id: "BEF-2026-09-14-CLEVO-002" organization: "Beffroi" title: "System Stability Investigation — Storage / Firmware Follow-up" subtitle: "PC Specialist / Clevo — exhaustive diagnostic record and current assessment" report_type: technical theme: default status: "open" severity: "high" metadata: Date: "2026-09-14" Device: "DESKTOP-SAU019H" Model: "PC Specialist / Clevo NH5x_NH7x_HHx_HJx_HKx" Motherboard: "CLEVO NHxxHJHK" BIOS: "Insyde 1.07.07TPCS1 — 2021-09-16" CPU: "Intel Core i7-11800H" Memory: "16 GB" GPU: "Intel UHD Graphics + NVIDIA GeForce GTX 1650" OS: "Windows 11 Pro — image version 10.0.26200.9168" SATAStorage: "128GB PCS 2.5-inch SSD — serial U9DI16295T" NVMeStorage: "1TB PCS PCIe M.2 SSD — Silicon Motion NVMe controller PCI VEN_126F DEV_2263" Author: "Maximilien Rigaut" Purpose: "Continuation and consolidation of the 2026-09-04 Clevo stability investigation" summary: > The investigation of DESKTOP-SAU019H was resumed on 2026-09-14 after earlier hard freezes, shutdown anomalies, an NVRAM-reset firmware warning, TPM / Measured Boot errors and repeated NVMe controller errors. Fast Startup was disabled by disabling hibernation. DISM successfully repaired the Windows component store and SFC found and repaired protected system-file corruption; a subsequent SFC scan after restart reported no remaining integrity violations. Storage enumeration before restart exposed only the 128 GB SATA SSD through Get-Disk and Get-PhysicalDisk, while the 1 TB NVMe remained known to Plug and Play. After restart, Windows positively enumerated the 1 TB PCS PCIe M.2 SSD and its Standard NVM Express Controller as present and working properly, with the controller linked at PCIe Gen3 x4, but Get-Disk still exposed only the 128 GB SATA SSD. A new stornvme Event 11 occurred at 16:23:21 on 2026-09-14, reporting a controller error on \Device\RaidPort2. Historical Event 129 resets and Event 11 controller errors against the same RaidPort2 are present on 2026-09-04 and repeatedly as far back as March 2026. The BIOS menus inspected during this session did not list the 1 TB NVMe/SATA device. The strongest current lead is therefore an intermittent or partial failure somewhere in the 1 TB NVMe storage path: the SSD itself, its M.2 connection, PCIe link / slot, controller behaviour, firmware interaction or power delivery. A failing SSD is not yet conclusively proven. The NVRAM / CMOS issue remains a separate open diagnostic track. timeline: - time: "2026-09-04 ~15:21" event: "NVMe controller error" notes: > stornvme Event 11 reported a controller error on \Device\RaidPort2. - time: "2026-09-04 ~16:30" event: "Firmware / NVRAM warning" notes: > Firmware displayed a first-boot or NVRAM-reset-condition warning. - time: "2026-09-04 ~16:32" event: "TPM / Measured Boot failure" notes: > TPM-WMI Event 1040 reported a critical pre-attestation health-check failure and referenced a MeasuredBoot diagnostic JSON file. - time: "2026-09-04 ~16:44" event: "Second confirmed NVMe controller error" notes: > stornvme Event 11 again reported a controller error on \Device\RaidPort2. - time: "2026-09-04 ~16:53" event: "Normal shutdown failure" notes: > Windows shutdown hung and a forced power-off was required. - time: "2026-09-04 later" event: "Restart test" notes: > A normal Windows restart completed successfully. - time: "2026-09-04 later" event: "Full-shutdown test bypassing Fast Startup" notes: > A full shutdown completed normally when Fast Startup was bypassed. - time: "2026-09-04 <17:23" event: "Hard freeze" notes: > A system freeze occurred before further configuration changes. - time: "2026-09-04 ~17:25" event: "Second hard freeze" notes: > A second freeze / hang was observed. - time: "2026-09-14 before storage retest" event: "Fast Startup disabled" notes: > powercfg /h off completed without error. Hibernation was disabled, which also disables Windows Fast Startup. - time: "2026-09-14 before storage retest" event: "DISM CheckHealth syntax attempt" notes: > DISM /CheckHealth returned Error 87 because /Online /Cleanup-Image were omitted. This was a command syntax issue and not a diagnostic failure. - time: "2026-09-14 before storage retest" event: "DISM component-store repair" notes: > DISM /Online /Cleanup-Image /RestoreHealth completed at 100 percent. DISM reported that the restore operation and overall operation completed successfully. - time: "2026-09-14 before storage retest" event: "SFC system-file repair" notes: > sfc /scannow found corrupt files and successfully repaired them. - time: "2026-09-14 before restart" event: "Initial disk inventory" notes: > Get-Disk exposed only Disk 0, the 128GB PCS 2.5-inch SATA SSD, Online and Healthy. The 1 TB NVMe was absent from Get-Disk. - time: "2026-09-14 before restart" event: "Initial physical-disk inventory" notes: > Get-PhysicalDisk exposed only the 128GB PCS 2.5-inch SATA SSD, Healthy / OK. The 1 TB NVMe was absent from this storage-provider view. - time: "2026-09-14 before restart" event: "PnP controller inventory" notes: > A Standard NVM Express Controller with hardware ID PCI\VEN_126F&DEV_2263 existed in Plug and Play history but was initially shown with status Unknown in the non-PresentOnly query. - time: "2026-09-14 before restart" event: "PnP disk inventory" notes: > A 1TB PCS PCIe M.2 SSD device entry existed in Plug and Play history but was initially shown with status Unknown in the non-PresentOnly query. Numerous historical USB storage entries were also present, demonstrating that this query included devices not currently attached. - time: "2026-09-14 before restart" event: "SATA reliability telemetry" notes: > Get-StorageReliabilityCounter returned data for DeviceId 0, the visible 128 GB SATA SSD: 2033 power-on hours, 40 C temperature, wear 0, ReadErrorsTotal 0, ReadErrorsUncorrected 0, maximum read latency 4253 ms and maximum write latency 6122 ms. This telemetry did not describe the 1 TB NVMe. - time: "2026-09-14 restart / BIOS inspection" event: "BIOS storage inspection" notes: > The BIOS menus inspected by the operator did not show a 1 TB NVMe or SATA storage device. This observation is retained as evidence, but the exact Clevo/Insyde BIOS menu capabilities have not been established; absence from a BIOS menu does not by itself prove failure to enumerate the PCIe function. - time: "2026-09-14 16:15:38" event: "NVMe controller arrival after reboot" notes: > Windows recorded the Standard NVM Express Controller LastArrivalDate as 2026-09-14 16:15:38. - time: "2026-09-14 ~16:15 onward" event: "PresentOnly controller enumeration" notes: > Get-PnpDevice -PresentOnly positively enumerated the Standard NVM Express Controller as present with Status OK. - time: "2026-09-14 ~16:15 onward" event: "PresentOnly 1 TB NVMe disk enumeration" notes: > Get-PnpDevice -PresentOnly -Class DiskDrive positively enumerated 1TB PCS PCIe M.2 SSD with Status OK alongside the 128 GB SATA SSD. - time: "2026-09-14 ~16:15 onward" event: "Get-Disk still omitted NVMe" notes: > Despite positive current PnP enumeration of the NVMe controller and disk device, Get-Disk still exposed only the 128 GB SATA SSD. - time: "2026-09-14 ~16:15 onward" event: "NVMe controller state inspected" notes: > The controller reported CM_PROB_NONE, ConfigManagerErrorCode CM_PROB_NONE, Status OK, Present True, HasProblem False and service stornvme. - time: "2026-09-14 ~16:15 onward" event: "NVMe PCIe link inspected" notes: > Device properties reported PCI bus 2, device 0, function 0, CurrentLinkSpeed 3, CurrentLinkWidth 4, MaxLinkSpeed 3 and MaxLinkWidth 4. The controller was therefore actively enumerated on a PCIe Gen3 x4 link at the time of inspection. - time: "2026-09-14 16:23:21" event: "New NVMe controller error" notes: > stornvme Event 11 reported: "The driver detected a controller error on \Device\RaidPort2." This event occurred during the current diagnostic session after the NVMe controller had arrived and been enumerated. - time: "2026-09-14 after 16:23" event: "Historical stornvme event review" notes: > Event-log review confirmed repeated Event 129 device resets and Event 11 controller errors on \Device\RaidPort2 on 2026-09-04 and repeatedly in June and March 2026. The pattern therefore predates the September incident. - time: "2026-09-14 after event review" event: "Post-repair SFC verification" notes: > A second sfc /scannow completed with "Windows Resource Protection did not find any integrity violations." findings: - title: "Fast Startup disabled for diagnosis" severity: "info" status: "completed" evidence: > powercfg /h off completed successfully. assessment: > Disabling hibernation disables Windows Fast Startup and removes the previously observed Fast Startup path from subsequent shutdown and boot tests. This is an appropriate diagnostic control and should remain in place until the investigation is complete. - title: "Windows component-store corruption" severity: "medium" status: "repaired" evidence: > DISM /Online /Cleanup-Image /RestoreHealth completed successfully. assessment: > A Windows integrity problem existed and was repaired. The evidence does not establish whether corruption contributed to the hard freezes, resulted from forced power-offs, or was unrelated. - title: "Protected Windows system-file corruption" severity: "medium" status: "repaired" evidence: > The first sfc /scannow found corrupt files and repaired them. A later sfc /scannow after restart reported no integrity violations. assessment: > Windows protected-file integrity is currently clean. Recurrent immediate corruption was not observed after repair and reboot. - title: "1 TB NVMe absent from Get-Disk / Get-PhysicalDisk" severity: "high" status: "confirmed" evidence: > Before restart, both Get-Disk and Get-PhysicalDisk exposed only the 128 GB PCS SATA SSD. After restart, Get-Disk still exposed only the SATA SSD even though PresentOnly PnP enumeration showed the NVMe controller and 1 TB NVMe disk as currently present. assessment: > Windows can enumerate the PCIe NVMe function and child disk device but does not currently expose the 1 TB NVMe as a normal usable disk through the storage-management layer. This strongly supports a problem in the NVMe storage path rather than a simple stale Device Manager entry. - title: "1 TB NVMe positively present on PCIe after reboot" severity: "high" status: "confirmed" evidence: > Get-PnpDevice -PresentOnly reported the Standard NVM Express Controller and 1TB PCS PCIe M.2 SSD as present with Status OK. The controller reported Present True, HasProblem False, CM_PROB_NONE and a PCIe Gen3 x4 link. assessment: > The controller was physically enumerated at the PCIe level during the session. Therefore the device was not simply absent from the PCIe bus at the moment these properties were queried. This does not demonstrate reliable access to the NVMe media. - title: "Current-session stornvme controller error" severity: "high" status: "confirmed" evidence: > stornvme Event 11 occurred at 2026-09-14 16:23:21 and reported a controller error on \Device\RaidPort2. assessment: > This is direct evidence of an active storage-controller/path failure during the current session. It materially raises the priority of the 1 TB NVMe path as a potential explanation for the historical freezes. - title: "Long-running RaidPort2 reset / error pattern" severity: "high" status: "confirmed" evidence: > Historical logs contain repeated stornvme Event 129 resets and Event 11 controller errors on \Device\RaidPort2 on 2026-09-04 and repeatedly in June and March 2026. assessment: > The storage-path problem is recurrent and predates the September 2026 hard-freeze investigation. This makes a one-off Windows corruption event an insufficient explanation for the controller-error history. - title: "NVMe physical failure not yet proven" severity: "high" status: "open" evidence: > The NVMe controller and child disk can currently enumerate on PCIe and report PnP Status OK, while stornvme simultaneously records controller errors and the disk remains absent from Get-Disk. assessment: > The evidence is compatible with a failing NVMe SSD, but does not yet distinguish the SSD itself from the M.2 connection, motherboard slot / PCIe path, controller firmware, system firmware interaction or power delivery. The drive should not yet be condemned solely from Event 11. - title: "BIOS does not list 1 TB storage device in inspected menus" severity: "medium" status: "observed" evidence: > During the 2026-09-14 BIOS inspection, no 1 TB NVMe/SATA device was visible to the operator. assessment: > Retain this as an important observation. However, the available evidence does not establish whether this specific Insyde BIOS exposes NVMe device inventory in its UI. Windows later demonstrated active PCIe enumeration, so absence from the visible BIOS menus must not be treated as conclusive proof that firmware failed to enumerate the controller. - title: "NVRAM / CMOS retention issue" severity: "high" status: "open" evidence: > The original firmware warning reported a first-boot or NVRAM-reset condition. assessment: > The warning remains unexplained. A failed RTC/CMOS battery is possible but not proven. A controlled unpowered retention test of BIOS time and settings remains required. This issue should remain diagnostically separate from the NVMe fault unless further evidence connects them. - title: "TPM / Measured Boot pre-attestation failure" severity: "medium" status: "open" evidence: > TPM-WMI Event 1040 previously reported a critical pre-attestation health-check failure and referenced a MeasuredBoot JSON diagnostic file. assessment: > The temporal association with the NVRAM warning makes a firmware-state relationship plausible, but no causal relationship has been proven. The TPM should not be cleared merely as a diagnostic experiment. - title: "128 GB SATA SSD telemetry" severity: "info" status: "currently-normal" evidence: > The visible SATA SSD reported Healthy / OK, 40 C, wear 0, ReadErrorsTotal 0 and ReadErrorsUncorrected 0 in the supplied reliability counter output. assessment: > No direct read-error evidence was observed for the 128 GB SATA SSD in this checkpoint. The unusually high recorded maximum latency values are retained for completeness but are not by themselves sufficient to diagnose failure. sections: - title: "Commands and results — boot / shutdown control" content: - type: bullets items: - > powercfg /h off — completed without error; hibernation and Fast Startup disabled. - > Previous full-shutdown test bypassing Fast Startup — completed normally. - > Previous normal restart test — completed normally. - title: "Commands and results — Windows integrity" content: - type: bullets items: - > DISM /CheckHealth — returned Error 87 because the required /Online /Cleanup-Image arguments were omitted. This result has no diagnostic significance for Windows health. - > DISM /Online /Cleanup-Image /RestoreHealth — reached 100 percent and completed successfully. - > First sfc /scannow — found corrupt files and successfully repaired them. - > Second sfc /scannow after restart — no integrity violations found. - title: "Commands and results — initial storage inventory" content: - type: bullets items: - > Get-Disk — only Disk 0, 128GB PCS 2.5-inch SSD, SATA, Online, Healthy, 128035676160 bytes. - > Get-PhysicalDisk — only 128GB PCS 2.5-inch SSD, SSD/SATA, Healthy / OK, 128035676160 bytes. - > Get-PnpDevice -Class SCSIAdapter — historical list included a Standard NVM Express Controller with PCI VEN_126F DEV_2263, initially shown as Unknown without -PresentOnly. - > Get-PnpDevice -Class DiskDrive — historical list included 1TB PCS PCIe M.2 SSD, initially shown as Unknown without -PresentOnly, along with many disconnected historical USB devices. - > Get-StorageReliabilityCounter — returned telemetry only for the visible DeviceId 0 / 128 GB SATA SSD: 2033 power-on hours, 40 C, wear 0, no reported read errors. - title: "Commands and results — post-reboot NVMe enumeration" content: - type: bullets items: - > Get-PnpDevice -PresentOnly -Class SCSIAdapter — Standard NVM Express Controller present, Status OK. - > Get-PnpDevice -PresentOnly -Class DiskDrive — 1TB PCS PCIe M.2 SSD present, Status OK; 128GB PCS 2.5-inch SSD also present. - > Get-Disk — still exposed only the 128 GB SATA SSD. - > PresentOnly PCI VEN_126F DEV_2263 query — controller Status OK, class SCSIAdapter. - > Controller detailed properties — CM_PROB_NONE, ConfigManagerErrorCode CM_PROB_NONE, Present True, HasProblem False, Service stornvme. - > Controller PCI properties — bus 2, device 0, function 0; current and maximum link speed 3; current and maximum link width 4; AER capability present. - > Controller LastArrivalDate — 2026-09-14 16:15:38. - > Controller bus relation and child relationship both reference SCSI\Disk&Ven_NVMe&Prod_1TB_PCS_PCIe_M.2. - title: "Commands and results — storage event history" content: - type: bullets items: - > 2026-09-14 16:23:21 — stornvme Event 11: controller error on \Device\RaidPort2. - > 2026-09-04 17:22:40 — stornvme Event 11: controller error on \Device\RaidPort2. - > 2026-09-04 17:20:46 — stornvme Event 129: reset to \Device\RaidPort2 issued. - > 2026-09-04 17:12:39 — stornvme Event 129: reset to \Device\RaidPort2 issued. - > 2026-09-04 16:57:14 — stornvme Event 11: controller error on \Device\RaidPort2. - > 2026-09-04 16:55:20 — stornvme Event 129: reset to \Device\RaidPort2 issued. - > 2026-09-04 16:44:57 — stornvme Event 11: controller error on \Device\RaidPort2. - > 2026-09-04 16:43:03 — stornvme Event 129: reset to \Device\RaidPort2 issued. - > 2026-09-04 15:21:28 — stornvme Event 11: controller error on \Device\RaidPort2. - > 2026-09-04 15:19:32 — stornvme Event 129: reset to \Device\RaidPort2 issued. - > Additional Event 11 / Event 129 occurrences were confirmed on 2026-06-09, 2026-06-08, 2026-03-26, 2026-03-21, 2026-03-20, 2026-03-19 and 2026-03-18. - title: "BIOS / firmware observations" content: - type: bullets items: - > Earlier investigation observed: "A first boot or NVRAM reset condition has been detected." - > BIOS used during this investigation: Insyde 1.07.07TPCS1. - > Operator inspection on 2026-09-14 found no visible 1 TB NVMe/SATA device in the BIOS menus examined. - > No BIOS update or firmware flash has been performed during this diagnostic sequence. - > No TPM clear has been performed. - title: "Interpretation at this checkpoint" content: - type: callout title: "Current state" kind: "warning" text: > Windows software integrity has been repaired and verified clean, but the storage problem remains active. The 1 TB NVMe controller and child disk can enumerate on PCIe while the disk remains absent from Get-Disk, and a new Event 11 controller error was generated during the current session. This substantially strengthens the NVMe / M.2 / PCIe storage-path hypothesis for the hard-freeze investigation. It still does not identify which physical component in that path is defective. - type: bullets items: - "Fast Startup is disabled and should remain disabled during diagnosis." - "Windows component-store corruption has been repaired." - "Windows protected system files are currently clean." - "The 128 GB SATA SSD is the only disk exposed by Get-Disk." - "The 1 TB NVMe controller is currently enumerated on PCIe Gen3 x4." - "The 1 TB NVMe child disk is currently visible to PresentOnly PnP." - "The 1 TB NVMe is nevertheless absent from Get-Disk." - "A new stornvme Event 11 occurred during this diagnostic session." - "RaidPort2 Event 11 / 129 faults recur over several months." - "The 1 TB SSD itself is a primary suspect but is not yet conclusively proven defective." - "M.2 seating, slot / motherboard path and power delivery remain plausible alternatives." - "The NVRAM / CMOS issue remains open and should not yet be merged with the NVMe diagnosis." - "TPM / Measured Boot evidence remains preserved and unresolved." - "A BIOS flash is not currently justified as a first remediation step." actions_title: "Recommended next diagnostic steps" actions: - title: "Back up important data" status: "priority" text: > Ensure all important data is backed up before further manipulation or testing. Recurrent NVMe controller resets/errors justify treating the storage path as potentially unreliable. - title: "Perform controlled full shutdown" status: "pending" text: > With Fast Startup already disabled, perform a full shutdown before any physical inspection. commands: - "shutdown /s /t 0" - title: "Physically inspect and reseat the 1 TB M.2 NVMe" status: "pending" text: > Disconnect AC power and, where service procedure permits, disconnect the internal battery before handling the M.2 drive. Photograph the SSD label and exact part number. Inspect the retaining screw, seating, contacts and socket; remove and firmly reseat the SSD. Do not change unrelated firmware settings during the same test. - title: "Retest storage enumeration after reseat" status: "pending" text: > After reseating, boot the machine and determine whether the 1 TB NVMe becomes a normal disk and whether new stornvme errors appear. commands: - "Get-Disk" - "Get-PhysicalDisk" - "Get-PnpDevice -PresentOnly -Class SCSIAdapter" - "Get-PnpDevice -PresentOnly -Class DiskDrive" - title: "Review post-reseat stornvme events" status: "pending" text: > Check the System log for fresh Event 11 or Event 129 entries immediately after the test boot. commands: - "Get-WinEvent -FilterHashtable @{ LogName='System'; ProviderName='stornvme'; StartTime=(Get-Date).AddMinutes(-30) } | Format-List TimeCreated, Id, LevelDisplayName, Message" - title: "Isolate SSD versus motherboard path if errors persist" status: "pending" text: > If the NVMe remains absent from Get-Disk or Event 11 / 129 recurs after reseating, test the SSD in another known-good compatible M.2 slot, external NVMe enclosure or another system where practical. Conversely, test a known-good compatible NVMe in the Clevo slot if available. The purpose is to distinguish drive failure from slot / motherboard / power failure before replacing parts. - title: "Perform NVRAM / CMOS retention test" status: "pending" text: > Record BIOS date/time and relevant settings, fully power down, disconnect AC for an extended interval, then boot directly into BIOS and verify retention. If settings or time are lost, investigate the RTC/CMOS battery, its connection and the motherboard RTC power path. - title: "Preserve TPM / Measured Boot evidence" status: "pending" text: > Keep the referenced MeasuredBoot JSON before any TPM, Secure Boot or firmware reset operation. Do not clear the TPM solely as a diagnostic experiment. commands: - 'C:\WINDOWS\Logs\MeasuredBoot\0000000053-0000000000.json' - title: "Defer BIOS flashing" status: "hold" text: > Do not flash BIOS/EC firmware while the machine has an unresolved intermittent storage fault unless a later, model-specific remediation path explicitly requires it and power/storage stability has first been established. closing: > At this checkpoint the Windows repair is complete and verified, while the storage-path fault remains active and repeatable. The strongest evidence is the coexistence of a currently enumerated PCIe Gen3 x4 NVMe controller, failure of the 1 TB NVMe to appear in Get-Disk, and a fresh stornvme Event 11 on \Device\RaidPort2 during the 2026-09-14 session, backed by months of historical Event 11 and Event 129 activity. The immediate priority is preservation of data followed by physical M.2 inspection/reseat and controlled device-versus-slot isolation. The firmware NVRAM/CMOS and TPM/Measured Boot findings remain unresolved parallel tracks and should not yet be assumed to share the same root cause.