Re: [PATCH/RFC] Add MM_DEAD flag to struct mm_struct

From: Benjamin Herrenschmidt
Date: Fri Jul 27 2007 - 02:42:18 EST


On Fri, 2007-07-27 at 12:01 +1000, Benjamin Herrenschmidt wrote:
> Some architectures like sparc can do useful optimizations when knowing
> that an entire MM is being destroyed. At the moment, they rely on
> fullmm in the mmu_gather structure. However, that doesn't always work
> out very well with some of the changes we are doing. Among other things,
> the TLB flushing on sparc64 is done using per-CPU tracking data, making
> the batch not per-CPU will break that link.
>
> So instead, we add a new flag to struct mm_struct that indicates that
> the mm is going away, for those archs to use. It also allows to use it
> in situations (such as PTE ops) where the batch isn't accessible (or
> there is no batch)

And the patch forgets to clear it in mm_init ... will send a new one
later.

Cheers,
Ben.


-
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/