Struct MemoryLayout 

pub struct MemoryLayout { /* private fields */ }

Describes the memory layout of a guest.

Implementations

impl MemoryLayout

pub fn new( ram_size: u64, mmio_gaps: &[MemoryRange], pci_ecam_gaps: &[MemoryRange], pci_mmio_gaps: &[MemoryRange], vtl2_range: Option<MemoryRange>, ) -> Result<Self, Error>

Makes a new memory layout for a guest with ram_size bytes of memory and MMIO gaps at the locations specified by gaps.

ram_size must be a multiple of the page size. Each mmio and device reserved gap must be non-empty, and the gaps must be in order and non-overlapping.

vtl2_range describes a range of memory reserved for VTL2. It is not reported in ram.

All RAM is assigned to NUMA node 0.

pub fn new_with_numa( numa_mem_sizes: &[u64], mmio_gaps: &[MemoryRange], pci_ecam_gaps: &[MemoryRange], pci_mmio_gaps: &[MemoryRange], vtl2_range: Option<MemoryRange>, ) -> Result<Self, Error>

Like Self::new(), but distributes RAM across NUMA nodes according to the per-node sizes in numa_mem_sizes.

numa_mem_sizes[i] is the number of RAM bytes for vnode i. Each size must be page-aligned and non-zero. The sum of all sizes is the total guest RAM.

RAM is placed sequentially around MMIO gaps, filling each node’s budget in order. When a node’s budget is exhausted mid-chunk, the chunk is split and the next node continues from that address.

pub fn new_from_ranges( memory: &[MemoryRangeWithNode], gaps: &[MemoryRange], ) -> Result<Self, Error>

Makes a new memory layout for a guest with the given mmio gaps and memory ranges.

memory and gaps ranges must be in sorted order and non-overlapping, and describe page aligned ranges.

pub fn new_from_resolved_ranges( ram: Vec<MemoryRangeWithNode>, mmio_gaps: Vec<MemoryRange>, pci_ecam_gaps: Vec<MemoryRange>, pci_mmio_gaps: Vec<MemoryRange>, vtl2_range: Option<MemoryRange>, ) -> Result<Self, Error>

Makes a new memory layout from already-resolved RAM and fixed ranges.

Each individual range must be non-empty, but the lists themselves may be empty (e.g. no PCIe root complexes means empty PCI ECAM/MMIO vectors). Ranges within each list must be sorted and non-overlapping. MMIO gaps may contain empty placeholder ranges to preserve positional indexing (e.g. mmio()[0] = low, mmio()[1] = high); empty entries are ignored during validation. The combined layout is also validated for overlaps, including the optional VTL2 range.

pub fn mmio(&self) -> &[MemoryRange]

The MMIO gap ranges.

pub fn ram(&self) -> &[MemoryRangeWithNode]

The populated RAM ranges. This does not include the vtl2_range.

pub fn vtl2_range(&self) -> Option<MemoryRange>

A special memory range for VTL2, if any. This memory range is treated like RAM, but is only used to hold VTL2 and is located above ram and mmio.

pub fn ram_size(&self) -> u64

The total RAM size in bytes. This is not contiguous.

pub fn end_of_ram(&self) -> u64

One past the last byte of RAM.

pub fn ram_below_4gb(&self) -> u64

The bytes of RAM below 4GB.

pub fn ram_above_4gb(&self) -> u64

The bytes of RAM at or above 4GB.

pub fn ram_above_high_mmio(&self) -> Option<u64>

The bytes of RAM above the high MMIO gap.

Returns None if there aren’t exactly 2 MMIO gaps.

pub fn max_ram_below_4gb(&self) -> Option<u64>

The ending RAM address below 4GB.

Returns None if there is no RAM mapped below 4GB.

pub fn end_of_layout(&self) -> u64

One past the last byte of RAM, MMIO, PCI ECAM, or PCI MMIO.

pub fn probe_address(&self, address: u64) -> Option<AddressType>

Probe a given address to see if it is in the memory layout described by self. Returns the AddressType of the address if it is in the layout.

This does not check the vtl2_range.

Trait Implementations

impl Clone for MemoryLayout

fn clone(&self) -> MemoryLayout

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for MemoryLayout

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Inspect for MemoryLayout

fn inspect(&self, req: Request<'_>)

Inspects the object.