Michael Brown

Which kernel module is providing the SPI controller driver in your setup?

It should take only an hour or so to narrow down the root cause using ftrace (or by adding printk() statements to the spi-imx.c driver), given access to the hardware....

On 20/11/2018 09:53, jonathanmuller wrote: > But dw1000_tx_data_complete() is called before spin_unlock_irqrestore(), > and the check catches the same error -22 : > > |if (unlikely(tx->data.status != 0)) { dev_err(dw->dev,...

On 20/11/2018 15:30, jonathanmuller wrote: > vmalloced_buf is 0 > kmap_buf is 0 > virt_addr_valid(buf) is 0 The TX data messages comprises multiple SPI transfer sets, each of which has...

Could you try the patch at #15 to see if this fixes the problem? Thanks, Michael

Thank you. Could you also check that the "sensors" command is able to read a power supply voltage and operating temperature from the dw1000 chip? (The same bug was present...

Just install the "lm_sensors" package and run "/usr/bin/sensors": that should be all you need to test it.

I think we need to have the veto take effect only if `CONSOLE_SERIAL` is enabled, in the same way that `efi_veto_hp_xhci()` will veto the HP xHCI driver only if we...

> Unfortunate, it's more an ugly workaround then a real fix, as I assume the intention of serial_iskey is to return as fast as possible and do blocking call in...

> @mcb30 Hmm... ok, your explanation makes sense, but from below logs I don't see any `terminal` or `serial` type of efi driver that can be unload: Build with `DEBUG=efi_veto:3`...