linux-kernel - Re: [PATCH 1/2] trace: Add trap entry/exit tracepoints

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:	Fri, 29 Apr 2011 02:33:04 +0200 (CEST)
From:	Thomas Gleixner <tglx@...utronix.de>
To:	Vaibhav Nagarnaik <vnagarnaik@...gle.com>
cc:	Steven Rostedt <rostedt@...dmis.org>,
	Ingo Molnar <mingo@...hat.com>,
	Michael Rubin <mrubin@...gle.com>,
	David Sharp <dhsharp@...gle.com>, linux-kernel@...r.kernel.org,
	x86@...nel.org, Jiaying Zhang <jiayingz@...gle.com>
Subject: Re: [PATCH 1/2] trace: Add trap entry/exit tracepoints

On Fri, 22 Apr 2011, Vaibhav Nagarnaik wrote:
>  #include <asm/uaccess.h>
>  #include <asm/pgtable.h>
>  #include <asm/system.h>
> @@ -1330,8 +1332,10 @@ void send_sigtrap(struct task_struct *tsk, struct pt_regs *regs,
>  	struct siginfo info;
>  
>  	fill_sigtrap_info(tsk, regs, error_code, si_code, &info);
> +	trace_trap_entry(tsk->thread.trap_no);

What the heck? How is that a trap? The code is sending SIGTRAP not
entering a trap at all. What are you trying to measure ? The time it
takes to send SIGTRAP? So how is that useful as an extra event? 

>  	/* Send us the fake SIGTRAP */
>  	force_sig_info(SIGTRAP, &info, tsk);
> +	trace_trap_exit(tsk->thread.trap_no);
>  }
>  
>  
> diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
> index 4857ff6..d450349 100644
> --- a/arch/x86/kernel/traps.c
> +++ b/arch/x86/kernel/traps.c
> @@ -33,6 +33,9 @@
>  #include <linux/io.h>
>  #include <trace/events/irq.h>
>  
> +#define CREATE_TRACE_POINTS
> +#include <trace/events/trap.h>
> +
>  #ifdef CONFIG_EISA
>  #include <linux/ioport.h>
>  #include <linux/eisa.h>
> @@ -123,6 +126,7 @@ do_trap(int trapnr, int signr, char *str, struct pt_regs *regs,
>  {
>  	struct task_struct *tsk = current;
>  
> +	trace_trap_entry(trapnr);

While the event of do_trap() itself might be interesting, it does not
matter at all how long it takes to handle it. That code is really not
so interesting.

> @@ -286,7 +293,9 @@ do_general_protection(struct pt_regs *regs, long error_code)
>  		printk("\n");
>  	}
>  
> +	trace_trap_entry(tsk->thread.trap_no);
>  	force_sig(SIGSEGV, tsk);
> +	trace_trap_exit(tsk->thread.trap_no);

We really do not care how long the force_sig() call takes. It's
irrelevant. The only interesting thing here is that we ran into a GP
trap.

> +
> +dotraplinkage void __kprobes
> +do_page_fault(struct pt_regs *regs, unsigned long error_code)
> +{
> +	trace_trap_entry(14);

Yuck. Magic number 14 ? Why not 42 ?

> +	__do_page_fault(regs, error_code);
> +	trace_trap_exit(14);
> +}

That page fault thing is the only interesting event in terms of
runtime, but I have yet to see a proper rationale for the whole thing
aside of that completly bogus changelog which tells what output I can
produce, but not why the hell it is a good idea to add all that trace
points.

Thanks,

	tglx
 
--
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/