From: "Yang, Shunyong" <shunyong.yang@hxt-semitech.com>
To: "robin.murphy@arm.com" <robin.murphy@arm.com>,
"thunder.leizhen@huawei.com" <thunder.leizhen@huawei.com>,
"will.deacon@arm.com" <will.deacon@arm.com>
Cc: "linux-arm-kernel@lists.infradead.org"
<linux-arm-kernel@lists.infradead.org>,
"joro@8bytes.org" <joro@8bytes.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"iommu@lists.linux-foundation.org"
<iommu@lists.linux-foundation.org>,
"jean-philippe.brucker@arm.com" <jean-philippe.brucker@arm.com>
Subject: Re: [PATCH v3 4/6] iommu/io-pgtable-arm: add support for non-strict mode
Date: Wed, 15 Aug 2018 01:43:37 +0000 [thread overview]
Message-ID: <7a2dedda98aa9e677eb7f85b6b55e34e0128d2d9.camel@hxt-semitech.com> (raw)
In-Reply-To: <feabf4a3-db1e-a22b-daab-fad831b473f9@arm.com>
Hi, Robin,
On Tue, 2018-08-14 at 11:02 +0100, Robin Murphy wrote:
> On 14/08/18 09:35, Will Deacon wrote:
> > On Tue, Aug 14, 2018 at 04:33:41PM +0800, Leizhen (ThunderTown)
> > wrote:
> > > On 2018/8/6 9:32, Yang, Shunyong wrote:
> > > > On 2018/7/26 22:37, Robin Murphy wrote:
> > > > > Because DMA code is not the only caller of iommu_map/unmap.
> > > > > It's
> > > > > perfectly legal in the IOMMU API to partially unmap a
> > > > > previous mapping
> > > > > such that a block entry needs to be split. The DMA API,
> > > > > however, is a
> > > > > lot more constrined, and thus by construction the iommu-dma
> > > > > layer will
> > > > > never generate a block-splitting iommu_unmap() except as a
> > > > > result of
> > > > > illegal DMA API usage, and we obviously do not need to
> > > > > optimise for that
> > > > > (you will get a warning about mismatched unmaps under dma-
> > > > > debug, but
> > > > > it's a bit too expensive to police in the general case).
> > > > >
> > > >
> > > > When I was reading the code around arm_lpae_split_blk_unmap(),
> > > > I was
> > > > curious in which scenario a block will be split. Now with your
> > > > comments
> > > > "Because DMA code is not the only caller of iommu_map/unmap",
> > > > it seems
> > > > depending on the user.
> > > >
> > > > Would you please explain this further? I mean besides DMA,
> > > > which user
> > > > will use iommu_map/umap and how it split a block.
> > >
> > > I also think that arm_lpae_split_blk_unmap() scenario is not
> > > exist, maybe
> > > we should remove it, and give a warning for this wrong usage.
> >
> > Can't it happen with VFIO?
>
> ...or GPU drivers, or anyone else managing their own IOMMU domain
> directly. A sequence like this is perfectly legal:
>
> iommu_map(domain, iova, paddr, SZ_8M, prot);
> ...
> iommu_unmap(domain, iova + SZ_1M * 5, SZ_1M * 3);
>
> where if iova and paddr happen to be suitably aligned, the map will
> lay
> down blocks, and the unmap will then have to split one of them into
> pages to remove half of it. We don't tear our hair out maintaining
> split_blk_unmap() for the fun of it :(
Thank you for the GPU example. But for VFIO, I remember all memory will
be pinned in the early stage of emulator (such as qemu) start. So,
the split will occur at which operation? Maybe virtio balloon inflate?
Thanks.
Shunyong.
>
> Robin.
WARNING: multiple messages have this Message-ID (diff)
From: "Yang, Shunyong" <shunyong.yang-PT9Dzx9SjPiXmMXjJBpWqg@public.gmane.org>
To: "robin.murphy-5wv7dgnIgG8@public.gmane.org"
<robin.murphy-5wv7dgnIgG8@public.gmane.org>,
"thunder.leizhen-hv44wF8Li93QT0dZR+AlfA@public.gmane.org"
<thunder.leizhen-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>,
"will.deacon-5wv7dgnIgG8@public.gmane.org"
<will.deacon-5wv7dgnIgG8@public.gmane.org>
Cc: "jean-philippe.brucker-5wv7dgnIgG8@public.gmane.org"
<jean-philippe.brucker-5wv7dgnIgG8@public.gmane.org>,
"iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org"
<iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org>,
"linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org"
<linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
"linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org"
<linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org>
Subject: Re: [PATCH v3 4/6] iommu/io-pgtable-arm: add support for non-strict mode
Date: Wed, 15 Aug 2018 01:43:37 +0000 [thread overview]
Message-ID: <7a2dedda98aa9e677eb7f85b6b55e34e0128d2d9.camel@hxt-semitech.com> (raw)
In-Reply-To: <feabf4a3-db1e-a22b-daab-fad831b473f9-5wv7dgnIgG8@public.gmane.org>
Hi, Robin,
On Tue, 2018-08-14 at 11:02 +0100, Robin Murphy wrote:
> On 14/08/18 09:35, Will Deacon wrote:
> > On Tue, Aug 14, 2018 at 04:33:41PM +0800, Leizhen (ThunderTown)
> > wrote:
> > > On 2018/8/6 9:32, Yang, Shunyong wrote:
> > > > On 2018/7/26 22:37, Robin Murphy wrote:
> > > > > Because DMA code is not the only caller of iommu_map/unmap.
> > > > > It's
> > > > > perfectly legal in the IOMMU API to partially unmap a
> > > > > previous mapping
> > > > > such that a block entry needs to be split. The DMA API,
> > > > > however, is a
> > > > > lot more constrined, and thus by construction the iommu-dma
> > > > > layer will
> > > > > never generate a block-splitting iommu_unmap() except as a
> > > > > result of
> > > > > illegal DMA API usage, and we obviously do not need to
> > > > > optimise for that
> > > > > (you will get a warning about mismatched unmaps under dma-
> > > > > debug, but
> > > > > it's a bit too expensive to police in the general case).
> > > > >
> > > >
> > > > When I was reading the code around arm_lpae_split_blk_unmap(),
> > > > I was
> > > > curious in which scenario a block will be split. Now with your
> > > > comments
> > > > "Because DMA code is not the only caller of iommu_map/unmap",
> > > > it seems
> > > > depending on the user.
> > > >
> > > > Would you please explain this further? I mean besides DMA,
> > > > which user
> > > > will use iommu_map/umap and how it split a block.
> > >
> > > I also think that arm_lpae_split_blk_unmap() scenario is not
> > > exist, maybe
> > > we should remove it, and give a warning for this wrong usage.
> >
> > Can't it happen with VFIO?
>
> ...or GPU drivers, or anyone else managing their own IOMMU domain
> directly. A sequence like this is perfectly legal:
>
> iommu_map(domain, iova, paddr, SZ_8M, prot);
> ...
> iommu_unmap(domain, iova + SZ_1M * 5, SZ_1M * 3);
>
> where if iova and paddr happen to be suitably aligned, the map will
> lay
> down blocks, and the unmap will then have to split one of them into
> pages to remove half of it. We don't tear our hair out maintaining
> split_blk_unmap() for the fun of it :(
Thank you for the GPU example. But for VFIO, I remember all memory will
be pinned in the early stage of emulator (such as qemu) start. So,
the split will occur at which operation? Maybe virtio balloon inflate?
Thanks.
Shunyong.
>
> Robin.
WARNING: multiple messages have this Message-ID (diff)
From: shunyong.yang@hxt-semitech.com (Yang, Shunyong)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH v3 4/6] iommu/io-pgtable-arm: add support for non-strict mode
Date: Wed, 15 Aug 2018 01:43:37 +0000 [thread overview]
Message-ID: <7a2dedda98aa9e677eb7f85b6b55e34e0128d2d9.camel@hxt-semitech.com> (raw)
In-Reply-To: <feabf4a3-db1e-a22b-daab-fad831b473f9@arm.com>
Hi, Robin,
On Tue, 2018-08-14 at 11:02 +0100, Robin Murphy wrote:
> On 14/08/18 09:35, Will Deacon wrote:
> > On Tue, Aug 14, 2018 at 04:33:41PM +0800, Leizhen (ThunderTown)
> > wrote:
> > > On 2018/8/6 9:32, Yang, Shunyong wrote:
> > > > On 2018/7/26 22:37, Robin Murphy wrote:
> > > > > Because DMA code is not the only caller of iommu_map/unmap.
> > > > > It's
> > > > > perfectly legal in the IOMMU API to partially unmap a
> > > > > previous mapping
> > > > > such that a block entry needs to be split. The DMA API,
> > > > > however, is a
> > > > > lot more constrined, and thus by construction the iommu-dma
> > > > > layer will
> > > > > never generate a block-splitting iommu_unmap() except as a
> > > > > result of
> > > > > illegal DMA API usage, and we obviously do not need to
> > > > > optimise for that
> > > > > (you will get a warning about mismatched unmaps under dma-
> > > > > debug, but
> > > > > it's a bit too expensive to police in the general case).
> > > > >
> > > >
> > > > When I was reading the code around arm_lpae_split_blk_unmap(),
> > > > I was
> > > > curious in which scenario a block will be split. Now with your
> > > > comments
> > > > "Because DMA code is not the only caller of iommu_map/unmap",
> > > > it seems
> > > > depending on the user.
> > > >
> > > > Would you please explain this further? I mean besides DMA,
> > > > which user
> > > > will use iommu_map/umap and how it split a block.
> > >
> > > I also think that arm_lpae_split_blk_unmap() scenario is not
> > > exist, maybe
> > > we should remove it, and give a warning for this wrong usage.
> >
> > Can't it happen with VFIO?
>
> ...or GPU drivers, or anyone else managing their own IOMMU domain
> directly. A sequence like this is perfectly legal:
>
> iommu_map(domain, iova, paddr, SZ_8M, prot);
> ...
> iommu_unmap(domain, iova + SZ_1M * 5, SZ_1M * 3);
>
> where if iova and paddr happen to be suitably aligned, the map will
> lay
> down blocks, and the unmap will then have to split one of them into
> pages to remove half of it. We don't tear our hair out maintaining
> split_blk_unmap() for the fun of it :(
Thank you for the GPU example. But for VFIO, I remember all memory will
be pinned in the early stage of emulator (such as qemu) start. So,
the split will occur at which operation? Maybe virtio balloon inflate?
Thanks.
Shunyong.
>
> Robin.
next prev parent reply other threads:[~2018-08-15 1:43 UTC|newest]
Thread overview: 66+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-07-12 6:18 [PATCH v3 0/6] add non-strict mode support for arm-smmu-v3 Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-12 6:18 ` [PATCH v3 1/6] iommu/arm-smmu-v3: fix the implementation of flush_iotlb_all hook Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-12 6:18 ` [PATCH v3 2/6] iommu/dma: add support for non-strict mode Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-24 22:01 ` Robin Murphy
2018-07-24 22:01 ` Robin Murphy
2018-07-24 22:01 ` Robin Murphy
2018-07-26 4:15 ` Leizhen (ThunderTown)
2018-07-26 4:15 ` Leizhen (ThunderTown)
2018-07-26 4:15 ` Leizhen (ThunderTown)
2018-07-12 6:18 ` [PATCH v3 3/6] iommu/amd: use default branch to deal with all non-supported capabilities Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-12 6:18 ` [PATCH v3 4/6] iommu/io-pgtable-arm: add support for non-strict mode Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-24 22:25 ` Robin Murphy
2018-07-24 22:25 ` Robin Murphy
2018-07-26 7:20 ` Leizhen (ThunderTown)
2018-07-26 7:20 ` Leizhen (ThunderTown)
2018-07-26 14:35 ` Robin Murphy
2018-07-26 14:35 ` Robin Murphy
2018-08-06 1:32 ` Yang, Shunyong
2018-08-06 1:32 ` Yang, Shunyong
2018-08-06 1:32 ` Yang, Shunyong
2018-08-14 8:33 ` Leizhen (ThunderTown)
2018-08-14 8:33 ` Leizhen (ThunderTown)
2018-08-14 8:33 ` Leizhen (ThunderTown)
2018-08-14 8:35 ` Will Deacon
2018-08-14 8:35 ` Will Deacon
2018-08-14 8:35 ` Will Deacon
2018-08-14 10:02 ` Robin Murphy
2018-08-14 10:02 ` Robin Murphy
2018-08-15 1:43 ` Yang, Shunyong [this message]
2018-08-15 1:43 ` Yang, Shunyong
2018-08-15 1:43 ` Yang, Shunyong
2018-08-15 7:33 ` Will Deacon
2018-08-15 7:33 ` Will Deacon
2018-08-15 7:33 ` Will Deacon
2018-08-15 7:35 ` Will Deacon
2018-08-15 7:35 ` Will Deacon
2018-08-15 7:35 ` Will Deacon
2018-08-16 0:43 ` Yang, Shunyong
2018-08-16 0:43 ` Yang, Shunyong
2018-08-16 0:43 ` Yang, Shunyong
2018-07-12 6:18 ` [PATCH v3 5/6] iommu/arm-smmu-v3: " Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-12 6:18 ` [PATCH v3 6/6] iommu/arm-smmu-v3: add bootup option "iommu_strict_mode" Zhen Lei
2018-07-12 6:18 ` Zhen Lei
2018-07-24 22:46 ` Robin Murphy
2018-07-24 22:46 ` Robin Murphy
2018-07-26 7:41 ` Leizhen (ThunderTown)
2018-07-26 7:41 ` Leizhen (ThunderTown)
2018-07-24 21:51 ` [PATCH v3 0/6] add non-strict mode support for arm-smmu-v3 Robin Murphy
2018-07-24 21:51 ` Robin Murphy
2018-07-26 3:44 ` Leizhen (ThunderTown)
2018-07-26 3:44 ` Leizhen (ThunderTown)
2018-07-26 3:44 ` Leizhen (ThunderTown)
2018-07-26 14:16 ` Robin Murphy
2018-07-26 14:16 ` Robin Murphy
2018-07-27 2:49 ` Leizhen (ThunderTown)
2018-07-27 2:49 ` Leizhen (ThunderTown)
2018-07-27 9:37 ` Will Deacon
2018-07-27 9:37 ` Will Deacon
2018-07-27 9:37 ` Will Deacon
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=7a2dedda98aa9e677eb7f85b6b55e34e0128d2d9.camel@hxt-semitech.com \
--to=shunyong.yang@hxt-semitech.com \
--cc=iommu@lists.linux-foundation.org \
--cc=jean-philippe.brucker@arm.com \
--cc=joro@8bytes.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=robin.murphy@arm.com \
--cc=thunder.leizhen@huawei.com \
--cc=will.deacon@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.