b-rad: ok, thanks ukleinek: https://patchwork.kernel.org/patch/11031383/ just code refactoring, no functional changes intended ukleinek: I'd appreciate a review BTW, it's easy to mess up these kinds of patches ;-) mchehab: if syoung is maintainer of DVB parts, shouldn't that show up somewhere in MAINTAINERS for get_maintainers to pick up? (I can send a patch if you want) marc|gonzalez: the change is si2168_ts_bus_ctrl is a bit ugly, as the internals are not completely hidden. cmd.args iss till modified. marc|gonzalez: not a constructive critisism though, no better idea marc|gonzalez: you have mail ukleinek: thanks I appreciate it. b-rad: syoung: what do you think? Should I spin a v4? ukleinek: I found a bug in si2168 that might affect your si2157 thing In si2168_probe, we memcpy the struct dvb_frontend, and zero the tuner.ops, so if the si2157_probe function occurs during si2168_probe, then bad things happen marc|gonzalez: if you provide a patch, I can test that ukleinek: OK, just a sec hmm, this won't explain my problem though. http://paste.debian.net/1091340/ is my current understanding of the problem ukleinek: https://pastebin.ubuntu.com/p/JXycPKSVXz/ probably too simple to explain your issue, though but it fixed an issue for me it's in the memcpy... ->ops.tuner_ops Unable to handle kernel NULL pointer dereference at virtual address 00000020 *config->i2c_adapter = dev->muxc->adapter[0]; before dev->muxc looks wrong ukleinek: my patch breaks linux? OK interesting will do it more carefully then Duh, you're right, I see why it didn't bomb for me marc|gonzalez: ack Let me spin it differently I still fail to see your issue ah, the i2c stuff might already make use of &dev->fe.ops I assume? ukleinek: https://pastebin.ubuntu.com/p/S4c6gdFxht/ ukleinek: right, in DT mode, i2c_mux_add instantiates the tuner driver e.g. si2157 and so will populate the fe->tuner_ops field which the memcpy promptly zeroes but since you're using the old way of init'ing 68 then 57, it's probably not your issue at all sorry for noise completely unrelated b-rad: my contact from silabs couldn't really help, too. I'll have to wait until my logic analyzer that I ordered is here. Then I'll check the i2c lines and clocking not terribly surprised ukleinek do you have a firmware that came with the device? b-rad: I'd have to check if I still have the CD ... b-rad: but I don't expect that this yield interesting facts. Another guy already traced how Windows behaves with the stick and there are just no i2c-reads after the demod is booted