LIO SCSI Target Framework (TCM)
An in-kernel framework that turns a Linux server into a SCSI storage target, exporting local block devices, files, or pass-through SCSI devices to other machines over fabrics like iSCSI, Fibre Channel, FireWire (SBP-2), and loopback. It is the engine behind the targetcli tool and is widely used to build software-defined SAN appliances and shared storage for virtualization clusters.
recommendation
It should stay because LIO is the standard in-kernel iSCSI/SCSI target on Linux and remains in active use on current enterprise distributions like Red Hat Enterprise Linux 8 and Debian, configured through targetcli. Upstream maintenance is healthy, with recent fixes and new feature work (such as backend-context command completion) landing in 2025-2026 from Oracle's Michael Christie, and there is no replacement framework on the horizon.
repository signals
sources
- lore.kernel.org
Recent 2026 fix traffic in target core shows active upstream maintenance, not abandonment.
- lore.kernel.org
Recent 2026 functional work added backend-context command completion support in scsi target core.
- access.redhat.com
Current enterprise distro documentation still covers configuring iSCSI targets with targetcli and LIO backstores including block, fileio, pscsi, and ramdisk.
- wiki.debian.org
Debian wiki documents LIO as the in-kernel iSCSI target and describes targetcli-fb usage, indicating ongoing real deployments.
codex reasoning notes (technical)
`drivers/target` is a real driver/subsystem, not an internal helper: local shell read of `drivers/target/Kconfig` shows target-core and fabric modules (iblock, fileio, pscsi, iscsi, loopback, tcm_fc, sbp, tcm_remote). Lore evidence came from `mcp__lore_http__.lore_activity` on `drivers/target/target_core_device.c`, which returned 2025-2026 fixes and feature work via the cited lore URLs. A broader lore subject scan and a `lei` query were attempted for removal/deprecation discussion, but timed out / were blocked in this sandbox; with the available lore sample, I found active maintenance rather than removal signals. Deployment evidence came from web search results opened from Red Hat and Debian pages showing current docs for LIO/targetcli usage on supported distro releases. Because this is a software target framework used on current servers rather than a dead chipset-specific driver, there is no natural in-tree replacement and no meaningful hardware EOL year to assign.