Re: [Bug 10489] Kprobe smoke test lockdep warning

[Masami Hiramatsu <mhiramat@redhat.com>]

Peter Zijlstra wrote:
> On Tue, 2008-04-22 at 11:25 -0400, Masami Hiramatsu wrote:
>> Peter Zijlstra wrote:
>>> On Tue, 2008-04-22 at 15:09 +0200, Peter Zijlstra wrote:
>>>> On Mon, 2008-04-21 at 18:54 -0400, Masami Hiramatsu wrote:
>>>>> Thank you for reporting.
>>>>>
>>>>> Actually, kprobes tries to fixup thread's flags in post_kprobe_handler
>>>>> (which is called from kprobe_exceptions_notify) by
>>>>> trace_hardirqs_fixup_flags(pt_regs->flags). However, even the irq flag
>>>>> is set in pt_regs->flags, true hardirq is still off until returning
>>>>> from do_debug. Thus, lockdep assumes that hardirq is off without annotation.
>>> Ah, can you clarrify? pt_regs->flags will only be set when returning to
>>> the original trap site? in that case we should not need a lockdep
>>> annotation I guess, unless its allowed and exptected for the int3 site
>>> to change IRQ state.
>> As far as I took a look at the lockdep, your suggestion is correct.
>> post_kprobe_handler should not set a lockdep annotation, because
>> processor's IF is not changed yet in that time.
> 
> That much was clear; but when _will_ it be changed?

processor's IF bit is changed when an iret(q) is issued.
In other words, when returning from a debug/int3 exception.


-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com