linux-kernel - Re: ftrace: preemptoff selftest not working

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:	Tue, 18 Nov 2008 08:49:41 -0500 (EST)
From:	Steven Rostedt <rostedt@...dmis.org>
To:	Heiko Carstens <heiko.carstens@...ibm.com>
cc:	linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...e.hu>,
	Martin Schwidefsky <schwidefsky@...ibm.com>
Subject: Re: ftrace: preemptoff selftest not working


On Tue, 18 Nov 2008, Heiko Carstens wrote:

> Hi Steven,
> 
> I was wondering why the preemptoff and preemptirqsoff tracer selftests
> don't work on s390. After all its just that they get called from
> non-preemptible context:
> 
> kernel_init() will execute all initcalls, however the first line in
> kernel_init() is lock_kernel(), which causes the preempt_count to be
> increased. Any later calls to add_preempt_count() (especially those
> from the selftests) will therefore not result in a call to
> trace_preempt_off() since the check below in add_preempt_count()
> will be false:
> 
> 	if (preempt_count() == val)
> 		trace_preempt_off(CALLER_ADDR0, get_parent_ip(CALLER_ADDR1));
> 
> Hence the trace buffer will be empty.
> The patch below makes the selftests working for me, since then they run
> in preemptible context. But it is ugly and I'm not proposing it for
> upstream ;)
> 
> Just wanted to make you aware that there is a bug.

Yep, this might be a better answer than what I put into linux-tip (and my 
git repo).

See:

  ftrace: force pass of preemptoff selftest

The cause of the bug was the conversion of the BKL back to a spinlock, and 
making it non preempt.  The initcall code is called with the BKL applied 
which now means it can not preempt. This breaks the preempt tracer 
selftest.

My solution was to just force a pass if this is detected. Perhaps moving 
the test might be better.

-- Steve

> 
> ---
>  init/main.c                  |    4 ++++
>  kernel/trace/trace_irqsoff.c |    3 +--
>  2 files changed, 5 insertions(+), 2 deletions(-)
> 
> Index: linux-2.6/kernel/trace/trace_irqsoff.c
> ===================================================================
> --- linux-2.6.orig/kernel/trace/trace_irqsoff.c
> +++ linux-2.6/kernel/trace/trace_irqsoff.c
> @@ -482,7 +482,7 @@ static struct tracer preemptirqsoff_trac
>  # define register_preemptirqsoff(trace) do { } while (0)
>  #endif
>  
> -__init static int init_irqsoff_tracer(void)
> +int init_irqsoff_tracer(void)
>  {
>  	register_irqsoff(irqsoff_tracer);
>  	register_preemptoff(preemptoff_tracer);
> @@ -490,4 +490,3 @@ __init static int init_irqsoff_tracer(vo
>  
>  	return 0;
>  }
> -device_initcall(init_irqsoff_tracer);
> Index: linux-2.6/init/main.c
> ===================================================================
> --- linux-2.6.orig/init/main.c
> +++ linux-2.6/init/main.c
> @@ -789,6 +789,8 @@ static void run_init_process(char *init_
>  	kernel_execve(init_filename, argv_init, envp_init);
>  }
>  
> +extern int init_irqsoff_tracer(void);
> +
>  /* This is a non __init function. Force it to be noinline otherwise gcc
>   * makes it inline to init() and it becomes part of init.text section
>   */
> @@ -800,6 +802,8 @@ static int noinline init_post(void)
>  	system_state = SYSTEM_RUNNING;
>  	numa_default_policy();
>  
> +	init_irqsoff_tracer();
> +
>  	if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
>  		printk(KERN_WARNING "Warning: unable to open an initial console.\n");
>  
> 
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/