UVM_KM(9) Kernel Developer's Manual UVM_KM(9)

uvm_km
raw kernel memory or address space allocator

#include <sys/param.h>
#include <uvm/uvm.h>
vaddr_t
uvm_km_alloc(struct vm_map *map, vsize_t size, vsize_t align, uvm_flag_t flags);
void
uvm_km_free(struct vm_map *map, vaddr_t addr, vsize_t size, uvm_flag_t flags);
struct vm_map *
uvm_km_suballoc(struct vm_map *map, vaddr_t *min, vaddr_t *max, vsize_t size, int flags, bool fixed, struct vm_map *submap);

The UVM facility for allocation of kernel memory or address space in pages. Both wired and pageable memory can be allocated by this facility, as well as kernel address space. Note that this is a raw allocator. For general purpose memory allocation, kmem(9) interface should be used.

uvm_km_alloc() allocates a contiguous range of size bytes of kernel memory in map map and returns the virtual address of the range, or returns zero on failure. The first address of the allocated memory range will be aligned according to the align argument (specify 0 if no alignment is necessary). The alignment must be a multiple of page size. The flags is a bitwise inclusive OR of the allocation type and operation flags.
The allocation type should be one of:
 
 
UVM_KMF_WIRED
Wired memory.
 
 
UVM_KMF_PAGEABLE
Demand-paged zero-filled memory.
 
 
UVM_KMF_VAONLY
Virtual address only. No physical pages are mapped in the allocated region. If necessary, it is the caller's responsibility to enter page mappings. It is also the caller's responsibility to clean up the mappings before freeing the address range.
The following operation flags are available:
 
 
UVM_KMF_CANFAIL
Can fail even if UVM_KMF_NOWAIT is not specified and UVM_KMF_WAITVA is specified.
 
 
UVM_KMF_ZERO
Request zero-filled memory. Only supported for UVM_KMF_WIRED. Should not be used with other types.
 
 
UVM_KMF_EXEC
Request memory with executable rights.
 
 
UVM_KMF_TRYLOCK
Fail if cannot lock the map without sleeping.
 
 
UVM_KMF_NOWAIT
Fail immediately if no memory is available.
 
 
UVM_KMF_WAITVA
Sleep to wait for the virtual address resources if needed.
If neither UVM_KMF_NOWAIT nor UVM_KMF_CANFAIL are specified and UVM_KMF_WAITVA is specified, uvm_km_alloc() will never fail, but rather sleep indefinitely until the allocation succeeds.
Pageability of the pages allocated with UVM_KMF_PAGEABLE can be changed by uvm_map_pageable(). In that case, the entire range must be changed atomically. Changing a part of the range is not supported.
uvm_km_free() frees the memory range allocated by uvm_km_alloc(). addr must be an address returned by uvm_km_alloc(). map and size must be the same as the ones used for the corresponding uvm_km_alloc(). flags must be the allocation type used for the corresponding uvm_km_alloc(). Note that uvm_km_free() is the only way to free memory ranges allocated by uvm_km_alloc(). uvm_unmap() must not be used.
uvm_km_suballoc() allocates submap from map, creating a new map if submap is NULL. The addresses of the submap can be specified explicitly by setting the fixed argument to true, which causes the min argument to specify the beginning of the address in the submap. If fixed is false, any address of size size will be allocated from map and the start and end addresses returned in min and max. The flags are used to initialize the created submap. The following flags can be set:
 
 
VM_MAP_PAGEABLE
Entries in the map may be paged out.
 
 
VM_MAP_INTRSAFE
Map should be interrupt-safe.
 
 
VM_MAP_TOPDOWN
A top-down mapping should be arranged.

kmem(9), pmap(9), pool_cache(9), uvm(9), uvm_map(9), vmem(9)

UVM and uvm_km first appeared in NetBSD 1.4.
August 15, 2015 NetBSD-current