rust-raspberrypi-OS-tutorials/11_virtual_memory
2019-11-13 08:40:23 +01:00
..
.vscode Add code for tutorial 11 2019-11-05 21:23:36 +01:00
src Housekeeping 2019-11-07 22:40:08 +01:00
Cargo.lock New binaries: rustc 2019-11-04 2019-11-10 23:57:28 +01:00
Cargo.toml Add code for tutorial 11 2019-11-05 21:23:36 +01:00
kernel Makefile changes 2019-11-13 08:40:23 +01:00
kernel8.img Housekeeping 2019-11-07 22:40:08 +01:00
Makefile Makefile changes 2019-11-13 08:40:23 +01:00
README.md Util script cleanup; Enable pre-commit hook 2019-11-10 23:00:53 +01:00

Tutorial 11 - Virtual Memory

tl;dr

The MMU is turned on; A simple scheme is used: static 64 KiB page tables; For educational purposes, we write to a remapped UART.

Introduction

Virtual memory is an immensely complex, but important and powerful topic. In this tutorial, we start slow and easy by switching on the MMU, using static page tables and mapping everything at once.

MMU and paging theory

At this point, we will not re-invent the wheel and go into detailed descriptions of how paging in modern application-grade processors works. The internet is full of great resources regarding this topic, and we encourage you to read some of it to get a high-level understanding of the topic.

To follow the rest of this AArch64 specific tutorial, I strongly recommend that you stop right here and first read Chapter 12 of the ARM Cortex-A Series Programmer's Guide for ARMv8-A before you continue. This will set you up with all the AArch64-specific knowledge needed to follow along.

Back from reading Chapter 12 already? Good job 👍!

Approach

  • Everything is mapped using a 64 KiB granule.
  • Attributes of different regions (e.g. R/W, no-execute, cached/uncached, etc...) are set in a high-level data structure in BSP code.
  • Arch code picks up this high-level description and maps it using its specific MMU HW.

BSP: bsp/rpi/virt_mem_layout.rs

This file is used to describe our kernel's memory layout in a high-level abstraction using our own descriptor format. We can define ranges of arbitrary length and set respective attributes, for example if the bits and bytes in this range should be executable or not.

The descriptors we use here are agnostic of the hardware MMU's actual descriptors, and we are also agnostic of the paging granule the MMU will use. Having this distinction is less of a technical need and more a convenience feature for us in order to easily describe the kernel's memory layout, and hopefully it makes the whole concept a bit more graspable for the reader.

The file contains an instance of memory::KernelVirtualLayout,which stores these descriptors. The policy is to only store regions that are not ordinary, normal chacheable DRAM. However, nothing prevents you from defining those too if you wish to. Here is an example for the device MMIO region:

// Device MMIO.
RangeDescriptor {
    name: "Device MMIO",
    virtual_range: || {
        RangeInclusive::new(memory_map::mmio::BASE, memory_map::mmio::END_INCLUSIVE)
    },
    translation: Translation::Identity,
    attribute_fields: AttributeFields {
        mem_attributes: MemAttributes::Device,
        acc_perms: AccessPermissions::ReadWrite,
        execute_never: true,
    },
},

KernelVirtualLayout also provides the following method:

pub fn get_virt_addr_properties(
        &self,
        virt_addr: usize,
    ) -> Result<(usize, AttributeFields), &'static str>

It will be used by the arch's MMU code to request attributes for a virtual address and the translation of the address. The function scans for a descriptor that contains the queried address, and returns the respective findings for the first entry that is a hit. If no entry is found, it returns default attributes for normal chacheable DRAM and the input address, hence telling the MMU code that the requested address should be identity mapped.

Due to this default return, it is technicall not needed to define normal cacheable DRAM regions.

Arch: arch/aarch64/mmu.rs

This file contains the AArch64 specific code. It is a driver, if you like, and the paging granule is hardcoded here (64 KiB page descriptors).

The actual page tables are stored in a global instance of the PageTables struct:

// Two newtypes for added type safety, so that you cannot accidentally place a TableDescriptor into
// a PageDescriptor slot in `struct PageTables`, and vice versa.
#[derive(Copy, Clone)]
#[repr(transparent)]
struct RawTableDescriptor(u64);

#[derive(Copy, Clone)]
#[repr(transparent)]
struct RawPageDescriptor(u64);

/// Big monolithic struct for storing the page tables. Individual levels must be 64 KiB aligned,
/// hence the "reverse" order of appearance.
#[repr(C)]
#[repr(align(65536))]
struct PageTables<const N: usize> {
    // Page descriptors, covering 64 KiB windows per entry.
    lvl3: [[RawPageDescriptor; 8192]; N],
    // Table descriptors, covering 512 MiB windows.
    lvl2: [RawTableDescriptor; N],
}

/// Usually evaluates to 1 GiB for RPi3 and 4 GiB for RPi 4.
const ENTRIES_512_MIB: usize = bsp::addr_space_size() >> FIVETWELVE_MIB_SHIFT;

/// The page tables.
///
/// Supposed to land in `.bss`. Therefore, ensure that they boil down to all "0" entries.
static mut TABLES: PageTables<{ ENTRIES_512_MIB }> = PageTables {
    lvl3: [[RawPageDescriptor(0); 8192]; ENTRIES_512_MIB],
    lvl2: [RawTableDescriptor(0); ENTRIES_512_MIB],
};

They are populated using get_virt_addr_properties() and a bunch of utility functions that convert our own descriptors to the actual 64 bit integer entries needed by the MMU hardware for the page table arrays.

Each page table has an entry (AttrIndex) that indexes into the MAIR_EL1 register, which holds information about the cacheability of the respective page. We currently define normal cacheable memory and device memory (which is not cached).

/// Setup function for the MAIR_EL1 register.
fn set_up_mair() {
    // Define the memory types being mapped.
    MAIR_EL1.write(
        // Attribute 1 - Cacheable normal DRAM
        MAIR_EL1::Attr1_HIGH::Memory_OuterWriteBack_NonTransient_ReadAlloc_WriteAlloc
            + MAIR_EL1::Attr1_LOW_MEMORY::InnerWriteBack_NonTransient_ReadAlloc_WriteAlloc

            // Attribute 0 - Device
            + MAIR_EL1::Attr0_HIGH::Device
            + MAIR_EL1::Attr0_LOW_DEVICE::Device_nGnRE,
    );
}

Afterwards, the Translation Table Base Register 0 - EL1 is set up with the base address of the lvl2 and the Translation Control Register - EL1 is configured.

Finally, the MMU is turned on through the System Control Register - EL1. The last step also enables caching for data and instructions.

link.ld

We need to align the ro section to 64 KiB so that it doesn't overlap with the next section that needs read/write attributes. This blows up the binary in size, but is a small price to pay considering that it reduces the amount of static paging entries significantly, when compared to the classical 4 KiB granule.

Address translation examples

For educational purposes, a layout is defined which allows to access the UART via two different virtual addresses:

  • Since we identity map the whole Device MMIO region, it is accessible by asserting its physical base address (0x3F20_1000 or 0xFA20_1000 depending on which RPi you use) after the MMU is turned on.
  • Additionally, it is also mapped into the last 64 KiB entry of the lvl3 table, making it accessible through base address 0x1FFF_1000.

The following block diagram visualizes the underlying translation for the second mapping.

Address translation using a 64 KiB page descriptor

4 KiB translation block diagram

Zero-cost abstraction

The MMU init code is again a good example to see the great potential of Rust's zero-cost abstractions[1][2] for embedded programming.

Take this piece of code for setting up the MAIR_EL1 register using the cortex-a crate:

/// Setup function for the MAIR_EL1 register.
fn set_up_mair() {
    // Define the memory types being mapped.
    MAIR_EL1.write(
        // Attribute 1 - Cacheable normal DRAM
        MAIR_EL1::Attr1_HIGH::Memory_OuterWriteBack_NonTransient_ReadAlloc_WriteAlloc
            + MAIR_EL1::Attr1_LOW_MEMORY::InnerWriteBack_NonTransient_ReadAlloc_WriteAlloc

            // Attribute 0 - Device
            + MAIR_EL1::Attr0_HIGH::Device
            + MAIR_EL1::Attr0_LOW_DEVICE::Device_nGnRE,
    );
}

This piece of code is super expressive, and it makes use of traits, different types and constants to provide type-safe register manipulation.

In the end, this code sets the first four bytes of the register to certain values according to the data sheet. Looking at the generated code, we can see that despite all the type-safety and abstractions, it boils down to two assembly instructions:

00000000000818d8 kernel::arch::aarch64::mmu::init::h97006e19222f36e2:
   ...
   8191c: 88 e0 9f 52                   mov     w8, #0xff04
   ...
   81924: 08 a2 18 d5                   msr     MAIR_EL1, x8

Test it

make chainbot
[...]
### Listening on /dev/ttyUSB0
 __  __ _      _ _                 _
|  \/  (_)_ _ (_) |   ___  __ _ __| |
| |\/| | | ' \| | |__/ _ \/ _` / _` |
|_|  |_|_|_||_|_|____\___/\__,_\__,_|

           Raspberry Pi 3

[ML] Requesting binary
### sending kernel kernel8.img [65560 byte]
### finished sending
[ML] Loaded! Executing the payload now

[    5.791515] Booting on: Raspberry Pi 3
[    5.793767] MMU online. Special regions:
[    5.797674]       0x00080000 - 0x0008FFFF |  64 KiB | C   RO PX  | Kernel code and RO data
[    5.805922]       0x1FFF0000 - 0x1FFFFFFF |  64 KiB | Dev RW PXN | Remapped Device MMIO
[    5.813910]       0x3F000000 - 0x3FFFFFFF |  16 MiB | Dev RW PXN | Device MMIO
[    5.821117] Current privilege level: EL1
[    5.825024] Exception handling state:
[    5.828670]       Debug:  Masked
[    5.831883]       SError: Masked
[    5.835095]       IRQ:    Masked
[    5.838308]       FIQ:    Masked
[    5.841520] Architectural timer resolution: 52 ns
[    5.846209] Drivers loaded:
[    5.848987]       1. GPIO
[    5.851592]       2. PL011Uart
[    5.854630] Timer test, spinning for 1 second
[     !!!    ] Writing through the remapped UART at 0x1FFF_1000
[    6.863133] Echoing input now

Diff to previous


diff -uNr 10_privilege_level/src/arch/aarch64/mmu.rs 11_virtual_memory/src/arch/aarch64/mmu.rs
--- 10_privilege_level/src/arch/aarch64/mmu.rs
+++ 11_virtual_memory/src/arch/aarch64/mmu.rs
@@ -0,0 +1,316 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright (c) 2018-2019 Andre Richter <andre.o.richter@gmail.com>
+
+//! Memory Management Unit.
+//!
+//! Static page tables, compiled on boot; Everything 64 KiB granule.
+
+use crate::{
+    bsp, interface,
+    memory::{AccessPermissions, AttributeFields, MemAttributes},
+};
+use core::convert;
+use cortex_a::{barrier, regs::*};
+use register::register_bitfields;
+
+// A table descriptor, as per AArch64 Reference Manual Figure D4-15.
+register_bitfields! {u64,
+    STAGE1_TABLE_DESCRIPTOR [
+        /// Physical address of the next page table.
+        NEXT_LEVEL_TABLE_ADDR_64KiB OFFSET(16) NUMBITS(32) [], // [47:16]
+
+        TYPE  OFFSET(1) NUMBITS(1) [
+            Block = 0,
+            Table = 1
+        ],
+
+        VALID OFFSET(0) NUMBITS(1) [
+            False = 0,
+            True = 1
+        ]
+    ]
+}
+
+// A level 3 page descriptor, as per AArch64 Reference Manual Figure D4-17.
+register_bitfields! {u64,
+    STAGE1_PAGE_DESCRIPTOR [
+        /// Privileged execute-never
+        PXN      OFFSET(53) NUMBITS(1) [
+            False = 0,
+            True = 1
+        ],
+
+        /// Physical address of the next page table (lvl2) or the page descriptor (lvl3).
+        OUTPUT_ADDR_64KiB OFFSET(16) NUMBITS(32) [], // [47:16]
+
+        /// Access flag
+        AF       OFFSET(10) NUMBITS(1) [
+            False = 0,
+            True = 1
+        ],
+
+        /// Shareability field
+        SH       OFFSET(8) NUMBITS(2) [
+            OuterShareable = 0b10,
+            InnerShareable = 0b11
+        ],
+
+        /// Access Permissions
+        AP       OFFSET(6) NUMBITS(2) [
+            RW_EL1 = 0b00,
+            RW_EL1_EL0 = 0b01,
+            RO_EL1 = 0b10,
+            RO_EL1_EL0 = 0b11
+        ],
+
+        /// Memory attributes index into the MAIR_EL1 register
+        AttrIndx OFFSET(2) NUMBITS(3) [],
+
+        TYPE     OFFSET(1) NUMBITS(1) [
+            Block = 0,
+            Table = 1
+        ],
+
+        VALID    OFFSET(0) NUMBITS(1) [
+            False = 0,
+            True = 1
+        ]
+    ]
+}
+
+// Two newtypes for added type safety, so that you cannot accidentally place a TableDescriptor into
+// a PageDescriptor slot in `struct PageTables`, and vice versa.
+#[derive(Copy, Clone)]
+#[repr(transparent)]
+struct RawTableDescriptor(u64);
+
+#[derive(Copy, Clone)]
+#[repr(transparent)]
+struct RawPageDescriptor(u64);
+
+const SIXTYFOUR_KIB_SHIFT: usize = 16; //  log2(64 * 1024)
+const FIVETWELVE_MIB_SHIFT: usize = 29; // log2(512 * 1024 * 1024)
+
+/// Big monolithic struct for storing the page tables. Individual levels must be 64 KiB aligned,
+/// hence the "reverse" order of appearance.
+#[repr(C)]
+#[repr(align(65536))]
+struct PageTables<const N: usize> {
+    // Page descriptors, covering 64 KiB windows per entry.
+    lvl3: [[RawPageDescriptor; 8192]; N],
+    // Table descriptors, covering 512 MiB windows.
+    lvl2: [RawTableDescriptor; N],
+}
+
+/// Usually evaluates to 1 GiB for RPi3 and 4 GiB for RPi 4.
+const ENTRIES_512_MIB: usize = bsp::addr_space_size() >> FIVETWELVE_MIB_SHIFT;
+
+/// The page tables.
+///
+/// Supposed to land in `.bss`. Therefore, ensure that they boil down to all "0" entries.
+static mut TABLES: PageTables<{ ENTRIES_512_MIB }> = PageTables {
+    lvl3: [[RawPageDescriptor(0); 8192]; ENTRIES_512_MIB],
+    lvl2: [RawTableDescriptor(0); ENTRIES_512_MIB],
+};
+
+trait BaseAddr {
+    fn base_addr_u64(&self) -> u64;
+    fn base_addr_usize(&self) -> usize;
+}
+
+impl<T, const N: usize> BaseAddr for [T; N] {
+    fn base_addr_u64(&self) -> u64 {
+        self as *const T as u64
+    }
+
+    fn base_addr_usize(&self) -> usize {
+        self as *const T as usize
+    }
+}
+
+/// A descriptor pointing to the next page table.
+struct TableDescriptor(register::FieldValue<u64, STAGE1_TABLE_DESCRIPTOR::Register>);
+
+impl TableDescriptor {
+    fn new(next_lvl_table_addr: usize) -> TableDescriptor {
+        let shifted = next_lvl_table_addr >> SIXTYFOUR_KIB_SHIFT;
+
+        TableDescriptor(
+            STAGE1_TABLE_DESCRIPTOR::VALID::True
+                + STAGE1_TABLE_DESCRIPTOR::TYPE::Table
+                + STAGE1_TABLE_DESCRIPTOR::NEXT_LEVEL_TABLE_ADDR_64KiB.val(shifted as u64),
+        )
+    }
+}
+
+impl convert::From<TableDescriptor> for RawTableDescriptor {
+    fn from(desc: TableDescriptor) -> Self {
+        RawTableDescriptor(desc.0.value)
+    }
+}
+
+/// Convert the kernel's generic memory range attributes to HW-specific attributes of the MMU.
+impl convert::From<AttributeFields>
+    for register::FieldValue<u64, STAGE1_PAGE_DESCRIPTOR::Register>
+{
+    fn from(attribute_fields: AttributeFields) -> Self {
+        // Memory attributes
+        let mut desc = match attribute_fields.mem_attributes {
+            MemAttributes::CacheableDRAM => {
+                STAGE1_PAGE_DESCRIPTOR::SH::InnerShareable
+                    + STAGE1_PAGE_DESCRIPTOR::AttrIndx.val(mair::NORMAL)
+            }
+            MemAttributes::Device => {
+                STAGE1_PAGE_DESCRIPTOR::SH::OuterShareable
+                    + STAGE1_PAGE_DESCRIPTOR::AttrIndx.val(mair::DEVICE)
+            }
+        };
+
+        // Access Permissions
+        desc += match attribute_fields.acc_perms {
+            AccessPermissions::ReadOnly => STAGE1_PAGE_DESCRIPTOR::AP::RO_EL1,
+            AccessPermissions::ReadWrite => STAGE1_PAGE_DESCRIPTOR::AP::RW_EL1,
+        };
+
+        // Execute Never
+        desc += if attribute_fields.execute_never {
+            STAGE1_PAGE_DESCRIPTOR::PXN::True
+        } else {
+            STAGE1_PAGE_DESCRIPTOR::PXN::False
+        };
+
+        desc
+    }
+}
+
+/// A page descriptor with 64 KiB aperture.
+///
+/// The output points to physical memory.
+struct PageDescriptor(register::FieldValue<u64, STAGE1_PAGE_DESCRIPTOR::Register>);
+
+impl PageDescriptor {
+    fn new(output_addr: usize, attribute_fields: AttributeFields) -> PageDescriptor {
+        let shifted = output_addr >> SIXTYFOUR_KIB_SHIFT;
+
+        PageDescriptor(
+            STAGE1_PAGE_DESCRIPTOR::VALID::True
+                + STAGE1_PAGE_DESCRIPTOR::AF::True
+                + attribute_fields.into()
+                + STAGE1_PAGE_DESCRIPTOR::TYPE::Table
+                + STAGE1_PAGE_DESCRIPTOR::OUTPUT_ADDR_64KiB.val(shifted as u64),
+        )
+    }
+}
+
+impl convert::From<PageDescriptor> for RawPageDescriptor {
+    fn from(desc: PageDescriptor) -> Self {
+        RawPageDescriptor(desc.0.value)
+    }
+}
+
+/// Constants for indexing the MAIR_EL1.
+#[allow(dead_code)]
+mod mair {
+    pub const DEVICE: u64 = 0;
+    pub const NORMAL: u64 = 1;
+}
+
+/// Setup function for the MAIR_EL1 register.
+fn set_up_mair() {
+    // Define the memory types being mapped.
+    MAIR_EL1.write(
+        // Attribute 1 - Cacheable normal DRAM
+        MAIR_EL1::Attr1_HIGH::Memory_OuterWriteBack_NonTransient_ReadAlloc_WriteAlloc
+            + MAIR_EL1::Attr1_LOW_MEMORY::InnerWriteBack_NonTransient_ReadAlloc_WriteAlloc
+
+            // Attribute 0 - Device
+            + MAIR_EL1::Attr0_HIGH::Device
+            + MAIR_EL1::Attr0_LOW_DEVICE::Device_nGnRE,
+    );
+}
+
+/// Iterates over all static page table entries and fills them at once.
+///
+/// # Safety
+///
+/// - Modifies a `static mut`. Ensure it only happens from here.
+unsafe fn populate_pt_entries() -> Result<(), &'static str> {
+    for (l2_nr, l2_entry) in TABLES.lvl2.iter_mut().enumerate() {
+        *l2_entry = TableDescriptor::new(TABLES.lvl3[l2_nr].base_addr_usize()).into();
+
+        for (l3_nr, l3_entry) in TABLES.lvl3[l2_nr].iter_mut().enumerate() {
+            let virt_addr = (l2_nr << FIVETWELVE_MIB_SHIFT) + (l3_nr << SIXTYFOUR_KIB_SHIFT);
+
+            let (output_addr, attribute_fields) =
+                bsp::virt_mem_layout().get_virt_addr_properties(virt_addr)?;
+
+            *l3_entry = PageDescriptor::new(output_addr, attribute_fields).into();
+        }
+    }
+
+    Ok(())
+}
+
+/// Configure various settings of stage 1 of the EL1 translation regime.
+fn configure_translation_control() {
+    let ips = ID_AA64MMFR0_EL1.read(ID_AA64MMFR0_EL1::PARange);
+    TCR_EL1.write(
+        TCR_EL1::TBI0::Ignored
+            + TCR_EL1::IPS.val(ips)
+            + TCR_EL1::TG0::KiB_64
+            + TCR_EL1::SH0::Inner
+            + TCR_EL1::ORGN0::WriteBack_ReadAlloc_WriteAlloc_Cacheable
+            + TCR_EL1::IRGN0::WriteBack_ReadAlloc_WriteAlloc_Cacheable
+            + TCR_EL1::EPD0::EnableTTBR0Walks
+            + TCR_EL1::T0SZ.val(32), // TTBR0 spans 4 GiB total.
+    );
+}
+
+//--------------------------------------------------------------------------------------------------
+// Arch-public
+//--------------------------------------------------------------------------------------------------
+
+pub struct MMU;
+
+//--------------------------------------------------------------------------------------------------
+// OS interface implementations
+//--------------------------------------------------------------------------------------------------
+
+impl interface::mm::MMU for MMU {
+    /// Compile the page tables from the `BSP`-supplied `virt_mem_layout()`.
+    ///
+    /// # Safety
+    ///
+    /// - User must ensure that the hardware supports the paremeters being set here.
+    unsafe fn init(&self) -> Result<(), &'static str> {
+        // Fail early if translation granule is not supported. Both RPis support it, though.
+        if !ID_AA64MMFR0_EL1.matches_all(ID_AA64MMFR0_EL1::TGran64::Supported) {
+            return Err("64 KiB translation granule not supported");
+        }
+
+        // Prepare the memory attribute indirection register.
+        set_up_mair();
+
+        // Populate page tables.
+        populate_pt_entries()?;
+
+        // Set the "Translation Table Base Register".
+        TTBR0_EL1.set_baddr(TABLES.lvl2.base_addr_u64());
+
+        configure_translation_control();
+
+        // Switch the MMU on.
+        //
+        // First, force all previous changes to be seen before the MMU is enabled.
+        barrier::isb(barrier::SY);
+
+        // Enable the MMU and turn on data and instruction caching.
+        SCTLR_EL1.modify(SCTLR_EL1::M::Enable + SCTLR_EL1::C::Cacheable + SCTLR_EL1::I::Cacheable);
+
+        // Force MMU init to complete before next instruction
+        barrier::isb(barrier::SY);
+
+        Ok(())
+    }
+}

diff -uNr 10_privilege_level/src/arch/aarch64.rs 11_virtual_memory/src/arch/aarch64.rs
--- 10_privilege_level/src/arch/aarch64.rs
+++ 11_virtual_memory/src/arch/aarch64.rs
@@ -5,6 +5,7 @@
 //! AArch64.

 mod exception;
+mod mmu;
 pub mod sync;
 mod time;

@@ -77,6 +78,7 @@
 //--------------------------------------------------------------------------------------------------

 static TIMER: time::Timer = time::Timer;
+static MMU: mmu::MMU = mmu::MMU;

 //--------------------------------------------------------------------------------------------------
 // Implementation of the kernel's architecture abstraction code
@@ -136,3 +138,8 @@
         println!("      FIQ:    {}", to_mask_str(exception::is_masked::<FIQ>()));
     }
 }
+
+/// Return a reference to an `interface::mm::MMU` implementation.
+pub fn mmu() -> &'static impl interface::mm::MMU {
+    &MMU
+}

diff -uNr 10_privilege_level/src/bsp/rpi/link.ld 11_virtual_memory/src/bsp/rpi/link.ld
--- 10_privilege_level/src/bsp/rpi/link.ld
+++ 11_virtual_memory/src/bsp/rpi/link.ld
@@ -8,6 +8,7 @@
     /* Set current address to the value from which the RPi starts execution */
     . = 0x80000;

+    __ro_start = .;
     .text :
     {
         *(.text._start) *(.text*)
@@ -17,6 +18,8 @@
     {
         *(.rodata*)
     }
+    . = ALIGN(65536); /* Fill up to 64 KiB */
+    __ro_end = .;

     .data :
     {

diff -uNr 10_privilege_level/src/bsp/rpi/memory_map.rs 11_virtual_memory/src/bsp/rpi/memory_map.rs
--- 10_privilege_level/src/bsp/rpi/memory_map.rs
+++ 11_virtual_memory/src/bsp/rpi/memory_map.rs
@@ -4,6 +4,14 @@

 //! The board's memory map.

+#[cfg(feature = "bsp_rpi3")]
+#[rustfmt::skip]
+pub const END_INCLUSIVE:       usize =        0x3FFF_FFFF;
+
+#[cfg(feature = "bsp_rpi4")]
+#[rustfmt::skip]
+pub const END_INCLUSIVE:       usize =        0xFFFF_FFFF;
+
 /// Physical devices.
 #[rustfmt::skip]
 pub mod mmio {
@@ -15,4 +23,5 @@

     pub const GPIO_BASE:       usize = BASE + 0x0020_0000;
     pub const PL011_UART_BASE: usize = BASE + 0x0020_1000;
+    pub const END_INCLUSIVE:   usize = super::END_INCLUSIVE;
 }

diff -uNr 10_privilege_level/src/bsp/rpi/virt_mem_layout.rs 11_virtual_memory/src/bsp/rpi/virt_mem_layout.rs
--- 10_privilege_level/src/bsp/rpi/virt_mem_layout.rs
+++ 11_virtual_memory/src/bsp/rpi/virt_mem_layout.rs
@@ -0,0 +1,82 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright (c) 2018-2019 Andre Richter <andre.o.richter@gmail.com>
+
+//! The virtual memory layout.
+//!
+//! The layout must contain only special ranges, aka anything that is _not_ normal cacheable DRAM.
+//! It is agnostic of the paging granularity that the architecture's MMU will use.
+
+use super::memory_map;
+use crate::memory::*;
+use core::ops::RangeInclusive;
+
+//--------------------------------------------------------------------------------------------------
+// BSP-public
+//--------------------------------------------------------------------------------------------------
+
+pub const NUM_MEM_RANGES: usize = 3;
+
+pub static LAYOUT: KernelVirtualLayout<{ NUM_MEM_RANGES }> = KernelVirtualLayout::new(
+    memory_map::END_INCLUSIVE,
+    [
+        RangeDescriptor {
+            name: "Kernel code and RO data",
+            virtual_range: || {
+                // Using the linker script, we ensure that the RO area is consecutive and 4 KiB
+                // aligned, and we export the boundaries via symbols:
+                //
+                // [__ro_start, __ro_end)
+                extern "C" {
+                    // The inclusive start of the read-only area, aka the address of the first
+                    // byte of the area.
+                    static __ro_start: u64;
+
+                    // The exclusive end of the read-only area, aka the address of the first
+                    // byte _after_ the RO area.
+                    static __ro_end: u64;
+                }
+
+                unsafe {
+                    // Notice the subtraction to turn the exclusive end into an inclusive end.
+                    #[allow(clippy::range_minus_one)]
+                    RangeInclusive::new(
+                        &__ro_start as *const _ as usize,
+                        &__ro_end as *const _ as usize - 1,
+                    )
+                }
+            },
+            translation: Translation::Identity,
+            attribute_fields: AttributeFields {
+                mem_attributes: MemAttributes::CacheableDRAM,
+                acc_perms: AccessPermissions::ReadOnly,
+                execute_never: false,
+            },
+        },
+        RangeDescriptor {
+            name: "Remapped Device MMIO",
+            virtual_range: || {
+                // The last 64 KiB slot in the first 512 MiB
+                RangeInclusive::new(0x1FFF_0000, 0x1FFF_FFFF)
+            },
+            translation: Translation::Offset(memory_map::mmio::BASE + 0x20_0000),
+            attribute_fields: AttributeFields {
+                mem_attributes: MemAttributes::Device,
+                acc_perms: AccessPermissions::ReadWrite,
+                execute_never: true,
+            },
+        },
+        RangeDescriptor {
+            name: "Device MMIO",
+            virtual_range: || {
+                RangeInclusive::new(memory_map::mmio::BASE, memory_map::mmio::END_INCLUSIVE)
+            },
+            translation: Translation::Identity,
+            attribute_fields: AttributeFields {
+                mem_attributes: MemAttributes::Device,
+                acc_perms: AccessPermissions::ReadWrite,
+                execute_never: true,
+            },
+        },
+    ],
+);

diff -uNr 10_privilege_level/src/bsp/rpi.rs 11_virtual_memory/src/bsp/rpi.rs
--- 10_privilege_level/src/bsp/rpi.rs
+++ 11_virtual_memory/src/bsp/rpi.rs
@@ -5,9 +5,10 @@
 //! Board Support Package for the Raspberry Pi.

 mod memory_map;
+mod virt_mem_layout;

 use super::driver;
-use crate::interface;
+use crate::{interface, memory::KernelVirtualLayout};
 use core::fmt;

 pub const BOOT_CORE_ID: u64 = 0;
@@ -69,3 +70,13 @@
     // Configure PL011Uart's output pins.
     GPIO.map_pl011_uart();
 }
+
+/// Return the address space size in bytes.
+pub const fn addr_space_size() -> usize {
+    memory_map::END_INCLUSIVE + 1
+}
+
+/// Return a reference to the virtual memory layout.
+pub fn virt_mem_layout() -> &'static KernelVirtualLayout<{ virt_mem_layout::NUM_MEM_RANGES }> {
+    &virt_mem_layout::LAYOUT
+}

diff -uNr 10_privilege_level/src/bsp.rs 11_virtual_memory/src/bsp.rs
--- 10_privilege_level/src/bsp.rs
+++ 11_virtual_memory/src/bsp.rs
@@ -4,7 +4,7 @@

 //! Conditional exporting of Board Support Packages.

-mod driver;
+pub mod driver;

 #[cfg(any(feature = "bsp_rpi3", feature = "bsp_rpi4"))]
 mod rpi;

diff -uNr 10_privilege_level/src/interface.rs 11_virtual_memory/src/interface.rs
--- 10_privilege_level/src/interface.rs
+++ 11_virtual_memory/src/interface.rs
@@ -127,3 +127,11 @@
         fn spin_for(&self, duration: Duration);
     }
 }
+
+/// Memory Management interfaces.
+pub mod mm {
+    pub trait MMU {
+        /// Called by the kernel early during init.
+        unsafe fn init(&self) -> Result<(), &'static str>;
+    }
+}

diff -uNr 10_privilege_level/src/main.rs 11_virtual_memory/src/main.rs
--- 10_privilege_level/src/main.rs
+++ 11_virtual_memory/src/main.rs
@@ -19,6 +19,8 @@
 //! [Architecture-specific code]: arch/index.html
 //! [`kernel::interface`]: interface/index.html

+#![allow(incomplete_features)]
+#![feature(const_generics)]
 #![feature(format_args_nl)]
 #![feature(panic_info_message)]
 #![feature(trait_alias)]
@@ -36,6 +38,7 @@
 mod bsp;

 mod interface;
+mod memory;
 mod panic_wait;
 mod print;

@@ -46,8 +49,18 @@
 /// # Safety
 ///
 /// - Only a single core must be active and running this function.
-/// - The init calls in this function must appear in the correct order.
+/// - The init calls in this function must appear in the correct order:
+///     - Virtual memory must be activated first.
+///       - Without it, any atomic operations, e.g. the yet-to-be-introduced spinlocks in the device
+///         drivers (which currently employ NullLocks instead of spinlocks), will fail to work on
+///         the RPi SoCs.
 unsafe fn kernel_init() -> ! {
+    use interface::mm::MMU;
+
+    if let Err(string) = arch::mmu().init() {
+        panic!("MMU: {}", string);
+    }
+
     for i in bsp::device_drivers().iter() {
         if let Err(()) = i.init() {
             panic!("Error loading driver: {}", i.compatible())
@@ -67,6 +80,9 @@

     println!("Booting on: {}", bsp::board_name());

+    println!("MMU online. Special regions:");
+    bsp::virt_mem_layout().print_layout();
+
     println!(
         "Current privilege level: {}",
         arch::state::current_privilege_level()
@@ -87,6 +103,13 @@
     println!("Timer test, spinning for 1 second");
     arch::timer().spin_for(Duration::from_secs(1));

+    let remapped_uart = unsafe { bsp::driver::PL011Uart::new(0x1FFF_1000) };
+    writeln!(
+        remapped_uart,
+        "[     !!!    ] Writing through the remapped UART at 0x1FFF_1000"
+    )
+    .unwrap();
+
     println!("Echoing input now");
     loop {
         let c = bsp::console().read_char();

diff -uNr 10_privilege_level/src/memory.rs 11_virtual_memory/src/memory.rs
--- 10_privilege_level/src/memory.rs
+++ 11_virtual_memory/src/memory.rs
@@ -0,0 +1,147 @@
+// SPDX-License-Identifier: MIT
+//
+// Copyright (c) 2018-2019 Andre Richter <andre.o.richter@gmail.com>
+
+//! Memory Management.
+
+use core::{fmt, ops::RangeInclusive};
+
+#[derive(Copy, Clone)]
+pub enum Translation {
+    Identity,
+    Offset(usize),
+}
+
+#[derive(Copy, Clone)]
+pub enum MemAttributes {
+    CacheableDRAM,
+    Device,
+}
+
+#[derive(Copy, Clone)]
+pub enum AccessPermissions {
+    ReadOnly,
+    ReadWrite,
+}
+
+#[derive(Copy, Clone)]
+pub struct AttributeFields {
+    pub mem_attributes: MemAttributes,
+    pub acc_perms: AccessPermissions,
+    pub execute_never: bool,
+}
+
+impl Default for AttributeFields {
+    fn default() -> AttributeFields {
+        AttributeFields {
+            mem_attributes: MemAttributes::CacheableDRAM,
+            acc_perms: AccessPermissions::ReadWrite,
+            execute_never: true,
+        }
+    }
+}
+
+/// An architecture agnostic descriptor for a memory range.
+pub struct RangeDescriptor {
+    pub name: &'static str,
+    pub virtual_range: fn() -> RangeInclusive<usize>,
+    pub translation: Translation,
+    pub attribute_fields: AttributeFields,
+}
+
+/// Human-readable output of a RangeDescriptor.
+impl fmt::Display for RangeDescriptor {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        // Call the function to which self.range points, and dereference the result, which causes
+        // Rust to copy the value.
+        let start = *(self.virtual_range)().start();
+        let end = *(self.virtual_range)().end();
+        let size = end - start + 1;
+
+        // log2(1024).
+        const KIB_RSHIFT: u32 = 10;
+
+        // log2(1024 * 1024).
+        const MIB_RSHIFT: u32 = 20;
+
+        let (size, unit) = if (size >> MIB_RSHIFT) > 0 {
+            (size >> MIB_RSHIFT, "MiB")
+        } else if (size >> KIB_RSHIFT) > 0 {
+            (size >> KIB_RSHIFT, "KiB")
+        } else {
+            (size, "Byte")
+        };
+
+        let attr = match self.attribute_fields.mem_attributes {
+            MemAttributes::CacheableDRAM => "C",
+            MemAttributes::Device => "Dev",
+        };
+
+        let acc_p = match self.attribute_fields.acc_perms {
+            AccessPermissions::ReadOnly => "RO",
+            AccessPermissions::ReadWrite => "RW",
+        };
+
+        let xn = if self.attribute_fields.execute_never {
+            "PXN"
+        } else {
+            "PX"
+        };
+
+        write!(
+            f,
+            "      {:#010X} - {:#010X} | {: >3} {} | {: <3} {} {: <3} | {}",
+            start, end, size, unit, attr, acc_p, xn, self.name
+        )
+    }
+}
+
+/// Type for expressing the kernel's virtual memory layout.
+pub struct KernelVirtualLayout<const NUM_SPECIAL_RANGES: usize> {
+    max_virt_addr_inclusive: usize,
+    inner: [RangeDescriptor; NUM_SPECIAL_RANGES],
+}
+
+impl<const NUM_SPECIAL_RANGES: usize> KernelVirtualLayout<{ NUM_SPECIAL_RANGES }> {
+    pub const fn new(max: usize, layout: [RangeDescriptor; NUM_SPECIAL_RANGES]) -> Self {
+        Self {
+            max_virt_addr_inclusive: max,
+            inner: layout,
+        }
+    }
+
+    /// For a virtual address, find and return the output address and corresponding attributes.
+    ///
+    /// If the address is not found in `inner`, return an identity mapped default with normal
+    /// cacheable DRAM attributes.
+    pub fn get_virt_addr_properties(
+        &self,
+        virt_addr: usize,
+    ) -> Result<(usize, AttributeFields), &'static str> {
+        if virt_addr > self.max_virt_addr_inclusive {
+            return Err("Address out of range");
+        }
+
+        for i in self.inner.iter() {
+            if (i.virtual_range)().contains(&virt_addr) {
+                let output_addr = match i.translation {
+                    Translation::Identity => virt_addr,
+                    Translation::Offset(a) => a + (virt_addr - (i.virtual_range)().start()),
+                };
+
+                return Ok((output_addr, i.attribute_fields));
+            }
+        }
+
+        Ok((virt_addr, AttributeFields::default()))
+    }
+
+    /// Print the memory layout.
+    pub fn print_layout(&self) {
+        use crate::println;
+
+        for i in self.inner.iter() {
+            println!("{}", i);
+        }
+    }
+}