diff --git a/WORKAROUNDS.md b/WORKAROUNDS.md index 13c8b293..d1d7401b 100644 --- a/WORKAROUNDS.md +++ b/WORKAROUNDS.md @@ -29,27 +29,26 @@ A version or revision pinned for a reason that can go away. A setting that works around a bug or a hardware quirk. -| ID | What & where | Why | Removal condition | Added | Revisit | Status | -| ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | --------------------------------------------------------------------------------------------------- | ------ | -| WA-008 | Relaxed NVMe APST (`nvme_core.default_ps_max_latency_us=100000`) — `hosts/idols-ai/hardware-configuration.nix`, `hosts/idols-aquamarine/hardware-configuration.nix` | the KINGBANK KP260 (DRAM-less QLC) froze on slow APST wake-ups | Drop the override once kernel/firmware fixes or replacement hardware make default APST reliable | 2026-10-03 | watch NVMe timeout/reset/AER in the journal | active | -| WA-009 | USB-SATA bridge mitigations (`usbcore.autosuspend=-1`, `usb-storage.delay_use=10`) — `hosts/12kingdoms-youko/default.nix` | flaky JMicron JMS567 resets with UDMA-CRC and I/O errors | Drop once the bridge/controller is replaced; the signal is smartctl CRC/media errors | 2026-09-21 | on hardware change or CRC errors | active | -| WA-010 | Insta360 Link USB reset on S3 resume — `hosts/idols-ai/default.nix` | the camera stays enumerated with a stalled UVC endpoint after resume | Drop after the Insta360 firmware or the kernel `uvcvideo` fixes it | 2026-08-01 | on kernel or firmware change | active | -| WA-011 | Kernel module blacklist (`esp4`, `esp6`, `rxrpc`) — `hardening/kernel-hardening.nix` | attack surface for the Dirty Frag LPE (CVE-2026-43284 / CVE-2026-43500) | Kept as defence in depth; drop only if something needs those modules | 2026-10-03 | rarely | active | -| WA-017 | keyd right-hand modifier passthrough — `modules/nixos/desktop/peripherals.nix` | keyd folds right Ctrl/Shift/Meta into their left keys | Drop once keyd or fcitx5 keeps right-hand modifiers distinct | 2026-10-05 | on keyd bump | active | -| WA-018 | user-owned `~/.ssh/config` + disabled systemd ssh-proxy include — `home/base/tui/ssh.nix`, `modules/nixos/base/ssh.nix` | bwrap exposes root-owned store files as `nobody` | Drop once sandboxes preserve store ownership | 2026-10-06 | when sandbox layout changes | active | -| WA-019 | programs.steam.extraCompatPackages = [ pkgs.dwproton-bin pkgs.proton-ge-bin ] — modules/nixos/desktop/gaming.nix | ACE games fail on GE/mainline Proton, work with DW-Proton | Drop once mainline or GE-Proton runs these games (GE-Proton#433) | 2026-10-07 | on Proton/GE bump | active | -| WA-024 | Wi-Fi power saving disabled (`networking.networkmanager.wifi.powersave = false`) — `hosts/12kingdoms-shoukei/apple-silicon.nix` | On this BCM4378/brcmfmac laptop, powersave is a common source of intermittent ASSOC-REJECT / authentication-timeout failures; NetworkManager then mistakes the failure for a bad key, re-prompts, and the client (Noctalia, nmtui, nmcli) saves the answer as a duplicate ` 1` profile | Drop once the kernel carries the brcmfmac `port_authorized` fix (WA-025) and no association failures remain | 2026-10-07 | watch `journalctl -u wpa_supplicant` for ASSOC-REJECT / authentication timeout | active | -| WA-026 | guest deploys run `installOnHost` then `systemctl restart microvm@` on the physical host — `Justfile`, `lib/genMicrovmGuestModule.nix` | the sshSwitch step runs `nix-store --load-db` against the guest's read-only virtiofs `/nix/store` and dies on `cannot remount "/nix/store" writable` (nix 2.35) | Drop the manual path once microvm.nix or nix handles a read-only store switch. A host activation re-links `/var/lib/microvms//current` to the runner baked into the host's system, so switch the host after installing a guest from a newer tree | 2026-10-09 | on microvm.nix or nix bump | active | -| WA-028 | GitHub SSH pinned to IPv4 (`AddressFamily = "inet"`) — `home/base/tui/ssh.nix` | a v6 fake AAAA for `ssh.github.com` from the `2001:2::/64` pool (WA-029) is not proven on the 443 SSH tunnel yet | Drop once the v6 fake-ip pool is proven for GitHub SSH | 2026-10-10 | on a mihomo bump or a v6 pool change | active | -| WA-029 | mihomo `fake-ip-range6` is `2001:2::/64`, not the shipped ULA default — `modules/nixos/desktop/networking/mihomo/generate.nu` | the default `fdfe:dcba:9876::/64` sits in `fc00::/7`, which Local Network Access calls local, so a public page loaded over the v4 fake pool and pulling a dual-stack subresource prompts for local-network access (Chrome 154, Firefox). The same range also stalled ssh: a session to a fake AAAA authenticated and then never closed, on both `gvisor` and `mips`. `2001:2::/48` is RFC 5180 benchmarking -- the v6 twin of the `198.18.0.0/15` v4 default, absent from that table -- and both symptoms went with the move | Go back to the upstream default once mihomo ships a pool outside the LNA non-public table; the ULA default brings the ssh stall back | 2026-10-09 | on a browser or mihomo bump -- that table already grew to cover `2001:db8::/32` and `198.18.0.0/15` | active | -| WA-030 | mihomo unit sets `SKIP_SYSTEM_IPV6_CHECK=1` — `modules/nixos/desktop/networking/mihomo/default.nix` | mihomo checks at startup whether another NIC already holds an IPv6 address and, finding none, skips the TUN's v6 half -- but `strict-route` still installs its v6 blackhole (`ip -6 rule add priority 9000 unreachable`). On a cold boot that rule lands before networkd's Router Solicitation is answered, so the host can never send RS/NS, never gets a GUA, and mihomo never re-checks: self-locking until the core restarts. The env var forces the TUN v6 address up, which installs mihomo's normal v6 rules whose `lookup main suppress_prefixlength 0` keeps on-link neighbor discovery and the RA default route off the TUN | Drop once mihomo either re-runs the IPv6 check or skips the v6 strict-route blackhole when the v6 TUN is disabled | 2026-10-10 | on a mihomo bump | active | -| WA-031 | `iommu=pt` kernel parameter — `hosts/idols-aquamarine/hardware-amd-gpu.nix` | ROCm's system requirements warn that multi-GPU systems may need IOMMU passthrough to prevent application hangs, and this host has two Radeon AI PRO R9700 cards. Added before the first inference run; not yet observed on gfx1201. | Drop once a ROCm release states that the default IOMMU mode is safe for multi-GPU gfx1201, or when the host runs a single GPU | 2026-10-13 | on a ROCm bump, or on the first hang report | active | -| WA-032 | `amdgpu.runpm=0` kernel parameter — `hosts/idols-aquamarine/hardware-amd-gpu.nix` | this board's DSDT advertises ATPX power control (`\_SB.PC00.GFX0.ATPX`) but never defines the `PEG0.PEGP._OFF`/`_ON` symbols its body calls, so the first idle resume of the headless R9700 leaves it in `runtime_status = error` (`resume of IP block failed -62`) until reboot, and amdgpu discards the ACPI error. Root cause, kernel-source trail and the measured field comparison: `hosts/idols-aquamarine/ROCM.md`. | Drop once a BIOS update defines those ACPI methods (upstream releases 1023/1066 exist for this board; a flash changes PCR7, so re-enroll TPM2), or a kernel release stops trusting ATPX when its power methods are missing | 2026-10-14 | on a BIOS update or kernel bump | active | -| WA-036 | catppuccin eval-time ports kept in the home closure (`home.extraDependencies`) — `home/base/core/theme.nix` | catppuccin/nix reads some ports at eval time (`lib.importTOML "${sources.starship}/.toml"`) and inlines the result, so the built port, catppuccin's build/install hooks and the fetched port source stay unreachable: every gc deletes them, and the deploy after it rebuilds them (build logs: 2026-10-16). `home.extraDependencies` records the port in the generation, which is a gc root. Also read at eval time upstream: `bottom`, `element` (option `element-desktop`), `firefox`, `rio`, `palette` (options `skim`, `tty`), `spotify-player`. | Drop once catppuccin/nix stops reading port files at eval time, or those ports come from a binary cache | 2026-10-16 | on a catppuccin bump | active | -| WA-039 | niri's NVIDIA free-buffer pool bounded (`GLVidHeapReuseRatio` 0) — `hosts/idols-ai/hardware-nvidia.nix` | the driver does not return freed vidmem buffers to its free pool, so niri's VRAM grows to ~1.1 GiB instead of the ~100 MiB its NVIDIA notes expect (measured 1137 MiB after 24 h uptime; [NVIDIA/egl-wayland#126](https://github.com/NVIDIA/egl-wayland/issues/126)) | Drop once a driver stops growing niri's VRAM without the profile, or niri ships its own default | 2026-10-19 | on an nvidia driver bump | active | -| WA-041 | Monitor OSD "no-signal auto power-off" disabled — no code in this repo (the P32A2V OSD setting) | during a DPMS-off idle the panel drops and re-adds HPD every ~13 s, and after wake niri keeps scanning out while the panel stays dark until the monitor is power cycled; the panel's own auto power-off drives that state, so it is now disabled. Pinning the mode to 120 Hz did not change the symptom | Drop once a driver retrains the DP link after a DPMS wake, the monitor firmware stops cycling HPD, or this monitor is replaced | 2026-10-21 | confirm on the next full idle cycle, then on an nvidia or monitor firmware update | active | -| WA-042 | Colmena nodes set `system.nixos.versionSuffix` and `revision` from the flake's lib — `outputs/default.nix` (`rawColmenaHive`) | Colmena evaluates each node with nixpkgs' `eval-config.nix` and plain lib, which reports `26.11pre-git`, so without the override a Colmena deploy differs from the `nixosConfigurations` toplevel; the `colmena-parity` eval test guards the match | Drop once Colmena evaluates nodes with the flake's lib or accepts `nixosConfigurations` directly, keeping `colmena-parity` green | 2026-10-21 | on a colmena bump | active | -| WA-046 | xdg-desktop-portal-wlr pinned to 0.8.2 — `modules/nixos/desktop/xdg.nix` | niri's GNOME ScreenCast path exposes only DMA-BUF with modifiers, which Zoom 7.2.1 rejects; xdg-desktop-portal-wlr 0.8.2 provides the working SHM path, while 0.8.4 regresses Zoom's SelectSources. | Drop once niri's ScreenCast path works with Zoom or xdg-desktop-portal-wlr 0.8.4+ is fixed. | 2026-10-31 | on xdg-desktop-portal-wlr or Zoom updates | active | +| ID | What & where | Why | Removal condition | Added | Revisit | Status | +| ------ | --------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | --------------------------------------------------------------------------------------------------- | ------ | +| WA-009 | USB-SATA bridge mitigations (`usbcore.autosuspend=-1`, `usb-storage.delay_use=10`) — `hosts/12kingdoms-youko/default.nix` | flaky JMicron JMS567 resets with UDMA-CRC and I/O errors | Drop once the bridge/controller is replaced; the signal is smartctl CRC/media errors | 2026-09-21 | on hardware change or CRC errors | active | +| WA-010 | Insta360 Link USB reset on S3 resume — `hosts/idols-ai/default.nix` | the camera stays enumerated with a stalled UVC endpoint after resume | Drop after the Insta360 firmware or the kernel `uvcvideo` fixes it | 2026-08-01 | on kernel or firmware change | active | +| WA-011 | Kernel module blacklist (`esp4`, `esp6`, `rxrpc`) — `hardening/kernel-hardening.nix` | attack surface for the Dirty Frag LPE (CVE-2026-43284 / CVE-2026-43500) | Kept as defence in depth; drop only if something needs those modules | 2026-10-03 | rarely | active | +| WA-017 | keyd right-hand modifier passthrough — `modules/nixos/desktop/peripherals.nix` | keyd folds right Ctrl/Shift/Meta into their left keys | Drop once keyd or fcitx5 keeps right-hand modifiers distinct | 2026-10-05 | on keyd bump | active | +| WA-018 | user-owned `~/.ssh/config` + disabled systemd ssh-proxy include — `home/base/tui/ssh.nix`, `modules/nixos/base/ssh.nix` | bwrap exposes root-owned store files as `nobody` | Drop once sandboxes preserve store ownership | 2026-10-06 | when sandbox layout changes | active | +| WA-019 | programs.steam.extraCompatPackages = [ pkgs.dwproton-bin pkgs.proton-ge-bin ] — modules/nixos/desktop/gaming.nix | ACE games fail on GE/mainline Proton, work with DW-Proton | Drop once mainline or GE-Proton runs these games (GE-Proton#433) | 2026-10-07 | on Proton/GE bump | active | +| WA-024 | Wi-Fi power saving disabled (`networking.networkmanager.wifi.powersave = false`) — `hosts/12kingdoms-shoukei/apple-silicon.nix` | On this BCM4378/brcmfmac laptop, powersave is a common source of intermittent ASSOC-REJECT / authentication-timeout failures; NetworkManager then mistakes the failure for a bad key, re-prompts, and the client (Noctalia, nmtui, nmcli) saves the answer as a duplicate ` 1` profile | Drop once the kernel carries the brcmfmac `port_authorized` fix (WA-025) and no association failures remain | 2026-10-07 | watch `journalctl -u wpa_supplicant` for ASSOC-REJECT / authentication timeout | active | +| WA-026 | guest deploys run `installOnHost` then `systemctl restart microvm@` on the physical host — `Justfile`, `lib/genMicrovmGuestModule.nix` | the sshSwitch step runs `nix-store --load-db` against the guest's read-only virtiofs `/nix/store` and dies on `cannot remount "/nix/store" writable` (nix 2.35) | Drop the manual path once microvm.nix or nix handles a read-only store switch. A host activation re-links `/var/lib/microvms//current` to the runner baked into the host's system, so switch the host after installing a guest from a newer tree | 2026-10-09 | on microvm.nix or nix bump | active | +| WA-028 | GitHub SSH pinned to IPv4 (`AddressFamily = "inet"`) — `home/base/tui/ssh.nix` | a v6 fake AAAA for `ssh.github.com` from the `2001:2::/64` pool (WA-029) is not proven on the 443 SSH tunnel yet | Drop once the v6 fake-ip pool is proven for GitHub SSH | 2026-10-10 | on a mihomo bump or a v6 pool change | active | +| WA-029 | mihomo `fake-ip-range6` is `2001:2::/64`, not the shipped ULA default — `modules/nixos/desktop/networking/mihomo/generate.nu` | the default `fdfe:dcba:9876::/64` sits in `fc00::/7`, which Local Network Access calls local, so a public page loaded over the v4 fake pool and pulling a dual-stack subresource prompts for local-network access (Chrome 154, Firefox). The same range also stalled ssh: a session to a fake AAAA authenticated and then never closed, on both `gvisor` and `mips`. `2001:2::/48` is RFC 5180 benchmarking -- the v6 twin of the `198.18.0.0/15` v4 default, absent from that table -- and both symptoms went with the move | Go back to the upstream default once mihomo ships a pool outside the LNA non-public table; the ULA default brings the ssh stall back | 2026-10-09 | on a browser or mihomo bump -- that table already grew to cover `2001:db8::/32` and `198.18.0.0/15` | active | +| WA-030 | mihomo unit sets `SKIP_SYSTEM_IPV6_CHECK=1` — `modules/nixos/desktop/networking/mihomo/default.nix` | mihomo checks at startup whether another NIC already holds an IPv6 address and, finding none, skips the TUN's v6 half -- but `strict-route` still installs its v6 blackhole (`ip -6 rule add priority 9000 unreachable`). On a cold boot that rule lands before networkd's Router Solicitation is answered, so the host can never send RS/NS, never gets a GUA, and mihomo never re-checks: self-locking until the core restarts. The env var forces the TUN v6 address up, which installs mihomo's normal v6 rules whose `lookup main suppress_prefixlength 0` keeps on-link neighbor discovery and the RA default route off the TUN | Drop once mihomo either re-runs the IPv6 check or skips the v6 strict-route blackhole when the v6 TUN is disabled | 2026-10-10 | on a mihomo bump | active | +| WA-031 | `iommu=pt` kernel parameter — `hosts/idols-aquamarine/hardware-amd-gpu.nix` | ROCm's system requirements warn that multi-GPU systems may need IOMMU passthrough to prevent application hangs, and this host has two Radeon AI PRO R9700 cards. Added before the first inference run; not yet observed on gfx1201. | Drop once a ROCm release states that the default IOMMU mode is safe for multi-GPU gfx1201, or when the host runs a single GPU | 2026-10-13 | on a ROCm bump, or on the first hang report | active | +| WA-032 | `amdgpu.runpm=0` kernel parameter — `hosts/idols-aquamarine/hardware-amd-gpu.nix` | this board's DSDT advertises ATPX power control (`\_SB.PC00.GFX0.ATPX`) but never defines the `PEG0.PEGP._OFF`/`_ON` symbols its body calls, so the first idle resume of the headless R9700 leaves it in `runtime_status = error` (`resume of IP block failed -62`) until reboot, and amdgpu discards the ACPI error. Root cause, kernel-source trail and the measured field comparison: `hosts/idols-aquamarine/ROCM.md`. | Drop once a BIOS update defines those ACPI methods (upstream releases 1023/1066 exist for this board; a flash changes PCR7, so re-enroll TPM2), or a kernel release stops trusting ATPX when its power methods are missing | 2026-10-14 | on a BIOS update or kernel bump | active | +| WA-036 | catppuccin eval-time ports kept in the home closure (`home.extraDependencies`) — `home/base/core/theme.nix` | catppuccin/nix reads some ports at eval time (`lib.importTOML "${sources.starship}/.toml"`) and inlines the result, so the built port, catppuccin's build/install hooks and the fetched port source stay unreachable: every gc deletes them, and the deploy after it rebuilds them (build logs: 2026-10-16). `home.extraDependencies` records the port in the generation, which is a gc root. Also read at eval time upstream: `bottom`, `element` (option `element-desktop`), `firefox`, `rio`, `palette` (options `skim`, `tty`), `spotify-player`. | Drop once catppuccin/nix stops reading port files at eval time, or those ports come from a binary cache | 2026-10-16 | on a catppuccin bump | active | +| WA-039 | niri's NVIDIA free-buffer pool bounded (`GLVidHeapReuseRatio` 0) — `hosts/idols-ai/hardware-nvidia.nix` | the driver does not return freed vidmem buffers to its free pool, so niri's VRAM grows to ~1.1 GiB instead of the ~100 MiB its NVIDIA notes expect (measured 1137 MiB after 24 h uptime; [NVIDIA/egl-wayland#126](https://github.com/NVIDIA/egl-wayland/issues/126)) | Drop once a driver stops growing niri's VRAM without the profile, or niri ships its own default | 2026-10-19 | on an nvidia driver bump | active | +| WA-041 | Monitor OSD "no-signal auto power-off" disabled — no code in this repo (the P32A2V OSD setting) | during a DPMS-off idle the panel drops and re-adds HPD every ~13 s, and after wake niri keeps scanning out while the panel stays dark until the monitor is power cycled; the panel's own auto power-off drives that state, so it is now disabled. Pinning the mode to 120 Hz did not change the symptom | Drop once a driver retrains the DP link after a DPMS wake, the monitor firmware stops cycling HPD, or this monitor is replaced | 2026-10-21 | confirm on the next full idle cycle, then on an nvidia or monitor firmware update | active | +| WA-042 | Colmena nodes set `system.nixos.versionSuffix` and `revision` from the flake's lib — `outputs/default.nix` (`rawColmenaHive`) | Colmena evaluates each node with nixpkgs' `eval-config.nix` and plain lib, which reports `26.11pre-git`, so without the override a Colmena deploy differs from the `nixosConfigurations` toplevel; the `colmena-parity` eval test guards the match | Drop once Colmena evaluates nodes with the flake's lib or accepts `nixosConfigurations` directly, keeping `colmena-parity` green | 2026-10-21 | on a colmena bump | active | +| WA-046 | xdg-desktop-portal-wlr pinned to 0.8.2 — `modules/nixos/desktop/xdg.nix` | niri's GNOME ScreenCast path exposes only DMA-BUF with modifiers, which Zoom 7.2.1 rejects; xdg-desktop-portal-wlr 0.8.2 provides the working SHM path, while 0.8.4 regresses Zoom's SelectSources. | Drop once niri's ScreenCast path works with Zoom or xdg-desktop-portal-wlr 0.8.4+ is fixed. | 2026-10-31 | on xdg-desktop-portal-wlr or Zoom updates | active | | WA-048 | Nixpak GUI fontconfig comes from the host — `hardening/nixpaks/modules/desktop-base.nix` | Nixpak normally generates a standalone `/etc/fonts` and font package list for each sandbox; that @@ -60,10 +59,9 @@ preferences through the already-mounted `/nix/store`. | Drop once Nixpak can dir fontconfig or no longer overrides `/etc/fonts`, then remove the bind and restore the library-supported configuration path | 2026-11-01 | on a nixpak update | active | -If WA-008 freezes recur, the fallback is to restore the parameters commented out in -`hosts/idols-ai/hardware-configuration.nix` — -`nvme_core.default_ps_max_latency_us=0 nvme_core.io_timeout=4294967295 pcie_aspm=off`; this is a -recovery procedure, not its removal condition. +If the KINGBANK KP260 NVMe freezes on `ai` recur (formerly WA-008, now back at the kernel's APST +defaults), restore the parameters commented out in `hosts/idols-ai/hardware-configuration.nix` — +`nvme_core.default_ps_max_latency_us=0 nvme_core.io_timeout=4294967295 pcie_aspm=off`. If Wi-Fi wedges after a failed association, restart the stack instead of rebooting: `sudo systemctl restart wpa_supplicant`, then `sudo systemctl restart NetworkManager`. Deleting an diff --git a/hosts/idols-ai/hardware-configuration.nix b/hosts/idols-ai/hardware-configuration.nix index ffcb12f1..a69e3bc9 100644 --- a/hosts/idols-ai/hardware-configuration.nix +++ b/hosts/idols-ai/hardware-configuration.nix @@ -13,27 +13,23 @@ (modulesPath + "/installer/scan/not-detected.nix") ]; - boot.kernelParams = [ - # === NVMe power management (relaxed 2026-10) === - # - # These began as a hard workaround for NVMe freezes on the KINGBANK KP260 - # (a DRAM-less QLC drive, prone to slow APST wakeups): APST fully disabled, - # the command timeout pushed to the maximum, and PCIe ASPM turned off. - # Kernel logs from 2026-09-19..10-03 (8 boots) show no NVMe - # timeout/reset/AER, so it is relaxed to reclaim idle power and let the - # kernel recover a stuck command instead of hanging on it: - # - # - APST allowed again, capped at a wide 100 ms latency (was: disabled). - # - io_timeout back to the kernel default 30 s (was: ~49 days, which let - # a stuck command wait forever instead of resetting the controller). - # - pcie_aspm removed: PCIe link power saving is unrelated to the drive's - # own APST. - # - # If freezes return under heavy read/write, restore the old workaround: - # nvme_core.default_ps_max_latency_us=0 nvme_core.io_timeout=4294967295 pcie_aspm=off - # See https://community.frame.work/t/nvme-timeout-woes/54999 - "nvme_core.default_ps_max_latency_us=100000" - ]; + # === NVMe power management (relaxed 2026-10) === + # + # The KINGBANK KP260 (a DRAM-less QLC drive, prone to slow APST wakeups) once + # froze, so APST was fully disabled, the command timeout pushed to the maximum, + # and PCIe ASPM turned off. Kernel logs from 2026-09-19..10-03 (8 boots) show + # no NVMe timeout/reset/AER, so all three are back at the kernel defaults: + # APST capped at 100 ms latency (nvme_core.default_ps_max_latency_us), a 30 s + # io_timeout that resets a stuck controller, and ASPM, which is unrelated to + # the drive's own APST. + # + # If freezes return under heavy read/write, restore the old workaround: + # boot.kernelParams = [ + # "nvme_core.default_ps_max_latency_us=0" + # "nvme_core.io_timeout=4294967295" + # "pcie_aspm=off" + # ]; + # See https://community.frame.work/t/nvme-timeout-woes/54999 # Use the EFI boot loader. boot.loader.efi.canTouchEfiVariables = true; diff --git a/hosts/idols-aquamarine/hardware-configuration.nix b/hosts/idols-aquamarine/hardware-configuration.nix index 9f536773..32091135 100644 --- a/hosts/idols-aquamarine/hardware-configuration.nix +++ b/hosts/idols-aquamarine/hardware-configuration.nix @@ -17,13 +17,7 @@ ]; boot.kernelParams = [ - # Carried over from `ai`, where it caps NVMe APST wake latency for the - # DRAM-less KINGBANK KP260 (WA-008 in WORKAROUNDS.md). This disk (Fanxiang - # S790) has shown no APST problem; the cap stays as idle-power insurance. - "nvme_core.default_ps_max_latency_us=100000" - - # Keep PCIe ASPM disabled. The R9700 links have retraining/timeout issues, - # and this repo already uses the same knob for NVMe APST problems (WA-008). + # Keep PCIe ASPM disabled. The R9700 links have retraining/timeout issues. "pcie_aspm=off" ];