Re: [PATCH 0 of 4] Generic AIO by scheduling stacks

[Zach Brown]

> Does that mean that we might not have some cases where we'd need to  
> make
> sure we do things differently? Of course not. Something migt show up.

Might, and has.  For a good time, take journal_info out of  
per_call_chain() in the patch set and watch it helpfully and loudly  
wet itself.  There really really are bits of thread_struct which are  
strictly thread-local-storage, of a sort, for a kernel call path.    
Sharing them, even only through cooperate scheduling, is fatal.   
link_count is another obvious one.

They're also the only ones I've bothered to discover so far :).

> But
> this actually makes it very clear what the difference between "struct
> thread_struct" and "struct task_struct" are. One is shared between
> fibrils, the other isn't.

Indeed.

Right now the per-fibril uses of task_struct members are left inline  
in task_struct and are copied on fibril switches.

We *could* put them in thread_info, at the cost of stack pressure, or  
hang them off task_struct in their own struct to avoid the copies, at  
the cost of indirection.  I didn't like imposing a cost on paths that  
don't use fibrils, though, so I left them inline.

(I think you know all this.  I'm clarifying for the peanut gallery, I  
hope.)

- z
-
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/