Re: [RFC v3 8/9] kmsg: add predefined _PID, _TID, _COMM keywords to kmsg* log dict

On Tue, Oct 20, 2015 at 4:16 AM, Paul Osmialowski
[off-list ref] wrote:

Hi Andy,

On Mon, 19 Oct 2015, Andy Lutomirski wrote:

quoted

On Mon, Oct 19, 2015 at 5:58 AM, Paul Osmialowski
[off-list ref] wrote:

quoted

From: Marcin Niesluchowski <redacted>

kmsg* devices write operation wrote no dict along with message
Due to usage of kmsg devices in userspace dict has been added
identifying pid, tid and comm of writing process.

Does this affect even the normal /dev/kmsg?

Yes.

Is this okay?  Comm at least is a bit odd.

quoted

quoted

-static int kmsg_sys_write(int minor, int level, const char *fmt, ...)
+static size_t set_kmsg_dict(char *buf)
+{
+       size_t len;
+
+       len = sprintf(buf, "_PID=%d", task_tgid_nr(current)) + 1;
+       len += sprintf(buf + len, "_TID=%d", task_pid_nr(current)) + 1;
+       memcpy(buf + len, "_COMM=", 6);
+       len += 6;
+       get_task_comm(buf + len, current);
+       while (buf[len] != '\0')
+               len++;

len += strlen(buf); ?

Is it obvious for some reason that this doesn't overflow buf?

KMSG_DICT_MAX_LEN sets architecture-intepentent max size.

And how many things are written and what asserts that it's big enough?

quoted

Why is task_pid_nr acceptable here?  Isn't this intended for use in
namespaces?

task_tgid_nr - process id (pid as seen in userspace),
task_pid_nr - thread id

These are the ids as seen by the global namespace.  If the goal is to
create one of these nodes in some other namespace, then this is the
wrong thing to do.

--Andy