[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20190727013333.11260-1-john.ogness@linutronix.de>
Date:   Sat, 27 Jul 2019 03:33:31 +0200
From:   John Ogness <john.ogness@...utronix.de>
To:     linux-kernel@...r.kernel.org
Cc:     Peter Zijlstra <peterz@...radead.org>,
        Petr Mladek <pmladek@...e.com>,
        Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Andrea Parri <andrea.parri@...rulasolutions.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Brendan Higgins <brendanhiggins@...gle.com>
Subject: [RFC PATCH v3 0/2] printk: new ringbuffer implementation
Hello,
This is a follow-up RFC on the work to re-implement much of
the core of printk. The threads for the previous RFC versions
are here: v1[0], v2[1].
As was planned[2], this is only the first piece: a new
lockless ringbuffer.
Changes from v2:
- Moved all code into kernel/printk/. Let's keep it private
  for now.
- Split the ringbuffer into 3 components:
  * a data ringbuffer (dataring) to manage the raw data and
    data descriptors
  * a numbered list (numlist) to manage committed entries and
    their sequence numbers
  * the printk_ringbuffer, which is the high-level structure
    providing the reader/writer API and glue for the other
    structures
  Splitting the components apart helped to document their
  roles and their related memory barriers (and will hopefully
  also simplify the review process).
- Renamed most functions, structures, and variables based on
  v2 feedback.
- Rewrote and reformatted nearly all comments (particularly
  the memory barrier comments) based on v2 feedback.
- Addressed implementation issues with v2:
  * invalid data blocks potentially becoming valid because of
    overflows
  * weak associations between data blocks and descriptors
  * excessive freeing of data blocks due to unavailable
    descriptors
- Improved error handling and data integrity checks in the test
  module.
For the memory barrier work I wrote a litmus test for nearly
every memory barrier. I did not include these in the series.
Should I? If yes, where should they be placed?
I would like to point out that Petr Mladek posted a
proof-of-concept[3] alternate implementation. I wanted to base my
v3 on his work, but ran into too many problems getting it to
run acceptably. I will address those issues in that thread. This
is why my v3 is based directly on my v2.
John Ogness
[0] https://lkml.kernel.org/r/20190212143003.48446-1-john.ogness@linutronix.de
[1] https://lkml.kernel.org/r/20190607162349.18199-1-john.ogness@linutronix.de
[2] https://lkml.kernel.org/r/87y35hn6ih.fsf@linutronix.de
[3] https://lkml.kernel.org/r/20190704103321.10022-1-pmladek@suse.com
John Ogness (2):
  printk-rb: add a new printk ringbuffer implementation
  printk-rb: add test module
 kernel/printk/Makefile     |   5 +
 kernel/printk/dataring.c   | 761 ++++++++++++++++++++++++++++++++++++++++++
 kernel/printk/dataring.h   |  95 ++++++
 kernel/printk/numlist.c    | 375 +++++++++++++++++++++
 kernel/printk/numlist.h    |  72 ++++
 kernel/printk/ringbuffer.c | 800 +++++++++++++++++++++++++++++++++++++++++++++
 kernel/printk/ringbuffer.h | 288 ++++++++++++++++
 kernel/printk/test_prb.c   | 256 +++++++++++++++
 8 files changed, 2652 insertions(+)
 create mode 100644 kernel/printk/dataring.c
 create mode 100644 kernel/printk/dataring.h
 create mode 100644 kernel/printk/numlist.c
 create mode 100644 kernel/printk/numlist.h
 create mode 100644 kernel/printk/ringbuffer.c
 create mode 100644 kernel/printk/ringbuffer.h
 create mode 100644 kernel/printk/test_prb.c
-- 
2.11.0
Powered by blists - more mailing lists
 
