drivers/scsi/hisi_sas

HiSilicon Hip05/Hip06/Hip07 on-chip SAS host adapters

SAS storage controllers built into HiSilicon's Hip05, Hip06, and Hip07 ARM server SoCs, used in Huawei TaiShan and XR-series enterprise servers from roughly the late 2010s onward to attach SAS and SATA disks directly to the CPU without a discrete HBA card.

keep conf=0.79 last_sold=2022 deploy=low replacement=none subsystem=scsi category=storage-scsi-ata
79%

recommendation

It should stay because the hardware is still in the field on Huawei ARM servers and the driver is actively maintained, with bug fixes still landing in stable trees as recently as 2026. Although the silicon is no longer sold new, there is no alternative driver for these on-SoC SAS blocks, so removing it would strand existing deployments.

repository signals

7 files
14,208 source lines
167 commits, 5y
+2,108 / −1,887 lines added / removed, 5y
30 authors, 5y
monthly commits · 2021-04-21 → 2026-04-21 · 167 total · active in 42/61 months
2021 2022 2023 2024 2025 2026 2021-04: 0 commits · +0 −0 2021-05: 3 commits · +22 −22 2021-06: 6 commits · +82 −33 2021-07: 0 commits · +0 −0 2021-08: 6 commits · +36 −43 2021-09: 1 commit · +1 −0 2021-10: 4 commits · +99 −64 2021-11: 4 commits · +4 −15 2021-12: 13 commits · +226 −203 2022-01: 3 commits · +6 −18 2022-02: 19 commits · +161 −282 2022-03: 2 commits · +171 −320 2022-04: 0 commits · +0 −0 2022-05: 4 commits · +40 −38 2022-06: 2 commits · +9 −2 2022-07: 4 commits · +32 −37 2022-08: 1 commit · +3 −7 2022-09: 6 commits · +41 −36 2022-10: 5 commits · +86 −34 2022-11: 3 commits · +9 −22 2022-12: 1 commit · +1 −1 2023-01: 2 commits · +2 −2 2023-02: 1 commit · +1 −2 2023-03: 9 commits · +296 −76 2023-04: 1 commit · +4 −0 2023-05: 4 commits · +24 −23 2023-06: 0 commits · +0 −0 2023-07: 5 commits · +22 −14 2023-08: 6 commits · +11 −21 2023-09: 4 commits · +70 −197 2023-10: 0 commits · +0 −0 2023-11: 0 commits · +0 −0 2023-12: 5 commits · +19 −11 2024-01: 4 commits · +25 −9 2024-02: 0 commits · +0 −0 2024-03: 2 commits · +15 −51 2024-04: 3 commits · +22 −11 2024-05: 0 commits · +0 −0 2024-06: 0 commits · +0 −0 2024-07: 0 commits · +0 −0 2024-08: 1 commit · +2 −1 2024-09: 1 commit · +1 −1 2024-10: 16 commits · +223 −72 2024-11: 0 commits · +0 −0 2024-12: 1 commit · +2 −3 2025-01: 1 commit · +1 −1 2025-02: 2 commits · +31 −12 2025-03: 2 commits · +39 −4 2025-04: 5 commits · +249 −176 2025-05: 2 commits · +7 −6 2025-06: 1 commit · +6 −6 2025-07: 0 commits · +0 −0 2025-08: 1 commit · +5 −9 2025-09: 0 commits · +0 −0 2025-10: 0 commits · +0 −0 2025-11: 0 commits · +0 −0 2025-12: 0 commits · +0 −0 2026-01: 0 commits · +0 −0 2026-02: 0 commits · +0 −0 2026-03: 1 commit · +2 −2 2026-04: 0 commits · +0 −0

sources

  1. lore.kernel.org

    Stable backport in March 2026: "scsi: hisi_sas: Fix NULL pointer exception during user_scan()", evidence of ongoing upstream maintenance rather than abandonment.

  2. lore.kernel.org

    Another March 2026 stable backport: "scsi: hisi_sas: Use macro instead of magic number", reinforcing active bug-fix traffic.

  3. support.huawei.com

    Huawei XR320 server-node guide says HiSilicon server CPUs on the mainboard provide SAS ports, showing real server deployment of HiSilicon-integrated SAS.

  4. info.support.huawei.com

    Huawei Cloud Stack 8.5.1 documentation (published in 2025 per search snippet) still carries HiSilicon-specific ARM server guidance, indicating an ongoing deployed base even if this older SAS block is not clearly still sold new.

codex reasoning notes (technical)

Lore evidence came from `lore_file_timeline` on `drivers/scsi/hisi_sas/hisi_sas_main.c`, which showed heavy 2021-2026 activity and recent 2026 stable fixes; I used the two returned lore URLs above. Local tree inspection via `exec_command` found `hisilicon,hip05-sas-v1`, `hisilicon,hip06-sas-v2`, and `hisilicon,hip07-sas-v2` compatible strings plus v1/v2/v3 driver variants, supporting the chipset-family label. Removal-discussion probe via lore subject regex timed out and a fallback `lei` query was blocked by local daemon permissions, so I rely on the timeline: it shows maintenance traffic and no obvious removal wave. Huawei deployment URLs were obtained by web search; they support a conclusion of legacy-but-real field use. No clear natural replacement driver exists for the same on-SoC HBA, so removal/deprecation is not justified.