Why am I able to passthrough pcie devices without enabling IOMMU?

[WidePeepoHappy]

Edit:
intel_iommu seems to be on by default now. Does this mean iommu=pt is also on by default?
https://pve.proxmox.com/wiki/Roadmap#Proxmox_VE_8.2

Kernel: intel_iommu now defaults to on​

The intel_iommu parameter defaults to on in the kernel 6.8 series. Enabling IOMMU can cause problems with older hardware, or systems with not up to date BIOS, due to bugs in the BIOS.

The issue can be fixed by explicitly disabling intel_iommu on the kernel commandline (intel_iommu=off) following the reference documentation.

Why am I able to passthrough pcie devices without putting intel_iommu=on iommu=pt in /etc/default/grub ? Both my dedicated gpu, igpu, and nvme seems to be working in the vm natively with pcie passthrough.

When I run dmesg | grep -e DMAR -e IOMMU there is no "DMAR: IOMMU enabled"

[    0.009280] ACPI: DMAR 0x0000000073E6E000 000088 (v01 INTEL  EDK2     00000002      01000013)
[    0.009304] ACPI: Reserving DMAR table memory at [mem 0x73e6e000-0x73e6e087]
[    0.264543] DMAR: Host address width 39
[    0.264543] DMAR: DRHD base: 0x000000fed90000 flags: 0x0
[    0.264548] DMAR: dmar0: reg_base_addr fed90000 ver 4:0 cap 1c0000c40660462 ecap 29a00f0505e
[    0.264549] DMAR: DRHD base: 0x000000fed91000 flags: 0x1
[    0.264552] DMAR: dmar1: reg_base_addr fed91000 ver 5:0 cap d2008c40660462 ecap f050da
[    0.264554] DMAR: RMRR base: 0x0000007c000000 end: 0x000000807fffff
[    0.264555] DMAR-IR: IOAPIC id 2 under DRHD base  0xfed91000 IOMMU 1
[    0.264556] DMAR-IR: HPET id 0 under DRHD base 0xfed91000
[    0.264557] DMAR-IR: Queued invalidation will be enabled to support x2apic and Intr-remapping.
[    0.266114] DMAR-IR: Enabled IRQ remapping in x2apic mode
[    0.417350] pci 0000:00:02.0: DMAR: Skip IOMMU disabling for graphics
[    0.485464] DMAR: No ATSR found
[    0.485465] DMAR: No SATC found
[    0.485466] DMAR: IOMMU feature fl1gp_support inconsistent
[    0.485467] DMAR: IOMMU feature pgsel_inv inconsistent
[    0.485467] DMAR: IOMMU feature nwfs inconsistent
[    0.485468] DMAR: IOMMU feature dit inconsistent
[    0.485469] DMAR: IOMMU feature sc_support inconsistent
[    0.485469] DMAR: IOMMU feature dev_iotlb_support inconsistent
[    0.485471] DMAR: dmar0: Using Queued invalidation
[    0.485472] DMAR: dmar1: Using Queued invalidation
[    0.488788] DMAR: Intel(R) Virtualization Technology for Directed I/O

I'm trying to understand if I really need to enable this to fully passthrough nvme's and gpu's to the VM. As I want to keep the host as clean as possible without altering the host files.

CPU: i7-13700k
Motherboard: mpg z690 edge wifi ddr4
Have PVE (6.8.4-2-pve) installed in ZFS, using grub as bootloader with secure boot.

 

[leesteken]

Edit:
intel_iommu seems to be on by default now.

Yes, but Proxmox has not updated the manual (or Wiki) yey.

Does this mean iommu=pt is also on by default?

No, but that only does something for non-passed through devices and is almost never needed. And I think the Proxmox manual is not entirely correct on what that does.