[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-ppc] [PATCH qemu v7 06/14] spapr_iommu: Introduce "enabled" st
From: |
Paolo Bonzini |
Subject: |
Re: [Qemu-ppc] [PATCH qemu v7 06/14] spapr_iommu: Introduce "enabled" state for TCE table |
Date: |
Mon, 6 Jul 2015 13:16:28 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.0.1 |
On 04/07/2015 03:12, Alexey Kardashevskiy wrote:
>>
>>> One step back :) Whole dance is what here? There are:
>>> 1) del+set_size(0)
>>> 2) set_size(not zero)+add
>>
>> Then no need for begin/commit. :)
>
> I got a new problem here - set_size(0) + set_size(not 0) do not invoke
> region_del/region_add which does not seem right. As the result,
> vfio_listener_region_del() never gets called and the IOMMU MR notifier
> stays in the container->giommu_list. What is the correct solution to this?
>
> Incorrect (most likely :) ) solution is in
> [PATCH qemu v9 13/13] spapr_pci/spapr_pci_vfio: Support Dynamic DMA
> Windows (DDW)
> where I explicitely do memory_region_del_subregion() +
> memory_region_add_subregion() which seems to do the right job.
I would like a comment there explaining the problem (i.e. what you would
have done, what you expected and what were the actual results: it's not
even that clear from this email), but it doesn't seem like an awful
thing to do.
Paolo