NVMe/MVMEeOF: Patch NVMe/NVMeOF driver to support GDS

Bug #1982518 reported by Brad Figg
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
linux-nvidia (Ubuntu)
Invalid
Undecided
Unassigned
Jammy
Fix Released
Medium
Brad Figg

Bug Description

 [Impact]
    Adding these changes will enable GDS functionality for NVMe and NVMeOF drivers.

 [Fix]
    This is a not a fix but a new feature being added to NVMe and NVMeOF driver.

 [Test]
    Tested the NVMe driver on a hpe system as I did not have a setup with BASEOS6.
     1) Installed 5.15.39 kernel on the system (this is the kernel that optimized kernel is on currently).
     2) Downloaded the optimized kernel.
     3) Applied the patches to the optimized kernel
     4) Replaced the nvme modules on the system with the one's built on optimized kernel.
     5) Ran gds tests on ext4 file system with the patched NVMe driver. All tests went fine.
     6) The boot drive is on NVMe and the system is up and running which indicates no regression on the standard NVMe path as well.

 Attached are the results with NVMe
 **************************************************
NVFS IOCTL negative Tests, : 23 / 23 tests passed
**************************************************
Testsuite : 249 / 249 tests passed
End: nvidia-fs:
GDS Version: 1.4.0.31
NVFS statistics(ver: 4.0)
NVFS Driver(version: 2.13.5)
Mellanox PeerDirect Supported: False
IO stats: Disabled, peer IO stats: Disabled
Logging level: info

Compat mode Tests
==================
**************************************************
API Tests, : 72 / 72 tests passed
**************************************************
Testsuite : 211 / 211 tests passed
done tests:Thu Jul 21 08:12:09 PM UTC 2022

Could not test the same with NVMeOF as a setup was not available.

Stefan Bader (smb)
Changed in linux-nvidia (Ubuntu Jammy):
assignee: nobody → Brad Figg (brad-figg)
importance: Undecided → Medium
status: New → In Progress
Changed in linux-nvidia (Ubuntu):
status: New → Invalid
Revision history for this message
Launchpad Janitor (janitor) wrote :
Download full text (118.0 KiB)

This bug was fixed in the package linux-nvidia - 5.15.0-1007.7

---------------
linux-nvidia (5.15.0-1007.7) jammy; urgency=medium

  * jammy/linux-nvidia: 5.15.0-1007.7 -proposed tracker (LP: #1990418)

  * GDS: Add NFS patches to optimized kernel (LP: #1982519)
    - NFS: Patch NFS driver to support GDS

  * NVMe/MVMEeOF: Patch NVMe/NVMeOF driver to support GDS (LP: #1982518)
    - NVMe/MVMEeOF: Patch NVMe/NVMeOF driver to support GDS

linux-nvidia (5.15.0-1006.6) jammy; urgency=medium

  * jammy/linux-nvidia: 5.15.0-1006.6 -proposed tracker (LP: #1988357)

  * Ubuntu 22.04 and 20.04 DPC Fixes for Failure Cases of DownPort Containment
    events (LP: #1965241)
    - [Config] Enable config option CONFIG_PCIE_EDR

  * disable Intel DMA remapping by default (LP: #1971699)
    - Revert "UBUNTU: [Config] enable Intel DMA remapping options by default"

  * Jammy update: v5.15.49 upstream stable release (LP: #1983149)
    - [Config] updateconfigs for LIB_MEMNEQ

  * GPIO character device v1 API not enabled in kernel (LP: #1953613)
    - [Config] Enable CONFIG_GPIO_CDEV_V1

  * Jammy update: v5.15.46 upstream stable release (LP: #1981864)
    - [Config] updateconfigs for IMA_TEMPLATE
    - [Packaging] add python3-dev as build dependency

  * Jammy update: v5.15.44 upstream stable release (LP: #1981649)
    - [Config] update configs for CRYPTO_LIB_BLAKE2S

  * Miscellaneous Ubuntu changes
    - [Packaging] Enable nvidia-lowlatency for amd64
    - remove debian.nvidia/control.stub which was inadvertently committed
    - [Config] CONFIG_LIBMEMNEQ -> CONFIG_LIB_MEMNEQ

  [ Ubuntu: 5.15.0-48.54 ]

  * jammy/linux: 5.15.0-48.54 -proposed tracker (LP: #1987775)
  * System freeze after resuming from suspend due to PCI ASPM settings
    (LP: #1980829)
    - SAUCE: PCI/ASPM: Save/restore L1SS Capability for suspend/resume
    - SAUCE: whitelist platforms that needs save/restore ASPM L1SS for
      suspend/resume
  * [SRU][J/OEM-5.17][PATCH 0/1] Fix oled brightness set above frame-average
    luminance (LP: #1978986)
    - SAUCE: drm: New function to get luminance range based on static hdr metadata
    - SAUCE: drm/amdgpu_dm: Rely on split out luminance calculation function
    - SAUCE: drm/i915: Use luminance range calculated during edid parsing
  * Jammy: Add OVS Internal Port HW Offload to mlx5 driver (LP: #1983498)
    - net/mlx5e: Refactor rx handler of represetor device
    - net/mlx5e: Use generic name for the forwarding dev pointer
    - net/mlx5: E-Switch, Add ovs internal port mapping to metadata support
    - net/mlx5e: Support accept action
    - net/mlx5e: Accept action skbedit in the tc actions list
    - net/mlx5e: Offload tc rules that redirect to ovs internal port
    - net/mlx5e: Offload internal port as encap route device
    - net/mlx5e: Enable TC offload for ingress MACVLAN
    - net/mlx5e: Add indirect tc offload of ovs internal port
    - net/mlx5e: Term table handling of internal port rules
    - net/mlx5: Support internal port as decap route device
    - net/mlx5: Fix some error handling paths in 'mlx5e_tc_add_fdb_flow()'
    - net/mlx5e: TC, Fix memory leak with rules with internal port
    - net/mlx5e: Fix skb memory leak when TC cl...

Changed in linux-nvidia (Ubuntu Jammy):
status: In Progress → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.