Removing Giant from VFS in 10.0 (was: Re: skipping locks, mutex_owned, usb)

[Benjamin Kaduk]

On Sat, 27 Aug 2011, Robert N. M. Watson wrote:

>
> On 27 Aug 2011, at 02:43, Benjamin Kaduk wrote:
>
>>
>> For what it's worth, OpenAFS already passes MNTK_MPSAFE /* solid steel */, with what I expect is nearly-correct locking.
>
> Excellent!
>
> (Although I guess OpenAFS internally has the moral equivalent of a Giant 
> lock that protects its own structures, but that's an entirely 
> independent problem that the OpenAFS community is already interested 
> in?)

This is true.  Though it may not actually be feasible to tackle until 
better integration with a lock-tracking framework such as WITNESS or 
Linux's (GPL-only) lock debugging framework.

-Ben