Re: [PATCH 1/2] tracing: Add $cpu and $current probe-vars

[Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>]

(2013/11/27 2:23), Oleg Nesterov wrote:
> On 11/26, Masami Hiramatsu wrote:
>>
>> (2013/11/26 4:29), Oleg Nesterov wrote:
>>> +#define PSEUDO_REG_OFFSET	4096	/* arbitrary value > MAX_REG_OFFSET */
>>> +
>>> +static unsigned long probe_get_register(struct pt_regs *regs, unsigned long offset)
>>> +{
>>> +	if (offset < PSEUDO_REG_OFFSET)
>>> +		return regs_get_register(regs, offset);
>>> +
>>> +	return pseudo_reg_table[offset - PSEUDO_REG_OFFSET].fetch();
>>> +}
>>
>>
>> Hmm, I don't like this, since fetch_reg is the most frequently
>> used fetch method, and it actually uses the offset in different
>> way.
> 
> Sure, this overloads the usage of FETCH_MTD_reg/offset.
> 
> And btw, yes, the naming is ugly (I mean pseudo_*). But why this
> is bad? This is cheap and simple.

I think it's not simple. The code looks short, but not well
self-described. It is "hidden" in the structure, and double-meaning.

> And,
> 
>> Why don't you make another fetch function for vars?
> 
> This is what I tried to avoid ;) I do not want to add another
> FETCH_MTD_. To me, this looks like unnecessary complication and
> bloat (but see below).

I see, perhaps, it is a time to introduce independent fetch
method implementation. Current one is not sophisticated
nor generic. We need a kind of "bytecode" implementation
for more complex expression, especially for the combination
of several registers/variables on stacks. This feature will
also be useful for some other scriptable tracers like ktap,
systemtap etc.

Thank you,

-- 
Masami HIRAMATSU
IT Management Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: masami.hiramatsu.pt@hitachi.com