Re: [patch 06/41] x86/traps: Simplify pagefault tracing logic

[Steven Rostedt]

On Fri, 25 Aug 2017 12:31:09 +0200
Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:

>  dotraplinkage void notrace
> -trace_do_page_fault(struct pt_regs *regs, unsigned long error_code)
> +do_page_fault(struct pt_regs *regs, unsigned long error_code)
>  {
> +	unsigned long address = read_cr2(); /* Get the faulting address */
> +	enum ctx_state prev_state;
> +
>  	/*
> -	 * The exception_enter and tracepoint processing could
> -	 * trigger another page faults (user space callchain
> -	 * reading) and destroy the original cr2 value, so read
> -	 * the faulting address now.
> +	 * We must have this function tagged with __kprobes, notrace and call
> +	 * read_cr2() before calling anything else. To avoid calling any kind
> +	 * of tracing machinery before we've observed the CR2 value.
> +	 *
> +	 * exception_{enter,exit}() contain all sorts of tracepoints.
>  	 */
> -	unsigned long address = read_cr2();
> -	enum ctx_state prev_state;
> +	if (trace_irqvectors_enabled())
> +		trace_page_fault_entries(address, regs, error_code);
>  
>  	prev_state = exception_enter();
> -	trace_page_fault_entries(address, regs, error_code);

I believe you need to keep the tracing after the "exception_enter()", as
for NO_HZ_FULL, that enables RCU again, and the tracepoint needs RCU
enabled.

-- Steve


>  	__do_page_fault(regs, error_code, address);
>  	exception_exit(prev_state);
>  }
> -NOKPROBE_SYMBOL(trace_do_page_fault);
> -#endif /* CONFIG_TRACING */
> +NOKPROBE_SYMBOL(do_page_fault);
>