| Message ID | 20240530171600.259495-1-hdegoede@redhat.com |
|---|---|
| Headers | show |
| Series |
|
| Related | show |
On 30/05/2024 18:15, Hans de Goede wrote: > Hi All, > > Here is v2 of my patch-series to add /dev/udmabuf dma-buffer allocation > support. This is based on: > https://patchwork.libcamera.org/patch/18922/ > > Changes in v2: > - New patch: Add linux/udmabuf.h to libcamera's local kernel headers > - New patch: Rename DmaHeap class to DmaBufAllocator > - libcamera: DmaBufAllocator: Support allocating from /dev/udmabuf : > - Reword the commit message > - Add a new DmaBufAllocator::DmaBufAllocatorFlag::UDmaBuf type for udmabuf > - Drop unnecessary size != size check > - Reword log messages to be more like the DMA heap alloc path > - Move UniqueFD(ret) up so as to not leak the fd on errors > -New patch: software_isp: Allow using dma-buffers from /dev/udmabuf > > I have also pushed these patches to: > https://gitlab.freedesktop.org/camera/libcamera-softisp/-/commits/udmabuf-v2 > > for CI, CI is happy with it except for the lint task which does not like > the kernel headers sync, which seems to be a false positive. > > Regards, > > Hans For the series Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> # Lenovo-x13s
Hi, On 5/31/24 5:02 PM, Bryan O'Donoghue wrote: > On 30/05/2024 18:15, Hans de Goede wrote: >> Hi All, >> >> Here is v2 of my patch-series to add /dev/udmabuf dma-buffer allocation >> support. This is based on: >> https://patchwork.libcamera.org/patch/18922/ >> >> Changes in v2: >> - New patch: Add linux/udmabuf.h to libcamera's local kernel headers >> - New patch: Rename DmaHeap class to DmaBufAllocator >> - libcamera: DmaBufAllocator: Support allocating from /dev/udmabuf : >> - Reword the commit message >> - Add a new DmaBufAllocator::DmaBufAllocatorFlag::UDmaBuf type for udmabuf >> - Drop unnecessary size != size check >> - Reword log messages to be more like the DMA heap alloc path >> - Move UniqueFD(ret) up so as to not leak the fd on errors >> -New patch: software_isp: Allow using dma-buffers from /dev/udmabuf >> >> I have also pushed these patches to: >> https://gitlab.freedesktop.org/camera/libcamera-softisp/-/commits/udmabuf-v2 >> >> for CI, CI is happy with it except for the lint task which does not like >> the kernel headers sync, which seems to be a false positive. >> >> Regards, >> >> Hans > > For the series > > Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> # Lenovo-x13s Thank you for testing. As for the small conflict with Milan's series I guess we'll just wait and see which series gets applied to the master branch first and then the other one can be rebased on top. Regards, Hans >
On Fri, May 31, 2024 at 06:45:00PM +0200, Hans de Goede wrote: > On 5/31/24 5:02 PM, Bryan O'Donoghue wrote: > > On 30/05/2024 18:15, Hans de Goede wrote: > >> Hi All, > >> > >> Here is v2 of my patch-series to add /dev/udmabuf dma-buffer allocation > >> support. This is based on: > >> https://patchwork.libcamera.org/patch/18922/ > >> > >> Changes in v2: > >> - New patch: Add linux/udmabuf.h to libcamera's local kernel headers > >> - New patch: Rename DmaHeap class to DmaBufAllocator > >> - libcamera: DmaBufAllocator: Support allocating from /dev/udmabuf : > >> - Reword the commit message > >> - Add a new DmaBufAllocator::DmaBufAllocatorFlag::UDmaBuf type for udmabuf > >> - Drop unnecessary size != size check > >> - Reword log messages to be more like the DMA heap alloc path > >> - Move UniqueFD(ret) up so as to not leak the fd on errors > >> -New patch: software_isp: Allow using dma-buffers from /dev/udmabuf > >> > >> I have also pushed these patches to: > >> https://gitlab.freedesktop.org/camera/libcamera-softisp/-/commits/udmabuf-v2 > >> > >> for CI, CI is happy with it except for the lint task which does not like > >> the kernel headers sync, which seems to be a false positive. > >> > >> Regards, > >> > >> Hans > > > > For the series > > > > Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> # Lenovo-x13s > > Thank you for testing. > > As for the small conflict with Milan's series I guess we'll just > wait and see which series gets applied to the master branch first > and then the other one can be rebased on top. I think I'll merge Milan's serie first, as it has been fully reviewed, while this series depends on a patch I have resent today.
On Sun, Jun 02, 2024 at 02:04:18AM +0300, Laurent Pinchart wrote: > On Fri, May 31, 2024 at 06:45:00PM +0200, Hans de Goede wrote: > > On 5/31/24 5:02 PM, Bryan O'Donoghue wrote: > > > On 30/05/2024 18:15, Hans de Goede wrote: > > >> Hi All, > > >> > > >> Here is v2 of my patch-series to add /dev/udmabuf dma-buffer allocation > > >> support. This is based on: > > >> https://patchwork.libcamera.org/patch/18922/ > > >> > > >> Changes in v2: > > >> - New patch: Add linux/udmabuf.h to libcamera's local kernel headers > > >> - New patch: Rename DmaHeap class to DmaBufAllocator > > >> - libcamera: DmaBufAllocator: Support allocating from /dev/udmabuf : > > >> - Reword the commit message > > >> - Add a new DmaBufAllocator::DmaBufAllocatorFlag::UDmaBuf type for udmabuf > > >> - Drop unnecessary size != size check > > >> - Reword log messages to be more like the DMA heap alloc path > > >> - Move UniqueFD(ret) up so as to not leak the fd on errors > > >> -New patch: software_isp: Allow using dma-buffers from /dev/udmabuf > > >> > > >> I have also pushed these patches to: > > >> https://gitlab.freedesktop.org/camera/libcamera-softisp/-/commits/udmabuf-v2 > > >> > > >> for CI, CI is happy with it except for the lint task which does not like > > >> the kernel headers sync, which seems to be a false positive. > > >> > > >> Regards, > > >> > > >> Hans > > > > > > For the series > > > > > > Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> # Lenovo-x13s > > > > Thank you for testing. > > > > As for the small conflict with Milan's series I guess we'll just > > wait and see which series gets applied to the master branch first > > and then the other one can be rebased on top. > > I think I'll merge Milan's serie first, as it has been fully reviewed, > while this series depends on a patch I have resent today. I've finished reviewing this series. Comments are minor, I expect v3 will be the final version. Could you base it on top of Milan's series ?
On Sun, Jun 02, 2024 at 02:38:09AM +0300, Laurent Pinchart wrote: > On Sun, Jun 02, 2024 at 02:04:18AM +0300, Laurent Pinchart wrote: > > On Fri, May 31, 2024 at 06:45:00PM +0200, Hans de Goede wrote: > > > On 5/31/24 5:02 PM, Bryan O'Donoghue wrote: > > > > On 30/05/2024 18:15, Hans de Goede wrote: > > > >> Hi All, > > > >> > > > >> Here is v2 of my patch-series to add /dev/udmabuf dma-buffer allocation > > > >> support. This is based on: > > > >> https://patchwork.libcamera.org/patch/18922/ > > > >> > > > >> Changes in v2: > > > >> - New patch: Add linux/udmabuf.h to libcamera's local kernel headers > > > >> - New patch: Rename DmaHeap class to DmaBufAllocator > > > >> - libcamera: DmaBufAllocator: Support allocating from /dev/udmabuf : > > > >> - Reword the commit message > > > >> - Add a new DmaBufAllocator::DmaBufAllocatorFlag::UDmaBuf type for udmabuf > > > >> - Drop unnecessary size != size check > > > >> - Reword log messages to be more like the DMA heap alloc path > > > >> - Move UniqueFD(ret) up so as to not leak the fd on errors > > > >> -New patch: software_isp: Allow using dma-buffers from /dev/udmabuf > > > >> > > > >> I have also pushed these patches to: > > > >> https://gitlab.freedesktop.org/camera/libcamera-softisp/-/commits/udmabuf-v2 > > > >> > > > >> for CI, CI is happy with it except for the lint task which does not like > > > >> the kernel headers sync, which seems to be a false positive. > > > >> > > > >> Regards, > > > >> > > > >> Hans > > > > > > > > For the series > > > > > > > > Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> # Lenovo-x13s > > > > > > Thank you for testing. > > > > > > As for the small conflict with Milan's series I guess we'll just > > > wait and see which series gets applied to the master branch first > > > and then the other one can be rebased on top. > > > > I think I'll merge Milan's serie first, as it has been fully reviewed, > > while this series depends on a patch I have resent today. > > I've finished reviewing this series. Comments are minor, I expect v3 > will be the final version. Could you base it on top of Milan's series ? Milan's series is now in the libcamera master branch, so you can base v3 on top of that. The kernel headers update has also been merged.
Hi, On 6/3/24 11:51 AM, Laurent Pinchart wrote: > On Sun, Jun 02, 2024 at 02:38:09AM +0300, Laurent Pinchart wrote: >> On Sun, Jun 02, 2024 at 02:04:18AM +0300, Laurent Pinchart wrote: >>> On Fri, May 31, 2024 at 06:45:00PM +0200, Hans de Goede wrote: >>>> On 5/31/24 5:02 PM, Bryan O'Donoghue wrote: >>>>> On 30/05/2024 18:15, Hans de Goede wrote: >>>>>> Hi All, >>>>>> >>>>>> Here is v2 of my patch-series to add /dev/udmabuf dma-buffer allocation >>>>>> support. This is based on: >>>>>> https://patchwork.libcamera.org/patch/18922/ >>>>>> >>>>>> Changes in v2: >>>>>> - New patch: Add linux/udmabuf.h to libcamera's local kernel headers >>>>>> - New patch: Rename DmaHeap class to DmaBufAllocator >>>>>> - libcamera: DmaBufAllocator: Support allocating from /dev/udmabuf : >>>>>> - Reword the commit message >>>>>> - Add a new DmaBufAllocator::DmaBufAllocatorFlag::UDmaBuf type for udmabuf >>>>>> - Drop unnecessary size != size check >>>>>> - Reword log messages to be more like the DMA heap alloc path >>>>>> - Move UniqueFD(ret) up so as to not leak the fd on errors >>>>>> -New patch: software_isp: Allow using dma-buffers from /dev/udmabuf >>>>>> >>>>>> I have also pushed these patches to: >>>>>> https://gitlab.freedesktop.org/camera/libcamera-softisp/-/commits/udmabuf-v2 >>>>>> >>>>>> for CI, CI is happy with it except for the lint task which does not like >>>>>> the kernel headers sync, which seems to be a false positive. >>>>>> >>>>>> Regards, >>>>>> >>>>>> Hans >>>>> >>>>> For the series >>>>> >>>>> Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> # Lenovo-x13s >>>> >>>> Thank you for testing. >>>> >>>> As for the small conflict with Milan's series I guess we'll just >>>> wait and see which series gets applied to the master branch first >>>> and then the other one can be rebased on top. >>> >>> I think I'll merge Milan's serie first, as it has been fully reviewed, >>> while this series depends on a patch I have resent today. >> >> I've finished reviewing this series. Comments are minor, I expect v3 >> will be the final version. Could you base it on top of Milan's series ? > > Milan's series is now in the libcamera master branch, so you can base v3 > on top of that. The kernel headers update has also been merged. Yes I noticed when I just rebased, thank you. I'm preparing a v3 of this series now. Regards, Hans