On Wed, May 16, 2018 at 2:32 PM Tejun Heo <tj@xxxxxxxxxx> wrote:
> The least disruptive thing likely is silently skipping !printable
> chars. I'm not sure about the cost vs. benefit tho. It's silly to
> include !printables in comm but maybe people use utf-whatever or other
> encodings for comm somewhere in the world which could be incorrectly
> skipped? I'm leaning more towards just letting it be and letting
> userland deal with it. What do you think?
The utf-8 argument is certainly valid. I'd hate to cause trouble for
non-English-speaking people.
And at least procps-ng seems to do the printability right in user space,
and apparently it's not a big deal for systemd either, so I guess it's not
worth worrying about it.
Also, if we actually just make /proc/<pid>/cmdline always fall back to
using p->comm[], then it doesn't even matter: only kernel threads will have
an empty cmdline, and kernel threads won't have unprintable characters
unless we've screwed up majorly.
And people are bound to already be careful with /proc/<pid>/cmdline, since
that is always binary (at the very least, it has the NUL characters for
argument boundaries)
Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-api" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
