Hi Alex, Following up on your suggestion to check for corruption; I've verified the firmware file is byte-for-byte identical to upstream.
Local file: /lib/firmware/amdgpu/raven2_rlc.bin SHA256: 370b453daf80ae22b40451b98454a4a2c9d1dd3168a5ab664509a25493070e73 Size: 38324 bytes Package: linux-firmware-amdgpu 20251125-2 (Manjaro) Upstream (git.kernel.org, commit ba491f60469c, "amdgpu: Update raven2 firmware from 21.50", Feb 2022): SHA256: 370b453daf80ae22b40451b98454a4a2c9d1dd3168a5ab664509a25493070e73 Size: 38324 bytes The files are identical. The firmware blob on disk is not corrupted. The PSP on this Raven2 silicon is rejecting a perfectly valid firmware binary with error 0xFFFF300F on LOAD_IP_FW(0x6) for all three RLC_RESTORE_LIST_* ucodes (CNTL, GPM_MEM, SRM_MEM). There has been no raven2 firmware update in linux-firmware since 21.50 (Feb 2022) — over 4 years. There is no newer firmware to try. This appears to be a silicon-level PSP signature validation rejection that AMD has not acknowledged. The GPU still initializes and functions (Display Core v3.2.340 on DCN 1.0.1 works), but the missing RLC restore-list ucodes may cause instability during suspend/resume power state restoration. Is there a known workaround, or is this a firmware binary that needs to be regenerated for certain Raven2 silicon revisions? Hardware: AMD Radeon Graphics (Raven 2, DRM 3.62.0, 6.17.3-1-MANJARO) Firmware version: 21.50 (raven2_rlc.bin, last updated Feb 2022) Thanks ________________________________________ De : Alex Deucher <[email protected]> Envoyé : vendredi 27 février 2026 15:15 À : outlook user <[email protected]> Cc : [email protected] <[email protected]> Objet : Re: Inquiry: Raven Ridge `RLC_RESTORE_LIST_CNTL` failure (0xFFFF300F) on Kernel 6.17 - Missing GFX9 fix? On Fri, Feb 27, 2026 at 4:29 AM outlook user <[email protected]> wrote: > > > Hi team, > > > > I am running a Raven Ridge APU (Ryzen 5 3500U, device 0x15D8) on kernel > > 6.17. > > > > **The Issue:** > > I see persistent firmware load failures at boot: > > `amdgpu: failed to load ucode RLC_RESTORE_LIST_CNTL` > > `psp gfx command LOAD_IP_FW(0x6) failed and response status is (0xFFFF300F)` > > > > **Context:** > > I noticed the recent patch *[PATCH] drm/amdgpu: fixing rlc firmware loading > > failure issue* (Aug 2024) which fixed header size mismatch issues for GFX10 > > by switching to `request_firmware`. > > > > **My Question:** > > Should this fix also apply to GFX9 (gfx_v9_0.c)? > > > > My system is currently using `amdgpu_ucode_request` for the "kicker" path, > > and the hardware rejects the firmware (likely due to the signature/header > > mismatch mentioned in the GFX10 patch). This leaves the GPU without RLC > > loaded, leading to intermittent system freezes. > > > > Is this a known gap for Raven Ridge, or is there a specific firmware > > version I should be using? > > That gfx10 patch is unrelated and not applicable. It sounds more like your firmware file on disk or in your initrd may have gotten corrupted. If that's not the issue, can you try older (or newer depending on what versions you are using) firmware files for your GPU? Alex
