Re: [PATCH 1/5] perf trace: add support for pagefault tracing

[Stanislav Fomichev]

> So make it default to the most common case :-)
Ok, agreed, will make --pf=all|maj|min and will default it to maj.

> That is my expectation, yes, if I ask for the map where address N is, it
> should return just that, where have you found this bug?
> 
> The thread__find_addr_map will end up calling maps__find() and it does
> this:
> 
>                 if (ip < m->start)
>                         p = &(*p)->rb_left;
>                 else if (ip > m->end)
>                         p = &(*p)->rb_right;
>                 else
>                         return m;
> 
> Where is the problem?
You're right, we don't need to check the range, if map is non-null it's a
valid one. I'll remove valid_dso and will just check al.map for null.

> Understood the intent, but the test here is really:
> 
> 		if (evsel->attr.type == PERF_TYPE_TRACEPOINT &&
> 		    sample.raw_data == NULL)
> 
> Ok?
Yes, makes more sense.

> Also, have you considered using:
> 
> [root@zoo ~]# perf list exceptions:page_fault*
>   exceptions:page_fault_user                         [Tracepoint event]
>   exceptions:page_fault_kernel                       [Tracepoint event]
> [root@zoo ~]#
> 
> Instead? I need to check if they're completely equivalent to what we
> need here...
Hm, didn't know there are two kinds of fault events :-/
But it seems these are x86 only.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/