The Limine Boot Protocol

The Limine boot protocol is a modern, portable, featureful, and extensible boot protocol.

This file serves as the official centralised collection of features that the Limine boot protocol is comprised of. Other bootloaders may support extra unofficial features, but it is strongly recommended to avoid fragmentation and submit new features by opening a pull request to the Limine repository.

The limine.h file provides an implementation of all the structures and constants described in this document, for the C and C++ languages.

General Notes

All pointers are 64-bit wide. All non-NULL pointers point to the object with the higher half direct map offset already added to them, unless otherwise noted.

All responses and associated data structures are placed in bootloader-reclaimable memory regions.

The calling convention matches the C ABI for the specific architecture (SysV for x86, AAPCS LP64 for aarch64, LP64 for riscv64).

Base protocol revisions

The Limine boot protocol comes in several base revisions; so far, 4 base revisions are specified: 0 through 3.

Base protocol revisions change certain behaviours of the Limine boot protocol outside any specific feature. The specifics are going to be described as needed throughout this specification.

Base revision 0 through 2 are considered deprecated. Base revision 0 is the default base revision a kernel is assumed to be requesting and complying to if no base revision tag is provided by the kernel, for backwards compatibility.

A base revision tag is a set of 3 64-bit values placed somewhere in the loaded kernel image on an 8-byte aligned boundary; the first 2 values are a magic number for the bootloader to be able to identify the tag, and the last value is the requested base revision number. Lack of base revision tag implies revision 0.

#define LIMINE_BASE_REVISION(N) \
    uint64_t limine_base_revision[3] = { 0xf9562b2d5c95a6c8, 0x6a7b384944536bdc, (N) };

If a bootloader drops support for an older base revision, the bootloader must fail to boot a kernel requesting such base revision. If a bootloader does not yet support a requested base revision (i.e. if the requested base revision is higher than the maximum base revision supported), it must boot the kernel using any arbitrary revision it supports, and communicate failure to comply to the kernel by leaving the 3rd component of the base revision tag unchanged. On the other hand, if the kernel's requested base revision is supported, the 3rd component of the base revision tag must be set to 0 by the bootloader.

Note: this means that unlike when the bootloader drops support for an older base revision and it is responsible for failing to boot the kernel, in case the bootloader does not yet support the kernel's requested base revision, it is up to the kernel itself to fail (or handle the condition otherwise).

For any Limine-compliant bootloader supporting base revision 3, it is mandatory to load kernels requesting higher unsupported base revisions with at least base revision 3, and it is mandatory for it to always set the 2nd component of the base revision tag to the base revision actually used to load the kernel, regardless of whether it was the requested one or not.

Features

The protocol is centered around the concept of request/response - collectively named "features" - where the kernel requests some action or information from the bootloader, and the bootloader responds accordingly, if it is capable of doing so.

In C terms, a feature is comprised of 2 structures: the request, and the response.

A request has 3 mandatory members at the beginning of the structure:

struct limine_example_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_example_response *response;
    ... optional members follow ...
};

id - The ID of the request. This is an 8-byte aligned magic number that the bootloader will scan for inside the loaded executable image to find requests. Requests may be located anywhere inside the loaded executable image as long as they are 8-byte aligned. There may only be 1 of the same request. The bootloader will refuse to boot an executable with multiple of the same request IDs. Alternatively, it is possible to provide a list of requests explicitly via an executable file section. See "Limine Requests Section". (Note: this is deprecated and removed in base revision 1)
revision - The revision of the request that the kernel provides. This starts at 0 and is bumped whenever new members or functionality are added to the request structure. Bootloaders process requests in a backwards compatible manner, always. This means that if the bootloader does not support the revision of the request, it will process the request as if were the highest revision that the bootloader supports.
response - This field is filled in by the bootloader at load time, with a pointer to the response structure, if the request was successfully processed. If the request is unsupported or was not successfully processed, this field is left untouched, meaning that if it was set to NULL, it will stay that way.

A response has only 1 mandatory member at the beginning of the structure:

struct limine_example_response {
    uint64_t revision;
    ... optional members follow ...
};

revision - Like for requests, bootloaders will instead mark responses with a revision number. This revision is not coupled between requests and responses, as they are bumped individually when new members are added or functionality is changed. Bootloaders will set the revision to the one they provide, and this is always backwards compatible, meaning higher revisions support all that lower revisions do.

This is all there is to features. For a list of official Limine features, read the "Feature List" section below.

Requests Delimiters

The bootloader can be told to start and/or stop searching for requests (including base revision tags) in an executable's loaded image by placing start and/or end markers, on an 8-byte aligned boundary.

The bootloader will only accept requests placed between the last start marker found (if there happen to be more than 1, which there should not, ideally) and the first end marker found.

#define LIMINE_REQUESTS_START_MARKER \
    uint64_t limine_requests_start_marker[4] = { 0xf6b8f4b39de7d1ae, 0xfab91a6940fcb9cf, \
                                                 0x785c6ed015d3e316, 0x181e920a7852b9d9 };

#define LIMINE_REQUESTS_END_MARKER \
    uint64_t limine_requests_end_marker[2] = { 0xadc0e0531bb10d03, 0x9572709f31764c62 };

For base revisions 0 and 1, the requests delimiters are hints. The bootloader can still search for requests and base revision tags outside the delimited area if it doesn't support the hints.

Base revision 2's sole difference compared to base revision 1 is that support for request delimiters has to be provided and the delimiters must be honoured, if present, rather than them just being a hint.

Limine Requests Section

Note: This behaviour is deprecated and removed as of base protocol revision 1

For kernels requesting deprecated base revision 0, if the executable kernel file contains a .limine_reqs section, the bootloader will, instead of scanning the executable for requests, fetch the requests from a NULL-terminated array of pointers to the provided requests, contained inside said section.

Entry memory layout

The protocol mandates kernels to load themselves at or above 0xffffffff80000000. Lower half kernels are not supported. For relocatable kernels asking to be loaded at address 0, a minimum slide of 0xffffffff80000000 is applied.

At handoff, the kernel will be properly loaded and mapped with appropriate MMU permissions, as supervisor, at the requested virtual memory address (provided it is at or above 0xffffffff80000000).

No specific physical memory placement is guaranteed, except that the loaded kernel image is guaranteed to be physically contiguous. In order to determine where the kernel is loaded in physical memory, see the Kernel Address feature below.

Alongside the loaded kernel, the bootloader will set up memory mappings as such:

 Base Physical Address |                               | Base Virtual Address
 ----------------------+-------------------------------+-----------------------
                       |    (4 GiB - 0x1000) and any   |
  0x0000000000001000   |  additional memory map region |  0x0000000000001000
                       |    (Base revision 0 only)     |
 ----------------------+-------------------------------+-----------------------
                       |     4 GiB and additional      |
  0x0000000000000000   |  memory map regions depending |      HHDM start
                       |       on base revision        |

Where "HHDM start" is returned by the Higher Half Direct Map feature (see below). These mappings are supervisor, read, write, execute (-rwx).

For base revision 0, the above-4GiB identity and HHDM mappings cover any memory map region.

For base revisions 1 and 2, the above-4GiB HHDM mappings do not comprise memory map regions of types:

Reserved
Bad memory

For base revision 3, the only memory map regions mapped to the HHDM are:

Usable
Bootloader reclaimable
Kernel and modules
Framebuffer

For base revision 3, the unconditional direct map of the first 4GiB is dropped, and only memory map regions of complying types are mapped in.

The bootloader page tables are in bootloader-reclaimable memory (see Memory Map feature below), and their specific layout is undefined as long as they provide the above memory mappings.

If the kernel is a position independent executable, the bootloader is free to relocate it as it sees fit, potentially performing KASLR (as specified by the config).

Caching

x86-64

The kernel executable, loaded at or above 0xffffffff80000000, sees all of its segments mapped using write-back (WB) caching at the page tables level.

All HHDM and identity map memory regions are mapped using write-back (WB) caching at the page tables level, except framebuffer regions which are mapped using write-combining (WC) caching at the page tables level (if the CPU support the PAT, see below).

If the CPU supports the PAT (Page Attribute Table), its layout is specified to be as follows:

PAT0 -> WB
PAT1 -> WT
PAT2 -> UC-
PAT3 -> UC
PAT4 -> WP
PAT5 -> WC
PAT6 -> unspecified
PAT7 -> unspecified

The MTRRs are left as the firmware set them up.

aarch64

The kernel executable, loaded at or above 0xffffffff80000000, sees all of its segments mapped using Normal Write-Back RW-Allocate non-transient caching mode.

All HHDM and identity map memory regions are mapped using the Normal Write-Back RW-Allocate non-transient caching mode, except for the framebuffer regions, which are mapped in using an unspecified caching mode, correct for use with the framebuffer on the platform.

The MAIR_EL1 register will at least contain entries for the above-mentioned caching modes, in an unspecified order.

In order to access MMIO regions, the kernel must ensure the correct caching mode is used on its own.

riscv64

If the Svpbmt extension is available, all framebuffer memory regions are mapped with PBMT=NC to enable write-combining optimizations. The kernel executable, loaded at or above 0xffffffff80000000, and all HHDM and identity map memory regions are mapped with the default PBMT=PMA.

If the Svpbmt extension is not available, no PMAs can be overridden (effectively, everything is mapped with PBMT=PMA).

loongarch64

The kernel executable, loaded at or above 0xffffffff80000000, sees all of its segments mapped using the Coherent Cached (CC) memory access type (MAT).

All HHDM and identity map memory regions are mapped using the Coherent Cached (CC) MAT, except for the framebuffer regions, which are mapped in using the Weakly-ordered UnCached (WUC) MAT.

Machine state at entry

x86-64

rip will be the entry point as defined as part of the executable file format, unless the Entry Point feature is requested (see below), in which case, the value of rip is going to be taken from there.

At entry all segment registers are loaded as 64 bit code/data segments, limits and bases are ignored since this is 64-bit mode.

The GDT register is loaded to point to a GDT, in bootloader-reclaimable memory, with at least the following entries, starting at offset 0:

Null descriptor
16-bit code descriptor. Base = 0, limit = 0xffff. Readable.
16-bit data descriptor. Base = 0, limit = 0xffff. Writable.
32-bit code descriptor. Base = 0, limit = 0xffffffff. Readable.
32-bit data descriptor. Base = 0, limit = 0xffffffff. Writable.
64-bit code descriptor. Base and limit irrelevant. Readable.
64-bit data descriptor. Base and limit irrelevant. Writable.

The IDT is in an undefined state. Kernel must load its own.

IF flag, VM flag, and direction flag are cleared on entry. Other flags undefined.

PG is enabled (cr0), PE is enabled (cr0), PAE is enabled (cr4), WP is enabled (cr0), LME is enabled (EFER), NX is enabled (EFER) if available. If 5-level paging is requested and available, then 5-level paging is enabled (LA57 bit in cr4).

The A20 gate is opened.

Legacy PIC (if available) and IO APIC IRQs (only those with delivery mode fixed (0b000) or lowest priority (0b001)) are all masked.

If booted by EFI/UEFI, boot services are exited.

rsp is set to point to a stack, in bootloader-reclaimable memory, which is at least 64KiB (65536 bytes) in size, or the size specified in the Stack Size Request (see below). An invalid return address of 0 is pushed to the stack before jumping to the kernel.

All other general purpose registers are set to 0.

aarch64

PC will be the entry point as defined as part of the executable file format, unless the Entry Point feature is requested (see below), in which case, the value of PC is going to be taken from there.

The contents of the VBAR_EL1 register are undefined, and the kernel must load its own.

The MAIR_EL1 register contents are described above, in the caching section.

All interrupts are masked (PSTATE.{D, A, I, F} are set to 1).

The kernel is entered in little-endian AArch64 EL1t (EL1 with PSTATE.SP set to 0, PSTATE.E set to 0, and PSTATE.nRW set to 0).

Other fields of PSTATE are undefined.

At entry: the MMU (SCTLR_EL1.M) is enabled, the I-Cache and D-Cache (SCTLR_EL1.{I, C}) are enabled, data alignment checking (SCTLR_EL1.A) is disabled. SP alignment checking (SCTLR_EL1.{SA, SA0}) is enabled. Other fields of SCTLR_EL1 are reset to 0 or to their reserved value.

Higher ELs do not interfere with accesses to vector or floating point instructions or registers.

Higher ELs do not interfere with accesses to the generic timer and counter.

The used translation granule size for both TTBR0_EL1 and TTBR1_EL1 is 4KiB.

TCR_EL1.{T0SZ, T1SZ} are set to 16 under 4-level paging, or 12 under 5-level paging. Additionally, for 5-level paging, TCR_EL1.DS is set to 1.

TTBR1_EL1 points to the bootloader-provided higher half page tables. For base revision 0, TTBR0_EL1 points to the bootloader-provided identity mapping page tables, and is unspecified for all other base revisions and can thus be freely used by the kernel.

If booted by EFI/UEFI, boot services are exited.

SP is set to point to a stack, in bootloader-reclaimable memory, which is at least 64KiB (65536 bytes) in size, or the size specified in the Stack Size Request (see below).

All other general purpose registers (including X29 and X30) are set to 0. Vector registers are in an undefined state.

riscv64

At entry the machine is executing in Supervisor mode.

pc will be the entry point as defined as part of the executable file format, unless the Entry Point feature is requested (see below), in which case, the value of pc is going to be taken from there.

x1(ra) is set to 0, the kernel must not return from the entry point.

x2(sp) is set to point to a stack, in bootloader-reclaimable memory, which is at least 64KiB (65536 bytes) in size, or the size specified in the Stack Size Request (see below).

x3(gp) is set to 0, kernel must load its own global pointer if needed.

All other general purpose registers, with the exception of x5(t0), are set to 0.

If booted by EFI/UEFI, boot services are exited.

stvec is in an undefined state. sstatus.SIE and sie are set to 0.

sstatus.FS and sstatus.XS are both set to Off.

Paging is enabled with the paging mode specified by the Paging Mode feature (see below).

The (A)PLIC, if present, is in an undefined state.

loongarch64

At entry the machine is executing in PLV0.

$pc will be the entry point as defined as part of the executable file format, unless the Entry Point feature is requested (see below), in which case, the value of $pc is going to be taken from there.

$r1($ra) is set to 0, the kernel must not return from the entry point.

$r3($sp) is set to point to a stack, in bootloader-reclaimable memory, which is at least 64KiB (65536 bytes) in size, or the size specified in the Stack Size Request (see below).

All other general purpose registers, with the exception of $r12($t0), are set to 0.

If booted by EFI/UEFI, boot services are exited.

CSR.EENTRY, CSR.MERRENTRY and CSR.DWM0-3 are in an undefined state.

PG in CSR.CRMD is 1, DA is 0, IE is 0 and PLV is 0 but is otherwise unspecified.

CSR.TLBRENTRY is filled with a provided TLB refill handler.

Feature List

Request IDs are composed of 4 64-bit unsigned integers, but the first 2 are common to every request:

#define LIMINE_COMMON_MAGIC 0xc7b1dd30df4c8b88, 0x0a82e883a194f07b

Bootloader Info Feature

ID:

#define LIMINE_BOOTLOADER_INFO_REQUEST { LIMINE_COMMON_MAGIC, 0xf55038d8e2a1202f, 0x279426fcf5f59740 }

Request:

struct limine_bootloader_info_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_bootloader_info_response *response;
};

Response:

struct limine_bootloader_info_response {
    uint64_t revision;
    char *name;
    char *version;
};

name and version are 0-terminated ASCII strings containing the name and version of the loading bootloader.

Firmware Type Feature

ID:

#define LIMINE_FIRMWARE_TYPE_REQUEST { LIMINE_COMMON_MAGIC, 0x8c2f75d90bef28a8, 0x7045a4688eac00c3 }

Request:

struct limine_firmware_type_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_firmware_type_response *response;
};

Response:

struct limine_firmware_type_response {
    uint64_t revision;
    uint64_t firmware_type;
};

firmware_type is an enumeration that can have one of the following values:

#define LIMINE_FIRMWARE_TYPE_X86BIOS 0
#define LIMINE_FIRMWARE_TYPE_UEFI32 1
#define LIMINE_FIRMWARE_TYPE_UEFI64 2

Stack Size Feature

ID:

#define LIMINE_STACK_SIZE_REQUEST { LIMINE_COMMON_MAGIC, 0x224ef0460a8e8926, 0xe1cb0fc25f46ea3d }

Request:

struct limine_stack_size_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_stack_size_response *response;
    uint64_t stack_size;
};

stack_size - The requested stack size in bytes (also used for SMP processors).

Response:

struct limine_stack_size_response {
    uint64_t revision;
};

HHDM (Higher Half Direct Map) Feature

ID:

#define LIMINE_HHDM_REQUEST { LIMINE_COMMON_MAGIC, 0x48dcf1cb8ad2b852, 0x63984e959a98244b }

Request:

struct limine_hhdm_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_hhdm_response *response;
};

Response:

struct limine_hhdm_response {
    uint64_t revision;
    uint64_t offset;
};

offset - the virtual address offset of the beginning of the higher half direct map.

Framebuffer Feature

ID:

#define LIMINE_FRAMEBUFFER_REQUEST { LIMINE_COMMON_MAGIC, 0x9d5827dcd881dd75, 0xa3148604f6fab11b }

Request:

struct limine_framebuffer_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_framebuffer_response *response;
};

Response:

struct limine_framebuffer_response {
    uint64_t revision;
    uint64_t framebuffer_count;
    struct limine_framebuffer **framebuffers;
};

framebuffer_count - How many framebuffers are present.
framebuffers - Pointer to an array of framebuffer_count pointers to struct limine_framebuffer structures.

// Constants for `memory_model`
#define LIMINE_FRAMEBUFFER_RGB 1

struct limine_framebuffer {
    void *address;
    uint64_t width;
    uint64_t height;
    uint64_t pitch;
    uint16_t bpp; // Bits per pixel
    uint8_t memory_model;
    uint8_t red_mask_size;
    uint8_t red_mask_shift;
    uint8_t green_mask_size;
    uint8_t green_mask_shift;
    uint8_t blue_mask_size;
    uint8_t blue_mask_shift;
    uint8_t unused[7];
    uint64_t edid_size;
    void *edid;

    /* Response revision 1 */
    uint64_t mode_count;
    struct limine_video_mode **modes;
};

modes is an array of mode_count pointers to struct limine_video_mode describing the available video modes for the given framebuffer.

edid points to the screen's EDID blob, if available, else NULL.

struct limine_video_mode {
    uint64_t pitch;
    uint64_t width;
    uint64_t height;
    uint16_t bpp;
    uint8_t memory_model;
    uint8_t red_mask_size;
    uint8_t red_mask_shift;
    uint8_t green_mask_size;
    uint8_t green_mask_shift;
    uint8_t blue_mask_size;
    uint8_t blue_mask_shift;
};

Paging Mode Feature

The Paging Mode feature allows the kernel to control which paging mode is enabled before control is passed to it.

ID:

#define LIMINE_PAGING_MODE_REQUEST { LIMINE_COMMON_MAGIC, 0x95c1a0edab0944cb, 0xa4e5cb3842f7488a }

Request:

struct limine_paging_mode_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_paging_mode_response *response;
    uint64_t mode;
    /* Request revision 1 and above */
    uint64_t max_mode;
    uint64_t min_mode;
};

The mode, max_mode, and min_mode fields take architecture-specific values as described below.

mode is the preferred paging mode by the OS; the bootloader should always aim to pick this mode unless unavailable or overridden by the user in the bootloader's configuration file.

max_mode is the highest paging mode in numerical order that the OS supports. The bootloader will refuse to boot the OS if no paging modes of this type or lower (but equal or greater than min_mode) are available.

min_mode is the lowest paging mode in numerical order that the OS supports. The bootloader will refuse to boot the OS if no paging modes of this type or greater (but equal or lower than max_mode) are available.

If no Paging Mode Request is provided, the values of mode, max_mode, and min_mode that the bootloader assumes are LIMINE_PAGING_MODE_DEFAULT, LIMINE_PAGING_MODE_DEFAULT, and LIMINE_PAGING_MODE_MIN, respectively.

If request revision 0 is used, the values of max_mode and min_mode that the bootloader assumes are the value of mode and LIMINE_PAGING_MODE_MIN, respectively.

Response:

struct limine_paging_mode_response {
    uint64_t revision;
    uint64_t mode;
};

The response indicates which paging mode was actually enabled by the bootloader. Kernels must be prepared to handle cases where the provided paging mode is not supported.

x86-64

Values assignable to mode, max_mode, and min_mode:

#define LIMINE_PAGING_MODE_X86_64_4LVL 0
#define LIMINE_PAGING_MODE_X86_64_5LVL 1

#define LIMINE_PAGING_MODE_DEFAULT LIMINE_PAGING_MODE_X86_64_4LVL
#define LIMINE_PAGING_MODE_MIN LIMINE_PAGING_MODE_X86_64_4LVL

aarch64

Values assignable to mode, max_mode, and min_mode:

#define LIMINE_PAGING_MODE_AARCH64_4LVL 0
#define LIMINE_PAGING_MODE_AARCH64_5LVL 1

#define LIMINE_PAGING_MODE_DEFAULT LIMINE_PAGING_MODE_AARCH64_4LVL
#define LIMINE_PAGING_MODE_MIN LIMINE_PAGING_MODE_AARCH64_4LVL

riscv64

Values assignable to mode, max_mode, and min_mode:

#define LIMINE_PAGING_MODE_RISCV_SV39 0
#define LIMINE_PAGING_MODE_RISCV_SV48 1
#define LIMINE_PAGING_MODE_RISCV_SV57 2

#define LIMINE_PAGING_MODE_DEFAULT LIMINE_PAGING_MODE_RISCV_SV48
#define LIMINE_PAGING_MODE_MIN LIMINE_PAGING_MODE_RISCV_SV39

loongarch64

Values assignable to mode, max_mode, and min_mode:

#define LIMINE_PAGING_MODE_LOONGARCH64_4LVL 0

#define LIMINE_PAGING_MODE_DEFAULT LIMINE_PAGING_MODE_LOONGARCH64_4LVL
#define LIMINE_PAGING_MODE_MIN LIMINE_PAGING_MODE_LOONGARCH64_4LVL

SMP (multiprocessor) Feature

ID:

#define LIMINE_SMP_REQUEST { LIMINE_COMMON_MAGIC, 0x95a67b819a1b857e, 0xa0b61b723b6a73e0 }

Request:

struct limine_smp_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_smp_response *response;
    uint64_t flags;
};

flags - Bit 0: Enable X2APIC, if possible. (x86-64 only)

x86-64:

Response:

struct limine_smp_response {
    uint64_t revision;
    uint32_t flags;
    uint32_t bsp_lapic_id;
    uint64_t cpu_count;
    struct limine_smp_info **cpus;
};

flags - Bit 0: X2APIC has been enabled.
bsp_lapic_id - The Local APIC ID of the bootstrap processor.
cpu_count - How many CPUs are present. It includes the bootstrap processor.
cpus - Pointer to an array of cpu_count pointers to struct limine_smp_info structures.

Note: The presence of this request will prompt the bootloader to bootstrap the secondary processors. This will not be done if this request is not present.

Note: The MTRRs of APs will be synchronised by the bootloader to match the BSP, as Intel SDM requires (Vol. 3A, 12.11.5).

struct limine_smp_info;

typedef void (*limine_goto_address)(struct limine_smp_info *);

struct limine_smp_info {
    uint32_t processor_id;
    uint32_t lapic_id;
    uint64_t reserved;
    limine_goto_address goto_address;
    uint64_t extra_argument;
};

processor_id - ACPI Processor UID as specified by the MADT
lapic_id - Local APIC ID of the processor as specified by the MADT
goto_address - An atomic write to this field causes the parked CPU to jump to the written address, on a 64KiB (or Stack Size Request size) stack. A pointer to the struct limine_smp_info structure of the CPU is passed in RDI. Other than that, the CPU state will be the same as described for the bootstrap processor. This field is unused for the structure describing the bootstrap processor. For all CPUs, this field is guaranteed to be NULL when control is first passed to the bootstrap processor.
extra_argument - A free for use field.

aarch64:

Response:

struct limine_smp_response {
    uint64_t revision;
    uint64_t flags;
    uint64_t bsp_mpidr;
    uint64_t cpu_count;
    struct limine_smp_info **cpus;
};

flags - Always zero
bsp_mpidr - MPIDR of the bootstrap processor (as read from MPIDR_EL1, with Res1 masked off).
cpu_count - How many CPUs are present. It includes the bootstrap processor.
cpus - Pointer to an array of cpu_count pointers to struct limine_smp_info structures.

Note: The presence of this request will prompt the bootloader to bootstrap the secondary processors. This will not be done if this request is not present.

struct limine_smp_info;

typedef void (*limine_goto_address)(struct limine_smp_info *);

struct limine_smp_info {
    uint32_t processor_id;
    uint32_t reserved1;
    uint64_t mpidr;
    uint64_t reserved;
    limine_goto_address goto_address;
    uint64_t extra_argument;
};

processor_id - ACPI Processor UID as specified by the MADT (always 0 on non-ACPI systems)
mpidr - MPIDR of the processor as specified by the MADT or device tree
goto_address - An atomic write to this field causes the parked CPU to jump to the written address, on a 64KiB (or Stack Size Request size) stack. A pointer to the struct limine_smp_info structure of the CPU is passed in X0. Other than that, the CPU state will be the same as described for the bootstrap processor. This field is unused for the structure describing the bootstrap processor.
extra_argument - A free for use field.

riscv64

Response:

struct limine_smp_response {
    uint64_t revision;
    uint64_t flags;
    uint64_t bsp_hartid;
    uint64_t cpu_count;
    struct limine_smp_info **cpus;
};

flags - Always zero
bsp_hartid - Hart ID of the bootstrap processor as reported by the UEFI RISC-V Boot Protocol or the SBI.
cpu_count - How many CPUs are present. It includes the bootstrap processor.
cpus - Pointer to an array of cpu_count pointers to struct limine_smp_info structures.

Note: The presence of this request will prompt the bootloader to bootstrap the secondary processors. This will not be done if this request is not present.

struct limine_smp_info;

typedef void (*limine_goto_address)(struct limine_smp_info *);

struct limine_smp_info {
    uint64_t processor_id;
    uint64_t hartid;
    uint64_t reserved;
    limine_goto_address goto_address;
    uint64_t extra_argument;
};

processor_id - ACPI Processor UID as specified by the MADT (always 0 on non-ACPI systems).
hartid - Hart ID of the processor as specified by the MADT or Device Tree.
goto_address - An atomic write to this field causes the parked CPU to jump to the written address, on a 64KiB (or Stack Size Request size) stack. A pointer to the struct limine_smp_info structure of the CPU is passed in x10(a0). Other than that, the CPU state will be the same as described for the bootstrap processor. This field is unused for the structure describing the bootstrap processor.
extra_argument - A free for use field.

Memory Map Feature

ID:

#define LIMINE_MEMMAP_REQUEST { LIMINE_COMMON_MAGIC, 0x67cf3d9d378a806f, 0xe304acdfc50c3c62 }

Request:

struct limine_memmap_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_memmap_response *response;
};

Response:

struct limine_memmap_response {
    uint64_t revision;
    uint64_t entry_count;
    struct limine_memmap_entry **entries;
};

entry_count - How many memory map entries are present.
entries - Pointer to an array of entry_count pointers to struct limine_memmap_entry structures.

// Constants for `type`
#define LIMINE_MEMMAP_USABLE                 0
#define LIMINE_MEMMAP_RESERVED               1
#define LIMINE_MEMMAP_ACPI_RECLAIMABLE       2
#define LIMINE_MEMMAP_ACPI_NVS               3
#define LIMINE_MEMMAP_BAD_MEMORY             4
#define LIMINE_MEMMAP_BOOTLOADER_RECLAIMABLE 5
#define LIMINE_MEMMAP_KERNEL_AND_MODULES     6
#define LIMINE_MEMMAP_FRAMEBUFFER            7

struct limine_memmap_entry {
    uint64_t base;
    uint64_t length;
    uint64_t type;
};

Note: All these memory entry types, besides usable and bootloader reclaimable, are meant to have an illustrative purpose only, and are not authoritative sources to be used as a means to find the addresses of kernel, modules, framebuffer, ACPI, or otherwise. Use the specific Limine features to do that, if available, or other discovery means.

Note: For base revisions <= 2, memory between 0 and 0x1000 is never marked as usable memory. The kernel and modules loaded are not marked as usable memory. They are marked as Kernel/Modules. The entries are guaranteed to be sorted by base address, lowest to highest. Usable and bootloader reclaimable entries are guaranteed to be 4096 byte aligned for both base and length. Usable and bootloader reclaimable entries are guaranteed not to overlap with any other entry. To the contrary, all non-usable entries (including kernel/modules) are not guaranteed any alignment, nor is it guaranteed that they do not overlap other entries.

Entry Point Feature

ID:

#define LIMINE_ENTRY_POINT_REQUEST { LIMINE_COMMON_MAGIC, 0x13d86c035a1cd3e1, 0x2b0caa89d8f3026a }

Request:

typedef void (*limine_entry_point)(void);

struct limine_entry_point_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_entry_point_response *response;
    limine_entry_point entry;
};

entry - The requested entry point.

Response:

struct limine_entry_point_response {
    uint64_t revision;
};

Kernel File Feature

ID:

#define LIMINE_KERNEL_FILE_REQUEST { LIMINE_COMMON_MAGIC, 0xad97e90e83f1ed67, 0x31eb5d1c5ff23b69 }

Request:

struct limine_kernel_file_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_kernel_file_response *response;
};

Response:

struct limine_kernel_file_response {
    uint64_t revision;
    struct limine_file *kernel_file;
};

kernel_file - Pointer to the struct limine_file structure (see below) for the kernel file.

Module Feature

ID:

#define LIMINE_MODULE_REQUEST { LIMINE_COMMON_MAGIC, 0x3e7e279702be32af, 0xca1c4f3bd1280cee }

Request:

#define LIMINE_INTERNAL_MODULE_REQUIRED (1 << 0)
#define LIMINE_INTERNAL_MODULE_COMPRESSED (1 << 1)

struct limine_internal_module {
    const char *path;
    const char *cmdline;
    uint64_t flags;
};

struct limine_module_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_module_response *response;

    /* Request revision 1 */
    uint64_t internal_module_count;
    struct limine_internal_module **internal_modules;
};

internal_module_count - How many internal modules are passed by the kernel.
internal_modules - Pointer to an array of internal_module_count pointers to struct limine_internal_module structures.

Note: Internal modules are honoured if the module response has revision >= 1.

As part of struct limine_internal_module:

path - Path to the module to load. This path is relative to the location of the kernel.
cmdline - Command line for the given module.
flags - Flags changing module loading behaviour:
- LIMINE_INTERNAL_MODULE_REQUIRED: Fail if the requested module is not found.
- LIMINE_INTERNAL_MODULE_COMPRESSED: Deprecated. Bootloader may not support it and panic instead (from Limine 8.x onwards). Alternatively: the module is GZ-compressed and should be decompressed by the bootloader. This is honoured if the response is revision 2 or greater.

Internal Limine modules are guaranteed to be loaded before user-specified (configuration) modules, and thus they are guaranteed to appear before user-specified modules in the modules array in the response.

Response:

struct limine_module_response {
    uint64_t revision;
    uint64_t module_count;
    struct limine_file **modules;
};

module_count - How many modules are present.
modules - Pointer to an array of module_count pointers to struct limine_file structures (see below).

File Structure

struct limine_uuid {
    uint32_t a;
    uint16_t b;
    uint16_t c;
    uint8_t d[8];
};

#define LIMINE_MEDIA_TYPE_GENERIC 0
#define LIMINE_MEDIA_TYPE_OPTICAL 1
#define LIMINE_MEDIA_TYPE_TFTP 2

struct limine_file {
    uint64_t revision;
    void *address;
    uint64_t size;
    char *path;
    char *cmdline;
    uint32_t media_type;
    uint32_t unused;
    uint32_t tftp_ip;
    uint32_t tftp_port;
    uint32_t partition_index;
    uint32_t mbr_disk_id;
    struct limine_uuid gpt_disk_uuid;
    struct limine_uuid gpt_part_uuid;
    struct limine_uuid part_uuid;
};

revision - Revision of the struct limine_file structure.
address - The address of the file. This is always at least 4KiB aligned.
size - The size of the file.
path - The path of the file within the volume, with a leading slash.
cmdline - A command line associated with the file.
media_type - Type of media file resides on.
tftp_ip - If non-0, this is the IP of the TFTP server the file was loaded from.
tftp_port - Likewise, but port.
partition_index - 1-based partition index of the volume from which the file was loaded. If 0, it means invalid or unpartitioned.
mbr_disk_id - If non-0, this is the ID of the disk the file was loaded from as reported in its MBR.
gpt_disk_uuid - If non-0, this is the UUID of the disk the file was loaded from as reported in its GPT.
gpt_part_uuid - If non-0, this is the UUID of the partition the file was loaded from as reported in the GPT.
part_uuid - If non-0, this is the UUID of the filesystem of the partition the file was loaded from.

RSDP Feature

ID:

#define LIMINE_RSDP_REQUEST { LIMINE_COMMON_MAGIC, 0xc5e77b6b397e7b43, 0x27637845accdcf3c }

Request:

struct limine_rsdp_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_rsdp_response *response;
};

Response:

struct limine_rsdp_response {
    uint64_t revision;
    void *address;
};

address - Address of the RSDP table. Physical for base revision >= 3.

SMBIOS Feature

ID:

#define LIMINE_SMBIOS_REQUEST { LIMINE_COMMON_MAGIC, 0x9e9046f11e095391, 0xaa4a520fefbde5ee }

Request:

struct limine_smbios_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_smbios_response *response;
};

Response:

struct limine_smbios_response {
    uint64_t revision;
    void *entry_32;
    void *entry_64;
};

entry_32 - Address of the 32-bit SMBIOS entry point. NULL if not present. Physical for base revision >= 3.
entry_64 - Address of the 64-bit SMBIOS entry point. NULL if not present. Physical for base revision >= 3.

EFI System Table Feature

ID:

#define LIMINE_EFI_SYSTEM_TABLE_REQUEST { LIMINE_COMMON_MAGIC, 0x5ceba5163eaaf6d6, 0x0a6981610cf65fcc }

Request:

struct limine_efi_system_table_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_efi_system_table_response *response;
};

Response:

struct limine_efi_system_table_response {
    uint64_t revision;
    void *address;
};

address - Address of EFI system table. Physical for base revision >= 3.

EFI Memory Map Feature

ID:

#define LIMINE_EFI_MEMMAP_REQUEST { LIMINE_COMMON_MAGIC, 0x7df62a431d6872d5, 0xa4fcdfb3e57306c8 }

Request:

struct limine_efi_memmap_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_efi_memmap_response *response;
};

Response:

struct limine_efi_memmap_response {
    uint64_t revision;
    void *memmap;
    uint64_t memmap_size;
    uint64_t desc_size;
    uint64_t desc_version;
};

memmap - Address (HHDM for base revision <= 2, else physical) of the EFI memory map.
memmap_size - Size in bytes of the EFI memory map.
desc_size - EFI memory map descriptor size in bytes.
desc_version - Version of EFI memory map descriptors.

Note: This feature provides data suitable for use with RT->SetVirtualAddressMap(), provided HHDM offset is subtracted from memmap.

Boot Time Feature

ID:

#define LIMINE_BOOT_TIME_REQUEST { LIMINE_COMMON_MAGIC, 0x502746e184c088aa, 0xfbc5ec83e6327893 }

Request:

struct limine_boot_time_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_boot_time_response *response;
};

Response:

struct limine_boot_time_response {
    uint64_t revision;
    int64_t boot_time;
};

boot_time - The UNIX time on boot, in seconds, taken from the system RTC.

Kernel Address Feature

ID:

#define LIMINE_KERNEL_ADDRESS_REQUEST { LIMINE_COMMON_MAGIC, 0x71ba76863cc55f63, 0xb2644a48c516a487 }

Request:

struct limine_kernel_address_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_kernel_address_response *response;
};

Response:

struct limine_kernel_address_response {
    uint64_t revision;
    uint64_t physical_base;
    uint64_t virtual_base;
};

physical_base - The physical base address of the kernel.
virtual_base - The virtual base address of the kernel.

Device Tree Blob Feature

ID:

#define LIMINE_DTB_REQUEST { LIMINE_COMMON_MAGIC, 0xb40ddb48fb54bac7, 0x545081493f81ffb7 }

Request:

struct limine_dtb_request {
    uint64_t id[4];
    uint64_t revision;
    struct limine_dtb_response *response;
};

Response:

struct limine_dtb_response {
    uint64_t revision;
    void *dtb_ptr;
};

dtb_ptr - Virtual pointer to the device tree blob.

Note: If the DTB cannot be found, the response will not be generated.

Note: Information contained in the /chosen node may not reflect the information given by bootloader tags, and as such the /chosen node properties should be ignored.

Note: If the DTB contained memory@... nodes, they will get removed. Kernels may not rely on these nodes and should use the Memory Map feature instead.

Files

PROTOCOL.md

Latest commit

History

PROTOCOL.md

File metadata and controls

The Limine Boot Protocol

General Notes

Base protocol revisions

Features

Requests Delimiters

Limine Requests Section

Entry memory layout

Caching

x86-64

aarch64

riscv64

loongarch64

Machine state at entry

x86-64

aarch64

riscv64

loongarch64

Feature List

Bootloader Info Feature

Firmware Type Feature

Stack Size Feature

HHDM (Higher Half Direct Map) Feature

Framebuffer Feature

Paging Mode Feature

x86-64

aarch64

riscv64

loongarch64

SMP (multiprocessor) Feature

x86-64:

aarch64:

riscv64

Memory Map Feature

Entry Point Feature

Kernel File Feature

Module Feature

File Structure

RSDP Feature

SMBIOS Feature

EFI System Table Feature

EFI Memory Map Feature

Boot Time Feature

Kernel Address Feature

Device Tree Blob Feature