Message-ID: <0e5b26b6-031a-c07c-0295-4cce0712779d@nvidia.com>
Date:   Sun, 19 Jul 2020 21:41:32 -0700
From:   Sowjanya Komatineni <skomatineni@...dia.com>
To:     Hans Verkuil <hverkuil@...all.nl>, <thierry.reding@...il.com>,
        <jonathanh@...dia.com>, <frankc@...dia.com>, <sakari.ailus@....fi>,
        <robh+dt@...nel.org>, <helen.koike@...labora.com>
CC:     <digetx@...il.com>, <sboyd@...nel.org>,
        <gregkh@...uxfoundation.org>, <linux-media@...r.kernel.org>,
        <devicetree@...r.kernel.org>, <linux-tegra@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>, <linux-i2c@...r.kernel.org>
Subject: Re: [RFC PATCH v3 00/18] Support for Tegra video capture from
 external sensor


On 7/17/20 10:48 AM, Sowjanya Komatineni wrote:
>
> On 7/17/20 10:38 AM, Sowjanya Komatineni wrote:
>>
>> On 7/17/20 10:23 AM, Sowjanya Komatineni wrote:
>>>
>>> On 7/17/20 10:08 AM, Hans Verkuil wrote:
>>>> On 17/07/2020 18:34, Sowjanya Komatineni wrote:
>>>>> On 7/17/20 3:54 AM, Hans Verkuil wrote:
>>>>>> Hi Sowjanya,
>>>>>>
>>>>>> On 15/07/2020 06:20, Sowjanya Komatineni wrote:
>>>>>>> This series adds support for video capture from external camera 
>>>>>>> sensor to
>>>>>>> Tegra video driver.
>>>>>>>
>>>>>>> Jetson TX1 has camera expansion connector and supports custom 
>>>>>>> camera module
>>>>>>> designed as per TX1 design specification.
>>>>>>>
>>>>>>> This series also enables camera capture support for Jetson Nano 
>>>>>>> which has
>>>>>>> Raspberry PI camera header.
>>>>>>>
>>>>>>> This series is tested with IMX219 camera sensor.
>>>>>>>
>>>>>>> This series include,
>>>>>>>
>>>>>>> VI I2C related fixes
>>>>>>> - Camera sensor programming happens through VI I2C which is on 
>>>>>>> host1x bus.
>>>>>>> - These patches includes device tree and I2C driver fixes for VI 
>>>>>>> I2C.
>>>>>>>
>>>>>>> Tegra video driver updates
>>>>>>> - TPG Vs Non-TPG based on Kconfig
>>>>>>> - Support for external sensor video capture based on device 
>>>>>>> graph from DT.
>>>>>>> - Support for selection ioctl operations
>>>>>>> - Tegra MIPI CSI pads calibration
>>>>>>> - CSI T-CLK and T-HS settle time computation based on clock rates.
>>>>>>>
>>>>>>> Host1x driver updates
>>>>>>> - Adds API to allow creating mipi device for specific device node.
>>>>>>> - Splits MIPI pads calibrate start and waiting for calibration 
>>>>>>> to be done.
>>>>>>>
>>>>>>> Device tree updates
>>>>>>> - Adds camera connector 2V8, 1V8, 1V2 regulator supplies to 
>>>>>>> Jetson TX1 DT.
>>>>>>> - Enabled VI and CSI support in Jetson Nano DT.
>>>>>> I'm doing a bit of stress testing with:
>>>>>>
>>>>>> while true; do v4l2-ctl --stream-mmap --stream-count=1; done
>>>>>>
>>>>>> and I see that the imx274 has often streaming failures:
>>>>>>
>>>>>> [  172.025144] IMX274 8-001a: s_stream failed
>>>>>> [  179.025192] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 3132 = 870 (2 bytes)
>>>>>> [  179.033575] IMX274 8-001a: s_stream failed
>>>>>> [  226.525378] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 3130 = 878 (2 bytes)
>>>>>> [  226.533761] IMX274 8-001a: s_stream failed
>>>>>> [  227.029325] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 30f6 = 107 (2 bytes)
>>>>>> [  227.037758] IMX274 8-001a: s_stream failed
>>>>>> [  247.025218] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 30f6 = 107 (2 bytes)
>>>>>> [  247.033658] IMX274 8-001a: s_stream failed
>>>>>> [  293.025517] IMX274 8-001a: s_stream failed
>>>>>> [  309.024727] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 30e0 = 0 (2 bytes)
>>>>>> [  309.032969] IMX274 8-001a: s_stream failed
>>>>>> [  309.529506] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 30f8 = 11d2 (3 bytes)
>>>>>> [  309.538103] IMX274 8-001a: imx274_set_frame_length error = -121
>>>>>> [  309.544102] IMX274 8-001a: imx274_set_frame_interval error = -121
>>>>>> [  309.550243] IMX274 8-001a: s_stream failed
>>>>>> [  314.025561] IMX274 8-001a: s_stream failed
>>>>>> [  329.025586] IMX274 8-001a: s_stream failed
>>>>>> [  340.529567] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 303a = f0c (2 bytes)
>>>>>> [  340.538009] IMX274 8-001a: s_stream failed
>>>>>> [  347.525627] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 30f6 = 107 (2 bytes)
>>>>>> [  347.534008] IMX274 8-001a: s_stream failed
>>>>>> [  365.033640] IMX274 8-001a: s_stream failed
>>>>>> [  437.525788] IMX274 8-001a: imx274_write_mbreg : i2c bulk write 
>>>>>> failed, 3038 = c (2 bytes)
>>>>>> [  437.533997] IMX274 8-001a: s_stream failed
>>>>>> [  456.029780] IMX274 8-001a: s_stream failed
>>>>>> [  472.025862] IMX274 8-001a: s_stream failed
>>>>>> [  498.025861] IMX274 8-001a: s_stream failed
>>>>>> [  500.025905] IMX274 8-001a: s_stream failed
>>>>>>
>>>>>> where v4l2-ctl returns:
>>>>>>
>>>>>>                   VIDIOC_STREAMON returned -1 (Remote I/O error)
>>>>>>
>>>>>> I don't see this with the imx219.
>>>>>>
>>>>>> I also see this occasionally:
>>>>>>
>>>>>> [Fri Jul 17 12:51:42 2020] video4linux video1: failed to run 
>>>>>> capture start kthread: -4
>>>>>>
>>>>>> Something is not stable here.
>>>>>>
>>>>>> Regards,
>>>>>>
>>>>>>     Hans
>>>>> Hi Hans,
>>>>>
>>>>> Running the same single frame continuous loop for more than 2 
>>>>> hours now
>>>>> and I don't see any failure.
>>>>>
>>>>> Above failure shows i2c bulk writes to IMX274 failure due to which
>>>>> s_stream also failed.
>>>>>
>>>>> Not sure if its due to i2c mux in the path to sensor on your module
>>>>> causing some issue when there is more i2c write traffic as we are 
>>>>> doing
>>>>> single stream in continuous loop. Also IMX219 does not show on 
>>>>> your side
>>>>> so something specific to IMX274 setup probably.
>>>> I'll take a closer look next week. Good to know that it works fine 
>>>> for you.
>>>>
>>>>>
>>>>> Regarding kthread_run failure where kthread_run() returned -EINTR 
>>>>> during
>>>>> capture start thread, I always see this happen at the point of 
>>>>> stopping
>>>>> the continuous single stream while loop by pressing ctrl+c after few
>>>>> loops of execution.
>>>> Hmm, if this is normal behavior, then should this message be a 
>>>> debug message
>>>> only? Or perhaps only show the message if the error code != EINTR.
>>>
>>> I believe its good to still show this as its reported by kthread_run 
>>> -> kthread_create_on_node.
>>>
>>> But not sure in real usecase we will ever use while true like this 
>>> and we should use script to also break while loop along with 
>>> v4l2-ctl termination when ctrl-c terminate request happens.
>>>
>> Hi Hans, As this happens only during this type of case, I can update 
>> to show message only when error code != EINTR.
>>
>> Thanks
>>
>> Sowjanya
>
>
> Sorry, Was thinking to not mask debug message for -EINTR in case if it 
> happens in any other valid scenarios. If you still want to mask, will 
> update in next version.
>
As we are running application that uses threads depending on when ctrl-c 
gets hit I see sigkill happens during kthread run even with break 
causing this.

Good way I see it working all the time is to use crtl-z and then kill 
application as below

Command to run

while true; do ./v4l2-ctl --stream-mmap --stream-count=1; done

To stop ,

press ctrl-z and then execute kill %%


>>
>>>
>>>>
>>>> Regards,
>>>>
>>>>     Hans
>>>>
>>>>> while true; do v4l2-ctl --stream-mmap --stream-count=1; done
>>>>>
>>>>> when we stop loop with ctrl+c, v4l2-ctl terminates but loop does not
>>>>> terminate immediately and probably SIGKILLed  is seen prior to 
>>>>> complete.
>>>>>
>>>>> Using below can help to terminate loop as well when we stop ctrl-c 
>>>>> and
>>>>> with this I don't see any repro of EINTR error from kthread_run 
>>>>> when run
>>>>> in infinite loop.
>>>>>
>>>>> while true; do ./v4l2-ctl --stream-mmap --stream-count=1 || break; 
>>>>> done
>>>>>
>>>>>
>>>>>
>>>>>>> Delta between patch versions:
>>>>>>>
>>>>>>> [v3]:    Includes v2 feedback
>>>>>>>     - Uses separate helper function for retrieving remote csi 
>>>>>>> subdevice
>>>>>>>       and source subdevice.
>>>>>>>     - Added check for presence of subdevice ops set/get_selection
>>>>>>>     - dropped vb2_queue_release from driver and using
>>>>>>>       vb2_video_unregister_device instead of 
>>>>>>> video_unregister_device.
>>>>>>>     - video device register should happen in the last after all 
>>>>>>> video
>>>>>>>       device related setup is done in the driver. This is being 
>>>>>>> addressed
>>>>>>>       in below RFC patch. Once proper implementation of this is 
>>>>>>> available
>>>>>>>       will update Tegra video driver to use split APIs and do 
>>>>>>> all setup
>>>>>>>       prior to device register. Added this as TODO in the driver.
>>>>>>> https://www.spinics.net/lists/linux-media/msg172761.html
>>>>>>>
>>>>>>>     Note:
>>>>>>>     Patch-0012 has compilation dependency on
>>>>>>>     https://patchwork.kernel.org/patch/11659521/
>>>>>>>
>>>>>>>
>>>>>>> [v2]:    Includes below changes based on v1 feedback
>>>>>>>     - dt-binding document and the driver update for device graph 
>>>>>>> to use
>>>>>>>       separate ports for sink endpoint and source endpoint for csi.
>>>>>>>     - Use data-lanes endpoint property for csi.
>>>>>>>     - Update tegra_mipi_request() to take device node pointer 
>>>>>>> argument
>>>>>>>       rather than adding extra API.
>>>>>>>     - Remove checking for clk pointer before clk_disable.
>>>>>>>
>>>>>>>
>>>>>>> Sowjanya Komatineni (18):
>>>>>>>     dt-bindings: i2c: tegra: Document Tegra210 VI I2C clocks and
>>>>>>>       power-domains
>>>>>>>     arm64: tegra: Add missing clocks and power-domains to 
>>>>>>> Tegra210 VI I2C
>>>>>>>     i2c: tegra: Don't mark VI I2C as IRQ safe runtime PM
>>>>>>>     i2c: tegra: Remove NULL pointer check before
>>>>>>>       clk_enable/disable/prepare/unprepare
>>>>>>>     i2c: tegra: Fix the error path in tegra_i2c_runtime_resume
>>>>>>>     i2c: tegra: Fix runtime resume to re-init VI I2C
>>>>>>>     i2c: tegra: Avoid tegra_i2c_init_dma() for Tegra210 vi i2c
>>>>>>>     media: tegra-video: Fix channel format alignment
>>>>>>>     media: tegra-video: Enable TPG based on kernel config
>>>>>>>     media: tegra-video: Update format lookup to offset based
>>>>>>>     dt-bindings: tegra: Update VI and CSI bindings with port info
>>>>>>>     media: tegra-video: Add support for external sensor capture
>>>>>>>     media: tegra-video: Add support for selection ioctl ops
>>>>>>>     gpu: host1x: mipi: Update tegra_mipi_request() to be node based
>>>>>>>     gpu: host1x: mipi: Use readl_relaxed_poll_timeout in 
>>>>>>> tegra_mipi_wait
>>>>>>>     gpu: host1x: mipi: Split tegra_mipi_calibrate and 
>>>>>>> tegra_mipi_wait
>>>>>>>     media: tegra-video: Add CSI MIPI pads calibration
>>>>>>>     media: tegra-video: Compute settle times based on the clock 
>>>>>>> rate
>>>>>>>
>>>>>>>    .../display/tegra/nvidia,tegra20-host1x.txt | 92 ++-
>>>>>>>    .../devicetree/bindings/i2c/nvidia,tegra20-i2c.txt | 19 +-
>>>>>>>    arch/arm64/boot/dts/nvidia/tegra210.dtsi | 6 +
>>>>>>>    drivers/gpu/drm/tegra/dsi.c | 9 +-
>>>>>>>    drivers/gpu/host1x/mipi.c | 37 +-
>>>>>>>    drivers/i2c/busses/i2c-tegra.c | 101 +--
>>>>>>>    drivers/staging/media/tegra-video/Kconfig | 7 +
>>>>>>>    drivers/staging/media/tegra-video/csi.c | 247 ++++++-
>>>>>>>    drivers/staging/media/tegra-video/csi.h | 8 +
>>>>>>>    drivers/staging/media/tegra-video/tegra210.c | 25 +-
>>>>>>>    drivers/staging/media/tegra-video/vi.c | 793 
>>>>>>> +++++++++++++++++++--
>>>>>>>    drivers/staging/media/tegra-video/vi.h | 25 +-
>>>>>>>    drivers/staging/media/tegra-video/video.c | 23 +-
>>>>>>>    include/linux/host1x.h | 4 +-
>>>>>>>    14 files changed, 1242 insertions(+), 154 deletions(-)
>>>>>>>

Powered by blists - more mailing lists