linux-kernel - [RFC PATCH 2/3] printk: add console_verbose

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Sun, 15 Mar 2020 18:09:02 +0100
From:   Eugeniu Rosca <roscaeugeniu@...il.com>
To:     linux-kernel@...r.kernel.org
Cc:     Petr Mladek <pmladek@...e.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Ingo Molnar <mingo@...nel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Peter Zijlstra <peterz@...radead.org>,
        Jisheng Zhang <Jisheng.Zhang@...aptics.com>,
        Valdis Kletnieks <valdis.kletnieks@...edu>,
        Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
        Andrew Gabbasov <andrew_gabbasov@...tor.com>,
        Dirk Behme <dirk.behme@...bosch.com>,
        Eugeniu Rosca <erosca@...adit-jv.com>
Subject: [RFC PATCH 2/3] printk: add console_verbose_{start,end}

Consider below example scenarios:
       - soft lockup
       - hard lockup
       - hung task
       - scheduling while atomic
       - rcu stall
       - oops
       - oom
       - WARN

In most of the above situations, it is up to the user to terminate the
execution with a panic (see panic_on_* in kernel/sysctl.c) or to
decide not to abort and try to recover.

A general concern applicable to the above use-cases is that, depending
on the console loglevel set by the user, precious information conveyed
by show_regs(), print_modules(), dump_stack() and friends may simply not
appear on the console. Below example commits tackle this exact concern
in the panic paths:

* commit 168e06f7937d96 ("kernel/hung_task.c: force console verbose before panic")
* commit 5b530fc1832460 ("panic: call console_verbose() in panic")

The approach behind the above commits is straightforward. Whenever
panic is imminent, they simply call console_verbose(). Unfortunately,
the same technique does not apply to non-panic paths. It requires a
counterpart of console_verbose(), which currently does not exist.

With that in mind, create a pair of functions named
console_verbose_start() and console_verbose_end() which turn the
console verbosity on and off in a lockless and SMP safe way.

Cc: Petr Mladek <pmladek@...e.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@...il.com>
Cc: Steven Rostedt <rostedt@...dmis.org>
Cc: Ingo Molnar <mingo@...nel.org>
Signed-off-by: Eugeniu Rosca <erosca@...adit-jv.com>
---
 include/linux/printk.h | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/include/linux/printk.h b/include/linux/printk.h
index 1e6108b8d15f..14755ef7b017 100644
--- a/include/linux/printk.h
+++ b/include/linux/printk.h
@@ -3,6 +3,7 @@
 #define __KERNEL_PRINTK__
 
 #include <stdarg.h>
+#include <linux/atomic.h>
 #include <linux/init.h>
 #include <linux/kern_levels.h>
 #include <linux/linkage.h>
@@ -77,6 +78,15 @@ static inline void console_verbose(void)
 		console_loglevel = CONSOLE_LOGLEVEL_MOTORMOUTH;
 }
 
+#ifdef CONFIG_PRINTK
+extern atomic_t ignore_loglevel;
+static inline void console_verbose_start(void) { atomic_inc(&ignore_loglevel); }
+static inline void console_verbose_end(void) { atomic_dec(&ignore_loglevel); }
+#else
+static inline void console_verbose_start(void) { }
+static inline void console_verbose_end(void) { }
+#endif
+
 /* strlen("ratelimit") + 1 */
 #define DEVKMSG_STR_MAX_SIZE 10
 extern char devkmsg_log_str[];
-- 
2.25.0