Pushing the Limits of Windows: Paged and Nonpaged Pool


In previous Pushing the Limits posts, I described the two most basic system resources, physical memory and virtual memory. This time I’m going to describe two fundamental kernel resources, paged pool and nonpaged pool, that are based on those, and that are directly responsible for many other system resource limits including the maximum number of processes, synchronization objects, and handles.

Here’s the index of the entire Pushing the Limits series. While they can stand on their own, they assume that you read them in order.

Pushing the Limits of Windows: Physical Memory

Pushing the Limits of Windows: Virtual Memory

Pushing the Limits of Windows: Paged and Nonpaged Pool

Pushing the Limits of Windows: Processes and Threads

Pushing the Limits of Windows: Handles

Pushing the Limits of Windows: USER and GDI Objects – Part 1

Pushing the Limits of Windows: USER and GDI Objects – Part 2

Paged and nonpaged pools serve as the memory resources that the operating system and device drivers use to store their data structures. The pool manager operates in kernel mode, using regions of the system’s virtual address space (described in the Pushing the Limits post on virtual memory) for the memory it sub-allocates. The kernel’s pool manager operates similarly to the C-runtime and Windows heap managers that execute within user-mode processes.  Because the minimum virtual memory allocation size is a multiple of the system page size (4KB on x86 and x64), these subsidiary memory managers carve up larger allocations into smaller ones so that memory isn’t wasted.

For example, if an application wants a 512-byte buffer to store some data, a heap manager takes one of the regions it has allocated and notes that the first 512-bytes are in use, returning a pointer to that memory and putting the remaining memory on a list it uses to track free heap regions. The heap manager satisfies subsequent allocations using memory from the free region, which begins just past the 512-byte region that is allocated.

Nonpaged Pool

The kernel and device drivers use nonpaged pool to store data that might be accessed when the system can’t handle page faults. The kernel enters such a state when it executes interrupt service routines (ISRs) and deferred procedure calls (DPCs), which are functions related to hardware interrupts. Page faults are also illegal when the kernel or a device driver acquires a spin lock, which, because they are the only type of lock that can be used within ISRs and DPCs, must be used to protect data structures that are accessed from within ISRs or DPCs and either other ISRs or DPCs or code executing on kernel threads. Failure by a driver to honor these rules results in the most common crash code, IRQL_NOT_LESS_OR_EQUAL.

Nonpaged pool is therefore always kept present in physical memory and nonpaged pool virtual memory is assigned physical memory. Common system data structures stored in nonpaged pool include the kernel and objects that represent processes and threads, synchronization objects like mutexes, semaphores and events, references to files, which are represented as file objects, and I/O request packets (IRPs), which represent I/O operations.

Paged Pool

Paged pool, on the other hand, gets its name from the fact that Windows can write the data it stores to the paging file, allowing the physical memory it occupies to be repurposed. Just as for user-mode virtual memory, when a driver or the system references paged pool memory that’s in the paging file, an operation called a page fault occurs, and the memory manager reads the data back into physical memory. The largest consumer of paged pool, at least on Windows Vista and later, is typically the Registry, since references to registry keys and other registry data structures are stored in paged pool. The data structures that represent memory mapped files, called sections internally, are also stored in paged pool.

Device drivers use the ExAllocatePoolWithTag API to allocate nonpaged and paged pool, specifying the type of pool desired as one of the parameters. Another parameter is a 4-byte Tag, which drivers are supposed to use to uniquely identify the memory they allocate, and that can be a useful key for tracking down drivers that leak pool, as I’ll show later.

Viewing Paged and Nonpaged Pool Usage

There are three performance counters that indicate pool usage:

  • Pool nonpaged bytes
  • Pool paged bytes (virtual size of paged pool – some may be paged out)
  • Pool paged resident bytes (physical size of paged pool)

However, there are no performance counters for the maximum size of these pools. They can be viewed with the kernel debugger !vm command, but with Windows Vista and later to use the kernel debugger in local kernel debugging mode you must boot the system in debugging mode, which disables MPEG2 playback.

So instead, use Process Explorer to view both the currently allocated pool sizes, as well as the maximum. To see the maximum, you’ll need to configure Process Explorer to use symbol files for the operating system. First, install the latest Debugging Tools for Windows package. Then run Process Explorer and open the Symbol Configuration dialog in the Options menu and point it at the dbghelp.dll in the Debugging Tools for Windows installation directory and set the symbol path to point at Microsoft’s symbol server:

image

After you’ve configured symbols, open the System Information dialog (click System Information in the View menu or press Ctrl+I) to see the pool information in the Kernel Memory section. Here’s what that looks like on a 2GB Windows XP system:

image

    2GB 32-bit Windows XP

Nonpaged Pool Limits

As I mentioned in a previous post, on 32-bit Windows, the system address space is 2GB by default. That inherently caps the upper bound for nonpaged pool (or any type of system virtual memory) at 2GB, but it has to share that space with other types of resources such as the kernel itself, device drivers, system Page Table Entries (PTEs), and cached file views.

Prior to Vista, the memory manager on 32-bit Windows calculates how much address space to assign each type at boot time. Its formulas takes into account various factors, the main one being the amount of physical memory on the system.  The amount it assigns to nonpaged pool starts at 128MB on a system with 512MB and goes up to 256MB for a system with a little over 1GB or more. On a system booted with the /3GB option, which expands the user-mode address space to 3GB at the expense of the kernel address space, the maximum nonpaged pool is 128MB. The Process Explorer screenshot shown earlier reports the 256MB maximum on a 2GB Windows XP system booted without the /3GB switch.

The memory manager in 32-bit Windows Vista and later, including Server 2008 and Windows 7 (there is no 32-bit version of Windows Server 2008 R2) doesn’t carve up the system address statically; instead, it dynamically assigns ranges to different types of memory according to changing demands. However, it still sets a maximum for nonpaged pool that’s based on the amount of physical memory, either slightly more than 75% of physical memory or 2GB, whichever is smaller. Here’s the maximum on a 2GB Windows Server 2008 system:

image

    2GB 32-bit Windows Server 2008

64-bit Windows systems have a much larger address space, so the memory manager can carve it up statically without worrying that different types might not have enough space. 64-bit Windows XP and Windows Server 2003 set the maximum nonpaged pool to a little over 400K per MB of RAM or 128GB, whichever is smaller. Here’s a screenshot from a 2GB 64-bit Windows XP system:

image 

    2GB 64-bit Windows XP

64-bit Windows Vista, Windows Server 2008, Windows 7 and Windows Server 2008 R2 memory managers match their 32-bit counterparts (where applicable – as mentioned earlier, there is no 32-bit version of Windows Server 2008 R2) by setting the maximum to approximately 75% of RAM, but they cap the maximum at 128GB instead of 2GB. Here’s the screenshot from a 2GB 64-bit Windows Vista system, which has a nonpaged pool limit similar to that of the 32-bit Windows Server 2008 system shown earlier.

image 

    2GB 32-bit Windows Server 2008

Finally, here’s the limit on an 8GB 64-bit Windows 7 system:

image 

    8GB 64-bit Windows 7

Here’s a table summarizing the nonpaged pool limits across different version of Windows:

  32-bit 64-bit
XP, Server 2003 up to 1.2GB RAM: 32-256 MB 
> 1.2GB RAM: 256MB
min( ~400K/MB of RAM, 128GB)
Vista, Server 2008,
Windows 7, Server 2008 R2
min( ~75% of RAM, 2GB) min(~75% of RAM, 128GB)
Windows 8, Server 2012 min( ~75% of RAM, 2GB) min( 2x RAM, 128GB)

Paged Pool Limits

The kernel and device drivers use paged pool to store any data structures that won’t ever be accessed from inside a DPC or ISR or when a spinlock is held. That’s because the contents of paged pool can either be present in physical memory or, if the memory manager’s working set algorithms decide to repurpose the physical memory, be sent to the paging file and demand-faulted back into physical memory when referenced again. Paged pool limits are therefore primarily dictated by the amount of system address space the memory manager assigns to paged pool, as well as the system commit limit.

On 32-bit Windows XP, the limit is calculated based on how much address space is assigned other resources, most notably system PTEs, with an upper limit of 491MB. The 2GB Windows XP System shown earlier has a limit of 360MB, for example:

image

   2GB 32-bit Windows XP

32-bit Windows Server 2003 reserves more space for paged pool, so its upper limit is 650MB.

Since 32-bit Windows Vista and later have dynamic kernel address space, they simply set the limit to 2GB. Paged pool will therefore run out either when the system address space is full or the system commit limit is reached.

64-bit Windows XP and Windows Server 2003 set their maximums to four times the nonpaged pool limit or 128GB, whichever is smaller. Here again is the screenshot from the 64-bit Windows XP system, which shows that the paged pool limit is exactly four times that of nonpaged pool:

image 

     2GB 64-bit Windows XP

Finally, 64-bit versions of Windows Vista, Windows Server 2008, Windows 7 and Windows Server 2008 R2 simply set the maximum to 128GB, allowing paged pool’s limit to track the system commit limit. Here’s the screenshot of the 64-bit Windows 7 system again:

image 

    8GB 64-bit Windows 7

Here’s a summary of paged pool limits across operating systems:

  32-bit 64-bit
XP, Server 2003 XP: up to 491MB
Server 2003: up to 650MB
min( 4 * nonpaged pool limit, 128GB)
Vista, Server 2008,
Windows 7, Server 2008 R2
min( system commit limit, 2GB) min( system commit limit, 128GB)
Windows 8, Server 2012 min( system commit limit, 2GB) min( system commit limit, 384GB)

Testing Pool Limits

Because the kernel pools are used by almost every kernel operation, exhausting them can lead to unpredictable results. If you want to witness first hand how a system behaves when pool runs low, use the Notmyfault tool. It has options that cause it to leak either nonpaged or paged pool in the increment that you specify. You can change the leak size while it’s leaking if you want to change the rate of the leak and Notmyfault frees all the leaked memory when you exit it:

image

Don’t run this on a system unless you’re prepared for possible data loss, as applications and I/O operations will start failing when pool runs out. You might even get a blue screen if the driver doesn’t handle the out-of-memory condition correctly (which is considered a bug in the driver). The Windows Hardware Quality Laboratory (WHQL) stresses drivers using the Driver Verifier, a tool built into Windows, to make sure that they can tolerate out-of-pool conditions without crashing, but you might have third-party drivers that haven’t gone through such testing or that have bugs that weren’t caught during WHQL testing.

I ran Notmyfault on a variety of test systems in virtual machines to see how they behaved and didn’t encounter any system crashes, but did see erratic behavior. After nonpaged pool ran out on a 64-bit Windows XP system, for example, trying to launch a command prompt resulted in this dialog:

image

On a 32-bit Windows Server 2008 system where I already had a command prompt running, even simple operations like changing the current directory and directory listings started to fail after nonpaged pool was exhausted:

image

On one test system, I eventually saw this error message indicating that data had potentially been lost. I hope you never see this dialog on a real system!

image

Running out of paged pool causes similar errors. Here’s the result of trying to launch Notepad from a command prompt on a 32-bit Windows XP system after paged pool had run out. Note how Windows failed to redraw the window’s title bar and the different errors encountered for each attempt:

image

And here’s the start menu’s Accessories folder failing to populate on a 64-bit Windows Server 2008 system that’s out of paged pool:

image

Here you can see the system commit level, also displayed on Process Explorer’s System Information dialog, quickly rise as Notmyfault leaks large chunks of paged pool and hits the 2GB maximum on a 2GB 32-bit Windows Server 2008 system:

image

The reason that Windows doesn’t simply crash when pool is exhausted, even though the system is unusable, is that pool exhaustion can be a temporary condition caused by an extreme workload peak, after which pool is freed and the system returns to normal operation. When a driver (or the kernel) leaks pool, however, the condition is permanent and identifying the cause of the leak becomes important. That’s where the pool tags described at the beginning of the post come into play.

Tracking Pool Leaks

When you suspect a pool leak and the system is still able to launch additional applications, Poolmon, a tool in the Windows Driver Kit, shows you the number of allocations and outstanding bytes of allocation by type of pool and the tag passed into calls of ExAllocatePoolWithTag. Various hotkeys cause Poolmon to sort by different columns; to find the leaking allocation type, use either ‘b’ to sort by bytes or ‘d’ to sort by the difference between the number of allocations and frees. Here’s Poolmon running on a system where Notmyfault has leaked 14 allocations of about 100MB each:

image

After identifying the guilty tag in the left column, in this case ‘Leak’, the next step is finding the driver that’s using it. Since the tags are stored in the driver image, you can do that by scanning driver images for the tag in question. The Strings utility from Sysinternals dumps printable strings in the files you specify (that are by default a minimum of three characters in length), and since most device driver images are in the %Systemroot%\System32\Drivers directory, you can open a command prompt, change to that directory and execute “strings * | findstr <tag>”. After you’ve found a match, you can dump the driver’s version information with the Sysinternals Sigcheck utility. Here’s what that process looks like when looking for the driver using “Leak”:

image

If a system has crashed and you suspect that it’s due to pool exhaustion, load the crash dump file into the Windbg debugger, which is included in the Debugging Tools for Windows package, and use the !vm command to confirm it. Here’s the output of !vm on a system where Notmyfault has exhausted nonpaged pool:

image

Once you’ve confirmed a leak, use the !poolused command to get a view of pool usage by tag that’s similar to Poolmon’s. !poolused by default shows unsorted summary information, so specify 1 as the the option to sort by paged pool usage and 2 to sort by nonpaged pool usage:

image 

Use Strings on the system where the dump came from to search for the driver using the tag that you find causing the problem.

So far in this blog series I’ve covered the most fundamental limits in Windows, including physical memory, virtual memory, paged and nonpaged pool. Next time I’ll talk about the limits for the number of processes and threads that Windows supports, which are limits that derive from these.


Comments (44)

  1. Anonymous says:

    excellent.

  2. Anonymous says:

    @Carl: Thanks for the feedback. I’ve only run into pool leaks a few times in the last 10 years. Device drivers of any kind – hardware, file system filter, antivirus – can be guilty of it.

  3. Crawford says:

    I get delayed write failures every once in a while, and usually the cause for me is a glitch in the hibernation process.  I’ve also discovered that you’re almost guaranteed one if you hibernate while a USB-connected NTFS volume is mounted (even if you resume with the same drive connected to the same port).  I’ve also gotten it once or twice by breaking my file server while writing to files.  Fortunately, it’s never actually caused any data loss for me.

  4. Carl Klemmer says:

    Great article and great series, Mark.  In 15 years of managing NT servers I’ve banged up against physical and virtual memory limitations plenty of times, but I’ve never had an issue that lead me to these pools.  Is exhausting them a common problem?  What kind of software most commonly leads to issues with these pools?

  5. Jonathan says:

    Server software that keeps many TCP/IP connections in the air can easily exhaust Non-paged pool. That happens on heavily-loaded ISA Servers for example – and is the reason why the next version of it is 64-bit only.

    Crawford: Maybe it’s because NTFS is always updated the last-access time of files? I think you can disable it with fsutil (fsutil behavior query disablelastaccess), though I don’t know whether it’s per-machine or per-volume.

  6. David says:

    Another great article. I learn so much, I recommend these articles to everyone.

  7. LookingForSolutions says:

    Earlier this week, I was troubleshooting a customer’s server where IIS6 had stopped accepting new connections. After checking the entries in the SYSTEM32LOGFILESHTTPERR logs, I searched on Google and found out it was due to available non-paged pool memory being less than 20MB, at which point IIS6 stops accepting new connections. Rebooting the server fixed the problem for now. When/if the problem returns, I’ll try the troubleshooting steps from this excellent blogpost!

    FYI, the related KB’s:

    http://support.microsoft.com/kb/934878

    http://support.microsoft.com/kb/820729

    "The kernel NonPagedPool memory has dropped below 20MB and http.sys has stopped receiving new connections"

  8. Carl says:

    It is my understanding that storing Outlook PST files on network drives can cause nonpaged memory pool issues on 32-bit file Windows 2003 file servers.

    More info on this in these articles:

    http://blogs.technet.com/askperf/archive/2007/01/21/network-stored-pst-files-don-t-do-it.aspx

    http://support.microsoft.com/?id=297019

    There is probably enough information in this article to identify the PST-network drive issue.  However, I’m guessing a lot of SysAdmins would really appreciate you taking those two links, combining what you’ve written here with the PST issue and explaining exactly how to identify and solve that issue.  

  9. Miral says:

    My 32-bit Vista machine gets *very* flaky when the non-paged pool hits 1GB or higher.  (On a couple of occasions it truncated files that were being written at the time.)

    In case anyone is interested: the leak was in the TdxA pool and was (possibly indirectly) caused by a poorly-written firewall/virusscan program.  (I tracked this down using poolmon a few months ago.)

  10. David says:

    I’ve gotten the "Delayed Write Failed" error most of the times I’ve tried to use ntbackup on Windows Server 2003. Extremely unacceptable for such an important piece of software.

  11. Nick says:

    Mark –

    The timing of your articles never ceases to amaze.  Just two weeks ago I was fighting with a server experiencing pool exhaustion, largely due to someone placing the /3GB switch into it’s boot options.  Process Explorer was (yet again) an invaluable tool to solve the problem.

    Thanks again for the great articles and utilities.

  12. Anonymous Coward says:

    Great article, Mark. I’m always amazed by the way you tackle these seemingly complicated things in such a straightforward manner.

  13. Chad says:

    Mark, great article & very timely for me as I’m getting 2019’s every 4.5 days & I’m assuming it’s 1 of many new print drivers that were recently loaded onto the server. I’ve begun using Poolman to try & diagnose which driver causes the leak, unfortunately when I attempt to use the poolmon -c switch to generate a "localtag.txt get the following error;

    "Poolmon: No localtag.txt in current directory

    Unable to load msvcr70.dll/msvcp70.dll, cannot create local tag file"

    So I receive a number of "unknown driver" tags…..any ideas on how to help me further isolate this driver leak or get poolmon to successfully create a localtag.txt file?

    Also, I fear it may be a event driven leak as I’ve been watching the nonpaged limit in process explorer for a couple of days & it’s a very steady & not climbing.

    Thanks again, the article was very straight-forward & helpful.

  14. Clovis says:

    Why is MPEG2 playback disabled if debugging is enabled? How do you debug MPEG2 drivers?

  15. dirbase says:

    Mark,

    Thanks for this post. This reminds me of a topic I started in the sysinternals forum back in January http://forum.sysinternals.com/forum_posts.asp?TID=17486 where I showed that on several XP (SP2/SP3) systems with NTFS partitions searching for a file via explorer search can take up a large portion of both page and non paged pools, which are not really released after the search has ended.This is presumably due to NTFS.sys.

    I’m still wondering if it’s an expected behaviour. What do you think?  

  16. dirbase says:

    Concerning my comment above about high pool usage following an explorer file search, since the MmSt paged pool tag is on top of the bytes list, I tend to think now that the basic reason is that the many searched files have been cached and the increased MmSt allocations correspond to their PTEs. This would explain why a repeat file search does not increase the pool usage.

  17. wtp says:

    Mark:

      We are getting back to the point IBM did in 1964 when it built the Stretch for the AEC – long words, meaning longer addresses and better control of memory (though I wish the next I7 or AMD similar would be a 100% 64-bit x86 rather than the 32 bits-with-bag-on-the-side models we’ve been seeing.

      Once you *can* address 4 TB of cheap RAM, then setting the various amounts allowed various jobs, as we used to say (a "job" is an active application) or daemons (read "TSRs" or inactive pre-loads), we could program (we did it with 1s and with 0s, and sometimes we ran out of 1s!)* without sloppy coding relying on stacks(or, for that matter, C, a language designed as a ‘portable’ single-pass compiler for two reasons 1) because compiling a job on a PDP-7 multi-pass device meant ‘ok, let’s start it up and break for lunch, then see if we had any errors) and mainframes were run, in part, by bean counters, who gave you so many ‘chocolate’ dollars in time to get your work done) and move back to writing all programs large and small in a ‘smart’ Assembler, yep in assembly language,** that would allow the construction or use of repeatedly-used optimized structures as "pseudo ops"***

       Which naturally brings up the question of why, guy, didn’t you write all your sisinternals to run under 64 bit extensions?

                       wtp

    *Tom Smith – filk writer "when I was a Boy"

    **See Steve Gibson on assembler GRC.com

    ***Of course, that was in the days before microcode and the need to build 1 circuit per class of instructions, The stunningly simple code for DEC’s 12-bit desktop machines fit on a 3×5 card, the absolutely awesome code for the PDP-10 room-filling "big iron" fit on an 8.5×11" sheet. Remember the study on VAXen and PDP-11s that produced RISC architecture? A lot of x86 code might also prove slow and redundant if given the same test, cutting down 3 vol.s of code manuals to , at most, 1.(wtp’s opinion, anyway) This might make for a physically larger Windows OS that ran fast enough to reduce CPU speed to MHz again – not 4-core hyperthreaded 3 GHz.+.And the assembled code might be way smaller than anything the most efficient C (+,++,# or.net) could produce.

  18. Chris says:

    The thing I don’t understand is this:

    When I install Windows (any version) on any of my machines, a quick look at the Task Manager shows me that a significant amount of memory is being paged.

    but when I install Ubuntu on that same hardware, the page drive is zero

    what’s the deal with that?

  19. Alex Ionescu says:

    @Clovis It’s disabled because the Protected Media Path (PMP) could be theoretically subverted by using the debugger APIs (which allow you to do things like patch the kernel from user-mode). When debugging PMP components, there’s a special internal build which keeps the components running but allows debugging (which is internal to Microsoft and their DRM partners) — this isn’t something a user can do.

  20. Friday says:

    @Clovis: If they told you they’d have to sue you. 🙂

    Oops, too late. 😀

  21. MPyle says:

    Great article. Very clear.

    But given that how does, Task-Manager define a per process count of Non-Paged-Pool memory usage?

    I have a problem where we see a per process memory-pool rise.

  22. GS says:

    So what do you if you find more then 1 driver matching symbol string? How do I identify which one is actually causing this?

  23. Timothy Davis says:

    @GS and @Chad

    You can use driver verifier to track pool usage by a driver.  See [http://msdn.microsoft.com/en-us/library/ms792856.aspx] for more information

  24. RW says:

    Great article.  I have a handle leak that accumulates in explorer.exe on my 4GB WinXP sp3 laptop.  This occurs multiple times per week and is happening as I type this.

    PoolMon shows:

    Even Nonp   12848369 ( 427)  11400450 ( 430)  1447919 69507456 (  -144)     48

    Total Handles 1,471,236 which of course means Windows is beginning to act up.

    The techniques so far have not led me to tell what is leaking the handles.  The Even tag I believes it is unknown.

    Explorer.exe handles total is 1,437,260

  25. LookingForSolutions says:

    Sure enough, the problem which I mentioned in my previous post has returned. This time I used poolmon to track the culprit. It turned out to be a faulty version of the Microsoft MPIO driver (version 1.21 leaks nonpaged pool memory -> http://support.microsoft.com/kb/961640). This issue was happening on one (Windows Server 2003 x86) cluster node, while the other cluster node did not exhibit any problems. The other cluster node turned out to be using a different version of the MPIO driver which didn’t have the memory leak bug.

  26. reporter says:

    Symantec Antivirus software is one of the software which i have seen allocating humungous amount of paged pool (almost ~70MB). it’s surprising how they got away with this so far given the huge instll base (both retail and corporate) they have.

  27. TechFreak says:

    Good information, but nothing creative here.

    For a deveoper like me, I rather liked the creativity and ingenuity of the idea in using pure Win32 and C++ to create Terabyte sized arrays on Win32 machines (without tweaking anything that too):

    http://blogs.msdn.com/gpalem/archive/2008/06/05/huge-arrays-with-file-mapping.aspx

    Thats what I call real pushing of limits. Whoever did it, I found it quite inspiring. Way to go Boys.

    -AMiCo

    NationalSemiConductors.

  28. Andrew Fidel says:

    Easier than installing the DDK just to get poolmon is to install the support tools for your OS. For XP SP2 it’s available at

    http://www.microsoft.com/downloads/details.aspx?FamilyId=49AE8576-9BB9-4126-9761-BA8011FABF38

  29. Travis says:

    Nice job.  By far the best, most comprehensive guide on the subject.

  30. elbutre says:

    hi

    i am running a windows 2008 64 bit server with 10gb of ram and exchange 2007 sp1.

    the exchange monitoring tool on the server says

    Paged pool memory on server uk-mailbox is over the error threshold of 220 MB. Current value: 234 MB.

    The maximum non-paged pool memory on server uk-mailbox is over the warning threshold of 100 MB. This may cause system instability. Current value: 110 MB.

    but having ran the process explorer tool it shows that it is nowhere near the limit of 128gb for 64bit 2008 server systems

    i also wanted to look at poolmon to see if any of the applications are leaking but i cannot see that poolmon is supported on windows 2008 64 bit server. is there a version for this OS?

    and should i be worried about the paged and non-paged pool memory warning?

    there are no event id’s in the system log and the server seems to be running fine. no crashes (yet)

  31. Remko Weijnen says:

    Mark,

    How about the other important resources (eg System PTE’s, Session View Space, Sesion Paged Pool, Desktop Heap), it would be great if you can discuss these as well!

  32. John says:

    Someone at microsoft needs to code up a utility that automatically does the steps outlined in the tracing pool leaks section of this article, and returns the offending leaky driver/system file. People could then run the tool if they’ve been getting "insufficient system resources" errors, and that would help them fix their problem. Optionally they could upload the results to microsoft and microsoft could pass along such results to the software vendor in question or post advisories about those software products.

  33. davetiye says:

    güzel davetiye örnekleri davetiye modelleri düğün davetiyeleri

  34. bryan says:

    Mark, awesome article, thank you very much!

  35. Lanruzehh says:

    Hi,

    Very nice article Mark.

    I have a couple of questions,

    1. Is it poosible to increase the Non Paged pool memory limit beyond 75% of RAM on Windows Server 2008 ?

    2. If the answer to question 1 is yes, how can you do that ?

    Thanks,

    Lanruzehh

  36. Ronald M says:

    I have been trying to find the reason  for a non paged pool leak on a windows server 2003 for quite some time now without much success. This happens randomly sometimes twice in a day. Here is the output from poolman when the non paged pool consumption was high. Is it possible to find from this output, the possible cause of the leak.

    Memory:16771724K Avail:15716652K  PageFlts: 76038   InRam Krnl: 2268K P:114036

    Commit: 765728K Limit:33273512K Peak: 968688K            Pool N:196572K P:1150

    System pool information

    Tag  Type     Allocs            Frees            Diff   Bytes       Per Alloc

    Irp  Nonp   15114034 (3654)  14980131 (3578)   133903 62270528 ( 33776)    465

    usbp Nonp     292346 ( 136)    159911 ( 136)   132435 41995160 (     0)    317

    Mdl  Nonp     578590 ( 286)    307149 ( 293)   271441 34807680 (    80)    128

    HidU Nonp     291648 ( 136)    159541 ( 136)   132107 13738616 (     0)    103

    MmCm Nonp       2463 (   0)      2305 (   0)      158 8316176 (     0)  52634

    MFE0 Nonp  210464239 (3115) 210434889 (3080)    29350 7450848 (  3568)    253

    HidC Nonp     264459 (   0)    132319 (   0)   132140 3175736 (     0)     24

    LSwi Nonp          1 (   0)         0 (   0)        1 2576384 (     0) 2576384

    Io   Nonp   95706488 (65071)  95574219 (65071)   132269 2191368 (     0)     1

    HdCl Nonp     264402 (   0)    132303 (   0)   132099 2116896 (     0)     16

    TPLA Nonp        512 (   0)         0 (   0)      512 2097152 (     0)   4096

    TCPt Nonp      37255 (   6)     37218 (   6)       37 1664536 (     0)  44987

    File Nonp   13742294 ( 368)  13734330 ( 358)     7964 1213584 (  1536)    152

    VxSb Nonp          6 (   0)         4 (   0)        2 1171456 (     0) 585728

    Mm   Nonp       1037 (   0)      1025 (   0)       12 1127512 (     0)  93959

    brcm Nonp         30 (   0)         0 (   0)       30  819184 (     0)  27306

    TCht Nonp       7652 (   0)      7456 (  58)      196  802816 (-237568)   4096

    bRcm Nonp          8 (   0)         0 (   0)        8  685840 (     0)  85730

    naFF Nonp        299 (   0)         1 (   0)      298  656504 (     0)   2203

    Thre Nonp     155086 (   7)    154036 (  11)     1050  655200 ( -2496)    624

    Pool Nonp          6 (   0)         3 (   0)        3  610304 (     0) 203434

    AfdC Nonp     149183 (  30)    146111 (  28)     3072  491520 (   320)    160

    LSwr Nonp        128 (   0)         0 (   0)      128  416768 (     0)   3256

    Even Nonp    3895544 ( 354)   3889534 ( 350)     6010  291088 (   192)     48

    Devi Nonp        705 (   0)       350 (   0)      355  270104 (     0)    760

    NDpp Nonp         95 (   0)        14 (   0)       81  267520 (     0)   3302

    TCPC Nonp      11370 (   6)      8215 (   0)     3155  265952 (   480)     84

    Vad  Nonp    2730397 (   9)   2725265 (   9)     5132  246336 (     0)     48

    Hal  Nonp     194560 ( 341)    194536 ( 340)       24  199968 (   368)   8332

    Ntf0 Nonp          3 (   0)         0 (   0)        3  196608 (     0)  65536

    Ntfr Nonp      20653 (   0)     17959 (   0)     2694  173384 (     0)     64

    AfdE Nonp     148107 (  30)    147506 (  28)      601  168280 (   560)    280

    TCPc Nonp     243685 (  53)    240390 (  49)     3295  158160 (   192)     48

    Sema Nonp    6766238 (2360)   6763555 (2360)     2683  150648 (     0)     56

    MmCi Nonp      27458 (   0)     26860 (   0)      598  141344 (     0)    236

    Dump Nonp          9 (   0)         1 (   0)        8  135600 (     0)  16950

    RceT Nonp          1 (   0)         0 (   0)        1  131072 (     0) 131072

    Muta Nonp     999419 ( 189)    997544 ( 189)     1875  125856 (     0)     67

    CcSc Nonp     255342 (  10)    254961 (   9)      381  121920 (   320)    320

    MmCa Nonp    1262482 (  11)   1261431 (  10)     1051  107792 (   112)    102

    VadS Nonp    1697218 (1113)   1694239 (1117)     2979   95328 (  -128)     32

    Vadl Nonp    2237198 ( 116)   2235790 ( 118)     1408   90112 (  -128)     64

    NtFs Nonp     874028 (  18)    872463 (   5)     1565   72712 (   520)     46

    Ntfi Nonp     115369 (  12)    115125 (   0)      244   66368 (  3264)    272

    Lfsr Nonp          6 (   0)         2 (   0)        4   65536 (     0)  16384

    PooL Nonp          8 (   0)         0 (   0)        8   65536 (     0)   8192

    AmlH Nonp          1 (   0)         0 (   0)        1   65536 (     0)  65536

  37. Paul Hickman says:

    Hi Mark,

    There seems to be one kind of memory allocation that I haven’t seen covered by your articles. How do virtualisation products such as Sun Virtual Box allocate their memory? It doesn’t show up in any of the process-level memory counters in task manager / process explorer / perfmon. The only place it seems to be counted is in the overall "Cached" memory on the performance tab of task manager and the equivilent in other tools.

  38. miguel q says:

    Looking at the Process Explorer in Windows Server 2003 guest in VMware. I am able to see the Paged Physical and Virtual values but the Paged Limit reads "no symbols"  What gives?

  39. nc says:

    @miguel q

    You need to configure symbols in Process Explorer so it can access the symbols for the kernel image.

  40. DaveH says:

    Hi Mark,

    First, great post. I’ve used this information to try and trace what was using all the non-paged memory on my server 2003 system – it’s an old system and still running SP0 so there could be many bugfixes which would explain my problem.

    I was getting volsnap events stating that it couldn’t allocate enough nonpaged memory to hold a bitmap for a snapshot of my C: drive.

    volsnap uses 16KB blocks in it’s bitmap, so on my 69GB drive, I calculated 552KB for the bitmap.

    It took me a while to get the symbols for the system so that I could see the nonpage pool limit.

    The system in question was using 100 MB of nonpaged pool memory. Procexp.exe gave a limit of 256 MB of nonpaged pool memory. More than enough free for the bitmap.

    I was confused! I removed the shadowstorage for the C: drive from it’s second volume D: so that it would just use the C: drive and the problem went away!

    Do you think this is a bug in the SP0 VSS implementation or do you really think I was running out of nonpaged memory?

    Thanks,

    Dave.

  41. Trana says:

    Great stuff, thanks.

    I used to say an IT admin/support persons job is 50% Internet search, 50% compare settings side by side. But I am changing it to 33% Search, 33% Settings comparing and 34% Using tools created by Mark R.

    Poolmon, Autoruns, Pagedefrag, ProcessExplorer, Disk2VHD, you name it…I love it!

  42. Hakan says:

    Helpful article, thanks.

    Also I would like to solicit opinion for the problem that I am debugging.

    My software uses a Jungo driver for which I do not have a source code. At some point when I try to allocate memory it returns insufficient resource error.

    I used a logger utility that came with the Jungo and extracted information that AllocateCommonBuffer fails which tells me that I am running out some memory resource somewhere. But I have plenty of physical memory ( 3 gb) which is available to me ( no special hardware hogging my memory making it unavailable to the system) . Since I am aware that kernel NonPaged pool is limited resource( 128 M on my system according to NonPagedPoolSize key in registery ) I have decided to  use poolmon to see how I am running out of NonPaged memory and who is the culprit also this would tell me the Tag for the Jungo driver since I know the amount of memory I am allocating (8 Mb). Here is the strange part. According to the Poolmon no one is allocating 8Mb memory and there is plenty of Paged and NonPaged memory at the time AllocateCommonBuffer fails.

    Now, I can hypothesize all kind of scenarios as to why the allocation fails. For example, maybe it needs something like NonPaged + Contiguous therefore it runs out of this flavor of memory or something. But I am not convinced by this kind of arguments since I do not have any way of verification of my hypothesis. I think the question I would like to ask is this : When I allocate 8Mb memory successfully with AllocateCommonBuffer where does the memory come from since I do not see anywhere Paged or NonPaged resources being reduced by this amount? Only indication of the allocation successful return of AllocateCommonBuffer and System PTE increased 2048 entries.

    For those of you who have suggestions. I thank you for your time and attention.

  43. Hakan says:

    I forgot to mention that I am using Windows 7.

  44. nran says:

    Awesome Article!!