[Bug 2137434] [NEW] Suspend / resume failure on Kubuntu 24.04, NVME with OPAL encryption

Public bug reported:

Suspend / resume failure on Kubuntu 24.04 with NVMe root filesystem
(Micron 3400) leading to NVMe I/O errors and EXT4 journal abort

System information

Machine: Lenovo IdeaPad 5 Pro 16ARH7

CPU platform: AMD (AMD-VI / IOMMU present)

Distribution: Kubuntu 24.04.3 LTS (Ubuntu 24.04 "noble")

Kernel (running):

6.14.0-37-generic


Kernel type: generic / HWE

Installed kernels:

linux-image-6.14.0-27-generic
linux-image-6.14.0-37-generic
linux-image-generic-hwe-24.04


Boot mode: UEFI

Dual boot: Yes (Windows + Linux)

Root filesystem: EXT4 on NVMe

Suspend mode: s2idle only

$ cat /sys/power/mem_sleep
[s2idle]


Hibernate: Not supported

systemctl hibernate
→ Sleep verb 'hibernate' is not configured or configuration is not supported by kernel Or i might have disabled something here in my attempts to fix the problem.

Storage device

NVMe controller: Micron 3400 NVMe SSD (Hendrix)

Firmware: 1006P7LN

PCI ID: 04:00.0

Kernel driver: nvme

Encryption configuration

NVMe drive encryption enabled via drive firmware (OPAL) / BIOS (pre-boot
authentication).

No Linux userspace encryption layer (e.g. no LUKS).

BIOS option disabling the password prompt was tested; behavior did not
change.

Problem description

When the system is suspended (s2idle) and later resumed:

The system does not recover cleanly.

The NVMe root filesystem becomes unavailable.

Kernel logs show NVMe, AMD-VI (IOMMU), and EXT4 errors.

The root filesystem is remounted read-only or the system becomes
unusable. (chat gpts working hypothesis eg no cd no ls in the console
window i still have open (if system wakes with gui) both commands (cd,
ls) just give io errors.)

A reboot (eg powercut) is required to recover. This is always the case
only how much is recovered when waking changes.

This occurs consistently when suspending from Linux.

The same hardware resumes successfully from suspend under Windows on the
same system.

Steps to reproduce

Boot Kubuntu normally.

Log in and allow the system to reach an idle state.

Trigger suspend (e.g. close lid or systemctl suspend).

Resume the system.

Expected result

System resumes normally.

NVMe device remains available.

Root filesystem remains writable.

No I/O or filesystem errors.

Actual result

Resume fails.

NVMe device reports errors.

EXT4 journal aborts.

Root filesystem remounts read-only or system becomes unusable.

Relevant kernel log excerpts (after failed resume)

Logs copied from a photographed console after resume failure; formatting
artifacts may be present.

USB / ACPI
ucsi_acpi USB0000:00: ucsi_handle_connector_change:
GET CONNECTOR STATUS failed (-110)

IOMMU (AMD-VI)
AMD-VI: Event logged [IO_PAGE_FAULT domain=0x0 address=0x71687000 flags=0x0000]
AMD-VI: Event logged [IO_PAGE_FAULT domain=0x0 address=0x716e7040 flags=0x0000]

NVMe
nvme 0000:04:00.0: Device not ready: aborting reset, CSTS=0x1

EXT4 filesystem (root device nvme0n1p5)
EXT4-fs error (device nvme0n1p5): ext4_find_entry: reading directory block
Buffer I/O error on device nvme0n1p5
Aborting journal on device nvme0n1p5
EXT4-fs error (device nvme0n1p5): Journal has aborted
EXT4-fs error (device nvme0n1p5): ext4_dirty_inode: Journal has aborted
EXT4-fs (nvme0n1p5): Remounting filesystem read-only
EXT4-fs (nvme0n1p5): I/O error while writing superblock


Multiple repeated buffer I/O errors follow for different logical blocks.

Additional observations

Disabling the pre-boot password prompt in BIOS does not change behavior.

The system only supports s2idle; deeper sleep states are not available.

Windows on the same hardware resumes from suspend successfully.


Reproducibility

Reproducible on every suspend/resume attempt under Linux.

Not observed under Windows on the same system.

Attachments / additional data available

Full nvme id-ctrl output

Full lspci -vv output for the NVMe device

/etc/default/grub

Additional kernel logs I hope I orderd the logs from most usefull to
most detailed as I can only attach a single file.

** Affects: linux (Ubuntu)
Importance: Undecided
Status: New


** Tags: suspend-resume

** Attachment added: "All kinds of logs I collected including a OCRd foto of the screen of the brokenly waked system"
https://bugs.launchpad.net/bugs/2137434/+attachment/5937324/+files/LogCollectionAndMore

--
You received this bug notification because you are subscribed to linux
in Ubuntu.
Matching subscriptions: Bgg, Bmail, Nb
https://bugs.launchpad.net/bugs/2137434

Title:
Suspend / resume failure on Kubuntu 24.04, NVME with OPAL encryption

Status in linux package in Ubuntu:
New

Bug description:
Suspend / resume failure on Kubuntu 24.04 with NVMe root filesystem
(Micron 3400) leading to NVMe I/O errors and EXT4 journal abort

System information

Machine: Lenovo IdeaPad 5 Pro 16ARH7

CPU platform: AMD (AMD-VI / IOMMU present)

Distribution: Kubuntu 24.04.3 LTS (Ubuntu 24.04 "noble")

Kernel (running):

6.14.0-37-generic


Kernel type: generic / HWE

Installed kernels:

linux-image-6.14.0-27-generic
linux-image-6.14.0-37-generic
linux-image-generic-hwe-24.04


Boot mode: UEFI

Dual boot: Yes (Windows + Linux)

Root filesystem: EXT4 on NVMe

Suspend mode: s2idle only

$ cat /sys/power/mem_sleep
[s2idle]


Hibernate: Not supported

systemctl hibernate
→ Sleep verb 'hibernate' is not configured or configuration is not supported by kernel Or i might have disabled something here in my attempts to fix the problem.

Storage device

NVMe controller: Micron 3400 NVMe SSD (Hendrix)

Firmware: 1006P7LN

PCI ID: 04:00.0

Kernel driver: nvme

Encryption configuration

NVMe drive encryption enabled via drive firmware (OPAL) / BIOS (pre-
boot authentication).

No Linux userspace encryption layer (e.g. no LUKS).

BIOS option disabling the password prompt was tested; behavior did not
change.

Problem description

When the system is suspended (s2idle) and later resumed:

The system does not recover cleanly.

The NVMe root filesystem becomes unavailable.

Kernel logs show NVMe, AMD-VI (IOMMU), and EXT4 errors.

The root filesystem is remounted read-only or the system becomes
unusable. (chat gpts working hypothesis eg no cd no ls in the console
window i still have open (if system wakes with gui) both commands (cd,
ls) just give io errors.)

A reboot (eg powercut) is required to recover. This is always the case
only how much is recovered when waking changes.

This occurs consistently when suspending from Linux.

The same hardware resumes successfully from suspend under Windows on
the same system.

Steps to reproduce

Boot Kubuntu normally.

Log in and allow the system to reach an idle state.

Trigger suspend (e.g. close lid or systemctl suspend).

Resume the system.

Expected result

System resumes normally.

NVMe device remains available.

Root filesystem remains writable.

No I/O or filesystem errors.

Actual result

Resume fails.

NVMe device reports errors.

EXT4 journal aborts.

Root filesystem remounts read-only or system becomes unusable.

Relevant kernel log excerpts (after failed resume)

Logs copied from a photographed console after resume failure;
formatting artifacts may be present.

USB / ACPI
ucsi_acpi USB0000:00: ucsi_handle_connector_change:
GET CONNECTOR STATUS failed (-110)

IOMMU (AMD-VI)
AMD-VI: Event logged [IO_PAGE_FAULT domain=0x0 address=0x71687000 flags=0x0000]
AMD-VI: Event logged [IO_PAGE_FAULT domain=0x0 address=0x716e7040 flags=0x0000]

NVMe
nvme 0000:04:00.0: Device not ready: aborting reset, CSTS=0x1

EXT4 filesystem (root device nvme0n1p5)
EXT4-fs error (device nvme0n1p5): ext4_find_entry: reading directory block
Buffer I/O error on device nvme0n1p5
Aborting journal on device nvme0n1p5
EXT4-fs error (device nvme0n1p5): Journal has aborted
EXT4-fs error (device nvme0n1p5): ext4_dirty_inode: Journal has aborted
EXT4-fs (nvme0n1p5): Remounting filesystem read-only
EXT4-fs (nvme0n1p5): I/O error while writing superblock


Multiple repeated buffer I/O errors follow for different logical blocks.

Additional observations

Disabling the pre-boot password prompt in BIOS does not change
behavior.

The system only supports s2idle; deeper sleep states are not
available.

Windows on the same hardware resumes from suspend successfully.


Reproducibility

Reproducible on every suspend/resume attempt under Linux.

Not observed under Windows on the same system.

Attachments / additional data available

Full nvme id-ctrl output

Full lspci -vv output for the NVMe device

/etc/default/grub

Additional kernel logs I hope I orderd the logs from most usefull to
most detailed as I can only attach a single file.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2137434/+subscriptions