Incorrect RAM Values

zapennac · August 6, 2025, 3:17pm

Summary

Fedora 42 kernel only exposes ~15 GB of system RAM out of 128 GB installed on Dell Precision 5860 Tower (Xeon w7-2475X, NVIDIA RTX 5000 Ada). Other OSes (Windows, RHEL, Nobara) recognize full memory.

Description

On a fresh install of Fedora 42 Workstation (GNOME), the system reports only ~14–16 GiB of available memory despite 128 GB ECC DDR5 RAM being installed and recognized by firmware diagnostics.

Dell BIOS diagnostics confirm all 128 GB is present and functional.
Nobara, RHEL, and Windows 11 all detect the full 128 GB without issue.

This seems to be an issue with Fedora’s kernel memory mapping.

Steps to Reproduce

Install Fedora Linux 42 (Workstation Edition) on Dell Precision 5860 Tower

CPU: Intel Xeon w7-2475X
GPU: 2× NVIDIA RTX 5000 Ada
RAM: 8 × 16 GB DDR5 ECC (128 GB total)

Boot into Fedora.
Check memory usage with free -h, inxi -m, or lsmem.

Actual Results

Only ~14.8 GiB of system RAM is available:

$ free -h
              total        used        free      shared  buff/cache   available
Mem:           14Gi       3.5Gi       6.2Gi       512Mi       4.6Gi        10Gi

$ inxi -m
Memory:
  System RAM: total: 128 GiB available: 14.85 GiB used: 3.81 GiB (25.7%)
  Array-1: capacity: 4 TiB slots: 8 modules: 8
  Device-1–8: 16 GiB each (DDR5 ECC)

$ lsmem
RANGE                                  SIZE  STATE REMOVABLE  BLOCK
0x0000000000000000-0x0000000047ffffff  1.1G online       yes    0-8
0x0000000058000000-0x000000005fffffff  128M online       yes     11
0x0000000100000000-0x000000048fffffff 14.3G online       yes 32-145

So only ~15 GB is mapped online.

Expected Results

Fedora should detect and expose the full 128 GB of installed system RAM.

dmesg | grep -i memory
Memory:
  System RAM: total: 128 GiB available: 14.85 GiB used: 5.18 GiB (34.9%)
  Message: For most reliable report, use superuser + dmidecode.
  Array-1: capacity: 4 TiB note: check slots: 8 modules: 8
    EC: Single-bit ECC
  Device-1: DIMM1 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s
  Device-2: DIMM5 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s
  Device-3: DIMM3 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s
  Device-4: DIMM7 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s
  Device-5: DIMM2 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s
  Device-6: DIMM6 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s
  Device-7: DIMM4 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s
  Device-8: DIMM8 type: DDR5 size: 16 GiB speed: spec: 5600 MT/s
    actual: 4400 MT/s

GPU Check

Not related to GPU VRAM:

glxinfo | grep -i 'Video memory'
Video memory: 33016 MB
Dedicated video memory: 33016 MB
Currently available dedicated video memory: 33011 MB

Additional Information

BIOS diagnostics: All 128 GB detected and passed memory test.
Other operating systems tested (Nobara, RHEL, Windows 11) see full memory.
Some documentation suggests setting a manual memory limit in GRUB, but this should not be necessary.

Environment

OS: Fedora 42 Workstation (GNOME 48.4)
Kernel: 6.15.9-201.fc42.x86_64
CPU: Intel Xeon w7-2475X (40 cores @ 4.80 GHz)
Memory: 8 × 16 GB DDR5 ECC (128 GB total)
GPU: 2 × NVIDIA RTX 5000 Ada Generation
Host: Dell Precision 5860 Tower

barryascott · August 6, 2025, 3:20pm

Have a look at the dmesg output and see what the kernel reports for memory. Are there any errors reported, maybe uefi bios errors?

Is your bios up to date?

computersavvy · August 6, 2025, 3:24pm

From dmesg and free I see the following

dmesg
[    0.572137] Memory: 32648472K/33481368K available (21657K kernel code, 4519K rwdata, 17664K rodata, 5060K init, 4268K bss, 808184K reserved, 0K cma-reserved)

$ free -h
               total        used        free      shared  buff/cache   available
Mem:            31Gi       2.8Gi        25Gi       117Mi       3.3Gi        28Gi
Swap:          8.0Gi          0B       8.0Gi

Fedora does not appear to limit memory.
This may be a hardware issue.
You seem to have 8 slots for RAM fully populated and I only have 4 slots with 2 populated.

zapennac · August 6, 2025, 3:26pm

As far as I amware I am up to date with bios:

sudo dmidecode -t bios | grep -i version
	Version: 3.0.2

sudo dmesg | grep -i memory
[    0.000000] DMI: Memory slots populated: 8/8
[    0.017466] ACPI: Reserving FACP table memory at [mem 0x5d6f4000-0x5d6f4113]
[    0.017468] ACPI: Reserving DSDT table memory at [mem 0x5d66e000-0x5d6efc9d]
[    0.017470] ACPI: Reserving FACS table memory at [mem 0x5c37b000-0x5c37b03f]
[    0.017471] ACPI: Reserving SSDT table memory at [mem 0x5d6fc000-0x5d6fcd93]
[    0.017472] ACPI: Reserving SSDT table memory at [mem 0x5d6fb000-0x5d6fb57f]
[    0.017473] ACPI: Reserving SSDT table memory at [mem 0x5d6fa000-0x5d6fa760]
[    0.017474] ACPI: Reserving SSDT table memory at [mem 0x5d6f9000-0x5d6f976d]
[    0.017475] ACPI: Reserving BOOT table memory at [mem 0x5d6f8000-0x5d6f8027]
[    0.017476] ACPI: Reserving BERT table memory at [mem 0x5d6f7000-0x5d6f702f]
[    0.017477] ACPI: Reserving ERST table memory at [mem 0x5d6f6000-0x5d6f622f]
[    0.017478] ACPI: Reserving BDAT table memory at [mem 0x5d6f5000-0x5d6f502f]
[    0.017479] ACPI: Reserving HPET table memory at [mem 0x5d6f3000-0x5d6f3037]
[    0.017480] ACPI: Reserving MCFG table memory at [mem 0x5d6f2000-0x5d6f203b]
[    0.017481] ACPI: Reserving MSCT table memory at [mem 0x5d6f1000-0x5d6f104d]
[    0.017482] ACPI: Reserving WDDT table memory at [mem 0x5d6f0000-0x5d6f003f]
[    0.017483] ACPI: Reserving APIC table memory at [mem 0x5d66d000-0x5d66d2dd]
[    0.017485] ACPI: Reserving SRAT table memory at [mem 0x5d66b000-0x5d66ce2f]
[    0.017486] ACPI: Reserving SLIT table memory at [mem 0x5d66a000-0x5d66a02c]
[    0.017487] ACPI: Reserving HMAT table memory at [mem 0x5d669000-0x5d6690f7]
[    0.017488] ACPI: Reserving OEM4 table memory at [mem 0x5d607000-0x5d668ed0]
[    0.017489] ACPI: Reserving OEM1 table memory at [mem 0x5d5c2000-0x5d606378]
[    0.017490] ACPI: Reserving OEM2 table memory at [mem 0x5d5b0000-0x5d5c1830]
[    0.017491] ACPI: Reserving SSDT table memory at [mem 0x5d595000-0x5d5af65d]
[    0.017492] ACPI: Reserving DBG2 table memory at [mem 0x5d6fd000-0x5d6fd05b]
[    0.017493] ACPI: Reserving SSDT table memory at [mem 0x5d593000-0x5d593631]
[    0.017494] ACPI: Reserving TPM2 table memory at [mem 0x5d592000-0x5d59204b]
[    0.017495] ACPI: Reserving HEST table memory at [mem 0x5d591000-0x5d59113b]
[    0.017496] ACPI: Reserving DMAR table memory at [mem 0x5d590000-0x5d5901bf]
[    0.017497] ACPI: Reserving FPDT table memory at [mem 0x5d58f000-0x5d58f043]
[    0.017498] ACPI: Reserving PHAT table memory at [mem 0x5d58d000-0x5d58d290]
[    0.017499] ACPI: Reserving ASF! table memory at [mem 0x5d58e000-0x5d58e073]
[    0.017500] ACPI: Reserving BGRT table memory at [mem 0x5d58c000-0x5d58c037]
[    0.018377] Early memory node ranges
[    0.053196] PM: hibernation: Registered nosave memory: [mem 0x00000000-0x00000fff]
[    0.053198] PM: hibernation: Registered nosave memory: [mem 0x0009f000-0x000fffff]
[    0.053200] PM: hibernation: Registered nosave memory: [mem 0x3633d000-0x363f9fff]
[    0.053201] PM: hibernation: Registered nosave memory: [mem 0x40091000-0x5d6fefff]
[    0.053203] PM: hibernation: Registered nosave memory: [mem 0x5d700000-0xffffffff]
[    0.192229] Freeing SMP alternatives memory: 56K
[    0.265010] Memory: 15398840K/15990976K available (21657K kernel code, 4519K rwdata, 17664K rodata, 5060K init, 4268K bss, 525248K reserved, 0K cma-reserved)
[    0.265682] x86/mm: Memory block size: 128MB
[    0.623945] acpi/hmat: Memory Flags:0001 Processor Domain:0 Memory Domain:0
[    0.914952] Freeing initrd memory: 96672K
[    0.951719] Non-volatile memory driver v1.3
[    1.734646] Freeing unused decrypted memory: 2028K
[    1.736686] Freeing unused kernel image (initmem) memory: 5060K
[    1.738119] Freeing unused kernel image (text/rodata gap) memory: 868K
[    1.738871] Freeing unused kernel image (rodata/data gap) memory: 768K
[   11.702402] systemd[1]: Listening on systemd-oomd.socket - Userspace Out-Of-Memory (OOM) Killer Socket.

sudo dmesg | grep -i e820
[    0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009efff] usable
[    0.000000] BIOS-e820: [mem 0x000000000009f000-0x00000000000fffff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000000100000-0x0000000040090fff] usable
[    0.000000] BIOS-e820: [mem 0x0000000040091000-0x000000004fd3afff] reserved
[    0.000000] BIOS-e820: [mem 0x000000004fd3b000-0x000000005c3f1fff] ACPI NVS
[    0.000000] BIOS-e820: [mem 0x000000005c3f2000-0x000000005d6fefff] ACPI data
[    0.000000] BIOS-e820: [mem 0x000000005d6ff000-0x000000005d6fffff] usable
[    0.000000] BIOS-e820: [mem 0x000000005d700000-0x000000007fffffff] reserved
[    0.000000] BIOS-e820: [mem 0x00000000fe010000-0x00000000fe010fff] reserved
[    0.000000] BIOS-e820: [mem 0x00000000ff0e0000-0x00000000ff0effff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000100000000-0x000000048fffffff] usable
[    0.000000] efi: Remove mem85: MMIO range=[0x70000000-0x7fffffff] (256MB) from e820 map
[    0.000000] e820: remove [mem 0x70000000-0x7fffffff] reserved
[    0.000000] efi: Not removing mem86: MMIO range=[0xfe010000-0xfe010fff] (4KB) from e820 map
[    0.000000] efi: Not removing mem87: MMIO range=[0xff0e0000-0xff0effff] (64KB) from e820 map
[    0.000034] e820: update [mem 0x00000000-0x00000fff] usable ==> reserved
[    0.000040] e820: remove [mem 0x000a0000-0x000fffff] usable
[    0.001653] e820: update [mem 0x6e000000-0xffffffff] usable ==> reserved
[    0.053158] e820: update [mem 0x3633d000-0x363f9fff] usable ==> reserved
[    0.465840] PCI: Ignoring E820 reservations for host bridge windows
[    0.629773] e820: reserve RAM buffer [mem 0x0009f000-0x0009ffff]
[    0.629775] e820: reserve RAM buffer [mem 0x3633d000-0x37ffffff]
[    0.629776] e820: reserve RAM buffer [mem 0x40091000-0x43ffffff]
[    0.629776] e820: reserve RAM buffer [mem 0x5d700000-0x5fffffff]

pg-tips · August 6, 2025, 3:27pm

Do you see the same behaviour with the Fedora Workstation 42 Live ISO?

That will have an older kernel version than your up-to-date system, so it might help to track down whether this is down to a recent kernel change.

zapennac · August 6, 2025, 3:28pm

So you want me to boot from a Live ISO to see if I see the same issues?
I can do that. May take a while to reply back

pg-tips · August 6, 2025, 3:30pm

It’s interesting that this problem doesn’t appear on Nobara, because it doesn’t seem to change the kernel in obviously relevant ways: Nobara Kernel Modifications | Nobara Project Wiki

barryascott · August 6, 2025, 3:31pm

That says you have 16GiB if I’m reading it right.
So the kernel cannot find the rest of the memory for some reason.

Trying an older kernel is worth a test to see if the isdue is with the newer kernel as @computersavvy suggested.

zapennac · August 6, 2025, 3:35pm

OK everyone I will try an older version from a live ISO and get back to you.
I will add this to the notes:
go to Settings → System → About
It shows 128.0 GiB in memory

# System Details Report
---

## Report details
- **Date generated:**                              2025-08-06 11:34:23

## Hardware Information:
- **Hardware Model:**                              Dell Inc. Precision 5860 Tower
- **Memory:**                                      128.0 GiB
- **Processor:**                                   Intel® Xeon® w7-2475X × 40
- **Graphics:**                                    zink Vulkan 1.4(NVIDIA RTX 5000 Ada Generation (NVK AD102))
- **Graphics 1:**                                  zink Vulkan 1.4(NVIDIA RTX 5000 Ada Generation (NVK AD102))
- **Disk Capacity:**                               36.1 TB

## Software Information:
- **Firmware Version:**                            3.0.2
- **OS Name:**                                     Fedora Linux 42 (Workstation Edition)
- **OS Build:**                                    (null)
- **OS Type:**                                     64-bit
- **GNOME Version:**                               48
- **Windowing System:**                            Wayland
- **Kernel Version:**                              Linux 6.15.9-201.fc42.x86_64

But the rest of the system says “NOPE”

zapennac · August 6, 2025, 4:02pm

I am back.
Bad news
I tried the Live ISO 42 and Live ISO 41-1.4

Both results in the same way.
under system properties 128 GiB
under Fastfetch or any other CLI command 16 GiB.

barryascott · August 6, 2025, 4:24pm

My guess is that the system properties is taking a value from the BIOS, not the kernel.

What is the memory size in /proc/meminfo?
This is for my 32GiB system.

$ cat /proc/meminfo
MemTotal:       32656796 kB
...

zapennac · August 6, 2025, 4:25pm

alincoln@localhost-live:~$ sudo cat /proc/meminfo
MemTotal:       15571128 kB
MemFree:         6734728 kB
MemAvailable:   11833400 kB
Buffers:            3588 kB
Cached:          5439788 kB
SwapCached:            0 kB
Active:          1575884 kB
Inactive:        6220408 kB
Active(anon):     891220 kB
Inactive(anon):  1596744 kB
Active(file):     684664 kB
Inactive(file):  4623664 kB
Unevictable:        6056 kB
Mlocked:            6056 kB
SwapTotal:       8388604 kB
SwapFree:        8388584 kB
Zswap:                 0 kB
Zswapped:              0 kB
Dirty:              1688 kB
Writeback:             8 kB
AnonPages:       2359212 kB
Mapped:           987324 kB
Shmem:            129852 kB
KReclaimable:      97104 kB
Slab:             490880 kB
SReclaimable:      97104 kB
SUnreclaim:       393776 kB
KernelStack:       28848 kB
PageTables:        54104 kB
SecPageTables:      6192 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:    16174168 kB
Committed_AS:   21340892 kB
VmallocTotal:   13743895347199 kB
VmallocUsed:      200264 kB
VmallocChunk:          0 kB
Percpu:            33600 kB
HardwareCorrupted:     0 kB
AnonHugePages:         0 kB
ShmemHugePages:        0 kB
ShmemPmdMapped:        0 kB
FileHugePages:         0 kB
FilePmdMapped:         0 kB
CmaTotal:              0 kB
CmaFree:               0 kB
Unaccepted:            0 kB
Balloon:               0 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
Hugetlb:               0 kB
DirectMap4k:      481864 kB
DirectMap2M:    10266624 kB
DirectMap1G:     6291456 kB

zapennac · August 6, 2025, 4:31pm

I think you are on to something with that last statement Barry:
Looking at this:

sudo dmesg | grep -i e820
[    0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009efff] usable
[    0.000000] BIOS-e820: [mem 0x000000000009f000-0x00000000000fffff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000000100000-0x0000000040090fff] usable
[    0.000000] BIOS-e820: [mem 0x0000000040091000-0x000000004fd3afff] reserved
[    0.000000] BIOS-e820: [mem 0x000000004fd3b000-0x000000005c3f1fff] ACPI NVS
[    0.000000] BIOS-e820: [mem 0x000000005c3f2000-0x000000005d6fefff] ACPI data
[    0.000000] BIOS-e820: [mem 0x000000005d6ff000-0x000000005d6fffff] usable
[    0.000000] BIOS-e820: [mem 0x000000005d700000-0x000000007fffffff] reserved
[    0.000000] BIOS-e820: [mem 0x00000000fe010000-0x00000000fe010fff] reserved
[    0.000000] BIOS-e820: [mem 0x00000000ff0e0000-0x00000000ff0effff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000100000000-0x000000048fffffff] usable
[    0.000000] efi: Remove mem85: MMIO range=[0x70000000-0x7fffffff] (256MB) from e820 map
[    0.000000] e820: remove [mem 0x70000000-0x7fffffff] reserved
[    0.000000] efi: Not removing mem86: MMIO range=[0xfe010000-0xfe010fff] (4KB) from e820 map
[    0.000000] efi: Not removing mem87: MMIO range=[0xff0e0000-0xff0effff] (64KB) from e820 map
[    0.000081] e820: update [mem 0x00000000-0x00000fff] usable ==> reserved
[    0.000110] e820: remove [mem 0x000a0000-0x000fffff] usable
[    0.006067] e820: update [mem 0x6e000000-0xffffffff] usable ==> reserved
[    0.219327] e820: update [mem 0x3633d000-0x363f9fff] usable ==> reserved
[    1.472232] PCI: Ignoring E820 reservations for host bridge windows
[    1.635276] e820: reserve RAM buffer [mem 0x0009f000-0x0009ffff]
[    1.635278] e820: reserve RAM buffer [mem 0x3633d000-0x37ffffff]
[    1.635279] e820: reserve RAM buffer [mem 0x40091000-0x43ffffff]
[    1.635280] e820: reserve RAM buffer [mem 0x5d700000-0x5fffffff]

what does this mean?

e820: update [mem 0x6e000000-0xffffffff] usable ==> reserved

Does this mean Fedora is reserving the rest of the memory?
Does that mean if I ever hit the 16 GiB Fedora will release more?

barryascott · August 6, 2025, 4:34pm

Again the kernel is reporting that you only have 16Gib.

Random thought, what is your kernel command line?
You can see it with sudo grubby --info=ALL?

zapennac · August 6, 2025, 4:36pm

sudo grubby --info=ALL
index=0
kernel="/boot/vmlinuz-6.15.9-201.fc42.x86_64"
args="ro rootflags=subvol=root rhgb quiet $tuned_params"
root="UUID=531f0abc-7e5c-4bbb-8c88-4e566166db99"
initrd="/boot/initramfs-6.15.9-201.fc42.x86_64.img $tuned_initrd"
title="Fedora Linux (6.15.9-201.fc42.x86_64) 42 (Workstation Edition)"
id="bcfebfffd08a403883e4c3d716e99c86-6.15.9-201.fc42.x86_64"
index=1
kernel="/boot/vmlinuz-6.14.0-63.fc42.x86_64"
args="ro rootflags=subvol=root rhgb quiet $tuned_params"
root="UUID=531f0abc-7e5c-4bbb-8c88-4e566166db99"
initrd="/boot/initramfs-6.14.0-63.fc42.x86_64.img $tuned_initrd"
title="Fedora Linux (6.14.0-63.fc42.x86_64) 42 (Workstation Edition)"
id="bcfebfffd08a403883e4c3d716e99c86-6.14.0-63.fc42.x86_64"
index=2
kernel="/boot/vmlinuz-0-rescue-bcfebfffd08a403883e4c3d716e99c86"
args="ro rootflags=subvol=root rhgb quiet"
root="UUID=531f0abc-7e5c-4bbb-8c88-4e566166db99"
initrd="/boot/initramfs-0-rescue-bcfebfffd08a403883e4c3d716e99c86.img"
title="Fedora Linux (0-rescue-bcfebfffd08a403883e4c3d716e99c86) 42 (Workstation Edition)"
id="bcfebfffd08a403883e4c3d716e99c86-0-rescue"

barryascott · August 6, 2025, 4:43pm

The kernel command line looks normal.

At this point I think you should report this as a kernel bug in the linux kernel bug tracker.
I don’t think yhou will get help reporting in the Fedora bugtracker.

It would be worth asking Dell about supported OS.
You should be told of the RHEL versions that are known to work on it.

zapennac · August 6, 2025, 5:18pm

Won’t lie, little disappointed, but I will reach out to Dell and see what they say. If I get a response or solution I will let you know.

Thanks

mattdm · August 6, 2025, 6:00pm

Have you tried RawhideKernelNodebug - Fedora Project Wiki ? Sometimes, newer is better…

zapennac · August 6, 2025, 7:24pm

Thanks for the suggestion. The instructions didn’t work for me I had to use a different method but I still got it installed:

cat /etc/yum.repos.d/rawhide-kernel-nodebug.repo 

[rawhide-kernel-nodebug]
name=Fedora Rawhide Kernel Nodebug
baseurl=https://dl.fedoraproject.org/pub/alt/rawhide-kernel-nodebug/$basearch/
enabled=1
gpgcheck=0
skip_if_unavailable=1

kernel-core-6.16.0-65.fc43.x86_64
kernel-modules-6.16.0-65.fc43.x86_64
kernel-6.16.0-65.fc43.x86_64

But sadly still the same results…Although I happen to come across the the fact that I am using the systemd-boot as the bootloader (not GRUB).

So thank you for that

barryascott · August 6, 2025, 9:04pm

You are running a customised install, not the Fedora one?

Topic		Replies	Views
Fedora Server 39 not using all of the available RAM on my HP EliteDesk 705 G2 Desktop Mini Ask Fedora f39 , server	18	1204	December 5, 2023
How to not crash Fedoras Server 41 with corrected ECC errors Ask Fedora	22	257	January 7, 2025
I'm new and trying to install Fedora 42 kde plasma and got an error Ask Fedora installation , grub , errors	23	1269	June 22, 2025
Fedora 39 (Random) Freezes Ask Fedora kde , bluetooth , audio , gnome , intel , lenovo , f39 , workstation	22	2353	March 14, 2024
Suddenly my system is out of memory all the time Ask Fedora	10	2920	November 17, 2022