drivers/gpu/drm/verisilicon

VeriSilicon DC8200 display controllers (StarFive JH7110)

VeriSilicon's DC-series display controller IP, most prominently the DC8200 block embedded in StarFive's JH7110 RISC-V SoC. That chip powers current single-board computers such as the VisionFive 2 and Pine64 Star64, where this code drives the on-chip display subsystem feeding HDMI and LCD outputs.

keep conf=0.91 deploy=low replacement=none subsystem=gpu category=graphics-display
91%

recommendation

It should stay because it is an actively developed, recently added upstream graphics driver tied to hardware still being sold in 2025, including popular RISC-V boards built around the StarFive JH7110. New feature work such as hardware cursor support was still landing in early 2026, no replacement driver covers the same IP block, and there is no removal discussion on the kernel mailing lists.

repository signals

0 commits, 5y
+0 / −0 lines added / removed, 5y
0 authors, 5y

sources

  1. lwn.net

    As of March 9, 2026, the upstream driver was still receiving new feature work; a patch series added hardware cursor support under drivers/gpu/drm/verisilicon.

  2. kernel.googlesource.com

    Upstream master Kconfig includes source "drivers/gpu/drm/verisilicon/Kconfig", with blame output showing the entry was added on 2026-01-29, indicating this is a recently added upstream DRM driver rather than abandoned legacy code.

  3. doc-en.rvspace.org

    StarFive's VisionFive 2 JH7110 SDK documentation places the Linux display driver in linux-5.15/drivers/gpu/drm/verisilicon, tying the driver to shipped JH7110 display hardware.

  4. doc-en.rvspace.org

    StarFive's JH7110 display documentation identifies compatible strings such as "verisilicon,display-subsystem" and "verisilicon,dc8200", confirming the hardware family covered by the driver.

  5. pine64.com

    A current retail product page lists the STAR64 single-board computer with the StarFive JH-7110 SoC as in stock, supporting the conclusion that this hardware class is still deployed and sold.

codex reasoning notes (technical)

Not an early-exit case: path is a real DRM driver directory. Local shell (`rg`) showed this kernel snapshot does not contain the directory, so evidence was gathered via `web.search_query`. That search found active upstream lore/LWN traffic in 2026 and a master Kconfig blame entry dated 2026-01-29, which outweigh the static '0 commits in 5y' metadata and indicates ongoing bring-up, not obsolescence. Vendor docs found via `web.search_query` map the driver to VeriSilicon DC8200 display blocks in StarFive JH7110 systems, and current retail availability of JH7110 boards implies ongoing niche embedded/RISC-V deployments. No natural in-tree replacement driver for the same IP block was found, and a targeted lore web search found no removal discussion, so `keep` is the defensible recommendation.