lkml.org 
[lkml]   [2020]   [Jun]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: syslog size unread: was: [PATCH v3 3/3] printk: use the lockless ringbuffer
Date
On 2020-06-25, Petr Mladek <pmladek@suse.com> wrote:
> On Thu 2020-06-18 16:55:19, John Ogness wrote:
>> --- a/kernel/printk/printk.c
>> +++ b/kernel/printk/printk.c
>> @@ -1609,11 +1633,15 @@ int do_syslog(int type, char __user *buf, int len, int source)
>> break;
>> /* Number of chars in the log buffer */
>> case SYSLOG_ACTION_SIZE_UNREAD:
>> + if (source != SYSLOG_FROM_PROC) {
>> + text = kmalloc(LOG_LINE_MAX + PREFIX_MAX, GFP_KERNEL);
>> + if (!text)
>> + return -ENOMEM;
>
> The buffer is needed only to count lines to count the size of added
> prefixes. Could we use the new prb_read_valid_info() that allows to
> get the number of lines without actually reading the buffer?

Yes!

For the next version I introduce a macro to iterate just the meta data:

#define prb_for_each_info(from, rb, s, i, lc) \
for ((s) = from; prb_read_valid_info(rb, s, i, lc); (s) = (i)->seq + 1)

This can be used in all 3 locations where prb_count_lines() is used. In
all three places, there is no need to be copying the data. And for
SYSLOG_ACTION_SIZE_UNREAD, there is no need for the record and kmalloc'd
buffer.

This also means that prb_count_lines() will become a private static
helper of the ringbuffer.

John Ogness

\
 
 \ /
  Last update: 2020-06-30 00:10    [W:0.243 / U:4.052 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site