Skip to content

Est. 2011·Microsoft Partner ID 7033487·Delivery under 3 min·Support 7 days a week

Your vault is empty.

License Error 0x00000149

REFS_FILE_SYSTEM 0x00000149 and file system filter driver bugchecks

12 min read Updated October 5, 2026 Windows Crashes & Boot Recovery

Fix it now

0x00000149 is a file system error in ReFS, reported the same way the other file system bug checks are: with an exception record and a context record in the parameters. Three of the codes grouped with it are documented filter driver faults, and those are the ones where removing or updating a filter is the fix.

Run these in an elevated Command Prompt to list the loaded minifilters

fltmc
fltmc help
  1. Read the fltmc output and identify every filter that is not part of Windows. Anti-malware, backup, encryption, DLP, replication and virtualisation products all attach here.
  2. Match each one to a product that is actually installed and supported on this Windows build. A filter belonging to something you removed is the first candidate.
  3. Take one out of the path with fltmc unload <name> and re-run the workload. Some products reload theirs from a service, so confirm with fltmc afterwards.
  4. Where a filter cannot be unloaded, use the vendor’s own removal utility rather than the installed programs list.
  5. Update whichever filters remain to builds their vendors support on your Windows version, then re-test.

On a ReFS volume in a storage pool, check the pool and virtual disk health too: Get-StoragePool and Get-VirtualDisk in PowerShell. A degraded virtual disk is a different problem wearing the same stop code.

If the machine is stable with the filter stack tidied, stop here. The next section separates the codes that are documented filter faults from the ones that are not.

Why it happens

A file system request does not go straight to the file system. It passes through the filter manager, which offers it to every registered minifilter in altitude order – anti-malware, encryption, backup, replication, virtualisation – each of which can inspect it, modify it, complete it or pass it on. Most of those filters come from third parties, and the file system below them assumes they follow the rules.

0x00000149 itself is the ReFS driver reporting a file system error, and its page carries no cause attribution at all. Parameter 1 is the source line, parameters 2 and 3 the exception and context records when RefsExceptionFilter is on the stack, and parameter 4 the exception address. The documented way in is .exr on parameter 2 to see the exception, then .cxr on parameter 3 and kb for a usable stack. 0x00000026, the CD file system, is the same structure with CdExceptionFilter – and it does have documented causes, the same two as the other file system codes: disk corruption and nonpaged pool depletion.

The three codes in this group that genuinely name filters are different. 0x00000108 is an unrecoverable problem in a third-party file system or file system filter, with parameter 1 identifying which – the only value Microsoft documents is 1, Polyserve’s psfs.sys – and the same two documented causes of disk corruption and nonpaged pool depletion. 0x000000F9 is a driver returning STATUS_REPARSE to an IRP_MJ_CREATE request that had no trailing names, which is only valid for creates that do have them; parameter 1 is the device object that was opened. 0x0000010C is a violation in the FsRtl Extra Create Parameter package, with parameter 1 naming the exact violation.

That 0x0000010C table is worth reading because it is unusually specific about what a filter did wrong: an invalid ECP signature from a bad pointer or memory corruption (0x1), undefined flags set (0x2), an ECP not allocated by the FsRtl (0x3), flags illegal for a parameter passed by a create caller (0x4), a size smaller than the header (0x5), an ECP freed while its list pointers are still non-empty (0x6), and the equivalent list-level violations at 0x11 to 0x15. Every one of them is a defect in whoever built the parameter, which on a real machine means a filter driver.

A third-party filter that is out of support on this Windows build

You have this one if fltmc names a filter from a product whose version predates the Windows build, and the code is 0x00000108, 0x000000F9 or 0x0000010C.

  1. Check the product’s supported platform list against the Windows version and build actually running.
  2. Update to a supported build, or remove the product with the vendor’s own removal utility.
  3. Confirm with fltmc after restarting, rather than trusting the uninstaller.
  4. Where the product is essential and no supported build exists, that is a procurement problem rather than a configuration one.

Several filters stacked on the same path

You have this one if fltmc shows filters from more than one product doing the same job – two anti-malware agents, an old backup agent beside a new one, DLP over encryption.

  1. Reduce to one product per function. Only one real-time scanner should own the file system filter path.
  2. Where two genuinely have to coexist, apply each vendor’s documented exclusions for the other.
  3. Unload one at a time with fltmc unload <name> and re-run the workload to find the pair that interact.
  4. Restart and confirm the resulting list is what you intended.

Nonpaged pool depletion

You have this one if The code is 0x00000108 or 0x00000026, the volume is healthy, and the crash follows a sustained job.

  1. In the debugger, !vm 1 for total pool usage and !poolused 2 for nonpaged pool by tag.
  2. Identify the driver behind the tag consuming the most and update or remove it.
  3. Microsoft’s published remedy for the depletion case on these codes is to add physical memory, which raises the nonpaged pool available to the kernel – but find the consumer first.
  4. Correlate the crashes with scheduled work: backups, replication and indexing all hold pool for the duration.

The storage underneath is degraded

You have this one if A ReFS volume on Storage Spaces, with a pool or virtual disk that is not healthy.

  1. Run Get-StoragePool and Get-VirtualDisk and read the health and operational status columns.
  2. Replace failed physical disks and let the virtual disk repair before drawing conclusions about the file system.
  3. ReFS performs its repairs online and keeps the volume available while handling most non-correctable corruption, with rare cases requiring it to go offline – so a volume that is offline is telling you something.
  4. Check the System log for storage errors in the same window as the crash.

Full reference

Which of these codes actually names a filter

Code Published meaning Filter attribution
0x00000149 A file system error has occurred in ReFS None published
0x00000026 A problem occurred in the CD file system None; documented causes are disk corruption and nonpaged pool depletion
0x00000108 An unrecoverable problem in a third-party file system or file system filter Yes, by name in the code itself
0x000000F9 A driver returned STATUS_REPARSE for a create with no trailing names Yes – a driver in the create path
0x0000010C A violation in the FsRtl Extra Create Parameter package Yes – whoever built or freed the ECP

0x0000010C parameter 1, the violation types

Value What was violated
0x1 The ECP signature is invalid, from a bad pointer or memory corruption
0x2 The ECP has undefined flags set
0x3 The ECP was not allocated by the FsRtl
0x4 The ECP has flags set that are illegal for a parameter passed by a create caller
0x5 The ECP is corrupted; its size is smaller than the header size
0x6 An ECP being freed has non-empty list pointers – it may still be on a list
0x11 to 0x15 The same five violations at ECP list level, including an empty list passed by a create caller

Working the dump on a file system bug check

Command What it does
.exr <parameter 2> Displays the exception record
.cxr <parameter 3> Sets the context to the context record
kb The stack backtrace, which after .cxr shows where the fault actually was
!analyze -v Automated analysis and the parameter breakdown
fltmc The loaded minifilters and their altitudes
fltmc unload <name> Takes one filter out of the path
Get-StoragePool, Get-VirtualDisk Storage Spaces health, where the ReFS volume sits on a pool

ReFS repairs itself, up to a point

ReFS is built to detect and repair corruption while the volume stays online, and to keep it available through most non-correctable corruption rather than requiring an offline check. That is a real difference from NTFS and it changes the shape of an investigation: there is no offline pass to run as a matter of routine, and a ReFS volume that has gone offline has done so because a rare case required it, not because somebody forgot to schedule a check. So the useful work is in the dump and in the filter stack, not in looking for a repair tool.

Reading the filter list without guessing

  1. Run fltmc and write down every filter name and altitude.
  2. Map each name to an installed product. Vendors publish their filter names; anything you cannot account for is worth chasing.
  3. Flag anything belonging to software that has been uninstalled – those are the ones that survive removal and keep running without the service that manages them.
  4. Flag duplicated functions: two real-time scanners, two backup agents, encryption plus a virtual drive tool.
  5. Test by unloading one at a time and re-running the workload, rather than removing several products at once and losing the ability to attribute the result.

When the filter stack is clean

  • Check nonpaged pool under the workload that triggers the crash, if the code is 0x00000108 or 0x00000026.
  • Read the System log for storage errors and for Storage Spaces health changes in the same window.
  • Confirm firmware on the drives and the controller is current, particularly in a storage pool.
  • Compare several dumps: a stable parameter 1 on 0x0000010C means the same violation each time, which is a specific defect to report.
  • Where the volume is served over SMB or a redirector, test locally as well to separate the network path from the file system.

When a licence is the actual fix

Where the filter in the path belongs to a product that is out of support on the Windows build you are running, no exclusion or setting will make it safe – it needs to be replaced or removed. Microsoft Defender ships with Windows, costs nothing, and leaves one fewer third-party filter on the create path, which is the right answer for a machine that does not need central management. Where managed policy, device control and reporting across a fleet are required, a current licence restores update delivery so the minifilter tracks the Windows builds in service. Arco supplies ESET PROTECT Entry endpoint licences and can check which seat count and term matches the estate you actually run.

Every code this article covers

Code What it points at Source
0x00000149 REFS_FILE_SYSTEM: a file system error has occurred in ReFS. Parameter 1 is the source line, parameters 2 and 3 the exception and context records when RefsExceptionFilter is on the stack, and parameter 4 the exception address. No cause is published Microsoft Learn
0x0000010C FSRTL_EXTRA_CREATE_PARAMETER_VIOLATION: a violation was detected in the FsRtl Extra Create Parameter package. Parameter 1 names the violation – invalid signature, undefined flags, an ECP not allocated by FsRtl, illegal caller flags, a size smaller than the header, an ECP freed while still on a list, and the same set at list level Microsoft Learn
0x000000F9 DRIVER_RETURNED_STATUS_REPARSE_FOR_VOLUME_OPEN: a driver returned STATUS_REPARSE to an IRP_MJ_CREATE request with no trailing names, which is only valid for creates that do have trailing names. Parameter 1 is the device object that was opened Microsoft Learn
0x00000108 THIRD_PARTY_FILE_SYSTEM_FAILURE: an unrecoverable problem occurred in a third-party file system or file system filter. Parameter 1 identifies the file system, and the only value Microsoft documents is 1, Polyserve (psfs.sys). Documented causes are disk corruption and nonpaged pool depletion Microsoft Learn
0x00000026 CDFS_FILE_SYSTEM: a problem occurred in the CD file system. Same parameter structure as the other file system bug checks with CdExceptionFilter, and the same documented causes of disk corruption and nonpaged pool depletion Microsoft Learn

Confirm the fix worked

  1. fltmc lists only filters belonging to products that are installed and supported on this Windows build.
  2. The workload that crashed the machine runs to completion twice with no new dump.
  3. Get-StoragePool and Get-VirtualDisk report healthy where the volume sits on a storage pool.
  4. The System log shows no storage or filter manager errors over several days.
  5. If a filter was unloaded for the test, the machine is now in the state you intend rather than in the test state.

Questions people ask about this

Is 0x00000149 always a filter driver problem?

No, and this is where a lot of advice overreaches. The ReFS page publishes no cause at all – only the parameters and how to read them. Filter attribution is documented for 0x00000108, 0x000000F9 and 0x0000010C, and those are the codes where a filter is the thing to work on.

Is there a chkdsk for ReFS?

Not in the way there is for NTFS. ReFS performs its repairs online and keeps the volume available through most non-correctable corruption, with rare cases requiring it to go offline. The investigation is the dump and the filter stack, not an offline check.

How do I know which filter to unload?

Start with anything belonging to a product that is no longer installed, then anything duplicating another product’s function. Unload one at a time with fltmc unload and re-run the workload, so the result can be attributed.

What does parameter 1 of 1 mean on 0x00000108?

Polyserve, psfs.sys. It is the only value Microsoft documents for that parameter, which tells you the field is meant to name the file system rather than to encode a cause.

Does adding memory fix 0x00000108?

It is Microsoft’s published remedy for the nonpaged pool depletion case on that code, because more physical memory raises the nonpaged pool available to the kernel. But a driver leaking pool will exhaust a larger pool too, so identify the consumer with !poolused 2 first.

Related error codes

Was this article helpful?

Your feedback helps us improve our documentation.

Related articles

License Error DRIVER_POWER_STATE_FAILURE 0x0000009F on sleep, hibernate or shutdown Free Fix 0xC0000001 boot loop and session initialisation failures on Windows Server Free Fix DPC_WATCHDOG_VIOLATION 0x00000133 – PC freezes then blue screens Free Fix UNEXPECTED_KERNEL_MODE_TRAP 0x0000007F – double fault and stack overflow
← Back to Knowledge Base