From: Auger Eric <eric.auger@redhat.com>
To: Peter Xu <peterx@redhat.com>
Cc: peter.maydell@linaro.org, kevin.tian@intel.com,
tnowicki@marvell.com, jean-philippe@linaro.org,
quintela@redhat.com, qemu-devel@nongnu.org, dgilbert@redhat.com,
bharatb.linux@gmail.com, qemu-arm@nongnu.org, mst@redhat.com,
eric.auger.pro@gmail.com
Subject: Re: [PATCH v12 05/13] virtio-iommu: Endpoint and domains structs and helpers
Date: Wed, 15 Jan 2020 14:00:22 +0100 [thread overview]
Message-ID: <f3da181d-094b-a647-31ea-bdb47eab5c1b@redhat.com> (raw)
In-Reply-To: <20200114180734.GB225163@xz-x1>
Hi Peter,
On 1/14/20 7:07 PM, Peter Xu wrote:
> On Tue, Jan 14, 2020 at 09:51:59AM +0100, Auger Eric wrote:
>> Hi Peter,
>
> Hi, Eric,
>
> [...]
>
>>>
>>>> +{
>>>> + VirtIOIOMMUEndpoint *ep;
>>>> +
>>>> + ep = g_tree_lookup(s->endpoints, GUINT_TO_POINTER(ep_id));
>>>> + if (ep) {
>>>> + return ep;
>>>> + }
>>>> + if (!virtio_iommu_mr(s, ep_id)) {
>>>
>>> Could I ask when this will trigger?
>>
>> This can happen when a device is attached to a domain and its RID does
>> not correspond to one of the devices protected by the iommu.
>
> So will it happen only because of a kernel driver bug?
Yes, at the moment, because virtio_iommu_mr() only gets called on device
attach to a domain.
The spec says:
"If the endpoint identified by endpoint doesn’t exist, the device MUST
reject the request and set status to VIRTIO_IOMMU_S_NOENT"
>
> Also, I think the name of "virtio_iommu_mr" is confusing on that it
> returned explicitly a MemoryRegion however it's never been used:
I use the same prototype as for smmu_iommu_mr(). Returning the iommu mr
will allow to proceed with further RID based operations like invalidations.
The same logic is used in vtd_context_device_invalidate.
>
> (since they're not in the same patch I'm pasting)
>
> static IOMMUMemoryRegion *virtio_iommu_mr(VirtIOIOMMU *s, uint32_t sid)
> {
> uint8_t bus_n, devfn;
> IOMMUPciBus *iommu_pci_bus;
> IOMMUDevice *dev;
>
> bus_n = PCI_BUS_NUM(sid);
> iommu_pci_bus = iommu_find_iommu_pcibus(s, bus_n);
> if (iommu_pci_bus) {
> devfn = sid & 0xFF;
> dev = iommu_pci_bus->pbdev[devfn];
> if (dev) {
> return &dev->iommu_mr;
> }
> }
> return NULL;
> }
>
> Maybe "return !!dev" would be enough, then make the return a boolean?
> Then we can rename it to virtio_iommu_has_device().
>
> PS. I think we can also drop IOMMU_PCI_DEVFN_MAX (after all you even
> didn't use it here!) and use PCI_DEVFN_MAX, and replace 0xFF.
well intel iommu and smmu use a similar constant (PCI_DEVFN_MAX,
SMMU_PCI_DEVFN_MAX resp.). I use it in virtio_iommu_find_add_as
Thanks
Eric
>
> Thanks,
>
next prev parent reply other threads:[~2020-01-15 13:02 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-09 14:43 [PATCH v12 00/13] VIRTIO-IOMMU device Eric Auger
2020-01-09 14:43 ` [PATCH v12 01/13] migration: Support QLIST migration Eric Auger
2020-01-09 14:43 ` [PATCH v12 02/13] virtio-iommu: Add skeleton Eric Auger
2020-01-13 19:53 ` Peter Xu
2020-01-09 14:43 ` [PATCH v12 03/13] virtio-iommu: Decode the command payload Eric Auger
2020-01-09 14:43 ` [PATCH v12 04/13] virtio-iommu: Add the iommu regions Eric Auger
2020-01-13 19:53 ` Peter Xu
2020-01-14 8:37 ` Auger Eric
2020-01-14 18:49 ` Peter Xu
2020-01-13 20:06 ` Peter Xu
2020-01-14 8:48 ` Auger Eric
2020-01-09 14:43 ` [PATCH v12 05/13] virtio-iommu: Endpoint and domains structs and helpers Eric Auger
2020-01-13 20:23 ` Peter Xu
2020-01-14 8:51 ` Auger Eric
2020-01-14 18:07 ` Peter Xu
2020-01-15 13:00 ` Auger Eric [this message]
2020-01-15 14:59 ` Peter Xu
2020-01-15 16:55 ` Auger Eric
2020-01-15 13:15 ` Auger Eric
2020-01-09 14:43 ` [PATCH v12 06/13] virtio-iommu: Implement attach/detach command Eric Auger
2020-01-09 14:43 ` [PATCH v12 07/13] virtio-iommu: Implement map/unmap Eric Auger
2020-01-14 18:55 ` Peter Xu
2020-01-09 14:43 ` [PATCH v12 08/13] virtio-iommu: Implement translate Eric Auger
2020-01-09 14:43 ` [PATCH v12 09/13] virtio-iommu: Implement fault reporting Eric Auger
2020-01-14 19:04 ` Peter Xu
2020-01-15 13:12 ` Auger Eric
2020-01-15 15:04 ` Peter Xu
2020-01-15 16:36 ` Auger Eric
2020-01-09 14:43 ` [PATCH v12 10/13] virtio-iommu-pci: Add virtio iommu pci support Eric Auger
2020-01-09 14:43 ` [PATCH v12 11/13] hw/arm/virt: Add the virtio-iommu device tree mappings Eric Auger
2020-01-09 14:43 ` [PATCH v12 12/13] virtio-iommu: Support migration Eric Auger
2020-01-14 19:07 ` Peter Xu
2020-02-10 12:33 ` Juan Quintela
2020-02-10 13:09 ` Auger Eric
2020-02-10 13:36 ` Auger Eric
2020-02-10 16:44 ` Juan Quintela
2020-01-09 14:43 ` [PATCH v12 13/13] tests: Add virtio-iommu test Eric Auger
2020-01-09 15:12 ` [PATCH v12 00/13] VIRTIO-IOMMU device no-reply
2020-01-09 16:07 ` Auger Eric
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=f3da181d-094b-a647-31ea-bdb47eab5c1b@redhat.com \
--to=eric.auger@redhat.com \
--cc=bharatb.linux@gmail.com \
--cc=dgilbert@redhat.com \
--cc=eric.auger.pro@gmail.com \
--cc=jean-philippe@linaro.org \
--cc=kevin.tian@intel.com \
--cc=mst@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=peterx@redhat.com \
--cc=qemu-arm@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=quintela@redhat.com \
--cc=tnowicki@marvell.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).