Hi,
We are currently porting our RGB IMX715 camera to JetPack6.1 on the Orin Nano. We see the following messages (dmesg) when recording the camera:
*[ 71.153000] imx715 9-001a: imx715_set_frame_rate: frame rate: 27fps*
*[ 71.163880] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.193452] tegra-mc 2c00000.memory-controller: vifalr: read @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.193470] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.193928] tegra-camrtc-capture-vi tegra-capture-vi: corr_err: discarding frame 0, flags: 0, err_data 131072*
*[ 71.199494] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.230462] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.236517] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.267487] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.273544] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
*[ 71.304513] tegra-mc 2c00000.memory-controller: vifalw: write @0x0000007ffc010000: EMEM address decode error (EMEM decode error)*
The RGB camera is currently generating test images using the test pattern mode. In the image you can see that the image data was correct in JetPack 5.1.2 (upper image), but in JetPack 6.1 (bottom image) the data is corrupt.
This problem is very similar to issue :https://forums.developer.nvidia.com/t/camera-vi-jp-6-0-r36-3-fixes-for-raw-image-corruption-line-artifacts-through-vi-mode/299834
The proposed patches are included in JP6.1 … this is ok after verifcation (Device tree + source address.c)
Do you have any idea what is going wrong? We suspect problems with the DMA in module VI.