virtio-pmem: Support PCI BAR-relative addresses
cover virtio-pmem: Support PCI BAR-relative addresses
From: Taylor Stark <[email protected]>
This patch updates the virtio-pmem spec to add support for describing the pmem region via PCI BARs. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
As virtio-pmem hasn't been added to the virtio spec yet (see this issue [1]), this patch is based off the RFC spec [2]. The linux driver implementation has been posted at [3].
[1] https://github.com/oasis-tcs/virtio-spec/issues/78
[2] https://lists.oasis-open.org/archives/virtio-dev/201903/msg00083.html
[3] https://lore.kernel.org/nvdimm/20210715223505.GA29329@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net
<Apologies for the resend - forgot to subscribe to virtio-comment so my original mail bounced>
Taylor Stark (1):
virtio_pmem: Support PCI BAR-relative addresses
virtio-pmem.tex | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-)
Signature · 18 lines
2.31.0.vfs.0.1 This publicly archived list offers a means to provide input to the OASIS Virtual I/O Device (VIRTIO) TC. In order to verify user consent to the Feedback License terms and to minimize spam in the list archive, subscription is required before posting. Subscribe: [email protected] Unsubscribe: [email protected] List help: [email protected] List archive: https://lists.oasis-open.org/archives/virtio-comment/ Feedback License: https://www.oasis-open.org/who/ipr/feedback_license.pdf List Guidelines: https://www.oasis-open.org/policies-guidelines/mailing-lists Committee: https://www.oasis-open.org/committees/virtio/ Join OASIS: https://www.oasis-open.org/join/
PATCH 1/1 virtio-pmem: Support PCI BAR-relative addresses
From: Taylor Stark <[email protected]>
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
--- virtio-pmem.tex | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-)
virtio-pmem.tex +17 -5
@@ -40,11 +40,21 @@ \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device Device hotplugs physical memory to guest address space. Persistent memory device is emulated with file backed memory at host side. +The device MUST indicate the guest physical address to the driver in one of two +ways: \begin{enumerate} -\item Guest vpmem start is read from \field{start}. -\item Guest vpmem end is read from \field{size}. +\item As a guest absolute address. +\item As a shared memory region. \end{enumerate} +If the guest physical address is indicated as an absolute address, the device +MUST set \field{start} to the absolute address and \field{size} to the size of +the address range, in bytes. + +If the guest physical address is indicated as a shared memory region, the shared +memory region MUST be shared memory region ID 0. The device SHOULD set +\field{start} and \field{size} to zero. + \devicenormative{\subsubsection}{Device Initialization}{Device Types / PMEM Device / Device Initialization} File backed memory SHOULD be memory mapped to guest address space with SHARED
@@ -52,9 +62,11 @@ \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device \subsection{Driver Initialization}\label{sec:Device Types / PMEM Driver / Driver Initialization} -Driver hotplugs the physical memory and registers associated -region with the pmem API. Also, configures a flush callback -function with the corresponding region. +The driver SHOULD query the physical address ranges where the pmem was mapped. +When performing the query, the driver MUST first query if shared memory region +ID 0 is supported by the device. If present, the driver MUST NOT use \field{start} +or \field{size}. If not present, the driver SHOULD fallback to reading the +physical address ranges from \field{start} and \field{size}. \drivernormative{\subsubsection}{Driver Initialization: Filesystem direct access}{Device Types / PMEM Driver / Driver Initialization / Direct access}
Signature · 18 lines
2.31.0.vfs.0.1 This publicly archived list offers a means to provide input to the OASIS Virtual I/O Device (VIRTIO) TC. In order to verify user consent to the Feedback License terms and to minimize spam in the list archive, subscription is required before posting. Subscribe: [email protected] Unsubscribe: [email protected] List help: [email protected] List archive: https://lists.oasis-open.org/archives/virtio-comment/ Feedback License: https://www.oasis-open.org/who/ipr/feedback_license.pdf List Guidelines: https://www.oasis-open.org/policies-guidelines/mailing-lists Committee: https://www.oasis-open.org/committees/virtio/ Join OASIS: https://www.oasis-open.org/join/
Quoted 7 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
Given that we already have pmem support out there (even though the spec has not been included yet, should this get a feature?
Quoted 11 lines
Signed-off-by Taylor Stark <[email protected]>--- virtio-pmem.tex | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-)virtio-pmem.tex +0
-0
virtio-pmem.tex \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device · line 44
Quoted 5 lines
virtio-pmem.tex +2
-0@@ -40,11 +40,21 @@ \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device Device hotplugs physical memory to guest address space. Persistent memory device is emulated with file backed memory at host side. +The device MUST indicate the guest physical address to the driver in one of two +ways:
This is not a normative section; better use "The device indicates...."?
Quoted 4 lines
virtio-pmem.tex +1
-2\begin{enumerate} -\item Guest vpmem start is read from \field{start}. -\item Guest vpmem end is read from \field{size}. +\item As a guest absolute address.
virtio-pmem.tex \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device · line 56
Quoted 10 lines
virtio-pmem.tex +8
-0+\item As a shared memory region. \end{enumerate} +If the guest physical address is indicated as an absolute address, the device +MUST set \field{start} to the absolute address and \field{size} to the size of +the address range, in bytes. + +If the guest physical address is indicated as a shared memory region, the shared +memory region MUST be shared memory region ID 0. The device SHOULD set +\field{start} and \field{size} to zero.
And these two paragraphs should go into a normative section.
Quoted 6 lines
virtio-pmem.tex +1
-0+ \devicenormative{\subsubsection}{Device Initialization}{Device Types / PMEM Device / Device Initialization} File backed memory SHOULD be memory mapped to guest address space with SHARED@@ -52,9 +62,11 @@ \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device
virtio-pmem.tex \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device · line 69
Quoted 10 lines
virtio-pmem.tex +5
-3\subsection{Driver Initialization}\label{sec:Device Types / PMEM Driver / Driver Initialization} -Driver hotplugs the physical memory and registers associated -region with the pmem API. Also, configures a flush callback -function with the corresponding region. +The driver SHOULD query the physical address ranges where the pmem was mapped. +When performing the query, the driver MUST first query if shared memory region +ID 0 is supported by the device. If present, the driver MUST NOT use \field{start} +or \field{size}. If not present, the driver SHOULD fallback to reading the +physical address ranges from \field{start} and \field{size}.
Reading this, I think we really need to have a feature to distinguish between the two modes; otherwise, old drivers will be confused.
Also, requirements belong in a normative section; it's better to just describe textually what the driver needs to do here and split out the normative statements.
virtio-pmem.tex \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device · line 71
Quoted 2 lines
virtio-pmem.tex +0
-0\drivernormative{\subsubsection}{Driver Initialization: Filesystem direct access}{Device Types / PMEM Driver / Driver Initialization / Direct access}
This publicly archived list offers a means to provide input to the OASIS Virtual I/O Device (VIRTIO) TC.
In order to verify user consent to the Feedback License terms and to minimize spam in the list archive, subscription is required before posting.
Subscribe: [email protected]
Unsubscribe: [email protected]
List help: [email protected]
List archive: https://lists.oasis-open.org/archives/virtio-comment/
Feedback License: https://www.oasis-open.org/who/ipr/feedback_license.pdf
List Guidelines: https://www.oasis-open.org/policies-guidelines/mailing-lists
Committee: https://www.oasis-open.org/committees/virtio/
Join OASIS: https://www.oasis-open.org/join/
Quoted 4 lines · Cornelia Huck
Quoted 2 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
Quoted 8 lines
Quoted 5 lines
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
Given that we already have pmem support out there (even though the spec has not been included yet, should this get a feature?
I wasn't sure if we needed to handle backwards compatibility given that pmem hasn't been merged into the spec yet. If we do, then yes I think it makes sense to add a feature bit. How about VIRTIO_PMEM_F_SHMEM_REGION?
Quoted 56 lines
Quoted 11 lines
Signed-off-by Taylor Stark <[email protected]>--- virtio-pmem.tex | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-)virtio-pmem.tex +0
-0virtio-pmem.tex \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device · line 44
Quoted 5 lines
virtio-pmem.tex +2
-0@@ -40,11 +40,21 @@ \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device Device hotplugs physical memory to guest address space. Persistent memory device is emulated with file backed memory at host side. +The device MUST indicate the guest physical address to the driver in one of two +ways:This is not a normative section; better use "The device indicates...."?
Quoted 4 lines
virtio-pmem.tex +1
-2\begin{enumerate} -\item Guest vpmem start is read from \field{start}. -\item Guest vpmem end is read from \field{size}. +\item As a guest absolute address.virtio-pmem.tex \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device · line 56
Quoted 10 lines
virtio-pmem.tex +8
-0+\item As a shared memory region. \end{enumerate} +If the guest physical address is indicated as an absolute address, the device +MUST set \field{start} to the absolute address and \field{size} to the size of +the address range, in bytes. + +If the guest physical address is indicated as a shared memory region, the shared +memory region MUST be shared memory region ID 0. The device SHOULD set +\field{start} and \field{size} to zero.And these two paragraphs should go into a normative section.
Quoted 6 lines
virtio-pmem.tex +1
-0+ \devicenormative{\subsubsection}{Device Initialization}{Device Types / PMEM Device / Device Initialization} File backed memory SHOULD be memory mapped to guest address space with SHARED@@ -52,9 +62,11 @@ \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Devicevirtio-pmem.tex \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / Device · line 69
Quoted 10 lines
virtio-pmem.tex +5
-3\subsection{Driver Initialization}\label{sec:Device Types / PMEM Driver / Driver Initialization} -Driver hotplugs the physical memory and registers associated -region with the pmem API. Also, configures a flush callback -function with the corresponding region. +The driver SHOULD query the physical address ranges where the pmem was mapped. +When performing the query, the driver MUST first query if shared memory region +ID 0 is supported by the device. If present, the driver MUST NOT use \field{start} +or \field{size}. If not present, the driver SHOULD fallback to reading the +physical address ranges from \field{start} and \field{size}.Reading this, I think we really need to have a feature to distinguish between the two modes; otherwise, old drivers will be confused.
Quoted 3 lines
Also, requirements belong in a normative section; it's better to just describe textually what the driver needs to do here and split out the normative statements.
Will do (ditto for your other comments). Thanks for the suggestions. It's a learning process for me - had to go and lookup what normative meant.. :)
Quoted 2 lines
Quoted 2 lines
virtio-pmem.tex +0
-0\drivernormative{\subsubsection}{Driver Initialization: Filesystem direct access}{Device Types / PMEM Driver / Driver Initialization / Direct access}
Quoted 11 lines · Taylor Stark
Quoted 10 lines · Cornelia Huck
Quoted 7 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
Quoted 6 lines
Quoted 2 lines
Given that we already have pmem support out there (even though the spec has not been included yet, should this get a feature?
I wasn't sure if we needed to handle backwards compatibility given that pmem hasn't been merged into the spec yet. If we do, then yes I think it makes sense to add a feature bit. How about
VIRTIO_PMEM_F_SHMEM_REGION?
Sounds good to me.
Quoted 6 lines
Quoted 3 lines
Also, requirements belong in a normative section; it's better to just describe textually what the driver needs to do here and split out the normative statements.
Will do (ditto for your other comments). Thanks for the suggestions. It's a learning process for me - had to go and lookup what normative meant.. :)
If you have any further questions (or any suggestions how we can make this easier for newcomers), feel free to follow up :)
This publicly archived list offers a means to provide input to the OASIS Virtual I/O Device (VIRTIO) TC.
In order to verify user consent to the Feedback License terms and to minimize spam in the list archive, subscription is required before posting.
Subscribe: [email protected]
Unsubscribe: [email protected]
List help: [email protected]
List archive: https://lists.oasis-open.org/archives/virtio-comment/
Feedback License: https://www.oasis-open.org/who/ipr/feedback_license.pdf
List Guidelines: https://www.oasis-open.org/policies-guidelines/mailing-lists
Committee: https://www.oasis-open.org/committees/virtio/
Join OASIS: https://www.oasis-open.org/join/
Quoted 11 lines · Taylor Stark
Quoted 10 lines · Cornelia Huck
Quoted 7 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
Quoted 6 lines
Quoted 2 lines
Given that we already have pmem support out there (even though the spec has not been included yet, should this get a feature?
I wasn't sure if we needed to handle backwards compatibility given that pmem hasn't been merged into the spec yet. If we do, then yes I think it makes sense to add a feature bit. How about
VIRTIO_PMEM_F_SHMEM_REGION?
The implementations that are upstream win, and usually we get a spec after the implementation is mature. The spec merely documents what the existing implementations are doing. Of course, one can plan ahead and propose additions to the spec that won't break existing setups.
The feature seems to be glued to PCI BARs, right? Maybe that should be part of the feature name.
Signature · 3 lines
Thanks, David / dhildenb
Quoted 19 lines · David Hildenbrand
Quoted 11 lines · Taylor Stark
Quoted 10 lines · Cornelia Huck
Quoted 7 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
Quoted 6 lines
Quoted 2 lines
Given that we already have pmem support out there (even though the spec has not been included yet, should this get a feature?
I wasn't sure if we needed to handle backwards compatibility given that pmem hasn't been merged into the spec yet. If we do, then yes I think it makes sense to add a feature bit. How about
VIRTIO_PMEM_F_SHMEM_REGION?
Quoted 13 lines
The implementations that are upstream win, and usually we get a spec after the implementation is mature. The spec merely documents what the existing implementations are doing. Of course, one can plan ahead and propose additions to the spec that won't break existing setups.
The feature seems to be glued to PCI BARs, right? Maybe that should be part of the feature name.
Signature · 3 lines
Thanks, David / dhildenb
I had to do a bit of research to develop an opinion on including BAR in the name. Using BARs is the implementation for the PCI transport, but there are other transports that have support for shared memory regions but don't use BARs (e.g. mmio - see vm_get_shm_region). Using shared memory regions was initially a convenient way to use PCI BARs, but I think it could be valuable for other transports as well (it adds similarity with other virtio devices). So I think it makes sense to leave BAR out of the name, since that's transport specific. Maybe I should retitle the patch? Let me know your thoughts (especially since I'm learning on the fly).
Thanks,
Taylor
Quoted 34 lines · Taylor Stark
Quoted 19 lines · David Hildenbrand
Quoted 11 lines · Taylor Stark
Quoted 10 lines · Cornelia Huck
Quoted 7 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
Quoted 6 lines
Quoted 2 lines
Given that we already have pmem support out there (even though the spec has not been included yet, should this get a feature?
I wasn't sure if we needed to handle backwards compatibility given that pmem hasn't been merged into the spec yet. If we do, then yes I think it makes sense to add a feature bit. How about
VIRTIO_PMEM_F_SHMEM_REGION?Quoted 13 lines
The implementations that are upstream win, and usually we get a spec after the implementation is mature. The spec merely documents what the existing implementations are doing. Of course, one can plan ahead and propose additions to the spec that won't break existing setups.
The feature seems to be glued to PCI BARs, right? Maybe that should be part of the feature name.
Signature · 3 lines
Thanks, David / dhildenb
Quoted 9 lines
I had to do a bit of research to develop an opinion on including BAR in the name. Using BARs is the implementation for the PCI transport, but there are other transports that have support for shared memory regions but don't use BARs (e.g. mmio - see
vm_get_shm_region). Using shared memory regions was initially a convenient way to use PCI BARs, but I think it could be valuable for other transports as well (it adds similarity with other virtio devices). So I think it makes sense to leave BAR out of the name, since that's transport specific. Maybe I should retitle the patch? Let me know your thoughts (especially since I'm learning on the fly).
Ah, the content doesn't actually talk about PCI BARs, that's my source of confusion :)
I know that virtio-fs also uses BARs in DAX mode, but I have no idea how they named features (if any?) or how they described it in their spec. Might be worth exploring.
Signature · 3 lines
Thanks, David / dhildenb
Quoted 45 lines · David Hildenbrand
Quoted 34 lines · Taylor Stark
Quoted 19 lines · David Hildenbrand
Quoted 11 lines · Taylor Stark
Quoted 10 lines · Cornelia Huck
Quoted 7 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
Update the virtio-pmem RFC spec to add support for describing the pmem region via PCI BARs. Shared memory windows are used to accomplish this, similar to virtio-fs and virtio-gpu. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
Quoted 6 lines
Quoted 2 lines
Given that we already have pmem support out there (even though the spec has not been included yet, should this get a feature?
I wasn't sure if we needed to handle backwards compatibility given that pmem hasn't been merged into the spec yet. If we do, then yes I think it makes sense to add a feature bit. How about
VIRTIO_PMEM_F_SHMEM_REGION?Quoted 13 lines
The implementations that are upstream win, and usually we get a spec after the implementation is mature. The spec merely documents what the existing implementations are doing. Of course, one can plan ahead and propose additions to the spec that won't break existing setups.
The feature seems to be glued to PCI BARs, right? Maybe that should be part of the feature name.
Signature · 3 lines
Thanks, David / dhildenbQuoted 9 lines
I had to do a bit of research to develop an opinion on including BAR in the name. Using BARs is the implementation for the PCI transport, but there are other transports that have support for shared memory regions but don't use BARs (e.g. mmio - see
vm_get_shm_region). Using shared memory regions was initially a convenient way to use PCI BARs, but I think it could be valuable for other transports as well (it adds similarity with other virtio devices). So I think it makes sense to leave BAR out of the name, since that's transport specific. Maybe I should retitle the patch? Let me know your thoughts (especially since I'm learning on the fly).
Quoted 12 lines
Ah, the content doesn't actually talk about PCI BARs, that's my source of confusion :)
I know that virtio-fs also uses BARs in DAX mode, but I have no idea how they named features (if any?) or how they described it in their spec. Might be worth exploring.
Signature · 3 lines
Thanks, David / dhildenb
After doing the research from my previous mail, I agree including BARs in the name is confusing. I think I'll update the patch to just refer to shared memory regions.
virtio-fs is actually how we landed on adding shared memory region support to virtio-pmem. We got virtio-fs working in Hyper-V, then tried virtio-pmem, ran into the guest absolute address problem, then went "well virtio-fs works, what does that do?", and here we are :)
virtio-fs doesn't have any relevant features. The presence of the shared memory region indicates to the driver that additional functionality is available, if desired. I initially did the same thing with this update to virtio-pmem, instead of using a feature. But a feature makes more sense for preserving backwards compatibility. And it's actually useful for Hyper-V as well, since we need to reject any drivers that don't support the feature since we have no way to configure them.
Thanks,
Taylor
Quoted 9 lines · On Wed, Jul 21 2021, [email protected] wrote:
From: Taylor Stark <[email protected]>
This patch updates the virtio-pmem spec to add support for describing the pmem region via PCI BARs. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
As virtio-pmem hasn't been added to the virtio spec yet (see this issue [1]), this patch is based off the RFC spec [2]. The linux driver implementation has been posted at [3].
What's the general status of the pmem spec? I had pinged in the issue, but I have not seen any progress.
Quoted 5 lines
[1] https://github.com/oasis-tcs/virtio-spec/issues/78
[2] https://lists.oasis-open.org/archives/virtio-dev/201903/msg00083.html
[3] https://lore.kernel.org/nvdimm/20210715223505.GA29329@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net
Quoted 7 lines
<Apologies for the resend - forgot to subscribe to virtio-comment so my original mail bounced>
Taylor Stark (1):
virtio_pmem: Support PCI BAR-relative addressesvirtio-pmem.tex | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-)
This publicly archived list offers a means to provide input to the OASIS Virtual I/O Device (VIRTIO) TC.
In order to verify user consent to the Feedback License terms and to minimize spam in the list archive, subscription is required before posting.
Subscribe: [email protected]
Unsubscribe: [email protected]
List help: [email protected]
List archive: https://lists.oasis-open.org/archives/virtio-comment/
Feedback License: https://www.oasis-open.org/who/ipr/feedback_license.pdf
List Guidelines: https://www.oasis-open.org/policies-guidelines/mailing-lists
Committee: https://www.oasis-open.org/committees/virtio/
Join OASIS: https://www.oasis-open.org/join/
Hi Cornelia,
Quoted 10 lines
Quoted 7 lines
This patch updates the virtio-pmem spec to add support for describing the pmem region via PCI BARs. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
As virtio-pmem hasn't been added to the virtio spec yet (see this issue [1]), this patch is based off the RFC spec [2]. The linux driver implementation has been posted at [3].
What's the general status of the pmem spec? I had pinged in the issue, but I have not seen any progress.
Appologies for missing your ping (As I was down for some time). I will try to post the
virtio-pmem spec new version in coming week.
Thanks,
Pankaj
Quoted 10 lines · Pankaj Gupta
Hi Cornelia,
Quoted 8 lines
Quoted 7 lines
This patch updates the virtio-pmem spec to add support for describing the pmem region via PCI BARs. This is required to support virtio-pmem in Hyper-V, since Hyper-V only allows PCI devices to operate on memory ranges defined via BARs.
As virtio-pmem hasn't been added to the virtio spec yet (see this issue [1]), this patch is based off the RFC spec [2]. The linux driver implementation has been posted at [3].
Quoted 6 lines
Quoted 2 lines
What's the general status of the pmem spec? I had pinged in the issue, but I have not seen any progress.
Appologies for missing your ping (As I was down for some time). I will try to post the
virtio-pmem spec new version in coming week.
Thanks, that's good news!
This publicly archived list offers a means to provide input to the OASIS Virtual I/O Device (VIRTIO) TC.
In order to verify user consent to the Feedback License terms and to minimize spam in the list archive, subscription is required before posting.
Subscribe: [email protected]
Unsubscribe: [email protected]
List help: [email protected]
List archive: https://lists.oasis-open.org/archives/virtio-comment/
Feedback License: https://www.oasis-open.org/who/ipr/feedback_license.pdf
List Guidelines: https://www.oasis-open.org/policies-guidelines/mailing-lists
Committee: https://www.oasis-open.org/committees/virtio/
Join OASIS: https://www.oasis-open.org/join/