Re: Making bpf MPSAFE (was Re: struct ifnet and ifaddr handling ...)

[Ryota Ozaki <ozaki-r%netbsd.org@localhost>]

On Thu, Aug 7, 2014 at 8:48 PM, Ryota Ozaki <ozaki-r%netbsd.org@localhost> 
wrote:
> On Thu, Aug 7, 2014 at 7:57 PM, Greg Troxel <gdt%ir.bbn.com@localhost> wrote:
>>
>> Ryota Ozaki <ozaki-r%netbsd.org@localhost> writes:
>>
>>> BTW, I worry about there is no easy way to
>>> know if a function in a critical section
>>> blocks/sleeps or not. So I wrote a patch to
>>> detect that: http://www.netbsd.org/~ozaki-r/debug-pserialize.diff
>>> Is it meaningful?
>>
>> It seems sensible, but it is very much undercommented.  Somewhere, the
>> point of the variable, what it means, and why the test is right should
>> be explained.
>
> Agreed. I'll write some notes.

http://www.netbsd.org/~ozaki-r/debug-pserialize.diff

I added some comments and made the new variable name
more meaningful. How does this look?

  ozaki-r

>
>>
>> I think the point is that sleep should not be called while pserialize is
>> active.  And probably preemption shouldn't happen either, so we won't
>> get false hits from other threads sleeping.
>>
>> I realize it's typical to reject adding such comments on the theory that
>> everyone should already understand all of this.  But the complexity of
>> our locking (and memory allocation) is high, and I think it helps
>> correctness to be much more explicit about the plan.
>
> Yes, I think we need some manual equivalent to locking(9) of FreeBSD.
>
>   ozaki-r