lkml.org 
[lkml]   [2006]   [Jun]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [Patch 3/6] statistics infrastructure - prerequisite: timestamp
Tim Bird wrote:
> Martin Peschke wrote:
>> And I think that I have fixed a printed_len related miscalculation.
>> printed_len needs to be increased if no valid log level has been found
>> and a log level prefix has been added by printk(). Otherwise, printed_len
>> must not be increased. The old code did it the other way around (in the
>> timestamp case).
>
> I'm not following your change here. I can't find the problem you
> mention in the original code. And your fix appears to mess up the
> printed_len.

I am sorry for not responding sooner ... and I am sorry for introducing
a bug here. I think you are right.

>> + if (new_line) {
>> + /* The log level token is first. */
>> + int loglev_char;
>> + if (p[0] == '<' && p[1] >='0' &&
>> + p[1] <= '7' && p[2] == '>') {
>> + loglev_char = p[1];
>> + p += 3;
>> + printed_len -= 3;
> Here you subtract from the printed_len to account for skipping
> the submitted log level.
>
>> + } else {
>> + loglev_char = default_message_loglevel + '0';
>> + }
>> + emit_log_char('<');
>> + emit_log_char(loglev_char);
>> + emit_log_char('>');
>
> But here you don't re-count the chars for the log-level
> you are adding back in. There's a discrepancy here.

Correct. My print_len loses 3 in the 'got log level'-case due to a surplus
substraction. It also loses 3 in the other case due to adding a log level
substring that is not entered in the books.

This is how I would fix it:

printed_len = vscnprintf(printk_buf, sizeof(printk_buf), fmt, args);

...

if (new_line) {
/* The log level token is first. */
int loglev_char;
if (p[0] == '<' && p[1] >='0' &&
p[1] <= '7' && p[2] == '>') {
loglev_char = p[1];
p += 3;
} else {
loglev_char = default_message_loglevel + '0';
print_len += 3;
}
emit_log_char('<');
emit_log_char(loglev_char);
emit_log_char('>');


Could you confirm, please. I will send a patch to Andrew then.


Or, Andrew, do you prefer a replacement patch for my
statistics-infrastructure-prerequisite-timestamp.patch
that introduces this miscalculation. I could put
statistics-infrastructure-make-printk_clock-a-generic-kernel-wide-nsec-resolution.patch
into that one as well, so that all the timestamp related printk-changes are in
one place.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2006-06-01 10:43    [W:0.097 / U:0.200 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site