CA-370578 use subsystemId in NVidia GPU matching #4826
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
This applies cleanly to the Yangtze branch for a potential backport. |
edwintorok
approved these changes
Oct 18, 2022
psafont
reviewed
Oct 18, 2022
psafont
approved these changes
Oct 18, 2022
|
Added comments as suggested. Will squash once accepted. |
So far, we identified a GPU by its deviceId; this turns out to be
ambiguous and vgpuConfig.xml may contain multiple GPUs with identical
deviceId, causing us to use more vGPU types than we can actually
support.
This patch introduces a new abstract type "gpu" in the VENDOR signature
such that its Nvidia instance can use more information to identify a
GPU.
When matching a GPU found on the PCI bus with GPUs listed in
vgpuConfig.xml we use:
deviceId match and subsystemId match
or deviceId match and subsystemId in vgpuConfig is 0
subsystemId=0 is used by Nvidia as a wildcard and thus subsystemId=0 is
ignored during a match.
The matching for other vendors remains unchanged (and is still based on
deviceId).
Signed-off-by: Christian Lindig <christian.lindig@citrix.com>
lindig
force-pushed
the
private/christianlin/CA-370578
branch
from
48a75b9
to
eb855fb
Compare
robhoes
approved these changes
Oct 18, 2022
| @@ -410,6 +413,16 @@ module Vendor_nvidia = struct | |||
| ; compatible_model_names_on_pgpu: string list | |||
| } | |||
|
|
|||
| type gpu = {device_id: int; subsys_id: int} | |||
There was a problem hiding this comment.
So we don't need the subsystem vendor id at this point?
There was a problem hiding this comment.
We do need it for Nvidia - that is the entire point of the PR.
There was a problem hiding this comment.
The is a subsystem device ID, and a subsystem vendor ID as well, which is not in the type now.
There was a problem hiding this comment.
Right. We don't use the subsystem vendor ID right now but could easily extend this to use it as well. We would expect it to match exactly (no wildcard mechanism).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
So far, we identified a GPU by its deviceId; this turns out to be ambiguous and vgpuConfig.xml may contain multiple GPUs with identical deviceId, causing us to use more vGPU types than we can actually support.
This patch introduces a new abstract type "gpu" in the VENDOR signature such that its Nvidia instance can use more information to identify a GPU. This is the core of this patch and fix.
When matching a GPU found on the PCI bus with GPUs listed in vgpuConfig.xml we use:
or deviceId match and subsystemId in vgpuConfig is 0
subsystemId=0 is used by Nvidia as a wildcard and thus subsystemId=0 is ignored during a match.
The matching for other vendors remains unchanged (and is still based on deviceId).
Signed-off-by: Christian Lindig christian.lindig@citrix.com