Daniel Barlow
67768004ee
erase full flash, it seems to help with jffs2
2023-04-26 22:19:01 +01:00
Daniel Barlow
26c6eac937
delete config boot.tftp.enable as it doesn't enable anything
2023-04-26 22:18:23 +01:00
Daniel Barlow
4cfaed7303
specify root device as /dev/mtdblockn
...
the (openwrt?) magic that autodetects it based on mtd labels
won't work with a two-stage boot
2023-04-26 22:16:15 +01:00
Daniel Barlow
6d6dbe1cbb
tftpboot: allow padding image with freeSpaceBytes
...
this is useful for writable filesystems so that there's more than
an erase block of space to write into
2023-04-26 22:15:19 +01:00
Daniel Barlow
151dc88c28
little evidence of thinking
2023-04-23 21:24:37 +01:00
Daniel Barlow
2f3b108ef1
tftpboot: set phram erase block size to match hardware
...
this is needed for tftp boot with a jffs2 filesystem
2023-04-23 21:24:37 +01:00
Daniel Barlow
6adab44fcf
rotuer; run chronyd for accurate time
2023-04-23 20:56:20 +01:00
Daniel Barlow
4342d3403f
make rotuer example build again
2023-04-23 20:56:20 +01:00
Daniel Barlow
28264febdb
add smaller-than-gnu "hello world" package
2023-04-23 20:56:20 +01:00
Daniel Barlow
2c4f8b823e
remove jffs2boot target
...
It's not needed now initramfs is compiled into the kernel.
2023-04-23 20:56:20 +01:00
Daniel Barlow
3595e36294
workaround for huge musl binaries
...
something in our configuration is causing gcc to generate binaries
in excess of 68K for "hello world". hardeningDisable seems to stop it
doing that, turning it on for s6 which writes _lots_ of binaries
but would be good to get to the bottom of this
https://discourse.nixos.org/t/crosscompilation-to-musl32-problems/3110
may have more
2023-04-22 23:04:03 +01:00
Daniel Barlow
245bf534e6
refactor: inline definition
2023-04-22 23:03:36 +01:00
Daniel Barlow
3feb8eb5f9
shrink hostap, build it without openssl
2023-04-22 22:17:35 +01:00
Daniel Barlow
c7aa8c7756
some notes on how we couldn't make it smaller
2023-04-19 22:03:18 +01:00
Daniel Barlow
c6b8a8488f
build s6 with shared libraries
2023-04-19 21:44:06 +01:00
Daniel Barlow
99af416e14
use /bin/sh for init scripts
2023-04-19 21:41:05 +01:00
Daniel Barlow
ad3ef53171
ntp: don't install manpages, docs etc
2023-04-18 23:57:15 +01:00
Daniel Barlow
65dfbad365
systemconfig: chown files if uid/gid > 0
2023-04-15 22:53:28 +01:00
Daniel Barlow
5dd0c6e3c0
rewrite preinit as very small C program
...
By using the kernel "nolibc" header to avoid requiring a C library, we
can bring the initramfs size to around 4k
This does involve a tiny bit of inline mips assembly which I'm not
sure about. gcc seems unwilling to generate the code to load $gp at
function entry of main(), so we do it by hand - but I'd rather find
out why gcc doesn't.
2023-04-15 18:27:39 +01:00
Daniel Barlow
b1a89ae8c3
jffs2: squash uids, enable lzo compression
2023-04-15 17:23:58 +01:00
Daniel Barlow
11f2715d18
mips-vm: enlarge mtd
2023-04-15 17:22:35 +01:00
Daniel Barlow
1cc0b13b57
rewrite systemconfig in C and link statically
...
systemconfig (a.k.a "activate") is run from the initramfs. Converting
it from a shell script to an executable means it doesn't depend on
there being a shell in the initramfs
2023-04-15 17:21:27 +01:00
Daniel Barlow
c744ef8c17
systemconfig: accept uid and gid options
...
all we do with them is assert they're zero, to unbreak CI.
This code is getting rewritten anyway
2023-04-14 23:19:56 +01:00
Brian McKenna
fb796e61e0
pseudofile: allow setting uid and gid of files
...
Necessary for Dropbear to accept non-root authorized_keys files.
2023-04-14 23:12:52 +01:00
Daniel Barlow
9f3c80e774
make initramfs busybox (very slightly) smaller
2023-04-11 23:08:02 +01:00
Daniel Barlow
29c7af5892
set qemu flash erase block size
2023-04-10 22:57:51 +01:00
Daniel Barlow
2566dc564c
remove need for symlinks in initramfs busybox
2023-04-10 22:57:50 +01:00
Daniel Barlow
0ad7b0e48b
don't hardcode mtd0 initramfs->second stage
2023-04-10 22:57:50 +01:00
Daniel Barlow
0a2881914b
set erase block size to something that works for jffs2
2023-04-10 22:57:50 +01:00
Daniel Barlow
4638092d3d
build initramfs into kernel
...
this doesn't matter for qemu, but there's no other way of
getting a hardware device to find it if the bootloader
won't co-operate
2023-04-10 22:57:50 +01:00
Daniel Barlow
633a4f8003
move BLK_DEV_INITRD to base.nix
2023-04-10 22:57:50 +01:00
Daniel Barlow
fe99abc450
shrink rsync
...
remove openssl requirement, it uses it for "optimised assembly
versions" of md5 - but only on x86/arm/sparc not mips anyway
2023-04-10 22:57:50 +01:00
Daniel Barlow
42725f56f9
make jffs2 module provide o.rootfs, conditionally
...
it's enabled if config.rootfsType == "jffs2"
2023-04-10 20:01:58 +01:00
Daniel Barlow
63a89f7d0d
add config.boot.initramfs.enable
2023-04-10 19:07:27 +01:00
Daniel Barlow
07eb3d66da
remove unused combined-image output
2023-04-10 18:15:48 +01:00
Daniel Barlow
0687ae7f5c
rename flashable->flashimage, o.squashfs to rootfs
...
"rootfs" describes what it is for, "squashfs" merely says
how it's implemented
(also, rootfs-as-jffs2 will soon be added)
2023-04-10 18:13:27 +01:00
Daniel Barlow
2e15acd61c
whitespace
2023-04-10 17:46:39 +01:00
Daniel Barlow
18b35b1ca6
thinking ...
2023-04-10 17:36:00 +01:00
Daniel Barlow
f5156425a2
correct jffs2 endian, remove unused bindings
2023-04-07 09:39:47 +01:00
Daniel Barlow
db54c877d6
add CI test for jffs2
2023-04-06 20:58:26 +01:00
Daniel Barlow
54a1ab3529
support jffs2, with initramfs
...
the jffs2 filesystem contains only /nix/store and a script which is
run in early init (initramfs) and is responsible for recreating
"traditional" directories (/bin /etc/**/* /var &c) based on the
configuration.
this is tested only in qemu so far and could use some cleanup
2023-04-04 23:35:49 +01:00
Daniel Barlow
25d9da967c
remove support for sockets in pseudofiles
...
(1) we can't make them on a real filesystem except by running
something that calls socket()
(2) whyever would we want to?
2023-04-04 23:23:22 +01:00
Daniel Barlow
aa1a2e5d75
package gen_init_cpio, tool for making an initramfs
2023-04-04 22:54:20 +01:00
Daniel Barlow
f02efa3fe3
refactor
2023-04-04 21:11:28 +01:00
Daniel Barlow
342c87b256
qemu: boot from mtd using mtd2block
...
doesn't make much difference for squashfs but this will make it much
simpler to test jffs2/ubifs
2023-04-04 21:07:02 +01:00
Daniel Barlow
07e7d63ade
fixup 98243d43da
2023-04-02 18:46:27 +01:00
Daniel Barlow
3194262eb3
it has not been a production-ready week
2023-03-31 23:44:49 +01:00
Daniel Barlow
98243d43da
add mtdutils mkfs.jffs2 --graft option
...
we'd like a bit more of the convenience of mksquashfs
(never thought I'd say _that_) for jffs2, in particular
not having to copy all the desired store paths into a
single directory just so we can create an image from them
2023-03-31 23:42:13 +01:00
Daniel Barlow
e0bd64b80f
week source
2023-03-28 16:14:00 +01:00
Daniel Barlow
a5630c1f30
add cautionary note about flashing a running device
2023-03-26 10:21:36 +01:00
