Fix it now
0x00000154 is the kernel memory store component catching an exception it was not expecting. Microsoft’s published troubleshooting for it is memory diagnostics, driver updates and system file repair – and the two related store bug checks that do carry a cause section both point at physical memory corruption rather than at the disk.
sfc /scannow
chkdsk C: /scan
- Test memory first. Type Windows Memory Diagnostics at Start, let it restart, press F1 during the test and set the test mix to Extended, then read the MemoryDiagnostics-Results entry in the System log afterwards.
- Check the System log for errors in the same window as the crash, and Device Manager for any device marked with an exclamation point.
- Update third-party drivers, particularly storage and endpoint drivers, from the hardware or software vendor rather than from Windows Update.
- Run
sfc /scannow, thenDISM /Online /Cleanup-Image /RestoreHealthif it reports files it could not repair. - Only after memory and drivers are clean, check the drive:
chkdsk C: /f /r, then the manufacturer’s own health tool, then a firmware update if one is published for that model.
If the crash only happens on a resume and never on a cold start, test with hibernation off – powercfg /hibernate off – to separate the two. That is a diagnostic to narrow the picture, not a documented fix.
If memory tests clean and the machine is stable with current drivers, stop here. The next section explains what the store is and which sibling code you are actually looking at.
Why it happens
Windows compresses memory rather than paging it out immediately. Pages the memory manager wants to reclaim are compressed and held in a store in RAM, and only written to disk if the pressure continues. The store manager owns that machinery: the compressed chunks, the page records that locate them, the heap those records live in. It is a large amount of bookkeeping, all of it in memory, and it is checked as it is used.
0x00000154 is the store component catching an exception it did not expect. Parameter 1 points at the store context or data manager and parameter 2 carries the exception information. Its own page publishes no cause section – the troubleshooting it gives is Event Viewer, Windows Memory Diagnostics, Device Manager and sfc /scannow, which is a memory-and-drivers list, not a storage one.
The two related codes that do carry a cause section are the ones worth reading, because they say the same thing as each other. 0x0000012B is the memory manager detecting corruption that could only have been caused by a component accessing memory using physical addressing, and 0x000001C7 is the store component detecting corruption in its own data structures. Both list the same three causes: defective RAM hardware, a driver or device incorrectly modifying physical pages through an incorrect DMA operation or MDL, and a hardware device or firmware corrupting memory – Microsoft’s example being firmware illegally modifying physical pages across a power transition.
0x0000012B has two parameter sets and the second one is this article’s subject exactly: a store manager memory error, which may be an authentication failure, a CRC failure or a decompression failure. The first set is the single-bit page-not-zero case. Microsoft also notes that the compressed store manager detects and corrects single-bit corruption on its own without bug checking, so if it has raised 0x12B the corruption was not a single-bit error – which makes it stronger evidence, not weaker.
Defective memory
You have this one if 0x00000154 alternating with 0x0000012B or 0x000001C7, or appearing alongside 0x0000001A. Crashes are not tied to one application.
- Run Windows Memory Diagnostics with the test mix set to Extended and read the result in the System log.
- Return firmware to stock settings first. A memory profile is an overclock and will change the result.
- Test one module at a time where more than one is fitted, and reseat each one.
- Run the manufacturer’s own diagnostics as well; several vendors’ tools test more thoroughly than the built-in one.
A driver or device writing to physical pages it should not
You have this one if 0x0000012B or 0x000001C7, with memory that tests clean.
- Read the published cause literally: a driver or device incorrectly modifying physical pages via an incorrect DMA operation or associated MDL.
- Update the drivers for devices that do their own DMA – storage controllers, network adapters, capture and acquisition hardware.
- Remove any add-in card that is not needed and see whether the crash follows.
- On a test machine, run the standard Driver Verifier option set, which includes DMA verification, against the suspect drivers.
Firmware corrupting pages across a power transition
You have this one if Crashes cluster on resume from sleep or hibernation rather than on cold starts.
- Update the system firmware from the machine or board vendor. Microsoft names firmware illegally modifying physical pages across a power transition as a cause of both store-related codes.
- As a diagnostic, turn hibernation off with
powercfg /hibernate offand see whether cold starts alone are stable. - Update the chipset and power management drivers from the same vendor.
- If the machine is stable only when it never sleeps, that is a firmware problem to report, not a setting to leave disabled forever.
Storage, once the memory case is closed
You have this one if Memory and drivers are clean, and the System log carries storage errors around each crash.
- Run
chkdsk C: /f /r, orRepair-Volume -DriveLetter C -OfflineScanAndFix, which Microsoft documents as the equivalent ofchkdsk /f. - Read the drive’s health data with the manufacturer’s tool and apply any firmware update it publishes.
- Check the System log for storage errors immediately before each bug check.
- Image the disk before running repairs if the drive is already producing read errors.
Full reference
The store bug checks, and what each publishes
| Code | Published meaning | Cause section |
|---|---|---|
0x00000154 |
The kernel memory store component caught an unexpected exception | None. Troubleshooting is memory diagnostics, drivers, Event Viewer and sfc |
0x0000012B |
Memory manager corruption that could only come from a component accessing memory by physical address | Defective RAM, incorrect DMA or MDL, firmware modifying pages across a power transition |
0x00000127 |
A page that should have been filled with zeros was not | A hardware error, or a privileged operating system component modifying a page after freeing it |
0x000001C7 |
The store component detected corruption in its own data structures | The same three causes as 0x0000012B |
0x00000191 |
Nothing published beyond the symbolic name; all four parameters are reserved | None |
0x000001C7’s corruption IDs
| Corruption ID | What was found |
|---|---|
0x0 |
A chunk heap buffer’s hash does not match |
0x1 |
An unhandled exception on the store thread, with a chunk heap buffer hash mismatch that is likely its source |
0x2 |
Page data appears corrupt during a read and the page record’s heap buffer hash does not match |
0x3 |
Page data appears corrupt during a read and the page record has changed since the read started |
Commands and cmdlets that apply here
| Command | What it does |
|---|---|
sfc /scannow |
Repairs missing or corrupted system files |
DISM /Online /Cleanup-Image /RestoreHealth |
Repairs the component store when sfc cannot |
chkdsk C: /scan |
Scans the volume online, without a restart |
chkdsk C: /f /r |
Fixes errors and locates bad sectors. /r includes /f and needs the volume locked, so on the system drive it offers to run at the next restart |
Repair-Volume -DriveLetter C -OfflineScanAndFix |
Documented as the equivalent of chkdsk /f |
powercfg /hibernate off |
Turns hibernation off, as a diagnostic for resume-only crashes |
Disable-MMAgent -MemoryCompression |
Turns memory compression off; a test, and a performance cost if left |
Repair-Volume -OfflineScanAndFix is documented as equivalent to chkdsk /f, not to chkdsk /f /r. If you need the bad-sector pass, run chkdsk with /r.
Turning memory compression off, and when not to
Disable-MMAgent -MemoryCompression stops Windows compressing memory, which removes the store from the picture entirely. As a diagnostic on a machine that keeps producing store bug checks it is legitimate: if the crashes stop, the store was the component catching the corruption. It is not a fix. The corruption that the store detected is still there, being detected by something less careful, and the machine pages to disk more instead. Turn it back on once the underlying memory, DMA or firmware problem is dealt with.
0x00000191, and what it is not
PF_DETECTED_CORRUPTION is published as a symbolic name with all four parameters marked reserved. There is no description, no cause and no resolution. Readings of it as corruption in page-file-backed data are inferences from the name. If it appears in a run of dumps alongside 0x154 or 0x12B, work the codes that are documented and treat this one as another symptom of the same underlying fault.
Order of work, and why it matters
- Test memory properly, with an extended pass and stock firmware settings.
- Update drivers for anything that does its own DMA, and the system firmware.
- Repair system files with sfc and DISM.
- Check the volume with chkdsk and read the drive’s health data.
- Only then consider replacing the drive – and if you replace it first and the crash returns, you have lost the machine’s history along with the money.
Every code this article covers
| Code | What it points at | Source |
|---|---|---|
0x00000154 |
UNEXPECTED_STORE_EXCEPTION: the kernel memory store component caught an unexpected exception. Parameter 1 points at the store context or data manager and parameter 2 carries the exception information. Published troubleshooting is Event Viewer, Windows Memory Diagnostics, Device Manager and sfc /scannow | Microsoft Learn |
0x0000012B |
FAULTY_HARDWARE_CORRUPTED_PAGE: the memory manager detected corruption that could only have been caused by a component accessing memory using physical addressing. Two parameter sets – a single-bit page-not-zero case and a compressed store manager error. Causes: defective RAM, an incorrect DMA operation or MDL, or firmware illegally modifying physical pages across a power transition | Microsoft Learn |
0x00000127 |
PAGE_NOT_ZERO: a page that should have been filled with zeros was not, because of a hardware error or because a privileged component of the operating system modified a page after freeing it | Microsoft Learn |
0x000001C7 |
STORE_DATA_STRUCTURE_CORRUPTION: the store component detected corruption in its data structures, with a corruption-ID table centred on chunk heap buffer hash mismatches, and the same physical-memory causes as 0x0000012B | Microsoft Learn |
0x00000191 |
PF_DETECTED_CORRUPTION: Microsoft publishes the symbolic name and lists all four parameters as reserved. No description, cause or resolution is given, so it is treated here as an unexplained corruption report to correlate with the documented codes | not published by the vendor |
Confirm the fix worked
- Windows Memory Diagnostics passes an extended run with the result recorded in the System log.
sfc /scannowreports no integrity violations.- The machine survives repeated sleep and resume cycles if that is when it used to crash.
- No new dump files appear in
%SystemRoot%\Minidumpover several days of normal use. - Memory compression is back on if you turned it off as a test –
Enable-MMAgent -MemoryCompression.
Questions people ask about this
Is this an SSD problem?
Not according to what Microsoft publishes. The 0x154 page’s troubleshooting is memory diagnostics, drivers, Event Viewer and sfc, and the two related store codes that do have a cause section both name physical memory corruption – defective RAM, bad DMA, or firmware modifying pages across a power transition. Check the drive, but not first.
Should I turn memory compression off?
As a test, yes. It removes the store from the picture and tells you whether that is where the corruption is being caught. As a permanent setting it costs performance and leaves the underlying fault in place.
What does 0x00000191 mean?
Nothing that Microsoft publishes. The page gives the name PF_DETECTED_CORRUPTION and marks all four parameters reserved. Work the documented codes it appears alongside.
Does fast startup cause this?
There is no published link. It is worth turning off as a diagnostic, because it separates a genuine cold start from a resume and the store codes do list firmware modifying pages across a power transition as a cause. Treat it as a way to narrow the picture rather than as the answer.
Do I need to reinstall Windows?
No. Reinstalling replaces system files, which sfc /scannow and DISM already do, and it does nothing about failing memory, a DMA-capable device misbehaving, or firmware. Work the memory and driver route first.
