drivers/rapidio/switches · keep-annotate

IDT/Renesas CPS and Tsi Serial RapidIO fabric switches

Switch chips that route traffic across Serial RapidIO fabrics, a high-speed interconnect used mainly in embedded telecom, defense, and industrial backplanes from the mid-2000s onward. Coverage spans IDT (now Renesas) CPS and Tsi parts across three generations, including the still-active 80HCPS1848 and the obsolete Tsi578.

keep-annotate conf=0.81 deploy=low replacement=none subsystem=rapidio category=bus-other

81%

recommendation

Worth keeping but documenting its niche, because RapidIO is a specialized interconnect with a small but real user base in embedded and telecom systems. Renesas still markets some of these switches as active products, and the code saw a fresh patch as recently as 2026, so it is not abandoned even though long-term commit activity is sparse and no alternative driver covers the same hardware.

repository signals

5 files

1,067 source lines

1 commits, 5y

+0 / −573 lines added / removed, 5y

1 authors, 5y

monthly commits · 2021-04-21 → 2026-04-21 · 1 total · active in 1/61 months

sources

lore.kernel.org
Upstream still sees recent attention: a 2026 patch touched `drivers/rapidio/switches/idt_gen3.c`.
renesas.com
At least one device covered by this directory, the IDT/Renesas 80HCPS1848 RapidIO switch, was still marketed as Active.
renesas.com
Older members of the family such as Tsi578 were marked Obsolete, indicating mixed lifecycle status within the directory's hardware coverage.

codex reasoning notes (technical)

Local source inspection via shell showed three real RapidIO switch drivers (`idtcps`, `idt_gen2`, `idt_gen3`) for IDT CPS/Tsi parts. `lore_file_timeline` on the directory returned no aggregate hits, so I drilled down with `lore_activity` per file; that yielded the cited 2026 `idt_gen3` patch, which argues against deprecation/removal despite sparse long-term churn. Web search produced the Renesas product pages: `80HCPS1848` is Active while `TSI578` is Obsolete, so the family is niche and partly legacy but not wholly dead. Deployment is likely low and specialized (embedded/telecom/industrial RapidIO fabrics), with no separate upstream replacement driver for the same use case.