On 2024-07-19, Andrew Halaney <ahalaney@xxxxxxxxxx> wrote:
>> Your thinking is not naive. It is correct (for PREEMPT_RT). printk()
>> stores messages locklessly into its ringbuffer and must work in all
>> cases (including NMI and scheduler). And AFAIK it does.
>>
>> However, printk() is also responsible for reliably triggering or
>> performing console printing. This separate responsibility is tricky. For
>> PREEMPT_RT the problem is supposed to be solved. Obviously it still has
>> some issues, so thank you for the report.
>>
>> For PREEMPT_RT, if you can put a printk() anywhere after the Linux
>> banner and cause a problem, the printk-folks need to know about it.
>
> Just to be sure I'm following correctly, when you say PREEMPT_RT above,
> are you meaning "the printk/nbcon/console solution in PREEMPT (independent of
> the current CONFIG_PREEMPT* setting)"? Or do you mean that + CONFIG_PREEMPT_RT?
First, the _complete_ printk/nbcon/console solution must be present in
the kernel (most likely by simply applying the PREEMPT_RT patchset).
Then, either
1. Enabling CONFIG_PREEMPT_RT
or
2. Enabling some other preemption model but only using nbcon consoles
(which currently is only the 8250 uart driver)
For !CONFIG_PREEMPT_RT using legacy consoles the
non-interference/reliability properties of the console printing is not
changed from current mainline. This is on purpose.
John
