50 for details. > > > > > > > > The code size has grown as a result. This is partly offset by code in > > > > the shared helpers that can be removed or simplified, but I haven't > > > > starting working on that. The helpers are now used for the i.MX6 IPUv3 > > > > only, so I will leave this exercise to anyone who would be interested in > > > > destaging that driver as well. > > > > > > > > Some of the items in the TODO file related to the imx7-media-csi driver > > > > have been addressed. The two remaining items are frame interval monitor > > > > support and restricting the list of supported formats to the SoC > > > > version. The former isn't a destaging blocker in my opinion, as the > > > > feature can be added later if desired (and frame interval monitoring > > > > should then be moved to the V4L2 core). I believe the latter could also > > > > be addressed after destaging the driver, but in any case, this is a > > > > discussion for a future destaging series (which may come as soon as this > > > > one is accepted). > > > > > > > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > > > > imx7/imx8mm media / csi patches" series. > > > > > > Thanks for putting me on CC. I'll try to get my system running your > > > patches > > > ASAP. > > > > As only minor changes were needed, I ended up posting v1.1 of the > > corresponding patches instead of spamming everybody with a full v2. For > > your convenience, I've pushed the result to > > > > git://linuxtv.org/pinchartl/media.git imx7/destage > > > > I know it's a bit of a short notice, but I'd like to send a pull request > > in time for v5.20, when do you think you would be able to test the > > series ? > > I happen to have my setup already available, so I could test your updated > series pretty fast. > Accessing imx327 still works for WIP config. So > Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> Thanks a lot for the quick testing. I've now sent the pull request. Do you plan to submit a new version of your patches on top of this ?
Hi Laurent, Am Donnerstag, 7. Juli 2022, 13:34:50 CEST schrieb Laurent Pinchart: > Hi Alexander, > > On Thu, Jul 07, 2022 at 12:45:01PM +0200, Alexander Stein wrote: > > Am Donnerstag, 7. Juli 2022, 02:23:55 CEST schrieb Laurent Pinchart: > > > On Wed, May 11, 2022 at 03:36:14PM +0200, Alexander Stein wrote: > > > > Hello Laurent, > > > > > > > > thanks for pushing out this huge set. > > > > > > > > Am Dienstag, 10. Mai 2022, 13:58:09 CEST schrieb Laurent Pinchart: > > > > > Hello, > > > > > > > > > > This patch series prepares the imx7-media-csi for destaging by > > > > > decoupling it from the helpers shared with the i.MX6 IPUv3. > > > > > > > > > > The strategy Paul and I have followed is to import copies of helper > > > > > code > > > > > and, refactor it within the imx7-media-csi driver, and repeat until > > > > > no > > > > > more shared helpers are used. There is still room for refactoring > > > > > and > > > > > simplification of the imx7-media-csi driver, but I believe it is now > > > > > in > > > > > a state clean enough to be moved out of staging. > > > > > > > > > > The series also includes a few fixes or improvements in supported > > > > > formats that are now made possible thanks to this refactoring. See > > > > > patches 45/50 and 46/50 for details. > > > > > > > > > > The code size has grown as a result. This is partly offset by code > > > > > in > > > > > the shared helpers that can be removed or simplified, but I haven't > > > > > starting working on that. The helpers are now used for the i.MX6 > > > > > IPUv3 > > > > > only, so I will leave this exercise to anyone who would be > > > > > interested in > > > > > destaging that driver as well. > > > > > > > > > > Some of the items in the TODO file related to the imx7-media-csi > > > > > driver > > > > > have been addressed. The two remaining items are frame interval > > > > > monitor > > > > > support and restricting the list of supported formats to the SoC > > > > > version. The former isn't a destaging blocker in my opinion, as the > > > > > feature can be added later if desired (and frame interval monitoring > > > > > should then be moved to the V4L2 core). I believe the latter could > > > > > also > > > > > be addressed after destaging the driver, but in any case, this is a > > > > > discussion for a future destaging series (which may come as soon as > > > > > this > > > > > one is accepted). > > > > > > > > > > Alexander, this also could greatly simplify your "[PATCH v3 0/8] > > > > > imx7/imx8mm media / csi patches" series. > > > > > > > > Thanks for putting me on CC. I'll try to get my system running your > > > > patches > > > > ASAP. > > > > > > As only minor changes were needed, I ended up posting v1.1 of the > > > corresponding patches instead of spamming everybody with a full v2. For > > > your convenience, I've pushed the result to > > > > > > git://linuxtv.org/pinchartl/media.git imx7/destage > > > > > > I know it's a bit of a short notice, but I'd like to send a pull request > > > in time for v5.20, when do you think you would be able to test the > > > series ? > > > > I happen to have my setup already available, so I could test your updated > > series pretty fast. > > Accessing imx327 still works for WIP config. So > > Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> > > Thanks a lot for the quick testing. I've now sent the pull request. > > Do you plan to submit a new version of your patches on top of this ? The patches regarding the imx[,7]-media-* are not needed anymore, they are obsolete now. There are only patches regarding DT overlay and the Vision Components FPGA and sensors I'm working on right now. Best regards, Alexander
Hello, This patch series prepares the imx7-media-csi for destaging by decoupling it from the helpers shared with the i.MX6 IPUv3. The strategy Paul and I have followed is to import copies of helper code and, refactor it within the imx7-media-csi driver, and repeat until no more shared helpers are used. There is still room for refactoring and simplification of the imx7-media-csi driver, but I believe it is now in a state clean enough to be moved out of staging. The series also includes a few fixes or improvements in supported formats that are now made possible thanks to this refactoring. See patches 45/50 and 46/50 for details. The code size has grown as a result. This is partly offset by code in the shared helpers that can be removed or simplified, but I haven't starting working on that. The helpers are now used for the i.MX6 IPUv3 only, so I will leave this exercise to anyone who would be interested in destaging that driver as well. Some of the items in the TODO file related to the imx7-media-csi driver have been addressed. The two remaining items are frame interval monitor support and restricting the list of supported formats to the SoC version. The former isn't a destaging blocker in my opinion, as the feature can be added later if desired (and frame interval monitoring should then be moved to the V4L2 core). I believe the latter could also be addressed after destaging the driver, but in any case, this is a discussion for a future destaging series (which may come as soon as this one is accepted). Alexander, this also could greatly simplify your "[PATCH v3 0/8] imx7/imx8mm media / csi patches" series. Laurent Pinchart (48): staging: media: imx: imx7-media-csi: Initialize locks early on staging: media: imx: imx7-media-csi: Split imx_media_dev from probe() staging: media: imx: imx7-media-csi: Import notifier helpers staging: media: imx: imx7-media-csi: Drop duplicate link creation staging: media: imx: imx7-media-csi: Drop the imx_media notifier staging: media: imx: imx7-media-csi: Don't populate vdev lists staging: media: imx: imx7-media-csi: Drop unused frame_interval staging: media: imx: imx7-media-csi: Move format init to probe time staging: media: imx: imx7-media-csi: Import video device helpers staging: media: imx: imx7-media-csi: Drop legacy video device support staging: media: imx: imx7-media-csi: Drop unused controls support staging: media: imx: imx7-media-csi: Reorganize imx7_csi structure staging: media: imx: imx7-media-csi: Fold capture_priv into imx7_csi staging: media: imx: imx7-media-csi: Ensure consistent function prefix staging: media: imx: imx7-media-csi: Don't set subdev group id staging: media: imx: imx7-media-csi: Import imx_media_dev_init() helper staging: media: imx: imx7-media-csi: Embed imx_media_dev in imx7_csi staging: media: imx: imx7-media-csi: Drop imx_media_add_video_device call staging: media: imx: imx7-media-csi: Don't initialize unused fields staging: media: imx: imx7-media-csi: Inline imx_media_pipeline_pad() staging: media: imx: imx7-media-csi: Import imx_media_pipeline_set_stream() staging: media: imx: imx7-media-csi: Avoid unnecessary casts staging: media: imx: imx7-media-csi: Inline pipeline start/stop staging: media: imx: imx7-media-csi: Fold imx_media_dev into imx7_csi staging: media: imx: imx7-media-csi: Decouple from imx_media_buffer staging: media: imx: imx7-media-csi: Fold imx_media_video_dev into imx7_csi staging: media: imx: imx7-media-csi: Store imx7_csi in drv data staging: media: imx: imx7-media-csi: Decouple from imx_media_dma_buf staging: media: imx: imx7-media-csi: Decouple from shared macros staging: media: imx: imx7-media-csi: Drop error message on alloc failure staging: media: imx: imx7-media-csi: Import format helpers staging: media: imx: imx7-media-csi: Replace ipu_color_space with bool yuv field staging: media: imx: imx7-media-csi: Drop IC support from imx7_csi_try_colorimetry() staging: media: imx: imx7-media-csi: Drop IPU-only formats staging: media: imx: imx7-media-csi: Drop unsupported YUV and RGB formats staging: media: imx: imx7-media-csi: Make default formats consistent staging: media: imx: imx7-media-csi: Define macro for default mbus code staging: media: imx: imx7-media-csi: Simplify default mbus code in try_fmt staging: media: imx: imx7-media-csi: Drop YUV/RGB/BAYER format selectors staging: media: imx: imx7-media-csi: Drop unneeded imx7_csi_pixfmt fields staging: media: imx: imx7-media-csi: Inline imx7_csi_init_mbus_fmt() staging: media: imx: imx7-media-csi: Simplify default format in try_fmt staging: media: imx: imx7-media-csi: Fix list of supported formats staging: media: imx: imx7-media-csi: Add V4L2_PIX_FMT_Y14 support staging: media: imx: imx7-media-csi: Drop unneeded pixel format validation staging: media: imx: imx7-media-csi: Inline imx7_csi_enum_pixel_formats() staging: media: imx: imx7-media-csi: Drop V4L2 events support staging: media: imx: imx7-media-csi: Drop usage of shared helpers Paul Elder (2): staging: media: imx: imx7-media-csi: Move misc init out of probe() staging: media: imx: imx7-media-csi: Remove imx_media_of_add_csi drivers/staging/media/imx/imx7-media-csi.c | 1370 +++++++++++++++++--- 1 file changed, 1172 insertions(+), 198 deletions(-) base-commit: c5eb0a61238dd6faf37f58c9ce61c9980aaffd7a