Hi,
在 2025/05/13 15:43, Christoph Hellwig 写道:
On Tue, May 13, 2025 at 03:14:03PM +0800, Yu Kuai wrote:
Yes, following change can work as well.
Just wonder, if the array is created by another array, and which is
created by another array ... In this case, the stack depth can be
huge. :( This is super weird case, however, should we keep the old code
in this case?
Yeah, that's a good question. Stacking multiple arrays using bitmaps
on top of each other is weird. But what if other block remappers
starting to use this for other remapping and they are stacked? That
seems much more likely unfotunately, so maybe we can't go down this
route after all, sorry for leading you to it.
I was thinking about record a stack dev depth in mddev to handle the
weird case inside raid. Is there other stack device have the same
problem? AFAIK, some dm targets like dm-crypt are using workqueue
to handle all IO.
I'm still interested because this can improve first write latency.
So instead just write a comment documenting why you switch to a
different stack using the workqueue.
Ok, I'll add comment if we keep using the workqueue.
Thanks,
Kuai
.
