Re: output: was: Re: [PATCH v4] printk: Userspace format enumeration support

[Chris Down <chris@chrisdown.name>]

Hey Petr,

Petr Mladek writes:
>This produces something like:
>
>3Warning: unable to open an initial console.
>3Failed to execute %s (error %d)
>6Kernel memory protection disabled.
>3Starting init: %s exists but couldn't execute it (error %d)
>6Run %s as init process
>7initcall %pS returned %d after %lld usecs
>7calling  %pS @ %i
>2initrd overwritten (0x%08lx < 0x%08lx) - disabling it.
>
>The loglevel is not well separated. It is neither human readable
>nor safe for a machine processing . It works only for single digit.
>[...]
>It looks in less like: [...]

Hmm, why is it important that debugfs output is human readable? My impression 
was that it's fine to have machine-readable stuff there.

Re: not being not safe for machine processing because it only works for a 
single digit, I'm a little confused. KERN_* levels are, as far as I know, only 
a single byte wide, and we rely on that already (eg. in printk_skip_header()). 
We also already have precedent for null-separation/control characters in (for 
example) /proc/pid/cmdline.

What am I missing? :-)

>Well, it still might be non-trivial to find the string in the code
>and see what exactly has changed. It might be pretty useful
>to mention even the source_file:line, for example:
>
><3> init/main.c:1489: Warning: unable to open an initial console.\n
><3> init/main.c:1446: Failed to execute %s (error %d)\n
><6> init/main.c:1398: Kernel memory protection disabled.\n
><3> init/main.c:1366: Starting init: %s exists but couldn't execute it (error %d)\n

Almost certainly a theoretical concern, but I am not a big fan of this format, 
because it relies on a character (":") which is legal in paths (although as 
you'd expect, we don't have any cases in the kernel right now). That's one of 
the reasons why I preferred to use nulls, which can't be in a filename.