VMWare Fusion 25H2 With M3 Memory Leak

[Furkan Yucel]

Hello there, The VM that I had no issues with in version 13.xx is now being forced to shut down due to unusual RAM consumption during use. It shuts down directly on macOS with the usage shown in the relevant screenshot, without any errors on VMware. Could this be a memory leak? How can I determine this?
Device and environment information: 
 Macbook pro m3 max 48gb MacOS Version 15.7.3 (24G419)  Running Win11 ARM64 on VMware Professional 25H2 (24995814),  allocating 20GB of the 48GB RAM for the VM.

-------------------------------------------

[Technogeezer]

It could be a memory leak but there is no way for you to determine that. The best advice I can offer is to post the log files found in the VM's bundle folder so that the community can see if there's anything that points to a cause. 

What build of Windows are you running in the VM? 

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[Furkan Yucel]

I am attaching the log files and Windows version information below. Thank you very much for your interest and support.

-------------------------------------------

Attachment(s)

mksSandbox.log   25 KB 1 version

vmware-0.log   395 KB 1 version

vmware-2.log   810 KB 1 version

vmware.log   334 KB 1 version

mksSandbox-0.log   25 KB 1 version

mksSandbox-1.log   26 KB 1 version

vmware-1.log   563 KB 1 version

mksSandbox-2.log   26 KB 1 version

[Technogeezer]

I've been looking through the logs and have a couple of questions:

You have several entries in the .vmx file that are not present in mine (nor are they included by default).

sched.mem.max = "10240"
sched.mem.min = "2048"
sched.mem.maxmemctl = "8192"
sched.mem.pshare.enable = "TRUE"
MemTrimRate = "0"
mainMem.useNamedFile = "FALSE"
prefvmx.minVmMemPct = "50"
isolation.tools.memSchedFakeSampleStats.disable = "FALSE"

Did you insert these? If so, why? They appear to be intended for a VMware product that runs the VMware hypervisor (e.g. vSphere/ESXi) and Fusion runs using the Apple hypervisor. 

Are you running with any snapshots active? (It doesn't look like it, but wanted to ask to make sure).

If you disable 3D support for the VM, does this still happen?

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[Furkan Yucel]

Since RAM consumption has been continuously increasing uncontrollably, this AI asked me to add these lines. My problem persisted even before these lines were added. Currently, even when I add `mks.enable3d = "FALSE"` with 3D support disabled, my RAM usage still appears high. Even now, without understanding how it happened, it shows 220GB usage, but it doesn't stop the VM. I don't understand what this means. What should my settings be for the most efficient VM usage on a device with 48GB RAM and an M3 Max processor? Thanks for your interest 

-------------------------------------------

[Technogeezer]

I would take any AI generated answer for a problem with VMware Fusion with a large grain of salt (that is, don't treat it as an accurate answer).  It most likely will reply with advice for ESXi which may be irrelevant for Fusion. Check the sources that the AI is using in its answer generation before doing anything.

What you're seeing is very unusual -- I don't have any "magic incantations" that I can recommend to fix the issue or optimize use of the product because we don't know what the issue really is. For the majority of us, Fusion works fine out of the box without runaway memory utilization. That's why we're going to need some Broadcom folks to weigh in on this if they're listening. 

Some random thoughts:

Do you have hard drive buffering enabled for this VM's Advanced settings. Try turning that off.

Are you using multiple monitors? Try using only 1. 

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[Duncan Epping]

I've experienced a similar memory leak issue, I've reported it internally, and they are trying to figure out if they can reproduce the issue now to be able to debug it.

-------------------------------------------

[Furkan Yucel]

Thank you for interest, I will follow updates.

-------------------------------------------

[Furkan Yucel]

I reverted the configurations created by AI, so the VM is now running with the settings it had when it was first created. The hard drive buffering value was set to Automatic. I disabled it and checked, but the RAM value continues to rise. Even if I don't perform any operations on the VM, the RAM usage continues to increase and never decreases. I don't know what to do about this. Thanks for support 

-------------------------------------------

[Duncan Epping]

I don't think you can do much to be honest, I tried everything as well, but nothing stops Fusion from bloating the mem at the moment. I do power off the vm regularly, and power it back on, just to start from 0 again.

-------------------------------------------

[Furkan Yucel]

While using VMware Fusion version 13, I created a backup with Time Machine. Can I resolve this issue by reverting to that backup? Thanks a lot!

-------------------------------------------

[krayzee]

I am running VMWare Fusion 25H2 on a 2019 Mac Pro (intel processor).  I have never been able to run without a crash for more than 2 days or so with 3D graphics acceleration enabled and the fix for that was to just disable acceleration.  However, since upgrading to Fusion 25H2 and Macos 26, I have had the exact same behavior occur as described in this forum at exactly the same point (between 300 GB and 350GB memory use for vmware-vmx).  Macos pauses the application and I basically have to reboot to get out of this mess.  So I have been suspending or rebooting my windows 11 vm every time the memory grows beyond 200 GB to avoid this.  All this to say, the issue exists on intel silicon as well. 

-------------------------------------------

[Ihar Andreyeu]

Same problem here. Fresh install of VMWare 25, no custom configs.
M4 Mac Pro.
The thing eats up through all of my 48 gigs of ram in around 40 minutes.

-------------------------------------------

[Furkan Yucel]

It's very sad to hear this. I hope the source of this problem is identified and resolved as soon as possible :(

-------------------------------------------

[Yoda Mann]

FWIW, I'm on a MacBook Pro M3 Max with 128 GB memory. Current OS, VMWare & Windows builds. Not seeing this issue. >>knock wood<<
OP said you were on, "MacOS Version 15.7.3 (24G419)" I'm on macOS Tahoe 26.2. Maybe it's time to upgrade from Sequoia?

-------------------------------------------

[Technogeezer]

If you're concerned about this, I'd stay on macOS Sequoia and Fusion 13.6.4 until someone figures out why this is happening. I would not try to run Fusion 13.6.4 on Tahoe as there are known issues with that combination. 

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[Furkan Yucel]

There doesn't seem to be any update for my m3 max processor device. Is there anything else I need to do for this?

-------------------------------------------

[Technogeezer]

Unfortunately we're going to have to wait for Broadcom to respond. It's likely that there's something with 25H2 going on here, so it's probably not something that an Apple update is going to address (that is, unless Apple has broken something).

@Duncan Epping (who's tagged as a Broadcom Employee) has stated that the issue has been reported internally. 

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[mappoint]

Similar problem here. MacBook Pro 16 2019 with 32 GB RAM, Tahoe 26.2, Fusion Player 25H2. I am running W10 VMs with 4 GB RAM.

Fusion allocates constantly more RAM (> 0.5 GB per hour). I also believe there is a dependency with the standby mode.

-------------------------------------------

[Furkan Yucel]

thanks for all, I will follow updates.

-------------------------------------------

[mappoint]

I reverted to Fusion 13.6.4 with VMware Tools 13.0.5 for now. The problem is (nearly) gone. 

But even with 13.6.4 the vmware-vmx processes are continously taking more memory, but much slower (monitored over several hours). A W7 VM with 2 GB virtual RAM doing nothing grew from 2,5 GB to 3 GB in 24 hours (8 hours in standby).

Is it normal that a W10 VM with 4 GB virtual RAM ist taking roughly 8 GB (plus Fusion (550 GB) itself and mksSandbox (750 GB)) on the host? I did not have a look to that before running out of memory with Fusion 25H2.

Disk buffering is disabled.

There are also some older reports about the same problem here in this forum:

https://community.broadcom.com/vmware-cloud-foundation/communities/community-home/digestviewer/viewthread?MessageKey=10c8ff4a-748f-4fc2-a943-e2b758c4aa24&CommunityKey=0c3a2021-5113-4ad1-af9e-018f5da40bc0#bm10c8ff4a-748f-4fc2-a943-e2b758c4aa24

-------------------------------------------

[Serje Kon]

Hi. Faced with same issue. Macbook Air M4, 24Gb, VMWare Fusion 25H2, Windows 11 25H2, after 1 hour.

-------------------------------------------

[Yoda Mann]

1 Hour

Windows 11 Pro 25H2 26200.7462

VMWare Fusion Professional 25H2 (24995814)

MacBook Pro Apple M3 Max 26.2 (25C56) 128 GB RAM

I just don't seem to be seeing it. Am I missing something, what's the difference?

-------------------------------------------

[Furkan Yucel]

Can you investigate your vmware-vmx process? Vmware Fusion process fine about RAM but vmware-vmx spending a lot of memory(exceed seperated memory)

-------------------------------------------

[Yoda Mann]

You bet. Launch vs 1 hour.

-------------------------------------------

[Furkan Yucel]

Thank you very much for your interest and quick response. Could you share the contents of the vmx file in txt format for reference? Perhaps there are issues with some of our definitions. Happy New Year!

-------------------------------------------

[Aranda Nalham Nastali]

VMX File attached

-------------------------------------------

Attachment(s)

Windows 11 64-bit Arm.vmx   4 KB 1 version

[Furkan Yucel]

Thank you very much for your interest and quick response. Could you share the contents of the vmx file in txt format for reference? Perhaps there are issues with some of our definitions.

-------------------------------------------

[Yoda Mann]

As I don't know what data is in the file, I'm reluctant to do so.

-------------------------------------------

[Duncan Epping]

it is just configuration data of the VM itself, nothing that would allow anyone to access your VM etc.

-------------------------------------------

[krayzee]

My vmx file is unedited but my virtual machine is very old (it has upgraded from windows 7 to 10 to 11).  With windows 11 I was forced to add the trusted platform module and encrypt everything.  It is my understanding that I now I cannot even edit or view the vmx file without unencrypting, which requires removing the trusted platform module, which would destroy the windows 11 virtual machine.  

-------------------------------------------

[Technogeezer]

If you chose partial encryption for that VM (the recommended setting for a Windows 11 VM in Fusion 13 and later), the .vmx file is in plain text (unencrypted).

Removing the TPM device then unencrypting a fully encrypted VM does not destroy the VM. If you used BitLocker encryption for the VM, you need to have the recovery key and knowing how to use it before continuing. Then remove the TPM device, decrypt the VM, do whatever you need to, re-encrypt, then add back the TPM device. Then treat the situation as if the TPM had been corrupted on a physical PC. If you didn't use  BitLocker, just power on the PC and be prepared to follow prompts to reset your Windows Hello PIN if you used one. If ypu did use BitLocker, now would be the time to perform recovery to allow the VM to access the disk. 

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[krayzee]

Unfortunately I did not use partial encryption for my VM, the .vmx file is definitely unreadable.  I don't recall how or why I did it this way, it was a large number of years ago.  In the windows 11 vm, using the Bitlocker Manager the option to turn on bitlocker is there, not to turn it off.  I believe I used EFS to perform the encryption because I can see an EFS certificate issued by me, with an issue date of 2016 and an expiration date of 2116.  

I am very worried that I will end up having to abandon my VM in order to avoid the memory leak.  This stuff is mostly all greek to me.  

-------------------------------------------

[Aranda Nalham Nastali]

Hi, I am also having the same issue.

See below VMX output:

vTpm and encryption has been removed.

####Start of VMX####

.encoding = "UTF-8"

config.version = "8"

virtualHW.version = "22"

mks.enable3d = "TRUE"

pciBridge4.present = "TRUE"

pciBridge4.virtualDev = "pcieRootPort"

pciBridge4.functions = "8"

pciBridge5.present = "TRUE"

pciBridge5.virtualDev = "pcieRootPort"

pciBridge5.functions = "8"

pciBridge6.present = "TRUE"

pciBridge6.virtualDev = "pcieRootPort"

pciBridge6.functions = "8"

pciBridge7.present = "TRUE"

pciBridge7.virtualDev = "pcieRootPort"

pciBridge7.functions = "8"

vmci0.present = "TRUE"

hpet0.present = "TRUE"

nvram = "Windows 11 64-bit Arm.nvram"

virtualHW.productCompatibility = "hosted"

powerType.powerOff = "soft"

powerType.powerOn = "soft"

powerType.suspend = "soft"

powerType.reset = "soft"

displayName = "Windows 11 64-bit Arm"

firmware = "efi"

sensor.location = "pass-through"

guestOS = "arm-windows11-64"

tools.syncTime = "TRUE"

tools.upgrade.policy = "upgradeAtPowerCycle"

sound.autoDetect = "TRUE"

sound.virtualDev = "hdaudio"

sound.fileName = "-1"

sound.present = "TRUE"

numvcpus = "4"

memsize = "8192"

sched.mem.maxmemctl = "0"

sata0.present = "TRUE"

nvme0.present = "TRUE"

nvme0:0.fileName = "Virtual Disk.vmdk"

sata0:1.deviceType = "cdrom-raw"

sata0:1.fileName = "auto detect"

sata0:1.present = "TRUE"

usb.present = "TRUE"

ehci.present = "TRUE"

usb_xhci.present = "TRUE"

svga.graphicsMemoryKB = "524288"

ethernet0.connectionType = "nat"

ethernet0.addressType = "generated"

ethernet0.virtualDev = "vmxnet3"

ethernet0.linkStatePropagation.enable = "TRUE"

ethernet0.present = "TRUE"

extendedConfigFile = "Windows 11 64-bit Arm.vmxf"

floppy0.present = "FALSE"

vmx.encryptionType = "partial"

vtpm.present = "TRUE"

vmxstats.filename = "Windows 11 64-bit Arm.scoreboard"

svga.vramSize = "268435456"

numa.autosize.cookie = "40012"

numa.autosize.vcpu.maxPerVirtualNode = "4"

uuid.bios = "56 4d 25 29 4e 82 cf 7e-0f 20 60 e7 34 c6 66 c4"

uuid.location = "56 4d 25 29 4e 82 cf 7e-0f 20 60 e7 34 c6 66 c4"

vm.genid = "-5396197438048241113"

vm.genidX = "8426568439511845617"

pciBridge4.pciSlotNumber = "21"

pciBridge5.pciSlotNumber = "22"

pciBridge6.pciSlotNumber = "23"

pciBridge7.pciSlotNumber = "24"

usb.pciSlotNumber = "16"

ethernet0.pciSlotNumber = "160"

sound.pciSlotNumber = "17"

ehci.pciSlotNumber = "18"

usb_xhci.pciSlotNumber = "192"

sata0.pciSlotNumber = "19"

nvme0.pciSlotNumber = "224"

nvme0:0.redo = ""

vmotion.checkpointFBSize = "4194304"

vmotion.checkpointSVGAPrimarySize = "268435456"

vmotion.svga.mobMaxSize = "1073741824"

vmotion.svga.graphicsMemoryKB = "524288"

vmotion.svga.supports3D = "1"

vmotion.svga.baseCapsLevel = "9"

vmotion.svga.maxPointSize = "189"

vmotion.svga.maxTextureSize = "16384"

vmotion.svga.maxVolumeExtent = "2048"

vmotion.svga.maxTextureAnisotropy = "16"

vmotion.svga.lineStipple = "0"

vmotion.svga.dxMaxConstantBuffers = "15"

vmotion.svga.dxProvokingVertex = "0"

vmotion.svga.sm41 = "1"

vmotion.svga.multisample2x = "1"

vmotion.svga.multisample4x = "1"

vmotion.svga.msFullQuality = "1"

vmotion.svga.logicOps = "1"

vmotion.svga.bc67 = "9"

vmotion.svga.sm5 = "1"

vmotion.svga.multisample8x = "1"

vmotion.svga.logicBlendOps = "1"

vmotion.svga.maxForcedSampleCount = "8"

vmotion.svga.gl43 = "1"

ethernet0.generatedAddress = "00:0C:29:C6:66:C4"

ethernet0.generatedAddressOffset = "0"

vmci0.id = "885417668"

nvme0.subnqnUUID = "52 fd ba b7 2f c8 69 7f-a2 a7 96 c1 b5 ae 28 77"

monitor.phys_bits_used = "36"

cleanShutdown = "TRUE"

softPowerOff = "TRUE"

usb_xhci:6.speed = "2"

usb_xhci:6.present = "TRUE"

usb_xhci:6.deviceType = "hub"

usb_xhci:6.port = "6"

usb_xhci:6.parent = "-1"

usb_xhci:7.speed = "4"

usb_xhci:7.present = "TRUE"

usb_xhci:7.deviceType = "hub"

usb_xhci:7.port = "7"

usb_xhci:7.parent = "-1"

sata0:1.startConnected = "FALSE"

toolsInstallManager.lastInstallError = "21004"

svga.guestBackedPrimaryAware = "TRUE"

gui.fitGuestUsingNativeDisplayResolution = "FALSE"

tools.capability.verifiedSamlToken = "TRUE"

tools.remindInstall = "FALSE"

toolsInstallManager.updateCounter = "4"

guestInfo.detailed.data = "architecture='Arm' bitness='32' buildNumber='26200' distroName='Windows' distroVersion='10.0' familyName='Windows' kernelVersion='26200.7462' prettyName='Windows 11 Pro, 32-bit (Build 26200.7462)'"

keyboardAndMouseProfile = "5243576e-5da8-d127-4c06-2f6b7025fd03"

gui.applyHostDisplayScalingToGuest = "FALSE"

checkpoint.vmState = ""

vmx.buildType = "release"

hard-disk.hostBuffer = "disabled"

isolation.tools.dnd.disable = "TRUE"

sata0:1.autodetect = "TRUE"

cpuid.coresPerSocket.cookie = "4"

nvme0:0.present = "TRUE"

gui.lastPoweredViewMode = "fullscreen"

gui.viewModeAtPowerOn = "fullscreen"

usb_xhci:5.present = "TRUE"

usb_xhci:5.deviceType = "hid"

usb_xhci:5.port = "5"

usb_xhci:5.parent = "-1"

usb_xhci:4.present = "TRUE"

usb_xhci:4.deviceType = "video"

usb_xhci:4.port = "4"

usb_xhci:4.parent = "-1"

####End of VMX####

-------------------------------------------

[Gaurav Joshi]

Hey folks,

If you're seeing high memory usage by vmware-vmx and experiencing system hangs because of it,
could you please share your Host OS version and Guest Os version and Macbook version (Intel/M3/M4 pro etc)?

Also, please run the following command and share the output with us:

>sudo leaks <pid of the vmware-vmx process>

It would help us determine whether there are actual memory leaks or if something else is going on.

Thanks,

Gaurav

-------------------------------------------

[krayzee]

Hello Gaurav, here is my response to your request.  Please bear with me and ask for more specifics as I am not very savvy with this stuff.

I believe the information is as follows.  My Host OS is MacOS 26.2.  My Guest OS is Windows 11 Pro version 25H2, OS build 26200.7462 with a 64 bit operating system.  I am using a 2019 Mac Pro workstation (cheese grater) which is intel based.  The guest is configured for 20 GB installed RAM, 8 processor cores, and 3D acceleration is disabled.  I am running VMware fusion 25H2 (24995814).  

The process vmware-vmx grows continuously starting out around 20 GB, and the MacOS halts the process somewhere between 300-350 GB.  If I suspend the guest OS and resume, the vmware-vmx process drops back down to almost nothing.  If I reboot the guest OS, it does not change anything unless I also restart fusion.  It takes a few days for the process to hit 300 GB in size.

When I run the sudo command I get the following message 

Process 92558 is not debuggable. Due to security restrictions, leaks can only show or save contents of readonly memory of restricted processes.

-------------------------------------------

[Gaurav Joshi]

Apologies, I completely forget we don't allow attaching debugger on `vmware-vmx` as its VMware fusion official released build (not debug enabled).

Btw, for me it is printing output on console on the same build. It may look like its stuck, but it should print something from read-only memory of process on console after some time. 

Anyways we are trying to repro this issue internally by tweaking few things so that the release build can print leaks from writable (heap) memory of the target process on our side and I can't share such build in public. 

(I am also on Intel Mac book pro 2019, MacOs 26.2)

-------------------------------------------

[Duncan Epping]

I just submitted the output from Leaks on the bug. Will do the same tomorrow after rebooting the VM and restarting Fusion Guarav. You should be getting the info you need asap.

-------------------------------------------

[Serje Kon]

My Scenario for Reproduction:

Hardware & Host OS:
MacBook Air M4 with 24 GB RAM, running macOS 15.5 (build 24F74).

Virtualization Setup:
VMware Fusion 25H2, with a Windows 11 ARM virtual machine (build 25H2), clean-installed from an ISO.

Launch the Windows 11 ARM virtual machine.

Steps to Reproduce:

1. Open Microsoft Edge.

2. Wait for the default page to load completely.

3. Close Microsoft Edge.

Observed Issue:
With each repetition of these steps, the memory consumption of the vmware-vmx process increases by 60–100 MB and does not release the accumulated memory after closing Edge.

-------------------------------------------

[Technogeezer]

Gaurav, can you make sure this is tested on an Apple Silicon Mac as well?

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[krayzee]

Hi Gaurav,  Just a quick clarification, I am using a Mac Pro workstation, not a laptop.  I did run the leaks command and got a large printout of leaks reported after waiting for a few minutes for it to run.  Here's the start of the output, I'll post the bottom of the output in my next post.  Let me know if you want me to put the entire output into a file and send it directly to you somehow ...

Process:         vmware-vmx [92558]

Path:            /Applications/VMware Fusion.app/Contents/Library/vmware-vmx

Load Address:    0x109b4a000

Identifier:      vmware-vmx

Version:         25.0.0 (0)

Code Type:       X86-64

Platform:        macOS

Parent Process:  launchd [1]

Target Type:     live task

 

Date/Time:       2026-01-07 15:29:29.207 -0800

Launch Time:     2026-01-06 17:43:04.881 -0800

OS Version:      macOS 26.2 (25C56)

Report Version:  7

Analysis Tool:   /usr/bin/leaks

 

Physical footprint:         64.8G

Physical footprint (peak):  64.8G

Idle exit:                  untracked

----

 

leaks Report Version: 4.0

Process 92558: 118678 nodes malloced for 545967 KB

Process 92558: 2830 leaks for 174704 total leaked bytes.

 

    2830 (171K) << TOTAL >>

 

      10 (736 bytes) ROOT LEAK: <IOSurface 0x600000910430> [16]

         9 (720 bytes) _impl --> <IOSurface._impl (malloc) 0x7fae3c284270> [400]

            5 (224 bytes) <NSDictionary 0x600000b30360> [32]

               3 (176 bytes) <NSDictionary (Storage) 0x6000020c0620> [112]

                  1 (32 bytes) <CFString 0x600000b30460> [32]

                  1 (32 bytes) <CFString 0x600000b32060> [32]

               1 (16 bytes) cow --> <NSDictionary.cow (struct __cow_state_t) 0x6000009108c0> [16]

            3 (96 bytes) <NSDictionary 0x600000b30ec0> [32]

               1 (48 bytes) <NSDictionary (Storage) 0x60000057e670> [48]

               1 (16 bytes) cow --> <NSDictionary.cow (struct __cow_state_t) 0x600000911480> [16]

 

      10 (736 bytes) ROOT LEAK: <IOSurface 0x600000910670> [16]

         9 (720 bytes) _impl --> <IOSurface._impl (malloc) 0x7fae0b835b00> [400]

            5 (224 bytes) <NSDictionary 0x600000b03ae0> [32]

               3 (176 bytes) <NSDictionary (Storage) 0x6000020b1c00> [112]

                  1 (32 bytes) <CFString 0x600000b01380> [32]

                  1 (32 bytes) <CFString 0x600000b036e0> [32]

               1 (16 bytes) cow --> <NSDictionary.cow (struct __cow_state_t) 0x600000913b00> [16]

            3 (96 bytes) <NSDictionary 0x600000b03420> [32]

               1 (48 bytes) <NSDictionary (Storage) 0x6000005da1c0> [48]

               1 (16 bytes) cow --> <NSDictionary.cow (struct __cow_state_t) 0x600000910fa0> [16]

-------------------------------------------

[Gaurav Joshi]

Hey Folks,

Thanks for the all the details shared here. We have got the information from Duncan Epping internally and we are looking into it. 

krayzee For your case as 

Physical footprint:         64.8G

Physical footprint (peak):  64.8G

Process 92558: 118678 nodes malloced for 545967 KB

Process 92558: 2830 leaks for 174704 total leaked bytes.

As the physical footprint is very large, and as you can see, the leaks are almost insignificant in comparison of that.

btw We are seeing similar behaviour wrt leaks in other cases also. 
Once you hit such large memory footprint, Could you please share the output of (in a file and upload here) :

sudo vmmap -summary <pid of vmx process>

Thanks,

Gaurav

-------------------------------------------

[krayzee]

-------------------------------------------

Attachment(s)

vmmap.rtf   8 KB 1 version

[krayzee]

Hi Gaurav, you can eliminate my from the list of people experiencing this issue, I have made the decision to start over and move forward with a different solution.  Clearly there are only a handful of people experiencing this issue, now one less.  

Know that is has nothing to do with you, but rather the list of showstopper issues above and beyond this one that I've been experiencing recently which I can't see a path to resolution.  Thanks for being very responsive and good luck moving forward.

-------------------------------------------

[krayzee]

Here is the last little part of the leaks report output (the output is quite huge)

      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009fe7d0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009fe840> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009fe970> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009fea90> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009fecb0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009fef10> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff0f0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff1d0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff220> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff2f0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff450> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff550> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff600> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff7a0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff7e0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ff920> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ffa00> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ffad0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ffbd0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ffc70> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ffd00> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ffd20> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009ffed0> [16]
      1 (16 bytes) ROOT LEAK: <IOSurface 0x6000009fff70> [16]

-------------------------------------------

[krayzee]

Just want to mention an interesting observation.  In my guest windows virtual machine I run a program called blueiris, which displays video streams from several video security cameras and writes the video to an external NAS device 24/7.  When I terminate the program, the speed at which the vmware-vmx grows slows down to about 20% of that when it is running.  I don't know if that means anything but I thought I'd mention it.  I don't know if this is consistent behavior.

-------------------------------------------

[Technogeezer]

Gaurav, Another set of data points. The attached zip file contains some traces from the vmmap command you suggested:

• vmmap_baseline.zip - a snapshot of memory footprint immediately after booting the Windows 11 ARM guest.
• vmmap_after_excel.txt - after I've edited an Excel spreadsheet. Note the memory usage has more than doubled.
• vmmap_after_more_stuff.txt - after performing some additional tasks (I think I opened Microsoft Edge and did some web browsing). Note that in the 25 or so minutes of usage, the memory footprint has added another 2 GB.
• vmmap_after_resume.txt - I suspended and resumed the guest. Note that memory footprint has decreased fro a high of 14GB to 8.1 GB. It still increases as I use the guest from that point. 

Something is seriously wrong here. Maybe a periodic suspend and resume can help mitigate the issue, but it doesn't seem right that a VM should continue to consume memory at the rate we're seeing. 

This behavior is a very big problem to users of Fusion on macOS Tahoe. It appears to be more widespread than I'd originally thought (it looks like it may be happening to others but they aren't noticing it). 

We really can't wait months for Broadcom to identify and fix this.  This isn't a cosmetic issue. 

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

Attachment(s)

FusionMemoryLeak.zip   447 KB 1 version

[Furkan Yucel]

I appricate to you. It's common issue but probably they never shared lot of memory from host. I hope they fix it quickly.

-------------------------------------------

[Gaurav Joshi]

Thanks Paul for more data points, we are actively looking into it.

-------------------------------------------

[Duncan Epping]

It is being worked on Paul, I've been testing various builds, even today I downloaded a new build and been testing it and it looks promising. Reporting back to engineering/PM and then they can decide on the next steps. And indeed, pause/resume resets the memory usage it seem, but this isn't a fix.

-------------------------------------------

[Technogeezer]

@Duncan Epping, Gaurav: Thanks, and it's good news to hear that it's being actively worked on. If there is anything that the community can do to help (e.g. provide more information), please let us know. 

Also glad to hear that a suspend/resume works for you as well to mitigate the issue. It's not a fix, but might help people get through the issue until a fix does appear. 

------------------------------
- Paul (technogeezer)
vExpert 2025
------------------------------

[Duncan Epping]

I've been testing various builds Paul, seems a fix was just checked in, so I am going to test the latest build to see if that fixes the issue.

-------------------------------------------

[shdwkeeper]

Experiencing the same exact issue on 25, have to keep shutting the VM down and starting over and I'm running out of App memory, I was at 150GB. and not doing anything.  When is the patch or fixed going to be released?

-------------------------------------------

[shdwkeeper]

I'm experiencing the same thing with Fusion 25, where its running up to 150+GB and I'm not even doing anything and a I get Running out of App Memory error, I have to keep shutting down and rebooting MY Win 11 VM.  Any fix available yet?

-------------------------------------------

[Technogeezer]

No fix is available publicly at the time of this reply. I suggest to periodically suspend and then resume the VM to work around this issue until a proper resolution is shipped.

I would hope that Broadcom would fix this sooner rather than later and we won't have to continue to deal with this. 

------------------------------
Paul Rockwell (technogeezer)
vExpert 2025
------------------------------

[krayzee]

I don't know if this is the time, or the place, to bring this up.   There's another significant, but unrelated issue I am experiencing many times when launching vmware that I have only seen with the latest version of vmware running on Tahoe.  Right after vmware launches, my bluetooth gets disabled.  In this event, I loose my mouse which connects via bluetooth along with other bluetooth devices.  I then use a usb mouse to regain control of my pointer.  If I then go to the top menu bar of macos to re-enable bluetooth with vmware still running, it does not show any bluetooth devices.  If I shut down vmware and then renable bluetooth, my devices all reappear as normal.  If I relaunch vmware, it then works and stops disabling bluetooth.

This behavior is not consistent, it doesn't happen every time I launch vmware fusion but it happens quite frequently, so I avoid relaunching at all costs.  

Sorry if this post is an unwelcome distraction, but I thought better to inform than to withhold.

-------------------------------------------

[Technogeezer]

@krayzee that's unusual - I'm running BT keyboards and mouse to my Mac and haven't seen this with any version of Fusion or  macOS (at least on Apple Silicon). I wonder what's different with your setup?

I'm assuming that 

• the Bluetooth is the built-in BT to the Mac that you're using, a
• you're not attempting to pass Bluetooth to the VM
• You don't have any overrides in the .vmx file to pass HID devices to the VM

You might want to start a new thread on this issue with more details about your configuration (such as the hardware you're using -- both mac and peripherals). Please include not only the logs in the VM's bundle, but the ~/Library/Logs/VMware\ Fusion/vmware-vmfusion*.log files as well.

------------------------------
----
Paul Rockwell (technogeezer)
------------------------------

[krayzee]

Hi Paul,

 

I am progressing to the competitor "paid for" alternative hypervisor, and plan to use VMware fusion just as an archival windows resource if needed.  As such I am no longer looking for resolution to the rather large list of issues (over and above these two) that I've encountered.  If there is value to supplying logs, etc in a non public way I am happy to do so, thanks. 

-------------------------------------------

[Jean Roy]

Hi,

I use a lot of VM on my Mac Pro 2019 since 2019.  I have VM running windows 10, windows 11 and different version of Mac OS. All of them are experiencing memory leak since VMware 25h2.

With VMware Fusion 13.6.4 I was able to leave VM running for weeks. Now, I have to shut down and restart them every few days because they take all my RAM. And I have 192GB !!

-------------------------------------------