Marvell PXA1908 power-domain controller
Manages the on-chip power domains of Marvell's PXA1908, a quad-core 64-bit ARM mobile processor announced in late 2014 for mass-market 4G LTE smartphones. It controls when blocks of the SoC are powered up or gated off to save energy, and is used today mainly by community mainlining projects keeping handsets like the Samsung Galaxy Core Prime VE LTE alive.
recommendation
Worth keeping but documenting as legacy silicon. The PXA1908 is a 2014-era Marvell mobile chip that powered budget LTE smartphones like the Samsung Galaxy Core Prime VE LTE, and the hardware hasn't been sold for new designs in roughly a decade. However, this code is brand new upstream (first appearing in Linux 6.18 after a v15 patch series landed in 2025), driven by hobbyist mainlining and postmarketOS efforts, so removing it now would discard work that just arrived.
repository signals
sources
- marvell.com
Marvell announced the PXA1908 in November 2014 for mass-market LTE smartphones, implying legacy mobile silicon rather than current new-design hardware.
- patchew.org
A 2025 upstream series added initial mainline support for Marvell PXA1908 and names Samsung Galaxy Core Prime VE LTE as a board using the SoC, showing fresh enablement work rather than removal.
- cateee.net
CONFIG_PXA1908_PM_DOMAINS first appears only in Linux 6.18+, confirming this PM-domain driver is newly upstreamed.
- wiki.postmarketos.org
The Samsung Galaxy Core Prime VE LTE, a 2014 PXA1908 handset, still has community postmarketOS/mainline activity, indicating residual hobbyist deployments.
codex reasoning notes (technical)
Local shell inspection showed this is a real auxiliary PM-domain driver for five PXA1908 domains and local git history showed a single add-only commit in September 2025, with no in-tree sign of removal. URLs were obtained via web search: Marvell newsroom for launch date/market, Patchew for 2025 upstream activity, LKDDb for kernel presence, and postmarketOS wiki for present-day community use. Conclusion: hardware is obsolete and not a 2025 new-sales target, but upstream support is brand new and tied to remaining enthusiast/mainlining use, so deprecation would be premature; keep it but annotate as legacy/niche silicon.