Go to file
KrekBuk b842720a70 FAT32: support directory entries larger than 1 sector
There was already a support for partitions with bigger cluster sizes but there was also a bug where the directory loading code would only look  at the first sector of each cluster and could incorrectly report some files as non-existing.
2020-05-02 15:47:29 +02:00
src FAT32: support directory entries larger than 1 sector 2020-05-02 15:47:29 +02:00
test Misc bug fixes and sanity checks 2020-04-24 18:26:08 +02:00
toolchain Bump gcc and binutils versions in make_toolchain.sh 2020-03-13 17:04:51 +01:00
.gitignore EXT2 and VFS support (#5) 2020-04-15 08:48:35 +02:00
bochsrc Initial linux protocol support 2020-04-19 10:14:49 +02:00
CONFIG.md Add CONFIG.md 2020-04-21 19:18:13 +02:00
LICENSE.md Added license 2019-05-30 19:30:15 +02:00
Makefile FAT32: support directory entries larger than 1 sector 2020-05-02 15:47:29 +02:00
qloader2-install Make qloader2-install less verbose 2020-04-21 15:12:00 +02:00
qloader2.bin Add partprobe to test targets in makefile 2020-05-01 18:23:49 +02:00
README.md Add FAT32 to list of supported filesystems in readme 2020-05-02 14:30:25 +02:00
STIVALE.md Add time stuff 2020-04-30 21:19:12 +02:00

qloader2

x86/x86_64 BIOS Bootloader

Supported boot protocols

  • Linux
  • stivale (qloader2's native boot protocol, see STIVALE.md for details)

Supported filesystems

  • ext2
  • echfs
  • FAT32

Supported partitioning schemes

  • MBR
  • GPT

How to use

This repository contains a prebuilt version of qloader2 so building it won't be necessary.

In order to install qloader2 on a MBR device (which can just be a raw image file), run the provided qloader2-install script as such:

./qloader2-install ./qloader2.bin <path to device/image>

If using a GPT formatted device, it will be necessary to create an extra partition (of at least 32K in size) to store stage 2 code. Then it will be necessary to tell the install script where this partition is located by specifying the start sector.

fdisk <device>    # Create bootloader partition using your favourite method
./qloader2-install ./qloader2.bin <path to device/image> <start sector of boot partition>

Then make sure the device/image contains at least 1 partition formatted in a supported filesystem containing a /qloader2.cfg or /boot/qloader2.cfg file and the kernel/modules one wants to load.

An example qloader2.cfg file can be found in test/qloader2.cfg.

More info on the format of qloader2.cfg can be found in CONFIG.md.

For example, to create an empty image file of 64MiB in size, 1 echfs partition on the image spanning the whole device, format it, copy the relevant files over, and install qloader2, one can do:

dd if=/dev/zero bs=1M count=0 seek=64 of=test.img
parted -s test.img mklabel msdos
parted -s test.img mkpart primary 1 100%
echfs-utils -m -p0 test.img quick-format 32768
echfs-utils -m -p0 test.img import path/to/qloader2.cfg qloader2.cfg
echfs-utils -m -p0 test.img import path/to/kernel.elf kernel.elf
echfs-utils -m -p0 test.img import <path to file> <path in image>
...
./qloader2-install $THIS_REPO/qloader2.bin test.img

One can get echfs-utils by installing https://github.com/qword-os/echfs.

Building from source

In order to hack qloader2, one must build the GCC toolchain from source first.

To do so, run the make_toolchain.sh script from within the toolchain directory; keep in mind that the script takes MAKEFLAGS as an argument.

cd toolchain
./make_toolchain.sh -j4

After that is done, simply run make in the root of the repo to generate src/qloader2.bin.