From: "Christoph Heiss" <c.heiss@proxmox.com>
To: "Dominik Csapak" <d.csapak@proxmox.com>
Cc: pve-devel@lists.proxmox.com
Subject: Re: [PATCH common/debcargo-conf/manager/proxmox-perl-rs/qemu-server 00/13] use NVML for vGPU info querying
Date: Thu, 19 Mar 2026 12:31:13 +0100 [thread overview]
Message-ID: <DH6Q43WRQQYO.2JRZNE4EFN2MR@proxmox.com> (raw)
In-Reply-To: <20260305091711.1221589-1-d.csapak@proxmox.com>
Built and tested on our RTX A5000 rig.
Works as advertised - the vGPU type descriptions are now actually shown
in the mdev type selector dropdown.
Please consider the series:
Reviewed-by: Christoph Heiss <c.heiss@proxmox.com>
Tested-by: Christoph Heiss <c.heiss@proxmox.com>
On Thu Mar 5, 2026 at 10:16 AM CET, Dominik Csapak wrote:
> This series replaces Christoph's previous series[0]
>
> It's more or less a rewrite, thus no v2 or similar, but some patches directly
> take code from Christoph's attempt, so i marked those commits as
> co-developed-by.
>
> Changes:
>
> Instead of using our own wrapper+high level abstraction use the already existing
> 'nvml-wrapper' and 'nvml-wrapper-sys' which had a very similar approach, but are
> already packaged in debian. (though not fully up-to-date to what we need, so
> i updated/backported the patches). Please tell me if when i should send
> these kind of patches differently.
>
> Also i expose two functions, 'creatable' and 'supported' for future proofing,
> since we want to have a more complete response in the api about which profiles
> are available.
>
> I moved the code that depends on the perlmod binding to qemu-server for now
> as it seemed fitting, we only need it there and in pve-manager
>
> If wanted I can split it out into it's own package (proxmox-pci?
> proxmox-hardware? proxmox-mdev?, not sure what's a good name for this), but
> that could also be done later when we need it.
>
> Also changed the format of the description from 'x,y,z' to 'x
> y
> z' since
> that's what was there previously and what the gui expects. Also the fields
> themselves can contain ',' so parsing would be unnecessarily hard when we do
> that.
>
> The pve-common cleanup patch is optional, and as an alternative we could
> just mark these deprecated and remove them with e.g. 10.0 so we can save us a
> breaks cyclye for pve-common <-> qemu-server.
>
> 0: https://lore.proxmox.com/pve-devel/20260120131319.949986-1-c.heiss@proxmox.com/
>
> debcargo-conf:
>
> Dominik Csapak (4):
> nvml-wrapper: Update to 0.12.0
> nvml-wrapper: add patch for vgpu ids
> backport nvml-wrapper-sys 0.9.0-1
> backport nvml-wrapper 0.12.0-1
>
> Matthias Geiger (4):
> nvml-wrapper-sys: Update to 0.9.0
> nvml-wrapper-sys: release 0.9.0-1
> nvml-wrapper: Update to 0.11.0
> nvml-wrapper: release 0.11.0-1
>
> src/nvml-wrapper-sys/debian/changelog | 12 +++++++
> .../debian/copyright.debcargo.hint | 7 ++++
> src/nvml-wrapper/debian/changelog | 20 ++++++++++++
> .../debian/copyright.debcargo.hint | 11 +++++--
> ...-method-for-getting-the-vGPU-type-id.patch | 32 +++++++++++++++++++
> src/nvml-wrapper/debian/patches/series | 1 +
> 6 files changed, 81 insertions(+), 2 deletions(-)
> create mode 100644 src/nvml-wrapper/debian/patches/add-method-for-getting-the-vGPU-type-id.patch
>
>
> proxmox-perl-rs:
>
> Dominik Csapak (1):
> pve: add binding for accessing vgpu info
>
> pve-rs/Cargo.toml | 1 +
> pve-rs/Makefile | 1 +
> pve-rs/src/bindings/mod.rs | 3 ++
> pve-rs/src/bindings/nvml.rs | 91 +++++++++++++++++++++++++++++++++++++
> 4 files changed, 96 insertions(+)
> create mode 100644 pve-rs/src/bindings/nvml.rs
>
>
> qemu-server:
>
> Dominik Csapak (2):
> pci: move mdev related code to own module
> pci: mdev: use PVE::RS::NVML for nvidia mdev information
>
> src/PVE/QemuServer.pm | 5 +-
> src/PVE/QemuServer/Makefile | 1 +
> src/PVE/QemuServer/PCI.pm | 58 +++--------
> src/PVE/QemuServer/PCI/Makefile | 9 ++
> src/PVE/QemuServer/PCI/Mdev.pm | 164 ++++++++++++++++++++++++++++++++
> 5 files changed, 190 insertions(+), 47 deletions(-)
> create mode 100644 src/PVE/QemuServer/PCI/Makefile
> create mode 100644 src/PVE/QemuServer/PCI/Mdev.pm
>
>
> pve-manager:
>
> Dominik Csapak (1):
> api: hardware: pci: use NVML for querying mdev information
>
> PVE/API2/Hardware/PCI.pm | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
>
> pve-common:
>
> Dominik Csapak (1):
> sysfs tools: remove moved code
>
> src/PVE/SysFSTools.pm | 111 ------------------------------------------
> 1 file changed, 111 deletions(-)
>
>
> Summary over all repositories:
> 17 files changed, 370 insertions(+), 163 deletions(-)
prev parent reply other threads:[~2026-03-19 11:31 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-03-05 9:16 Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 1/8] nvml-wrapper-sys: Update to 0.9.0 Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 2/8] nvml-wrapper-sys: release 0.9.0-1 Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 3/8] nvml-wrapper: Update to 0.11.0 Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 4/8] nvml-wrapper: release 0.11.0-1 Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 5/8] nvml-wrapper: Update to 0.12.0 Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 6/8] nvml-wrapper: add patch for vgpu ids Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 7/8] backport nvml-wrapper-sys 0.9.0-1 Dominik Csapak
2026-03-05 9:16 ` [PATCH debcargo-conf 8/8] backport nvml-wrapper 0.12.0-1 Dominik Csapak
2026-03-05 9:16 ` [PATCH proxmox-perl-rs 1/1] pve: add binding for accessing vgpu info Dominik Csapak
2026-03-19 11:16 ` Christoph Heiss
2026-03-05 9:16 ` [PATCH qemu-server 1/2] pci: move mdev related code to own module Dominik Csapak
2026-03-05 9:16 ` [PATCH qemu-server 2/2] pci: mdev: use PVE::RS::NVML for nvidia mdev information Dominik Csapak
2026-03-19 11:23 ` Christoph Heiss
2026-03-05 9:16 ` [PATCH manager 1/1] api: hardware: pci: use NVML for querying " Dominik Csapak
2026-03-05 9:16 ` [PATCH common 1/1] sysfs tools: remove moved code Dominik Csapak
2026-03-10 9:00 ` partially-applied: [PATCH common/debcargo-conf/manager/proxmox-perl-rs/qemu-server 00/13] use NVML for vGPU info querying Fabian Grünbichler
2026-03-19 11:31 ` Christoph Heiss [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DH6Q43WRQQYO.2JRZNE4EFN2MR@proxmox.com \
--to=c.heiss@proxmox.com \
--cc=d.csapak@proxmox.com \
--cc=pve-devel@lists.proxmox.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.