Upstream status update:

This issue now appears to be a device-specific SATA link power
management compatibility issue affecting the ADATA SU680 model, rather
than classic SSD media failure and not a generic Linux storage issue
affecting all SSDs.

A Linux/libata maintainer replied on linux-ide and indicated that this
device likely needs an ATA_QUIRK_NOLPM quirk entry in
drivers/ata/libata-core.c.

The exact IDENTIFY model string reported by hdparm -I /dev/sda for the
affected SSD is:

Model Number: ADATA SU680

This exact model string has now been sent upstream.

Current understanding:
- the issue appears tied to this specific SSD model / firmware interacting 
badly with SATA link power management
- the likely Linux-side fix is a libata ATA_QUIRK_NOLPM quirk for this model
- this does not look like classic SSD NAND/media failure based on SMART 
evidence, successful full-disk read testing, and the later successful extended 
SMART self-test

Important note:
This type of libata quirk is model-specific. If similar LPM issues are 
discovered on other SSD models, those models generally need to be added 
independently with their own quirk entries unless already matched by an 
existing rule.

Persistent workaround currently in use while waiting for the upstream
kernel fix:

File:
/etc/systemd/system/disable-sata-alpm.service

Contents:

[Unit]
Description=Disable SATA ALPM for AHCI hosts
After=multi-user.target

[Service]
Type=oneshot
ExecStart=/bin/sh -c 'for f in 
/sys/class/scsi_host/host*/link_power_management_policy; do echo 
max_performance > "$f"; done'
RemainAfterExit=yes

[Install]
WantedBy=multi-user.target

Then:

sudo systemctl daemon-reload
sudo systemctl enable disable-sata-alpm.service
sudo systemctl start disable-sata-alpm.service

This is a persistent workaround, not the actual fix. The proper fix
should be the upstream kernel quirk for the affected device model.

Current upstream progress:
- Linux/libata maintainer replied
- maintainer indicated that ATA_QUIRK_NOLPM is likely needed for this device
- exact model string has been provided upstream
- patch testing can be performed once a patch is proposed

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2144060

Title:
  ADATA SU680 causes repeated SATA resets and I/O errors on Ubuntu
  unless link power management is forced to max_performance

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux-hwe-6.17/+bug/2144060/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to