Hi Shimoda-san, I had a look now with Kieran's Spider and could reproduce your findings. It is not a problem, though, because the system behaves the same with plain v6.6-rc1. The only difference with my branch is that DEBUG is enabled for the I2C driver, so we see all the messages regarding EPROBE_DEFER with DMA. If you add DEBUG to plain v6.6-rc1, then you see the same messages. But still, the busses get probed and they also get DMA channels after some time when the DMA driver is finally loaded. So, this really has nothing to do with the patches themselves, only with the additions for debugging. Phew! Thank you for reporting and testing, still! Wolfram
Attachment:
signature.asc
Description: PGP signature
