Messages in this thread | | | Date | Tue, 8 Nov 2022 11:43:56 +0800 | Subject | Re: [RESEND PATCH v3] iommu/iova: Optimize alloc_iova with rbtree_augmented | From | Peng Zhang <> |
| |
> https://lore.kernel.org/linux-iommu/2b0ca6254dd0102bf559b2a73e9b51da089afbe3.1663764627.git.robin.murphy@arm.com/ I don't think this method solves the problem, because sometimes it's just that the allocation is very slow, it's not that it can't be allocated.
This is a test of the performance improvement of the network card by this patch: https://lore.kernel.org/lkml/42909903-5b6c-efe8-9ed3-3ac012f1a421@huawei.com/ At least this optimization makes sense.
The only downside of this optimization is that it is more complex, if you think it is difficult to maintain it, I can help maintain it.
Thanks, Peng.
| |