Messages in this thread | | | Date | Wed, 27 May 2015 18:12:59 +0600 | Subject | Re: [PATCH v8 0/3] x86/earlyprintk: setup serial earlyprintk as early as possible | From | Alexander Kuleshov <> |
| |
2015-05-27 18:01 GMT+06:00 Borislav Petkov <bp@suse.de>: > Yes, and I'm simply asking why do we even want to use it there at all. > What do you think would need to be dumped with early_printk() calls > there and would be useful? > > IOW, do you have a sensible use case for why we need it there or not?
Is early printk debugging sensible use case? We have memblock usage on the early stage which uses printk for debugging which does not work on the early stage, early ioremap intitialization, early cpu initialization, many others stuff and I'm not sure that this code between head{32,64}.c and parse_early_param will remain unchanged forever. Somebody will update/add/remove something from it and ealyprintk allows us to see what's going on.
Is it bad or we no need in it? I want to understand not only how kernel works, but also why it written so and I can't understand why we have 'earlyprintk' (or another feature) in one place and do not have in another. Maybe there is explanation and I very want to know about it.
Thank you Borislav.
| |