admin: Access legacy registers using admin commands

PATCH Superseded Parav Pandit 3 participants 39 messages Jul 6, 2023 – Jul 7, 2023

Share & tools

Parav Pandit

author

2023-07-06 04:20 UTC #0 lore raw

v10 cover admin: Access legacy registers using admin commands

This short series introduces legacy registers access commands for the owner group member access the legacy registers of the member VFs.
This short series introduces legacy region access commands by the group owner device for its member devices.
Currently it is applicable to the PCI PF and VF devices. If in future any SIOV devices to support legacy registers, they can be easily supported using same commands by using the group member identifiers of the future SIOV devices.

More details as overview, motivation, use case are further described below.

Patch summary:
--------------
patch-1 split rows of admin opcode tables by a line patch-2 fix section numbering
patch-3 add generic legacy region access commands patch-4 add pci specific definition

It uses the newly introduced administration command facility with 4 new commands and a new optional command to query the legacy notification region.

Usecase:
--------
1. A hypervisor/system needs to provide transitional virtio devices to the guest VM at scale of thousands, typically, one to eight devices per VM.

2. A hypervisor/system needs to provide such devices using a vendor agnostic driver in the hypervisor system.

3. A hypervisor system prefers to have single stack regardless of virtio device type (net/blk) and be future compatible with a single vfio stack using SR-IOV or other scalable device virtualization technology to map PCI devices to the guest VM. (as transitional or otherwise)

Motivation/Background:
----------------------
The existing virtio transitional PCI device is missing support for PCI SR-IOV based devices. Currently it does not work beyond PCI PF, or as software emulated device in reality. Currently it has below cited system level limitations:

[a] PCIe spec citation:
VFs do not support I/O Space and thus VF BARs shall not indicate I/O Space.

[b] cpu arch citiation:
Intel 64 and IA-32 Architectures Software Developer’s Manual:
The processor’s I/O address space is separate and distinct from the physical-memory address space. The I/O address space consists of 64K individually addressable 8-bit I/O ports, numbered 0 through FFFFH.

[c] PCIe spec citation:
If a bridge implements an I/O address range,...I/O address range will be aligned to a 4 KB boundary.

Overview:
---------
Above usecase requirements is solved by PCI PF group owner accessing its group member PCI VFs legacy registers using an admin virtqueue of the group owner PCI PF.

Two new admin virtqueue commands are added which read/write PCI VF registers.

Software usage example:
-----------------------
One way to use and map to the guest VM is by using vfio driver framework in Linux kernel.

                +----------------------+
                |pci_dev_id = 0x100X   |
+---------------|pci_rev_id = 0x0      |-----+
|vfio device    |BAR0 = I/O region     |     |
|               |Other attributes      |     |
|               +----------------------+     |
|                                            |
+   +--------------+     +-----------------+ |
|   |I/O BAR to AQ |     | Other vfio      | |
|   |rd/wr mapper  |     | functionalities | |
|   +--------------+     +-----------------+ |
|                                            |
+------+-------------------------+-----------+
       |                         |
   Legacy region            Driver notification
    access                       |
       |                         |
  +----+------------+       +----+------------+
  | +-----+         |       | PCI VF device A |
  | | AQ  |-------------+---->+-------------+ |
  | +-----+         |   |   | | legacy regs | |
  | PCI PF device   |   |   | +-------------+ |
  +-----------------+   |   +-----------------+
                        |
                        |   +----+------------+
                        |   | PCI VF device N |
                        +---->+-------------+ |
                            | | legacy regs | |
                            | +-------------+ |
                            +-----------------+

2. Virtio pci driver to bind to the listed device id and use it in the host.

3. Use it in a light weight hypervisor to run bare-metal OS.

Please review.

Alternatives considered:
========================
1. Exposing BAR0 as MMIO BAR that follows legacy registers template Pros:
a. Kind of works with legacy drivers as some of them have used API
which is agnostic to MMIO vs IOBAR.
b. Does not require hypervisor intervantion
Cons:
a. Device reset is extremely hard to implement in device at scale as
driver does not wait for device reset completion
b. Device register width related problems persist that hypervisor if
wishes, it cannot be fixed.

2. Accessing VF registers by tunneling it through new legacy PCI capability Pros:
a. Self contained, but cannot work with future PCI SIOV devices Cons:
a. Equally slow as AQ access
b. Still requires new capability for notification access
c. Requires hardware to build low level registers access which is not worth
for long term future

3. Accessing VF notification region using new PF BAR
Cons:
a. Requires hardware to build new PCI steering logic per PF to forward
notification from the PF to VF, requires double the amount of logic compared to today
b. Requires very large additional PF BAR whose size must be max_Vfs * BAR size.

4. Trapping CVQ, configuration region, LEGACY_HDR
Cons:
a. This does not fullfil the very basic requirement to not trap the
1.x objects (configuration registers, vqs)
b. Requires feature negotiations mediation in hypervisor software c. Requires constant device type specific knowledge in hypervisor driver
(Does not scale for 30+ device types)

4. F_LEACY_HDR, F_WRITE_MAC
Cons:
a. Requires device support to have read/write mac address which is
hard to implement on every member device.
b. such functionality is duplicate of existing cvq per device. c. config space is only for the initialization specific purpose. d. Requires mediation of 1.x objects, which is not good design. e. Solves only for the net device.
Pros:
a. May work for nested env

conclusion for picking AQ approach:
==================================
1. Overall AQ based access is simpler to implement with combination of best from software and device so that legacy registers do not get baked in the device hardware
2. AQ allows hypervisor software to intercept legacy registers and make corrections if needed
3. Provides trade-off between performance, device complexity vs spec, while still maintaining passthrough mode for the VFs with minimal hypervisor intercepts only for legacy registers access
4. AQ mechanism is designed for accessing other member devices registers as noted in AQ submission, it utilizes the existing infrastructure over other alternatives.
5. Uses existing driver notification region similar to legacy notification saves hardware resources

Fixes https://github.com/oasis-tcs/virtio-spec/issues/167

Signed-off-by Parav Pandit <[email protected]>

---
changelog:
v9->v10:
- added white space at end of line
- addressed below comments from Cornelia
- fixed errors related to article
- hardwire to hardwires
- replaced various to all
- added hardwire to zero
- fixed requirements for administration virtqueue section
- added missing articles
- reworded description for notification query command
- grammar fixes
- addressed below comments from Michael
- added description for member group id setting
- reworded device and driver conformance statements
- opcode table description updated
- fixed label for device read command
- length alignment restriction text added
- data length described for read write commands
- notification description added and refined
- reworded text around command specific result and data field usage v8->v9:
- add missing articles in notify query command
- replaced 'this notification' with 'such a notification'
- addressed below comments from Michael
- dropped 'Region' from the commands
- added 7 reserved pad bytes in config write commands
- rewrote from 'use following structure' to 'field' has the following struct..
- dropped mentioning to follow struct virtio_admin_cmd.
- added note about command limited to only sriov group type for now
- rewrote the description little differently
v7->v8:
- remove empty line at the end of file
- removed white space at the end
- addressed comments from Michael add link to pci
- renamed region to region_data
- made region_data width to be 16 bytes to cover for 8 bytes offset
- moved generic notification region related normative from pci to generic section
- addressed comments from Michael
- made bar offset 64-bit
- prefix legacy specific structure with _legacy
- moved generic normative from pci to generic section
- added link to virtio pci capabilities when referring to bar 0
- remove 'should' from generic description
v6->v7:
- addressed several comments from Michael
- use AQ command to query legacy notify region, dropped pci capability modifications
- moved most part of the text to the generic admin command section
- replace administrative to administration
- replace admin vq citation to admin commands
- added normatives for device and driver side
- made BAR0 to be not used at all when supporting legacy interface
- added normative around BAR0 and SR-IOV extended capability
- grammar corrections
v5->v6:
- fixed previous missed abbreviation of LCC and LD
- added text for the PCI capability for the group member device v4->v5:
- split pci transport and generic command section to new patch
- removed multiple references to the VF
- written the description of the command as generic with member
and group device terminology
- reflected many section names to remove VF
- split from pci transport specific patch
- split conformance to transport and generic sections
- written the description of the command as generic with member
and group device terminology
- reflected many section names to remove VF
- rename fields from register to region
- avoided abbreviation for legacy, device and config
v3->v4:
- moved noted to the conformance section details in next patch
- removed queue notify address query AQ command on Michael's suggestion, though it is fine. Instead replaced with extending virtio_pci_notify_cap to indicate that legacy queue notifications can be done on the notification location
- fixed spelling errors
- replaced administrative virtqueue to administration virtqueue
- moved legacy interface normative references to legacy conformance section
v2->v3:
- added new patch to split raws of admin vq opcode table
- adddressed Jason and Michael's comment to split single register access command to common config and device specific commands.
- dropped the suggetion to introduce enable/disable command as
admin command cap bit already covers it.
- added other alternative design considered and discussed in detail in v0, v1 and v2 v1->v2:
- addressed comments from Michael
- added theory of operation
- grammar corrections
- removed group fields description from individual commands as
it is already present in generic section
- added endianness normative for legacy device registers region
- renamed the file to drop vf and add legacy prefix
- added overview in commit log
- renamed subsection to reflect command
v0->v1:
- addressed comments, suggesetions and ideas from Michael Tsirkin and Jason Wang
- far more simpler design than MMR access
- removed complexities of MMR device ids
- removed complexities of MMR registers and extended capabilities
- dropped adding new extended capabilities because if if they are added, a pci device still needs to have existing capabilities
in the legacy configuration space and hypervisor driver do not
need to access them

Parav Pandit (4):
admin: Split opcode table rows with a line
admin: Fix section numbering
admin: Add group member legacy register access commands
transport-pci: Introduce group legacy group member config region
access

 admin-cmds-legacy-interface.tex | 262 ++++++++++++++++++++++++++++++++
 admin.tex                       |  24 ++-
 conformance.tex                 |   3 +
 transport-pci-legacy-regs.tex   |  42 +++++
 transport-pci.tex               |   2 +
 5 files changed, 328 insertions(+), 5 deletions(-)
 create mode 100644 admin-cmds-legacy-interface.tex
 create mode 100644 transport-pci-legacy-regs.tex

Signature · 18 lines

2.26.2


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/

Parav Pandit

author

2023-07-06 04:20 UTC #1 lore raw

PATCH v10 1/4 admin: Split opcode table rows with a line

Currently all opcode appears to be in a single row. Separate them with a line similar to other tables.

Signed-off-by Parav Pandit <[email protected]>

Reviewed-by Cornelia Huck <[email protected]>

---
changelog:
v2->v3:
- new patch

---
 admin.tex | 2 ++
 1 file changed, 2 insertions(+)

admin.tex +2 -0

index 2efd4d7..e51f9e6 100644
@@ -114,7 +114,9 @@ \subsection{Group administration commands}\label{sec:Basic Facilities of a Virti
 opcode & Name & Command Description \\
 \hline \hline
 0x0000 & VIRTIO_ADMIN_CMD_LIST_QUERY & Provides to driver list of commands supported for this group type    \\
+\hline
 0x0001 & VIRTIO_ADMIN_CMD_LIST_USE & Provides to device list of commands used for this group type \\
+\hline
 0x0002 - 0x7FFF & - & Commands using \field{struct virtio_admin_cmd}    \\
 \hline
 0x8000 - 0xFFFF & - & Reserved for future commands (possibly using a different structure)    \\

Signature · 18 lines

2.26.2


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/

Parav Pandit

author

2023-07-06 04:20 UTC #2 lore raw

PATCH v10 2/4 admin: Fix section numbering

In reply to Parav Pandit

Requirements are put one additional level down. Fix it.

Signed-off-by Parav Pandit <[email protected]>

---
changelog:
v9->v10:
- addressed comments from Cornelia
- fixed requirements for administration virtqueue section v4->v5:
- new patch

---
 admin.tex | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

admin.tex +4 -4

index e51f9e6..b0a1a91 100644
@@ -286,7 +286,7 @@ \subsection{Group administration commands}\label{sec:Basic Facilities of a Virti
 supporting multiple group types, the list of supported commands
 might differ between different group types.
 
-\devicenormative{\paragraph}{Group administration commands}{Basic Facilities of a Virtio Device / Device groups / Group administration commands}
+\devicenormative{\subsubsection}{Group administration commands}{Basic Facilities of a Virtio Device / Device groups / Group administration commands}
 
 The device MUST validate \field{opcode}, \field{group_type} and
 \field{group_member_id}, and if any of these has an invalid or

@@ -378,7 +378,7 @@ \subsection{Group administration commands}\label{sec:Basic Facilities of a Virti
 \field{VF Enable} refer to registers within the SR-IOV Extended
 Capability as specified by \hyperref[intro:PCIe]{[PCIe]}.
 
-\drivernormative{\paragraph}{Group administration commands}{Basic Facilities of a Virtio Device / Device groups / Group administration commands}
+\drivernormative{\subsubsection}{Group administration commands}{Basic Facilities of a Virtio Device / Device groups / Group administration commands}
 
 The driver MAY discover whether device supports a specific group type
 by issuing VIRTIO_ADMIN_CMD_LIST_QUERY with the matching

@@ -506,7 +506,7 @@ \section{Administration Virtqueues}\label{sec:Basic Facilities of a Virtio Devic
 tail of a structure, with the driver/device using the full
 structure without concern for versioning.
 
-\devicenormative{\paragraph}{Group administration commands}{Basic Facilities of a Virtio Device / Administration virtqueues}
+\devicenormative{\subsection}{Group administration commands}{Basic Facilities of a Virtio Device / Administration virtqueues}
 
 The device MUST support device-readable and device-writeable buffers
 shorter than described in this specification, by

@@ -555,7 +555,7 @@ \section{Administration Virtqueues}\label{sec:Basic Facilities of a Virtio Devic
 or VIRTIO_ADMIN_STATUS_ENOMEM, then the command MUST NOT
 have any side effects, making it safe to retry.
 
-\drivernormative{\paragraph}{Group administration commands}{Basic Facilities of a Virtio Device / Administration virtqueues}
+\drivernormative{\subsection}{Group administration commands}{Basic Facilities of a Virtio Device / Administration virtqueues}
 
 The driver MAY supply device-readable or device-writeable parts
 of \field{struct virtio_admin_cmd} that are longer than described in

Signature · 18 lines

2.26.2


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/

Cornelia Huck

Reviewed-by on v10 2/4

2023-07-06 13:39 UTC #3 lore raw

[virtio-comment] Re: [PATCH v10 2/4] admin: Fix section numbering

Quoted 10 lines · Parav Pandit

Requirements are put one additional level down. Fix it.
Signed-off-by Parav Pandit <[email protected]>
---
changelog:
v9->v10:
- addressed comments from Cornelia
- fixed requirements for administration virtqueue section v4->v5:
- new patch

Quoted 3 lines

---
 admin.tex | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

Reviewed-by Cornelia Huck <[email protected]>

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 04:20 UTC #4 lore raw

PATCH v10 3/4 admin: Add group member legacy register access commands

In reply to Parav Pandit

Introduce group member legacy common configuration and legacy device configuration access read/write commands.

Group member legacy registers access commands enable group owner driver software to access legacy registers on behalf of the guest virtual machine.

Usecase:
========
1. A hypervisor/system needs to provide transitional virtio devices to the guest VM at scale of thousands, typically, one to eight devices per VM.

2. A hypervisor/system needs to provide such devices using a vendor agnostic driver in the hypervisor system.

Motivation/Background:
=====================
The existing virtio transitional PCI device is missing support for PCI SR-IOV based devices. Currently it does not work beyond PCI PF, or as software emulated device in reality. Currently it has below cited system level limitations:

[a] PCIe spec citation:
VFs do not support I/O Space and thus VF BARs shall not indicate I/O Space.

[c] PCIe spec citation:
If a bridge implements an I/O address range,...I/O address range will be aligned to a 4 KB boundary.

Overview:
=========
Above usecase requirements is solved by PCI PF group owner accessing its group member PCI VFs legacy registers using the administration commands of the group owner PCI PF.

Two types of administration commands are added which read/write PCI VF registers.

Software usage example:
=======================

1. One way to use and map to the guest VM is by using vfio driver framework in Linux kernel.

                +----------------------+
                |pci_dev_id = 0x100X   |
+---------------|pci_rev_id = 0x0      |-----+
|vfio device    |BAR0 = I/O region     |     |
|               |Other attributes      |     |
|               +----------------------+     |
|                                            |
+   +--------------+     +-----------------+ |
|   |I/O BAR to AQ |     | Other vfio      | |
|   |rd/wr mapper& |     | functionalities | |
|   | forwarder    |     |                 | |
|   +--------------+     +-----------------+ |
|                                            |
+------+-------------------------+-----------+
       |                         |
   Config region                 |
     access                Driver notifications
       |                         |
  +----+------------+       +----+------------+
  | +-----+         |       | PCI VF device A |
  | | AQ  |-------------+---->+-------------+ |
  | +-----+         |   |   | | legacy regs | |
  | PCI PF device   |   |   | +-------------+ |
  +-----------------+   |   +-----------------+
                        |
                        |   +----+------------+
                        |   | PCI VF device N |
                        +---->+-------------+ |
                            | | legacy regs | |
                            | +-------------+ |
                            +-----------------+

2. Continue to use the virtio pci driver to bind to the listed device id and use it as in the host.

3. Use it in a light weight hypervisor to run bare-metal OS.

Fixes https://github.com/oasis-tcs/virtio-spec/issues/167

Signed-off-by Parav Pandit <[email protected]>

---
changelog:
v9->v10:
- added white space at end of line
- addressed below comments from Cornelia
- added missing articles
- reworded description for notification query command
- grammar fixes
- addressed below comments from Michael
- added description for member group id setting
- reworded device and driver conformance statements
- opcode table description updated
- fixed label for device read command
- length alignment restriction text added
- data length described for read write commands
- notification description added and refined
- reworded text around command specific result and data field usage v8->v9:
- add missing articles in notify query command
- replaced 'this notification' with 'such a notification'
- addressed below comments from Michael
- dropped 'Region' from the commands
- added 7 reserved pad bytes in config write commands
- rewrote from 'use following structure' to 'field' has the following struct..
- dropped mentioning to follow struct virtio_admin_cmd.
- added note about command limited to only sriov group type for now
- rewrote the description little differently
v7->v8:
- remove empty line at the end of file
- removed white space at the end
- addressed comments from Michael add link to pci
- renamed region to region_data
- made region_data width to be 16 bytes to cover for 8 bytes offset
- moved generic notification region related normative from pci to generic section
v6->v7:
- changed administrative to administration
- renamed admin-access.tex to admin-interface.tex
- large rewrite ad generic admin commands instead of pci
- added theory of operation section
- added driver notification region query command
v5->v6:
- fixed previous missed abbreviation of LCC and LD
v4->v5:
- split from pci transport specific patch
- split conformance to transport and generic sections
- written the description of the command as generic with member and group device terminology
- reflected many section names to remove VF
- rename fields from register to region
- avoided abbreviation for legacy, device and config

---
 admin-cmds-legacy-interface.tex | 261 ++++++++++++++++++++++++++++++++
 admin.tex                       |  14 +-
 conformance.tex                 |   2 +
 3 files changed, 276 insertions(+), 1 deletion(-)
 create mode 100644 admin-cmds-legacy-interface.tex

admin-cmds-legacy-interface.tex new file +261 -0

new file mode 100644
index 0000000..dd01c0a
@@ -0,0 +1,261 @@
+\subsubsection{Legacy Interfaces}\label{sec:Basic Facilities of a Virtio Device / Device groups / Group
+administration commands / Legacy Interface}
+
+In some systems, there is a need to support utilizing the legacy driver with
+the device that do not directly support the legacy interface. In such scenarios,
+a group owner device can provide the legacy interface functionality for the
+group member devices. The driver of an owner device can then access the legacy
+interface of a member device on behalf of the legacy member device driver.
+
+For example, with the SR-IOV group type, group members (VFs) can not present
+the legacy interface in an I/O BAR in BAR0 as expected by the legacy pci driver.
+If the legacy driver is running inside a virtual machine, the hypervisor
+executing the virtual machine can present a virtual device with an I/O BAR in
+BAR0. The hypervisor intercepts the legacy driver accesses to this I/O BAR and
+forwards them to the group owner device (PF) using group administration commands.
+
+The following commands support such legacy interface functionality:
+
+\begin{enumerate}
+\item Legacy Common Configuration Write Command
+\item Legacy Common Configuration Read Command
+\item Legacy Device Configuration Write Command
+\item Legacy Device Configuration Read Command
+\end{enumerate}
+
+These commands are currently only defined for the SR-IOV group type and
+have, generally, the same effect as member device accesses through a legacy
+interface listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Legacy Interfaces: A Note on PCI Device Layout} except that little endian format is assumed unconditionally.
+
+\paragraph{Legacy Common Configuration Write Command}\label{par:Basic Facilities of a Virtio Device / Device groups / Group
+administration commands / Legacy Interface / Legacy Common Configuration Write Command}
+
+This command has the same effect as writing into the virtio common configuration
+structure through the legacy interface. The \field{command_specific_data} is in
+the format \field{struct virtio_admin_cmd_legacy_common_cfg_wr_data} describing
+the access to be performed.
+
+\begin{lstlisting}
+struct virtio_admin_cmd_legacy_common_cfg_wr_data {
+        u8 offset; /* Starting byte offset within the common configuration structure to write */
+        u8 reserved[7];
+        u8 data[];
+};
+\end{lstlisting}
+
+The driver sets command \field{opcode} to VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE.
+The driver sets \field{group_member_id} which refers to the member device to be
+accessed. The driver sets \field{offset} which refers to the offset to write within the
+virtio common configuration structure, and excluding the device-specific configuration.
+
+The length of the data to write is simply the length of \field{data}.
+
+No length or alignment restrictions are placed on the value of the
+\field{offset} and the length of the \field{data}, except that the resulting
+access refers to a single field and is completely within the virtio common
+configuration structure, excluding the device-specific configuration.
+
+This command has no command specific result.
+
+\paragraph{Legacy Common Configuration Read Command}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Common Configuration Read Command}
+
+This command has the same effect as reading from the virtio common configuration
+structure through the legacy interface. The \field{command_specific_data} is in
+the format \field{struct virtio_admin_cmd_legacy_common_cfg_rd_data} describing
+the access to be performed.
+
+\begin{lstlisting}
+struct virtio_admin_cmd_legacy_common_cfg_rd_data {
+	u8 offset; /* Starting byte offset within the common configuration structure to read */
+};
+\end{lstlisting}
+
+The driver sets command \field{opcode} to VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ.
+
+The driver sets \field{offset} which refers to the offset to read from the
+virtio common configuration structure, and excluding the device-specific
+configuration.
+
+\begin{lstlisting}
+struct virtio_admin_cmd_legacy_common_cfg_rd_result {
+        u8 data[];
+};
+\end{lstlisting}
+
+When the command completes successfully, \field{command_specific_result}
+is in the format of \field{struct virtio_admin_cmd_legacy_common_cfg_rd_result}
+returned by the device.
+
+The length of the data read is simply the length of \field{data}.
+
+\paragraph{Legacy Device Configuration Write Command}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Device Configuration Write Command}
+
+This command has the same effect as writing into the virtio device-specific
+configuration through the legacy interface. The \field{command_specific_data} is in
+the format \field{struct virtio_admin_cmd_legacy_dev_reg_wr_data} describing
+the access to be performed.
+
+\begin{lstlisting}
+struct virtio_admin_cmd_legacy_dev_reg_wr_data {
+        u8 offset; /* Starting byte offset within the device-specific configuration to write */
+        u8 reserved[7];
+        u8 data[];
+};
+\end{lstlisting}
+
+The driver sets command \field{opcode} to VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE.
+The driver sets \field{group_member_id} which refers to the member device to be
+accessed. The driver sets \field{offset} which refers to the offset to write
+within the virtio device-specific configuration.
+
+The length of the data to write is simply the length of \field{data}.
+
+No length or alignment restrictions are placed on the value of the
+\field{offset} and the length of the \field{data}, except that the resulting
+access refers to a single field and is completely within the device-specific
+configuration.
+
+This command has no command specific result.
+
+\paragraph{Legacy Device Configuration Read Command}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Device Configuration Read Command}
+
+This command has the same effect as reading from the virtio device-specific
+configuration through the legacy interface. The \field{command_specific_data} is in
+the format \field{struct virtio_admin_cmd_legacy_common_cfg_rd_data} describing
+the access to be performed.
+
+\begin{lstlisting}
+struct virtio_admin_cmd_legacy_dev_cfg_rd_data {
+        u8 offset; /* Starting byte offset within the device-specific configuration to read */
+};
+\end{lstlisting}
+
+The driver sets command \field{opcode} to VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ.
+The driver sets \field{group_member_id} which refers to the member device to be
+accessed. The driver sets \field{offset} which refers to the offset to read from
+the virtio device-specific configuration.
+
+\begin{lstlisting}
+struct virtio_admin_cmd_legacy_dev_reg_rd_result {
+        u8 data[];
+};
+\end{lstlisting}
+
+When the command completes successfully, \field{command_specific_result} is in
+the format of \field{struct virtio_admin_cmd_legacy_dev_reg_rd_result}
+returned by the device.
+
+The length of the data read is simply the length of \field{data}.
+
+\paragraph{Legacy Driver Notification}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Driver Notifications}
+
+If the group owner device or the group member device support driver
+notifications via a memory-mapped operation or I/O operation, these
+notifications are sent to the device via accessing such a notification region
+using a memory or I/O operation instead of sending the notifications through
+the admistration command.
+
+The driver of the owner device can send a driver notification to the member
+device by executing VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE with the
+\field{offset} matching \field{Queue Notify} and the \field{data} containing
+the virtqueue index to be notified.
+
+However, as many administration commands are used for slow path configuration,
+a separate fast path mechanism for such notifications is desired. For the
+SR-IOV group type, the optional command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
+addresses this need by returning to the driver one or more addresses which to
+be used to send such driver notifications.
+
+The driver sets command \field{opcode} to VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY.
+The driver sets \field{group_member_id} which refers to the member device to be
+accessed. This command does not use \field{command_specific_data}.
+
+When the command completes successfully, \field{command_specific_result}
+is in the format of \field{struct virtio_admin_cmd_legacy_notify_query_result}.
+
+\begin{lstlisting}
+struct virtio_admin_cmd_legacy_notify_query_entry {
+        u8 data[16];
+};
+
+struct virtio_admin_cmd_legacy_notify_query_result {
+	struct virtio_virtio_admin_cmd_legacy_notify_query_entry entries[];
+};
+\end{lstlisting}
+
+The driver picks the suitable entry when multiple entries are supplied
+by the device.
+
+Refer to the specific transport section for the definition of the
+\field{data}.
+
+This command is currently only defined for the SR-IOV group type.
+
+\devicenormative{\paragraph}{Legacy Interface}{Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface}
+
+A device MUST either support all of, or none of
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ commands.
+
+For VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ commands, the device MUST decode and
+encode (respectively) the value of the \field{data} using the little-endian
+format.
+
+The device MUST fail VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ commands where the value of the
+\field{offset} and the length of the \field{data} does not refer to a
+refer to a single field or is not completely within the virtio common
+configuration structure excluding the device-specific configuration.
+
+The device MUST fail VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ commands where the value of the
+\field{offset} and the length of the \field{data} does not refer to a
+refer to a single field or is not completely within the
+device-specific configuration.
+
+If the device supports VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY it MUST
+also support all of VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ.
+
+The command VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE MUST have the same effect
+as writing into the virtio common configuration structure through the legacy
+interface.
+
+The command VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ MUST have the same effect as
+reading from the virtio common configuration structure through the legacy
+interface.
+
+The command VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE MUST have the same effect as
+writing into the virtio device configuration structure through the legacy
+interface.
+
+The command VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ MUST have the same effect as
+reading from the virtio device configuration structure through the legacy
+interface.
+
+\drivernormative{\paragraph}{Legacy Interface}{Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface}
+
+For VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ commands, the driver MUST encode and
+decod3 (respectively) the value of the \field{data} using the little-endian
+format.
+
+For VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ commands the driver SHOULD set
+\field{offset} and the length of the \field{data} to refer to a
+a single field within the virtio common configuration structure excluding
+the device-specific configuration.
+
+For VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ commands the driver SHOULD set
+\field{offset} and the length of the \field{data} to refer to a
+a single field within the virtio device-specific configuration.
+
+The group member driver SHOULD use the notification region supplied by the group
+owner device to send driver notifications to the device if
+VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY is supported.

admin.tex +13 -1

index b0a1a91..e2134e1 100644
@@ -117,7 +117,17 @@ \subsection{Group administration commands}\label{sec:Basic Facilities of a Virti
 \hline
 0x0001 & VIRTIO_ADMIN_CMD_LIST_USE & Provides to device list of commands used for this group type \\
 \hline
-0x0002 - 0x7FFF & - & Commands using \field{struct virtio_admin_cmd}    \\
+0x0002 & VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE & Writes into the legacy common configuration structure \\
+\hline
+0x0003 & VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ & Reads from the legacy common configuration structure  \\
+\hline
+0x0004 & VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE & Writes into the legacy device configuration structure \\
+\hline
+0x0005 & VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ & Reads into the legacy device configuration structure \\
+\hline
+0x0006 & VIRTIO_ADMIN_CMD_LEGACY_DEV_NOTIFY_QUERY & Query the notification region \\
+\hline
+0x0007 - 0x7FFF & - & Commands using \field{struct virtio_admin_cmd}    \\
 \hline
 0x8000 - 0xFFFF & - & Reserved for future commands (possibly using a different structure)    \\
 \hline

@@ -286,6 +296,8 @@ \subsection{Group administration commands}\label{sec:Basic Facilities of a Virti
 supporting multiple group types, the list of supported commands
 might differ between different group types.
 
+\input{admin-cmds-legacy-interface.tex}
+
 \devicenormative{\subsubsection}{Group administration commands}{Basic Facilities of a Virtio Device / Device groups / Group administration commands}
 
 The device MUST validate \field{opcode}, \field{group_type} and

conformance.tex +2 -0

index 01ccd69..dc00e84 100644
@@ -260,6 +260,8 @@ \section{Conformance Targets}\label{sec:Conformance / Conformance Targets}
 \item Section \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Legacy Interfaces: A Note on Virtqueue Layout}
 \item Section \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Legacy Interfaces: A Note on Virtqueue Endianness}
 \item Section \ref{sec:Basic Facilities of a Virtio Device / Virtqueues / Message Framing / Legacy Interface: Message Framing}
+\item Section \ref{devicenormative:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface}
+\item Section \ref{drivernormative:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface}
 \item Section \ref{sec:General Initialization And Device Operation / Device Initialization / Legacy Interface: Device Initialization}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Discovery / Legacy Interfaces: A Note on PCI Device Discovery}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Legacy Interfaces: A Note on PCI Device Layout}

Signature · 6 lines

2.26.2


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Cornelia Huck

2023-07-06 16:12 UTC #5 lore raw

[virtio-comment] Re: [PATCH v10 3/4] admin: Add group member legacy register access commands

admin-cmds-legacy-interface.tex line 5

Quoted 11 lines · Parav Pandit

admin-cmds-legacy-interface.tex new file +5 -0

new file mode 100644
index 0000000..dd01c0a
@@ -0,0 +1,261 @@
+\subsubsection{Legacy Interfaces}\label{sec:Basic Facilities of a Virtio Device / Device groups / Group
+administration commands / Legacy Interface}
+
+In some systems, there is a need to support utilizing the legacy driver with
+the device that do not directly support the legacy interface. In such scenarios,

Please use either Michael's or my suggestion from v9 here, using definite articles here isn't a good idea.

admin-cmds-legacy-interface.tex line 8

Quoted 3 lines

admin-cmds-legacy-interface.tex new file +3 -0

+a group owner device can provide the legacy interface functionality for the
+group member devices. The driver of an owner device can then access the legacy
+interface of a member device on behalf of the legacy member device driver.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 16:16 UTC #6 lore raw

[virtio-comment] RE: [PATCH v10 3/4] admin: Add group member legacy register access commands

Quoted 11 lines

From: Cornelia Huck <[email protected]>
Sent: Thursday, July 6, 2023 12:12 PM
Review
Quoted 3 lines
(quoted diff) +3 -0
+In some systems, there is a need to support utilizing the legacy
+driver with the device that do not directly support the legacy
+interface. In such scenarios,
Please use either Michael's or my suggestion from v9 here, using definite articles here isn't a good idea.

Yes, working on it.
I am taking, below suggested version.

In some systems, there is a need to support utilizing a legacy driver with a device that does not directly support the legacy interface.

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/

Parav Pandit

author

2023-07-06 04:20 UTC #7 lore raw

PATCH v10 4/4 transport-pci: Introduce group legacy group member config region access

In reply to Parav Pandit

This patch links how in a PCI transport a group owner can access group member (PCI VFs) legacy registers using a legacy registers access commands using administration virtqueue infrastructure.

Additionally it extend the PCI notification capability through which a PCI VF device indicates to the driver which PCI BAR region to be used for driver notifications.

Fixes https://github.com/oasis-tcs/virtio-spec/issues/167

Signed-off-by Parav Pandit <[email protected]>

---
changelog:
v9->v10:
- addressed comments from Cornelia
- fixed errors related to article
- hardwire to hardwires
- replaced various to all
- added hardwire to zero
v7->v8:
- addressed comments from Michael
- made bar offset 64-bit
- prefix legacy specific structure with _legacy
- moved generic normative from pci to generic section
- added link to virtio pci capabilities when referring to bar 0
- remove 'should' from generic description
v6->v7:
- addressed comments from Michael
- removed driver normative about I/O BAR emulation as it does not make much sense for the spec
- removed references to administration virtqueue
- rewrote driver legacy region access without guest and hypervisor wording
- added normative for notification region
- added normative for PCI IDs for device which support legacy commands v5->v6:
- aligned pci capability to 4B as required by PCI spec
- added text for the PCI capability for the group member device v4->v5:
- split pci transport and generic command section to new patch
- removed multiple references to the VF
- written the description of the command as generic with member and group device terminology
- reflected many section names to remove VF
v3->v4:
- moved noted to the conformance section details in next patch
- removed queue notify address query AQ command on Michael's suggestion, though it is fine. Instead replaced with extending virtio_pci_notify_cap to indicate that legacy queue notifications can be done on the notification location.
- fixed spelling errors.
- replaced administrative virtqueue to administration virtqueue
- added queue notification capability register to indicate legacy q notification supported
v2->v3:
- adddressed Jason and Michael's comment to split single register access command to common config and device specific commands.
- dropped the suggetion to introduce enable/disable command as admin command cap bit already covers it.
v1->v2:
- addressed comments from Michael
- added theory of operation
- grammar corrections
- removed group fields description from individual commands as it is already present in generic section
- added endianness normative for legacy device registers region
- renamed the file to drop vf and add legacy prefix

- added overview in commit log
- renamed subsection to reflect command

---
 admin-cmds-legacy-interface.tex |  3 ++-
 conformance.tex                 |  1 +
 transport-pci-legacy-regs.tex   | 42 +++++++++++++++++++++++++++++++++
 transport-pci.tex               |  2 ++
 4 files changed, 47 insertions(+), 1 deletion(-)
 create mode 100644 transport-pci-legacy-regs.tex

admin-cmds-legacy-interface.tex +2 -1

index dd01c0a..e67632b 100644
@@ -187,7 +187,8 @@ \subsubsection{Legacy Interfaces}\label{sec:Basic Facilities of a Virtio Device
 by the device.
 
 Refer to the specific transport section for the definition of the
-\field{data}.
+\field{data}. For the PCI transport refer to section
+\ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}.
 
 This command is currently only defined for the SR-IOV group type.

conformance.tex +1 -0

index dc00e84..b3f2c92 100644
@@ -267,6 +267,7 @@ \section{Conformance Targets}\label{sec:Conformance / Conformance Targets}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Legacy Interfaces: A Note on PCI Device Layout}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI-specific Initialization And Device Operation / Device Initialization / Virtio Device Configuration Layout Detection / Legacy Interface: A Note on Device Layout Detection}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI-specific Initialization And Device Operation / Device Initialization / Virtqueue Configuration / Legacy Interface: A Note on Virtqueue Configuration}
+\item Section \ref{devicenormative:Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over MMIO / Legacy interface}
 \item Section \ref{sec:Virtio Transport Options / Virtio over channel I/O / Device Initialization / Setting the Virtio Revision / Legacy Interfaces: A Note on Setting the Virtio Revision}
 \item Section \ref{sec:Virtio Transport Options / Virtio over channel I/O / Device Initialization / Configuring a Virtqueue / Legacy Interface: A Note on Configuring a Virtqueue}

transport-pci-legacy-regs.tex new file +42 -0

42 lines not on this page · raw message

transport-pci.tex +2 -0

index a5c6719..72c78f6 100644
@@ -1212,3 +1212,5 @@ \subsubsection{Driver Handling Interrupts}\label{sec:Virtio Transport Options /
         re-examine the configuration space to see what changed.
     \end{itemize}
 \end{itemize}
+
+\input{transport-pci-legacy-regs.tex}

Signature · 18 lines

2.26.2


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/

Cornelia Huck

2023-07-06 16:28 UTC #8 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

transport-pci-legacy-regs.tex line 3

Quoted 9 lines · Parav Pandit

transport-pci-legacy-regs.tex new file +3 -0

new file mode 100644
index 0000000..ceea28c
@@ -0,0 +1,42 @@
+\subsection{Legacy Interface: Group member device Configuration Region Access}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}
+
+The PCI owner device or the member device or both support driver notifications using

What about

"The PCI owner device, the member device, or both can choose to support..." ?

Quoted 8 lines

transport-pci-legacy-regs.tex new file +8 -0

+a notification region defined in \field{struct virtio_pci_legacy_notify_region}.
+
+In \field{struct virtio_virtio_admin_cmd_legacy_notify_query_entry},
+\field{region_data} is defined as following:
+
+\begin{lstlisting}
+struct virtio_pci_legacy_notify_region {
+        u8 owner;  /* When set to 1, notification region is of the owner device */

transport-pci-legacy-regs.tex line 21

Quoted 10 lines

transport-pci-legacy-regs.tex new file +10 -0

+        u8 bar;    /* BAR of the member or owner device */
+        u8 padding[6];
+        le64 offset; /* Offset within bar. */
+};
+\end{lstlisting}
+
+The group owner device hardwires VF BAR0 to zero in the SR-IOV Extended
+capability.
+
+The group member device does not use PCI BAR0 in all the Virtio PCI capabilities

"in [or for?] any of the...", I guess?

transport-pci-legacy-regs.tex line 30

Quoted 9 lines

transport-pci-legacy-regs.tex new file +9 -0

+listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.
+
+\devicenormative{\subsubsection}{Legacy Interface: Group Member Device Legacy Configuration Region Access}{Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}
+
+When a PCI SR-IOV group owner device supports
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE, VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE commands, the group owner device MUST
+hardwire VF BAR0 to zero in the SR-IOV Extended capability and the group memberi

s/memberi/member/

transport-pci-legacy-regs.tex line 34

Quoted 4 lines

transport-pci-legacy-regs.tex new file +4 -0

+device MUST NOT use BAR0 in any of the Virtio Structure PCI Capabilities.
+
+The group owner device or the group member device or both MAY support driver
+notifications region.

Make this "a driver notification region"?

transport-pci-legacy-regs.tex line 42

Quoted 8 lines

transport-pci-legacy-regs.tex new file +8 -0

+
+For the SR-IOV group type, the owner device supporting
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE, VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
+commands and its member device SHOULD follow the rules for the PCI Device ID,
+Revision ID and Subsystem Device ID of the non-transitional devices documented in
+section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Discovery}.

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.

Parav Pandit

author

2023-07-06 16:33 UTC #9 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 23 lines

From: Cornelia Huck <[email protected]>
Sent: Thursday, July 6, 2023 12:28 PM
transport-pci-legacy-regs.tex line 7
Quoted 13 lines · Parav Pandit
(quoted diff) +0 -0
b/transport-pci-legacy-regs.tex new file mode 100644 index 0000000..ceea28c
transport-pci-legacy-regs.tex new file +7 -0
@@ -0,0 +1,42 @@
+\subsection{Legacy Interface: Group member device Configuration
+Region Access}\label{sec:Virtio Transport Options / Virtio Over PCI
+Bus / Legacy Interface: Group Member Device Configuration Region
+Access}
+
+The PCI owner device or the member device or both support driver
+notifications using
What about

Quoted 2 lines

"The PCI owner device, the member device, or both can choose to support..." ?

Fine too.
Any thing wrong in having or as above, so I don't write it next time? Or that in current form reads better to me.

Quoted 10 lines

transport-pci-legacy-regs.tex line 8
Quoted 1 line
transport-pci-legacy-regs.tex new file +1 -0
+a notification region defined in \field{struct
virtio_pci_legacy_notify_region}.
transport-pci-legacy-regs.tex
Quoted 7 lines
transport-pci-legacy-regs.tex new file +7 -0
+
+In \field{struct virtio_virtio_admin_cmd_legacy_notify_query_entry},
+\field{region_data} is defined as following:
+
+\begin{lstlisting}
+struct virtio_pci_legacy_notify_region {
+        u8 owner;  /* When set to 1, notification region is of the owner device
*/

Quoted 11 lines

transport-pci-legacy-regs.tex line 24

Quoted 9 lines

transport-pci-legacy-regs.tex new file +9 -0

+        u8 bar;    /* BAR of the member or owner device */
+        u8 padding[6];
+        le64 offset; /* Offset within bar. */ }; \end{lstlisting}
+
+The group owner device hardwires VF BAR0 to zero in the SR-IOV
+Extended capability.
+
+The group member device does not use PCI BAR0 in all the Virtio PCI
+capabilities

"in [or for?] any of the...", I guess?

Will change to "Any of the ".

Quoted 15 lines

transport-pci-legacy-regs.tex line 25

Quoted 2 lines

transport-pci-legacy-regs.tex new file +1 -0
+listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus /

Virtio Structure PCI Capabilities}.

transport-pci-legacy-regs.tex line 36

Quoted 11 lines

transport-pci-legacy-regs.tex new file +11 -0

+
+\devicenormative{\subsubsection}{Legacy Interface: Group Member
+Device Legacy Configuration Region Access}{Virtio Transport Options /
+Virtio Over PCI Bus / Legacy Interface: Group Member Device
+Configuration Region Access}
+
+When a PCI SR-IOV group owner device supports
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE commands, the group

owner

Quoted 5 lines

transport-pci-legacy-regs.tex line 38
Quoted 2 lines
transport-pci-legacy-regs.tex new file +2 -0
+device MUST hardwire VF BAR0 to zero in the SR-IOV Extended
+capability and the group memberi
s/memberi/member/

Yes, already changed in my v11 where patch 3 and 4 content is merged to single generic section.

Quoted 7 lines

transport-pci-legacy-regs.tex
Quoted 4 lines
transport-pci-legacy-regs.tex new file +4 -0
+device MUST NOT use BAR0 in any of the Virtio Structure PCI Capabilities.
+
+The group owner device or the group member device or both MAY support
+driver notifications region.
Make this "a driver notification region"?

Notifications are generally more than one and spec has the section "driver notifications", so...

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.

Michael S. Tsirkin

2023-07-06 16:42 UTC #10 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 75 lines · Parav Pandit

Quoted 23 lines
From: Cornelia Huck <[email protected]>
Sent: Thursday, July 6, 2023 12:28 PM
transport-pci-legacy-regs.tex line 7
Quoted 13 lines · Parav Pandit
(quoted diff) +0 -0
b/transport-pci-legacy-regs.tex new file mode 100644 index 0000000..ceea28c
transport-pci-legacy-regs.tex new file +7 -0
@@ -0,0 +1,42 @@
+\subsection{Legacy Interface: Group member device Configuration
+Region Access}\label{sec:Virtio Transport Options / Virtio Over PCI
+Bus / Legacy Interface: Group Member Device Configuration Region
+Access}
+
+The PCI owner device or the member device or both support driver
+notifications using
What about
Quoted 2 lines
"The PCI owner device, the member device, or both can choose to support..." ?
Fine too.
Any thing wrong in having or as above, so I don't write it next time? Or that in current form reads better to me.
Quoted 10 lines
transport-pci-legacy-regs.tex line 8
Quoted 1 line
transport-pci-legacy-regs.tex new file +1 -0
+a notification region defined in \field{struct
virtio_pci_legacy_notify_region}.
transport-pci-legacy-regs.tex
Quoted 7 lines
transport-pci-legacy-regs.tex new file +7 -0
+
+In \field{struct virtio_virtio_admin_cmd_legacy_notify_query_entry},
+\field{region_data} is defined as following:
+
+\begin{lstlisting}
+struct virtio_pci_legacy_notify_region {
+        u8 owner;  /* When set to 1, notification region is of the owner device
*/
Quoted 11 lines
transport-pci-legacy-regs.tex line 24
Quoted 9 lines
transport-pci-legacy-regs.tex new file +9 -0
+        u8 bar;    /* BAR of the member or owner device */
+        u8 padding[6];
+        le64 offset; /* Offset within bar. */ }; \end{lstlisting}
+
+The group owner device hardwires VF BAR0 to zero in the SR-IOV
+Extended capability.
+
+The group member device does not use PCI BAR0 in all the Virtio PCI
+capabilities
"in [or for?] any of the...", I guess?
Will change to "Any of the ".
Quoted 15 lines
transport-pci-legacy-regs.tex line 25
Quoted 2 lines
transport-pci-legacy-regs.tex new file +1 -0
+listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus /
Virtio Structure PCI Capabilities}.
transport-pci-legacy-regs.tex line 36
Quoted 11 lines
transport-pci-legacy-regs.tex new file +11 -0
+
+\devicenormative{\subsubsection}{Legacy Interface: Group Member
+Device Legacy Configuration Region Access}{Virtio Transport Options /
+Virtio Over PCI Bus / Legacy Interface: Group Member Device
+Configuration Region Access}
+
+When a PCI SR-IOV group owner device supports
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE commands, the group
owner
Quoted 5 lines
transport-pci-legacy-regs.tex line 38
Quoted 2 lines
transport-pci-legacy-regs.tex new file +2 -0
+device MUST hardwire VF BAR0 to zero in the SR-IOV Extended
+capability and the group memberi
s/memberi/member/
Yes, already changed in my v11 where patch 3 and 4 content is merged to single generic section.

Quoted 9 lines

Quoted 7 lines
transport-pci-legacy-regs.tex
Quoted 4 lines
transport-pci-legacy-regs.tex new file +4 -0
+device MUST NOT use BAR0 in any of the Virtio Structure PCI Capabilities.
+
+The group owner device or the group member device or both MAY support
+driver notifications region.
Make this "a driver notification region"?
Notifications are generally more than one and spec has the section "driver notifications", so...

We will need to detail how notifications work though. The reason is that the legacy notifications are received by owner driver (which is not legacy!) and forwarded to device.
We need to explain exactly how this is done for people who don't read minds and did not follow 3 months of mailing list debates, and explain how this has the same effect as triggering driver notification through the legacy interface.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 16:58 UTC #11 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 5 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:42 PM

Quoted 7 lines

We will need to detail how notifications work though. The reason is that the legacy notifications are received by owner driver (which is not legacy!) and forwarded to device.
We need to explain exactly how this is done for people who don't read minds and did not follow 3 months of mailing list debates, and explain how this has the same effect as triggering driver notification through the legacy interface.

I am not going to write driver implementation in the spec here. But I agree some high-level text is needed.
With your suggestions
1. to merge pci and generic in one section,
2. add description to notification,
3. past rewrites provided in v9,

The draft of v11, looks like below. Please see how much more verbose to make it, if at all.

\paragraph{Legacy Driver Notification}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Driver Notifications}

If the group owner device or the group member device support driver notifications via a memory-mapped operation or I/O operation, these notifications are sent to the device via accessing such a notification region using a memory or I/O operation instead of sending the notifications through the administration command.

However, as many administration commands are used for slow path configuration, a separate fast path mechanism for such notifications is desired. For the SR-IOV group type, the optional command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO addresses this need by returning to the driver one or more addresses which to be used to send such driver notifications.

In one example, the driver in the hypervisor which intercepts I/O BAR accesses for the \field{Queue Notify} can access the notification area directly as memory or I/O access (depending on what the device reported) instead of slow administration command.

For the command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO the driver sets \field{opcode} to 0x6.
The driver sets \field{group_member_id} which refers to the member device to be accessed. This command does not use \field{command_specific_data}.

This command is currently only defined for the SR-IOV group type. When the
device supports VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO command, the group owner
device hardwires VF BAR0 to zero in the SR-IOV Extended capability and the
group member device does not use PCI BAR0 in all the Virtio PCI capabilities
listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.

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.

Michael S. Tsirkin

2023-07-06 17:35 UTC #12 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 21 lines · Parav Pandit

Quoted 5 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:42 PM
Quoted 7 lines
We will need to detail how notifications work though. The reason is that the legacy notifications are received by owner driver (which is not legacy!) and forwarded to device.
We need to explain exactly how this is done for people who don't read minds and did not follow 3 months of mailing list debates, and explain how this has the same effect as triggering driver notification through the legacy interface.
I am not going to write driver implementation in the spec here. But I agree some high-level text is needed.
With your suggestions
1. to merge pci and generic in one section,
2. add description to notification,
3. past rewrites provided in v9,
The draft of v11, looks like below. Please see how much more verbose to make it, if at all.

Quoted 7 lines

\paragraph{Legacy Driver Notification}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Driver Notifications}
If the group owner device or the group member device support driver notifications via a memory-mapped operation or I/O operation, these notifications are sent to the device via accessing such a notification region using a memory or I/O operation instead of sending the notifications through the administration command.

This paragraph adds very little information - just some vague hints at how things work.

Quoted 21 lines

The driver of the owner device can send a driver notification to the member device by executing VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified.
However, as many administration commands are used for slow path configuration, a separate fast path mechanism for such notifications is desired. For the SR-IOV group type, the optional command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO addresses this need by returning to the driver one or more addresses which to be used to send such driver notifications.
In one example, the driver in the hypervisor which intercepts I/O BAR accesses for the \field{Queue Notify} can access the notification area directly as memory or I/O access (depending on what the device reported) instead of slow administration command.
For the command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO the driver sets \field{opcode} to 0x6.
The driver sets \field{group_member_id} which refers to the member device to be accessed. This command does not use \field{command_specific_data}.

Quoted 5 lines

This command is currently only defined for the SR-IOV group type. When the
device supports VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO command, the group owner
device hardwires VF BAR0 to zero in the SR-IOV Extended capability and the
group member device does not use PCI BAR0 in all the Virtio PCI capabilities
listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.

After reading this I still have no clue at how these notifications work. maybe it's in some other patch. Let's see v11.

Just making a wild guess:

it is true that
to supply a driver notification to the device, it is possible
for the driver to write the vq index of the vq to be notified, in little endian format, using a single 16 byte write at the address provided ?

and I am guessing that
the write is an IO or a memory write, depending on the type of the BAR referred to?

then why don't we just say this?

Signature · 18 lines

MST


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/

Parav Pandit

author

2023-07-06 17:47 UTC #13 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 32 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 1:34 PM
Quoted 11 lines · Parav Pandit
Quoted 11 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:42 PM
We will need to detail how notifications work though. The reason is that the legacy notifications are received by owner driver (which is not legacy!) and forwarded to device.
We need to explain exactly how this is done for people who don't read minds and did not follow 3 months of mailing list debates, and explain how this has the same effect as triggering driver notification through
the legacy interface.
Quoted 8 lines
Quoted 1 line
I am not going to write driver implementation in the spec here. But I agree some high-level text is needed.
With your suggestions
1. to merge pci and generic in one section, 2. add description to notification, 3. past rewrites provided in v9,
The draft of v11, looks like below. Please see how much more verbose to make
it, if at all.
Quoted 5 lines
\paragraph{Legacy Driver Notification}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Driver Notifications}

Quoted 9 lines

Quoted 5 lines
If the group owner device or the group member device support driver notifications via a memory-mapped operation or I/O operation, these notifications are sent to the device via accessing such a notification region using a memory or I/O operation instead of sending the notifications through the administration command.
This paragraph adds very little information - just some vague hints at how things work.

A hunk added this paragraph back.
Going to remove it as description is duplicate of what is written below.

Quoted 18 lines

Quoted 2 lines
The driver of the owner device can send a driver notification to the member device by executing
VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE
Quoted 15 lines
with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified.
However, as many administration commands are used for slow path configuration, a separate fast path mechanism for such notifications is desired. For the SR-IOV group type, the optional command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO
addresses this need by returning to the driver one or more addresses which to be used to send such driver notifications.
In one example, the driver in the hypervisor which intercepts I/O BAR accesses for the \field{Queue Notify} can access the notification area directly as memory or I/O access (depending on what the device reported) instead of slow administration command.

Above paragraph is really more than enough for the reader to implement. I will merge your text of 16-bit notify vq line here.

Quoted 9 lines

Quoted 1 line
For the command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO the driver
sets
Quoted 3 lines
\field{opcode} to 0x6.
The driver sets \field{group_member_id} which refers to the member device to be accessed. This command does not use
\field{command_specific_data}.
Quoted 3 lines
This command is currently only defined for the SR-IOV group type. When the device supports VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO command,

Quoted 9 lines

the
Quoted 3 lines
group owner device hardwires VF BAR0 to zero in the SR-IOV Extended capability and the group member device does not use PCI BAR0 in all the Virtio PCI capabilities listed in section \ref{sec:Virtio Transport Options /
Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.
After reading this I still have no clue at how these notifications work. maybe it's in some other patch. Let's see v11.

There is only one patch now in v11 other than two fixes.

Quoted 7 lines

Just making a wild guess:
it is true that
to supply a driver notification to the device, it is possible for the driver
to write the vq index of the vq to be notified, in little endian format, using a single 16 byte write at the address provided ?

Quoted 5 lines

and I am guessing that
the write is an IO or a memory write, depending on the type of the BAR referred to?
then why don't we just say this?

All this while I assume that what you wrote above is already written in description for "queue notify" register in " Legacy Interfaces: A Note on PCI Device Layout". So no need to repeat.

But I realized that base spec is totally silent on it, but still we are supposed to write here. Ok.
So yes, let me add above line here, without touching existing legacy interface section.

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.

Michael S. Tsirkin

2023-07-06 18:06 UTC #14 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 57 lines · Parav Pandit

Quoted 32 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 1:34 PM
Quoted 11 lines · Parav Pandit
Quoted 11 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:42 PM
We will need to detail how notifications work though. The reason is that the legacy notifications are received by owner driver (which is not legacy!) and forwarded to device.
We need to explain exactly how this is done for people who don't read minds and did not follow 3 months of mailing list debates, and explain how this has the same effect as triggering driver notification through
the legacy interface.
Quoted 8 lines
Quoted 1 line
I am not going to write driver implementation in the spec here. But I agree some high-level text is needed.
With your suggestions
1. to merge pci and generic in one section, 2. add description to notification, 3. past rewrites provided in v9,
The draft of v11, looks like below. Please see how much more verbose to make
it, if at all.
Quoted 5 lines
\paragraph{Legacy Driver Notification}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Driver Notifications}
Quoted 9 lines
Quoted 5 lines
If the group owner device or the group member device support driver notifications via a memory-mapped operation or I/O operation, these notifications are sent to the device via accessing such a notification region using a memory or I/O operation instead of sending the notifications through the administration command.
This paragraph adds very little information - just some vague hints at how things work.
A hunk added this paragraph back.
Going to remove it as description is duplicate of what is written below.
Quoted 13 lines
Quoted 2 lines
The driver of the owner device can send a driver notification to the member device by executing
VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE
Quoted 10 lines
with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified.
However, as many administration commands are used for slow path configuration, a separate fast path mechanism for such notifications is desired. For the SR-IOV group type, the optional command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO
addresses this need by returning to the driver one or more addresses which to be used to send such driver notifications.

Quoted 7 lines

Quoted 5 lines
Quoted 5 lines
In one example, the driver in the hypervisor which intercepts I/O BAR accesses for the \field{Queue Notify} can access the notification area directly as memory or I/O access (depending on what the device reported) instead of slow administration command.
Above paragraph is really more than enough for the reader to implement. I will merge your text of 16-bit notify vq line here.

I don't have all of the context in front of me. We'll see in v11. Try to keep a style consistent with what we have written for the config space.

Quoted 31 lines

Quoted 9 lines
Quoted 1 line
For the command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO the driver
sets
Quoted 3 lines
\field{opcode} to 0x6.
The driver sets \field{group_member_id} which refers to the member device to be accessed. This command does not use
\field{command_specific_data}.
Quoted 3 lines
This command is currently only defined for the SR-IOV group type. When the device supports VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO command,
Quoted 9 lines
the
Quoted 3 lines
group owner device hardwires VF BAR0 to zero in the SR-IOV Extended capability and the group member device does not use PCI BAR0 in all the Virtio PCI capabilities listed in section \ref{sec:Virtio Transport Options /
Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.
After reading this I still have no clue at how these notifications work. maybe it's in some other patch. Let's see v11.
There is only one patch now in v11 other than two fixes.
Quoted 11 lines
Just making a wild guess:
it is true that
to supply a driver notification to the device, it is possible for the driver
to write the vq index of the vq to be notified, in little endian format, using a single 16 byte write at the address provided ?
and I am guessing that
the write is an IO or a memory write, depending on the type of the BAR referred to?

Quoted 7 lines

Quoted 1 line
then why don't we just say this?
All this while I assume that what you wrote above is already written in description for "queue notify" register in " Legacy Interfaces: A Note on PCI Device Layout". So no need to repeat.
But I realized that base spec is totally silent on it, but still we are supposed to write here. Ok.
So yes, let me add above line here, without touching existing legacy interface section.

It's not documented all that well. I think it was clearer originally but with all the VIRTIO_F_NOTIFICATION_DATA and such it bit-rotted.

But again even if you add it there, you can not claim it's exactly the same as legacy because the address is different, the address type is different, the driver is different (this is owner driver) and even the device is different.

Please also call out the unusual configuration where the type is "member" and then you have the owner driver access the memory of the member device. People might be confused.

I also think we should explain that order of entries is a hint to driver: use the 1st entry that you can.

Signature · 18 lines

MST


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/

Parav Pandit

author

2023-07-06 18:16 UTC #15 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 8 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 2:06 PM
But again even if you add it there, you can not claim it's exactly the same as legacy because the address is different, the address type is different, the driver is different (this is owner driver) and even the device is different.

You drafted below, that I included,

We extend this further by saying Queue notify notifications is sent via MMIO as,

The driver of the owner device can send a driver notification to the member device by executing VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified or by performing memory or I/O write in the any of the notification region at offset 0 supplied by the device in VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO result.

Quoted 6 lines

Please also call out the unusual configuration where the type is "member" and then you have the owner driver access the memory of the member device. People might be confused.
I also think we should explain that order of entries is a hint to driver: use the 1st entry that you can.

Driver really can choose any valid entry out of the 4 that driver likes.
I really don't see a need for overwriting this area as I fail to see why one will expose multiple entries from the device side in reality.

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.

Michael S. Tsirkin

2023-07-06 18:48 UTC #16 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 15 lines · Parav Pandit

Quoted 8 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 2:06 PM
But again even if you add it there, you can not claim it's exactly the same as legacy because the address is different, the address type is different, the driver is different (this is owner driver) and even the device is different.
You drafted below, that I included,
The driver of the owner device can send a driver notification to the member device by executing VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified.

Quoted 6 lines

We extend this further by saying Queue notify notifications is sent via MMIO as,
The driver of the owner device can send a driver notification to the member device by executing VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified or by performing memory or I/O write in

split to two sentences please. E.g.
Alternatively, the driver ...

Quoted 1 line

the any of the notification region at offset 0 supplied by the device in

the any is agrammatical

Quoted 11 lines

VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO result.
Quoted 6 lines
Please also call out the unusual configuration where the type is "member" and then you have the owner driver access the memory of the member device. People might be confused.
I also think we should explain that order of entries is a hint to driver: use the 1st entry that you can.
Driver really can choose any valid entry out of the 4 that driver likes.
I really don't see a need for overwriting this area as I fail to see why one will expose multiple entries from the device side in reality.

I do: one for owner one for member.
which is best for device is device specific.

I don't know what do you mean by "overwriting". Explaining in detail? You feel like this because you spent a lot of time thinking about the area and there is a specific solution is very clear in your head. First readers are not like this and second they might have a different solution.

Signature · 18 lines

MST


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/

Parav Pandit

author

2023-07-06 18:53 UTC #17 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 8 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 2:48 PM
Quoted 2 lines
The driver of the owner device can send a driver notification to the member device by executing

Quoted 8 lines

VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE
Quoted 3 lines
with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified or by performing memory or I/O write in
split to two sentences please. E.g.
Alternatively, the driver ...

Ack.
Moving to it dedicated para, as in this first para the notification region and command is not yet introduced.

Quoted 12 lines

Quoted 2 lines
the any of the notification region at offset 0 supplied by the device in
the any is agrammatical
Quoted 4 lines
VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO result.
Quoted 2 lines
Please also call out the unusual configuration where the type is "member" and then you have the owner driver access the memory of the
member device.
Quoted 2 lines
Quoted 2 lines
People might be confused.

Quoted 8 lines

Quoted 5 lines
Quoted 2 lines
I also think we should explain that order of entries is a hint to
driver: use the 1st entry that you can.
Driver really can choose any valid entry out of the 4 that driver likes. I really don't see a need for overwriting this area as I fail to see why one will
expose multiple entries from the device side in reality.
I do: one for owner one for member.

Yes, I am aware of choice or multi-entry. :)

Quoted 3 lines

which is best for device is device specific.
I don't know what do you mean by "overwriting". Explaining in detail?

:) being too verbose than needed.

Quoted 3 lines

You feel like this because you spent a lot of time thinking about the area and there is a specific solution is very clear in your head.
First readers are not like this and second they might have a different solution.

I am just saying to keep things simple. Let driver choose any entry it wants to use instead of device deciding on priority and hints etc.

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.

Michael S. Tsirkin

2023-07-06 18:56 UTC #18 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 43 lines · Parav Pandit

Quoted 8 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 2:48 PM
Quoted 2 lines
The driver of the owner device can send a driver notification to the member device by executing
Quoted 8 lines
VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE
Quoted 3 lines
with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified or by performing memory or I/O write in
split to two sentences please. E.g.
Alternatively, the driver ...
Ack.
Moving to it dedicated para, as in this first para the notification region and command is not yet introduced.
Quoted 12 lines
Quoted 2 lines
the any of the notification region at offset 0 supplied by the device in
the any is agrammatical
Quoted 4 lines
VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO result.
Quoted 2 lines
Please also call out the unusual configuration where the type is "member" and then you have the owner driver access the memory of the
member device.
Quoted 2 lines
Quoted 2 lines
People might be confused.
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
I also think we should explain that order of entries is a hint to
driver: use the 1st entry that you can.
Driver really can choose any valid entry out of the 4 that driver likes. I really don't see a need for overwriting this area as I fail to see why one will
expose multiple entries from the device side in reality.
I do: one for owner one for member.
Yes, I am aware of choice or multi-entry. :)
Quoted 2 lines
which is best for device is device specific.

Quoted 8 lines

Quoted 1 line
I don't know what do you mean by "overwriting". Explaining in detail?
:) being too verbose than needed.
Quoted 3 lines
You feel like this because you spent a lot of time thinking about the area and there is a specific solution is very clear in your head.
First readers are not like this and second they might have a different solution.
I am just saying to keep things simple. Let driver choose any entry it wants to use instead of device deciding on priority and hints etc.

Let's say driver can support both types which to choose? As the order is there anyway, why not just prescribe entries are used in order?

Signature · 6 lines

MST


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 19:00 UTC #19 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 8 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 2:57 PM
Quoted 1 line
I am just saying to keep things simple. Let driver choose any entry it wants to
use instead of device deciding on priority and hints etc.
Let's say driver can support both types which to choose?

Any entry that driver likes.

Quoted 1 line

As the order is there anyway, why not just prescribe entries are used in order?

I don't see any value in defining any order. It is an array of entries not a priority list.

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.

Michael S. Tsirkin

2023-07-06 19:43 UTC #20 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 5 lines · Parav Pandit

Quoted 5 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 2:57 PM
Quoted 1 line
I am just saying to keep things simple. Let driver choose any entry it wants to

Quoted 8 lines

Quoted 3 lines
use instead of device deciding on priority and hints etc.
Let's say driver can support both types which to choose?
Any entry that driver likes.
Quoted 1 line
As the order is there anyway, why not just prescribe entries are used in order?
I don't see any value in defining any order. It is an array of entries not a priority list.

I think we are losing out. For example I can see how access through member would be preferable for ordering reasons. However device might still allow access through PF for cases where driver can't access VF.

But I don't see any configs where leaving this to the driver's discretion is preferable. If you see one let me know.

Signature · 6 lines

MST


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 20:21 UTC #21 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 7 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 3:43 PM
Quoted 1 line
Quoted 1 line
As the order is there anyway, why not just prescribe entries are used in
order?
Quoted 1 line

Quoted 8 lines

Quoted 1 line
I don't see any value in defining any order. It is an array of entries not a priority
list.
I think we are losing out. For example I can see how access through member would be preferable for ordering reasons.
However device might still allow access through PF for cases where driver can't access VF.

So a driver can choose say I prefer order over accessibility over VF, so it choose PF. Device doesn't have the knowledge anyway whether driver can/cannot access the VF. So, device's preference vs driver's preference may be different.

Quoted 2 lines

But I don't see any configs where leaving this to the driver's discretion is preferable. If you see one let me know.

In doesn't need to be config.
It is the environment that chooses which is preferred by the driver. For example preference of accessibility over ordering.

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.

Michael S. Tsirkin

2023-07-06 20:28 UTC #22 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 10 lines · Parav Pandit

Quoted 10 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 3:43 PM
Quoted 1 line
Quoted 1 line
As the order is there anyway, why not just prescribe entries are used in
order?
Quoted 2 lines
I don't see any value in defining any order. It is an array of entries not a priority
list.

Quoted 8 lines

Quoted 5 lines
I think we are losing out. For example I can see how access through member would be preferable for ordering reasons.
However device might still allow access through PF for cases where driver can't access VF.
So a driver can choose say I prefer order over accessibility over VF, so it choose PF. Device doesn't have the knowledge anyway whether driver can/cannot access the VF. So, device's preference vs driver's preference may be different.

The driver has a final decision. Let's make it a SHOULD and then if driver knows best then it has the choice?

Quoted 6 lines

Quoted 2 lines
But I don't see any configs where leaving this to the driver's discretion is preferable. If you see one let me know.
In doesn't need to be config.
It is the environment that chooses which is preferred by the driver. For example preference of accessibility over ordering.

what does accessibility mean exactly? I definitely see OSes where owner driver can't access a member. So in that case naturally driver will skip the entry for member even if it's first. maybe there are configs where member access is possible but is very slow e.g. with lots of indirect function calls?
OK fine, but then it will be up to the driver to test and make damn sure the benefits outweight the costs.

IOW it's a hint for the driver. If you like you can say it explicitly even.

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.

Parav Pandit

author

2023-07-06 20:35 UTC #23 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 7 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 4:29 PM
The driver has a final decision. Let's make it a SHOULD and then if driver knows best then it has the choice?

As you said, the driver has the final say.
There is really no reason to complicate the spec for a narrow case where device cannot judge.

Quoted 11 lines

Quoted 7 lines
Quoted 3 lines
But I don't see any configs where leaving this to the driver's discretion is preferable. If you see one let me know.
In doesn't need to be config.
It is the environment that chooses which is preferred by the driver. For example preference of accessibility over ordering.
what does accessibility mean exactly? I definitely see OSes where owner driver can't access a member.

Accessibility = access a member

Quoted 8 lines

So in that case naturally driver will skip the entry for member even if it's first. maybe there are configs where member access is possible but is very slow e.g. with lots of indirect function calls?
OK fine, but then it will be up to the driver to test and make damn sure the benefits outweight the costs.
IOW it's a hint for the driver. If you like you can say it explicitly even.

Device doesnt know anything about those indirect function calls, so device cannot hint about driver environment.

Can we please avoid this over engineering?
Interface has the doors open for driver to make wise decision depending on its environment.

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.

Michael S. Tsirkin

2023-07-06 20:41 UTC #24 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 30 lines · Parav Pandit

Quoted 7 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 4:29 PM
The driver has a final decision. Let's make it a SHOULD and then if driver knows best then it has the choice?
As you said, the driver has the final say.
There is really no reason to complicate the spec for a narrow case where device cannot judge.
Quoted 11 lines
Quoted 7 lines
Quoted 3 lines
But I don't see any configs where leaving this to the driver's discretion is preferable. If you see one let me know.
In doesn't need to be config.
It is the environment that chooses which is preferred by the driver. For example preference of accessibility over ordering.
what does accessibility mean exactly? I definitely see OSes where owner driver can't access a member.
Accessibility = access a member
Quoted 6 lines
So in that case naturally driver will skip the entry for member even if it's first. maybe there are configs where member access is possible but is very slow e.g. with lots of indirect function calls?
OK fine, but then it will be up to the driver to test and make damn sure the benefits outweight the costs.

Quoted 6 lines

Quoted 2 lines
IOW it's a hint for the driver. If you like you can say it explicitly even.
Device doesnt know anything about those indirect function calls, so device cannot hint about driver environment.
Can we please avoid this over engineering?
Interface has the doors open for driver to make wise decision depending on its environment.

what if driver can access both with the same ease? this is the case that bothers me and I think it's practical since it will be common on linux.

Signature · 18 lines

MST


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/

Parav Pandit

author

2023-07-06 20:43 UTC #25 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 3 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 4:42 PM

[..]

Quoted 6 lines

Quoted 2 lines
Can we please avoid this over engineering?
Interface has the doors open for driver to make wise decision depending on
its environment.
what if driver can access both with the same ease? this is the case that bothers me and I think it's practical since it will be common on linux.

Then Linux can say my preference is order, so it picks member.

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.

Cornelia Huck

2023-07-07 08:12 UTC #26 lore raw

Re: [virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 4 lines · Parav Pandit

Quoted 2 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 4:42 PM
[..]

Quoted 8 lines

Quoted 6 lines
Quoted 2 lines
Can we please avoid this over engineering?
Interface has the doors open for driver to make wise decision depending on
its environment.
what if driver can access both with the same ease? this is the case that bothers me and I think it's practical since it will be common on linux.
Then Linux can say my preference is order, so it picks member.

Let's step back and consider what our goal is here: to provide a spec that someone wanting to implement a driver can follow and end up with a driver that works with devices that adhere to this spec.

I don't think expecting the driver to make "wise descisions" is helpful for the person wanting to implement a driver. "Do this, and in case this does not work in your environment, do that" seems much more helpful, and still gives enough flexibility to cover different environments. In short, make things simple for the consumer of the spec.

[Also, I notice that the discussion seems to get bogged down in tiny details, fundamental statements, or a mixture of both. It might sometimes be helpful to just step back, re-read the original proposal, and only then answer. Otherwise, this leads to frustration for everyone involved.]

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.

Michael S. Tsirkin

2023-07-07 08:32 UTC #27 lore raw

Re: [virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 5 lines · Cornelia Huck

[Also, I notice that the discussion seems to get bogged down in tiny details, fundamental statements, or a mixture of both. It might sometimes be helpful to just step back, re-read the original proposal, and only then answer. Otherwise, this leads to frustration for everyone involved.]

I feel in this instance we are actually good. All that's left is small details that's why we are discussing them.

Signature · 6 lines

MST


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Michael S. Tsirkin

2023-07-06 17:38 UTC #28 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

In reply to Parav Pandit

Quoted 50 lines · Parav Pandit

Quoted 5 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:42 PM
Quoted 7 lines
We will need to detail how notifications work though. The reason is that the legacy notifications are received by owner driver (which is not legacy!) and forwarded to device.
We need to explain exactly how this is done for people who don't read minds and did not follow 3 months of mailing list debates, and explain how this has the same effect as triggering driver notification through the legacy interface.
I am not going to write driver implementation in the spec here. But I agree some high-level text is needed.
With your suggestions
1. to merge pci and generic in one section,
2. add description to notification,
3. past rewrites provided in v9,
The draft of v11, looks like below. Please see how much more verbose to make it, if at all.
\paragraph{Legacy Driver Notification}\label{par:Basic Facilities of a Virtio Device / Device groups / Group administration commands / Legacy Interface / Legacy Driver Notifications}
If the group owner device or the group member device support driver notifications via a memory-mapped operation or I/O operation, these notifications are sent to the device via accessing such a notification region using a memory or I/O operation instead of sending the notifications through the administration command.
The driver of the owner device can send a driver notification to the member device by executing VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE with the \field{offset} matching \field{Queue Notify} and the \field{data} containing the virtqueue index to be notified.
However, as many administration commands are used for slow path configuration, a separate fast path mechanism for such notifications is desired. For the SR-IOV group type, the optional command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO addresses this need by returning to the driver one or more addresses which to be used to send such driver notifications.
In one example, the driver in the hypervisor which intercepts I/O BAR accesses for the \field{Queue Notify} can access the notification area directly as memory or I/O access (depending on what the device reported) instead of slow administration command.
For the command VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO the driver sets \field{opcode} to 0x6.
The driver sets \field{group_member_id} which refers to the member device to be accessed. This command does not use \field{command_specific_data}.

Quoted 5 lines

This command is currently only defined for the SR-IOV group type. When the
device supports VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_INFO command, the group owner
device hardwires VF BAR0 to zero in the SR-IOV Extended capability and the
group member device does not use PCI BAR0 in all the Virtio PCI capabilities
listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.

lots of comments wrt grammar. do post v11 and i will fix.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]

Cornelia Huck

2023-07-06 16:47 UTC #29 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

In reply to Parav Pandit

Quoted 21 lines · Parav Pandit

Quoted 21 lines
From: Cornelia Huck <[email protected]>
Sent: Thursday, July 6, 2023 12:28 PM
transport-pci-legacy-regs.tex line 7
Quoted 13 lines · Parav Pandit
(quoted diff) +0 -0
b/transport-pci-legacy-regs.tex new file mode 100644 index 0000000..ceea28c
transport-pci-legacy-regs.tex new file +7 -0
@@ -0,0 +1,42 @@
+\subsection{Legacy Interface: Group member device Configuration
+Region Access}\label{sec:Virtio Transport Options / Virtio Over PCI
+Bus / Legacy Interface: Group Member Device Configuration Region
+Access}
+
+The PCI owner device or the member device or both support driver
+notifications using
What about

Quoted 5 lines

Quoted 2 lines
"The PCI owner device, the member device, or both can choose to support..." ?
Fine too.
Any thing wrong in having or as above, so I don't write it next time? Or that in current form reads better to me.

Just a bare "support" does not really tell the reader if this is something that is required or optional. Dropping the first "or" makes it read better for me.

Quoted 3 lines

Quoted 3 lines
transport-pci-legacy-regs.tex line 8
Quoted 2 lines
transport-pci-legacy-regs.tex new file +1 -0
+a notification region defined in \field{struct
virtio_pci_legacy_notify_region}.

(...)

Quoted 6 lines

Quoted 5 lines
transport-pci-legacy-regs.tex
Quoted 2 lines
transport-pci-legacy-regs.tex new file +2 -0
+The group owner device or the group member device or both MAY support
+driver notifications region.
Make this "a driver notification region"?
Notifications are generally more than one and spec has the section "driver notifications", so...

I'd parse this as "a region for the purpose of notification" (and "notification region" is used above)... but in any case, we need the article here, I think.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 16:53 UTC #30 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 29 lines

From: Cornelia Huck <[email protected]>
Sent: Thursday, July 6, 2023 12:47 PM
Quoted 21 lines · Parav Pandit
Quoted 21 lines
From: Cornelia Huck <[email protected]>
Sent: Thursday, July 6, 2023 12:28 PM
transport-pci-legacy-regs.tex line 7
Quoted 13 lines · Parav Pandit
(quoted diff) +0 -0
b/transport-pci-legacy-regs.tex new file mode 100644 index 0000000..ceea28c
transport-pci-legacy-regs.tex new file +7 -0
@@ -0,0 +1,42 @@
+\subsection{Legacy Interface: Group member device Configuration
+Region Access}\label{sec:Virtio Transport Options / Virtio Over
+PCI Bus / Legacy Interface: Group Member Device Configuration
+Region Access}
+
+The PCI owner device or the member device or both support driver
+notifications using
What about
Quoted 1 line
Quoted 1 line
"The PCI owner device, the member device, or both can choose to

Quoted 9 lines

support..." ?
Quoted 4 lines
Quoted 1 line
Fine too.
Any thing wrong in having or as above, so I don't write it next time? Or that in current form reads better to me.
Just a bare "support" does not really tell the reader if this is something that is required or optional. Dropping the first "or" makes it read better for me.

Since this is not normative, Iwill write it as,

The PCI owner device or the member device or both can support

Quoted 11 lines

Quoted 3 lines
Quoted 3 lines
transport-pci-legacy-regs.tex line 8
Quoted 2 lines
transport-pci-legacy-regs.tex new file +1 -0
+a notification region defined in \field{struct
virtio_pci_legacy_notify_region}.
(...)
Quoted 5 lines
Quoted 5 lines
transport-pci-legacy-regs.tex
Quoted 2 lines
transport-pci-legacy-regs.tex new file +2 -0
+The group owner device or the group member device or both MAY
+support driver notifications region.
Make this "a driver notification region"?

Quoted 5 lines

Quoted 1 line
Notifications are generally more than one and spec has the section "driver
notifications", so...
I'd parse this as "a region for the purpose of notification" (and "notification region" is used above)... but in any case, we need the article here, I think.

Ah ok. I was missing the article, I was thinking to write "driver notifications" vs "driver notification". Will change to "a driver notifications region" as the article applies to region.

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.

Michael S. Tsirkin

2023-07-06 16:39 UTC #31 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

In reply to Cornelia Huck

Quoted 50 lines · Cornelia Huck

transport-pci-legacy-regs.tex line 3

Quoted 9 lines · Parav Pandit

transport-pci-legacy-regs.tex new file +3 -0

new file mode 100644
index 0000000..ceea28c
@@ -0,0 +1,42 @@
+\subsection{Legacy Interface: Group member device Configuration Region Access}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}
+
+The PCI owner device or the member device or both support driver notifications using

What about

"The PCI owner device, the member device, or both can choose to support..." ?

Quoted 8 lines

transport-pci-legacy-regs.tex new file +8 -0

+a notification region defined in \field{struct virtio_pci_legacy_notify_region}.
+
+In \field{struct virtio_virtio_admin_cmd_legacy_notify_query_entry},
+\field{region_data} is defined as following:
+
+\begin{lstlisting}
+struct virtio_pci_legacy_notify_region {
+        u8 owner;  /* When set to 1, notification region is of the owner device */

transport-pci-legacy-regs.tex line 21

Quoted 10 lines

transport-pci-legacy-regs.tex new file +10 -0

+        u8 bar;    /* BAR of the member or owner device */
+        u8 padding[6];
+        le64 offset; /* Offset within bar. */
+};
+\end{lstlisting}
+
+The group owner device hardwires VF BAR0 to zero in the SR-IOV Extended
+capability.
+
+The group member device does not use PCI BAR0 in all the Virtio PCI capabilities

"in [or for?] any of the...", I guess?

transport-pci-legacy-regs.tex line 30

Quoted 9 lines

transport-pci-legacy-regs.tex new file +9 -0

+listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.
+
+\devicenormative{\subsubsection}{Legacy Interface: Group Member Device Legacy Configuration Region Access}{Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}
+
+When a PCI SR-IOV group owner device supports
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE, VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE commands, the group owner device MUST
+hardwire VF BAR0 to zero in the SR-IOV Extended capability and the group memberi

s/memberi/member/

Quoted 6 lines

transport-pci-legacy-regs.tex line 34
Quoted 4 lines
transport-pci-legacy-regs.tex new file +4 -0
+device MUST NOT use BAR0 in any of the Virtio Structure PCI Capabilities.
+
+The group owner device or the group member device or both MAY support driver
+notifications region.
Make this "a driver notification region"?

I think in fact for conformance we can just refer to supporting the command. The command can return an array of structures, each referring to the memory of the owner device or the member device.

Quoted 6 lines

transport-pci-legacy-regs.tex line 40

Quoted 6 lines

transport-pci-legacy-regs.tex new file +6 -0

+
+For the SR-IOV group type, the owner device supporting
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE, VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
+commands and its member device SHOULD follow the rules for the PCI Device ID,

what does "its" mean here? Of the group?
Members of the SR-IOV group type are VFs. They can not follow the rules for the Device ID: the spec says:
This field in all VFs returns FFFFh when read.

Even if you somehow refer to the software it's extraneous anyway since the spec proceeds: VI software should return the Vendor ID value from the associated PF as the Vendor ID value for the VF.

We'll need a separate statement for Device ID.

Quoted 2 lines

transport-pci-legacy-regs.tex line 42
Quoted 2 lines
transport-pci-legacy-regs.tex new file +2 -0
+Revision ID and Subsystem Device ID of the non-transitional devices documented in
+section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Discovery}.

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.

Parav Pandit

author

2023-07-06 16:45 UTC #32 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 5 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:39 PM

Quoted 9 lines

Quoted 4 lines
Review
Quoted 2 lines
(quoted diff) +2 -0
+The group owner device or the group member device or both MAY
+support driver notifications region.
Make this "a driver notification region"?
I think in fact for conformance we can just refer to supporting the command. The command can return an array of structures, each referring to the memory of the owner device or the member device.

Array and referring to owner/member is already present in this patch.

Do you suggest to drop above statement?

Quoted 12 lines

Quoted 10 lines

Review

Quoted 10 lines

(quoted diff) +9 -0

+
+For the SR-IOV group type, the owner device supporting
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
+commands and its member device SHOULD follow the rules for the PCI
+Device ID,

what does "its" mean here? Of the group?

Yes, will change to the member device of the group.

Quoted 10 lines

Members of the SR-IOV group type are VFs. They can not follow the rules for the Device ID: the spec says:
This field in all VFs returns FFFFh when read.
Even if you somehow refer to the software it's extraneous anyway since the spec proceeds: VI software should return the Vendor ID value from the associated PF as the Vendor ID value for the VF.
We'll need a separate statement for Device ID.

Will split the line for Device ID.

Quoted 3 lines

Quoted 2 lines
Review
Quoted 2 lines
(quoted diff) +2 -0
+Revision ID and Subsystem Device ID of the non-transitional devices
+documented in section \ref{sec:Virtio Transport Options / Virtio Over PCI
Bus / PCI Device Discovery}.

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.

Michael S. Tsirkin

2023-07-06 16:51 UTC #33 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 10 lines · Parav Pandit

Quoted 10 lines
From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:39 PM
Quoted 4 lines
Review
Quoted 2 lines
(quoted diff) +2 -0
+The group owner device or the group member device or both MAY
+support driver notifications region.
Make this "a driver notification region"?

Quoted 5 lines

Quoted 4 lines
I think in fact for conformance we can just refer to supporting the command. The command can return an array of structures, each referring to the memory of the owner device or the member device.
Array and referring to owner/member is already present in this patch.

I don't see a statement that devices MAY support VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY with entries returning just owner, just member, or both.

Quoted 20 lines

Do you suggest to drop above statement?
Quoted 12 lines
Quoted 10 lines
Review
Quoted 10 lines
(quoted diff) +9 -0
+
+For the SR-IOV group type, the owner device supporting
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
+commands and its member device SHOULD follow the rules for the PCI
+Device ID,
what does "its" mean here? Of the group?
Yes, will change to the member device of the group.
Quoted 4 lines
Members of the SR-IOV group type are VFs. They can not follow the rules for the Device ID: the spec says:
This field in all VFs returns FFFFh when read.

Quoted 11 lines

Quoted 6 lines
Even if you somehow refer to the software it's extraneous anyway since the spec proceeds: VI software should return the Vendor ID value from the associated PF as the Vendor ID value for the VF.
We'll need a separate statement for Device ID.
Will split the line for Device ID.
Quoted 3 lines
Quoted 2 lines
Review
Quoted 2 lines
(quoted diff) +2 -0
+Revision ID and Subsystem Device ID of the non-transitional devices
+documented in section \ref{sec:Virtio Transport Options / Virtio Over PCI
Bus / PCI Device Discovery}.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 16:54 UTC #34 lore raw

[virtio-comment] RE: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 10 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 12:51 PM
I don't see a statement that devices MAY support VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
with entries returning just owner, just member, or both.

Ok. understood.
From the driver notifications wording it didn't imply that it is related to VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY. Will rewrite as above.

Quoted 24 lines

Quoted 24 lines
Do you suggest to drop above statement?
Quoted 12 lines
Quoted 10 lines
Review
Quoted 10 lines
(quoted diff) +9 -0
+
+For the SR-IOV group type, the owner device supporting
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and
+VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
+commands and its member device SHOULD follow the rules for the
+PCI Device ID,
what does "its" mean here? Of the group?
Yes, will change to the member device of the group.
Quoted 8 lines
Members of the SR-IOV group type are VFs. They can not follow the rules for the Device ID: the spec says:
This field in all VFs returns FFFFh when read.
Even if you somehow refer to the software it's extraneous anyway since the spec proceeds: VI software should return the Vendor ID value from the associated PF as the Vendor ID value for the VF.

Quoted 8 lines

Quoted 8 lines
Quoted 2 lines
We'll need a separate statement for Device ID.
Will split the line for Device ID.
Quoted 4 lines
Quoted 3 lines
Review
Quoted 3 lines
(quoted diff) +3 -0
+Revision ID and Subsystem Device ID of the non-transitional
+devices documented in section \ref{sec:Virtio Transport Options
+/ Virtio Over PCI
Bus / PCI Device Discovery}.

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.

Michael S. Tsirkin

2023-07-06 19:00 UTC #35 lore raw

[virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

In reply to Parav Pandit

Quoted 110 lines · Parav Pandit

This patch links how in a PCI transport a group owner can access group member (PCI VFs) legacy registers using a legacy registers access commands using administration virtqueue infrastructure.
Additionally it extend the PCI notification capability through which a PCI VF device indicates to the driver which PCI BAR region to be used for driver notifications.
Fixes https://github.com/oasis-tcs/virtio-spec/issues/167
Signed-off-by Parav Pandit <[email protected]>
---
changelog:
v9->v10:
- addressed comments from Cornelia
- fixed errors related to article
- hardwire to hardwires
- replaced various to all
- added hardwire to zero
v7->v8:
- addressed comments from Michael
- made bar offset 64-bit
- prefix legacy specific structure with _legacy
- moved generic normative from pci to generic section
- added link to virtio pci capabilities when referring to bar 0
- remove 'should' from generic description
v6->v7:
- addressed comments from Michael
- removed driver normative about I/O BAR emulation as it does not make much sense for the spec
- removed references to administration virtqueue
- rewrote driver legacy region access without guest and hypervisor wording
- added normative for notification region
- added normative for PCI IDs for device which support legacy commands v5->v6:
- aligned pci capability to 4B as required by PCI spec
- added text for the PCI capability for the group member device v4->v5:
- split pci transport and generic command section to new patch
- removed multiple references to the VF
- written the description of the command as generic with member and group device terminology
- reflected many section names to remove VF
v3->v4:
- moved noted to the conformance section details in next patch
- removed queue notify address query AQ command on Michael's suggestion, though it is fine. Instead replaced with extending virtio_pci_notify_cap to indicate that legacy queue notifications can be done on the notification location.
- fixed spelling errors.
- replaced administrative virtqueue to administration virtqueue
- added queue notification capability register to indicate legacy q notification supported
v2->v3:
- adddressed Jason and Michael's comment to split single register access command to common config and device specific commands.
- dropped the suggetion to introduce enable/disable command as admin command cap bit already covers it.
v1->v2:
- addressed comments from Michael
- added theory of operation
- grammar corrections
- removed group fields description from individual commands as it is already present in generic section
- added endianness normative for legacy device registers region
- renamed the file to drop vf and add legacy prefix
- added overview in commit log
- renamed subsection to reflect command
---
 admin-cmds-legacy-interface.tex |  3 ++-
 conformance.tex                 |  1 +
 transport-pci-legacy-regs.tex   | 42 +++++++++++++++++++++++++++++++++
 transport-pci.tex               |  2 ++
 4 files changed, 47 insertions(+), 1 deletion(-)
 create mode 100644 transport-pci-legacy-regs.tex
admin-cmds-legacy-interface.tex +2 -1
index dd01c0a..e67632b 100644
@@ -187,7 +187,8 @@ \subsubsection{Legacy Interfaces}\label{sec:Basic Facilities of a Virtio Device
 by the device.
 
 Refer to the specific transport section for the definition of the
-\field{data}.
+\field{data}. For the PCI transport refer to section
+\ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}.
 
 This command is currently only defined for the SR-IOV group type.
 
conformance.tex +1 -0
index dc00e84..b3f2c92 100644
@@ -267,6 +267,7 @@ \section{Conformance Targets}\label{sec:Conformance / Conformance Targets}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Layout / Legacy Interfaces: A Note on PCI Device Layout}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI-specific Initialization And Device Operation / Device Initialization / Virtio Device Configuration Layout Detection / Legacy Interface: A Note on Device Layout Detection}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI-specific Initialization And Device Operation / Device Initialization / Virtqueue Configuration / Legacy Interface: A Note on Virtqueue Configuration}
+\item Section \ref{devicenormative:Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}
 \item Section \ref{sec:Virtio Transport Options / Virtio Over MMIO / Legacy interface}
 \item Section \ref{sec:Virtio Transport Options / Virtio over channel I/O / Device Initialization / Setting the Virtio Revision / Legacy Interfaces: A Note on Setting the Virtio Revision}
 \item Section \ref{sec:Virtio Transport Options / Virtio over channel I/O / Device Initialization / Configuring a Virtqueue / Legacy Interface: A Note on Configuring a Virtqueue}
transport-pci-legacy-regs.tex new file +1 -0
new file mode 100644
index 0000000..ceea28c
@@ -0,0 +1,42 @@
+\subsection{Legacy Interface: Group member device Configuration Region Access}\label{sec:Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}

transport-pci-legacy-regs.tex line 11

Quoted 10 lines

transport-pci-legacy-regs.tex new file +10 -0

+
+The PCI owner device or the member device or both support driver notifications using
+a notification region defined in \field{struct virtio_pci_legacy_notify_region}.
+
+In \field{struct virtio_virtio_admin_cmd_legacy_notify_query_entry},
+\field{region_data} is defined as following:
+
+\begin{lstlisting}
+struct virtio_pci_legacy_notify_region {
+        u8 owner;  /* When set to 1, notification region is of the owner device */

I propose we rename this to flags:

	0x0 - end of list (driver should ignore following values until end of list)
	0x1 - owner
	0x2 - member

other values - reserved

and prescribe that driver ignores any reserved values.

basically we are following how the virtio capabilities work.

transport-pci-legacy-regs.tex line 22

Quoted 11 lines

transport-pci-legacy-regs.tex new file +11 -0

+        u8 bar;    /* BAR of the member or owner device */
+        u8 padding[6];
+        le64 offset; /* Offset within bar. */
+};
+\end{lstlisting}
+
+The group owner device hardwires VF BAR0 to zero in the SR-IOV Extended
+capability.
+
+The group member device does not use PCI BAR0 in all the Virtio PCI capabilities
+listed in section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / Virtio Structure PCI Capabilities}.

Just drop this last one. How can they use it if there's no VF BAR0?

Quoted 32 lines

transport-pci-legacy-regs.tex new file +20 -0

+
+\devicenormative{\subsubsection}{Legacy Interface: Group Member Device Legacy Configuration Region Access}{Virtio Transport Options / Virtio Over PCI Bus / Legacy Interface: Group Member Device Configuration Region Access}
+
+When a PCI SR-IOV group owner device supports
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE, VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE commands, the group owner device MUST
+hardwire VF BAR0 to zero in the SR-IOV Extended capability and the group memberi
+device MUST NOT use BAR0 in any of the Virtio Structure PCI Capabilities.
+
+The group owner device or the group member device or both MAY support driver
+notifications region.
+
+For the SR-IOV group type, the owner device supporting
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_COMMON_CFG_WRITE, VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_READ,
+VIRTIO_ADMIN_CMD_LEGACY_DEV_CFG_WRITE and VIRTIO_ADMIN_CMD_LEGACY_NOTIFY_QUERY
+commands and its member device SHOULD follow the rules for the PCI Device ID,
+Revision ID and Subsystem Device ID of the non-transitional devices documented in
+section \ref{sec:Virtio Transport Options / Virtio Over PCI Bus / PCI Device Discovery}.

transport-pci.tex +2 -0

index a5c6719..72c78f6 100644
@@ -1212,3 +1212,5 @@ \subsubsection{Driver Handling Interrupts}\label{sec:Virtio Transport Options /
         re-examine the configuration space to see what changed.
     \end{itemize}
 \end{itemize}
+
+\input{transport-pci-legacy-regs.tex}

Signature · 1 line

2.26.2

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.

Parav Pandit

author

2023-07-06 19:07 UTC #36 lore raw

RE: [virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 13 lines

From: [email protected] <[email protected] open.org> On Behalf Of Michael S. Tsirkin
Sent: Thursday, July 6, 2023 3:01 PM
Review
Quoted 4 lines
+\begin{lstlisting}
(quoted diff) +3 -0
+struct virtio_pci_legacy_notify_region {
+        u8 owner;  /* When set to 1, notification region is of the
+owner device */
I propose we rename this to flags:
	0x0 - end of list (driver should ignore following values until end of list)
	0x1 - owner
	0x2 - member

Quoted 7 lines

other values - reserved
and prescribe that driver ignores any reserved values.
basically we are following how the virtio capabilities work.

Looks good. Will change it.

Quoted 9 lines

Review
Quoted 5 lines
(quoted diff) +5 -0
+The group owner device hardwires VF BAR0 to zero in the SR-IOV
+Extended capability.
+
+The group member device does not use PCI BAR0 in all the Virtio PCI
+capabilities listed in section \ref{sec:Virtio Transport Options / Virtio Over
PCI Bus / Virtio Structure PCI Capabilities}.
Just drop this last one. How can they use it if there's no VF BAR0?

Hardwaring BAR0 is in BAR area of the VF.
But virtio pci capabilities without this can still report bar 0 and virtio capabilities.
Ofcourse, it is a broken device.
But skipping this line imply that "because VF BAR0 is hardwired", this metadata in pci capability must not expose it.

Why not write it extra thing instead of implying it?
We already wrote few duplicate things to make reader life easier.

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.

Michael S. Tsirkin

2023-07-06 19:59 UTC #37 lore raw

Re: [virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 29 lines · Parav Pandit

Quoted 13 lines
From: [email protected] <[email protected] open.org> On Behalf Of Michael S. Tsirkin
Sent: Thursday, July 6, 2023 3:01 PM
Review
Quoted 4 lines
+\begin{lstlisting}
(quoted diff) +3 -0
+struct virtio_pci_legacy_notify_region {
+        u8 owner;  /* When set to 1, notification region is of the
+owner device */
I propose we rename this to flags:
	0x0 - end of list (driver should ignore following values until end of list)
	0x1 - owner
	0x2 - member
Quoted 7 lines
other values - reserved
and prescribe that driver ignores any reserved values.
basically we are following how the virtio capabilities work.
Looks good. Will change it.
Quoted 7 lines
Review
Quoted 5 lines
(quoted diff) +5 -0
+The group owner device hardwires VF BAR0 to zero in the SR-IOV
+Extended capability.
+
+The group member device does not use PCI BAR0 in all the Virtio PCI
+capabilities listed in section \ref{sec:Virtio Transport Options / Virtio Over
PCI Bus / Virtio Structure PCI Capabilities}.

Quoted 4 lines

Quoted 2 lines
Just drop this last one. How can they use it if there's no VF BAR0?
Hardwaring BAR0 is in BAR area of the VF.
But virtio pci capabilities without this can still report bar 0 and virtio capabilities.

How do they do this?

Quoted 5 lines

Ofcourse, it is a broken device.
But skipping this line imply that "because VF BAR0 is hardwired", this metadata in pci capability must not expose it.
Why not write it extra thing instead of implying it?
We already wrote few duplicate things to make reader life easier.

If you really want to go ahead, but prefix it with "In consequence" and do not start a new paragraph, so reader knows it's not an extra requirement.

So I started writing it using correct grammar and spec terminology:

In consequence, non of the group member devices has BAR0, and in particular none of the virtio structure capabilities of a member device has \field{bar} with the value of 0.

However, this actually is kind of wrong (and so is your text). Not all caps have a RO bar value. virtio_pci_cfg_cap has bar that is RW by driver. So if we are going this route we also need to explain that it's true for all caps except virtio_pci_cfg_cap. And for virtio_pci_cfg_cap driver is not allowed to write 0 there.

Frankly too much trouble but if you want to, keep trying.

Signature · 6 lines

MST


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Parav Pandit

author

2023-07-06 20:28 UTC #38 lore raw

RE: [virtio-comment] Re: [PATCH v10 4/4] transport-pci: Introduce group legacy group member config region access

Quoted 10 lines

From: Michael S. Tsirkin <[email protected]>
Sent: Thursday, July 6, 2023 3:59 PM
Quoted 2 lines
Hardwaring BAR0 is in BAR area of the VF.
But virtio pci capabilities without this can still report bar 0 and virtio
capabilities.
How do they do this?

Device must not do it, but a broken device can report pci capabilities as garbage.
I will skip writing about it.

Quoted 17 lines

Quoted 2 lines
Ofcourse, it is a broken device.
But skipping this line imply that "because VF BAR0 is hardwired", this
metadata in pci capability must not expose it.
Quoted 3 lines
Why not write it extra thing instead of implying it?
We already wrote few duplicate things to make reader life easier.
If you really want to go ahead, but prefix it with "In consequence" and do not start a new paragraph, so reader knows it's not an extra requirement.
So I started writing it using correct grammar and spec terminology:
In consequence, non of the group member devices has BAR0, and in particular none of the virtio structure capabilities of a member device has \field{bar} with the value of 0.

Quoted 7 lines

However, this actually is kind of wrong (and so is your text). Not all caps have a RO bar value. virtio_pci_cfg_cap has bar that is RW by driver. So if we are going this route we also need to explain that it's true for all caps except virtio_pci_cfg_cap. And for virtio_pci_cfg_cap driver is not allowed to write 0 there.
Frankly too much trouble but if you want to, keep trying.

What I wrote still holds true because device wont have BAR0 and driver writing there is ignored. But it is some weird broken device who expose PCI capabilities, so I am going to skip writing this.

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.