Now we can do this, since nobody touches this file before this step.
(previously by mkarchroot)
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
This mode allow to build all 6 ISOs in just one step if build type is <all>.
So "build.sh build all all" -> (single-i686, single-x86_64, dual) X (netinstall, core).
Note that the <purge> command is executed between each build step to save space,
so do not use this mode if you want to keep all temporal files in work_dir.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
This allow to execute build.sh in 32-bit-compat without using a chroot
and doing tricks.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
* Remove devtools dependency.
* Better control over what files are touched inside chroot (root-image).
Now: NONE :)
* Two new commands:
+ init: To install {base} group and other needed packages (syslinux for now)
+ run: If we want to run some command inside chroot
(mkinitcpio, locale-gen, useradd, etc etc...)
* Renamed command: "created" to "install", says much better what does.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
Use new -d option in mkarchroot (mount /dev as devtmpfs), then avoid
mknod for loop dev. (Needs devtools >= 20120305)
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
Needs mkinitpcio >= 0.8.4
Next changes will be for mdadm and mkinitcpio-nfs-utils:
* /lib/initcpio/udev/11-dm-initramfs.rules
* /lib/initcpio/ipconfig , /lib/initcpio/nfsmount
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
* Add libraries from /usr/lib and /sbin/losetup.
* Use used_block_devices to detach loop devices.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
This list includes block devices of:
* Boot medium [archisodevice] (only if no copytoram= is used)
* Loop medium [img_dev] (only if no copytoram= is used)
* COW space [cowdevice] (only if cowdevice= is used (no tmpfs is used))
* Loop devices used for SquashFS images.
* Loop devices used for device-mapper devices (two per each dm-device, one RO{*.fs} and one RW{*.cow})
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
New losetup from util-linux 2.21 uses new /dev/loop-control from Linux 3.1.
Needs mkinitcpio-0.8.2+ (provides modules.devname, or workaround with earlymodules=loop)
Get advantages from it! Removing custom shell code.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
This is needed for initscripts 2012.01, at least the entry for /.
Otherwise / is remounted ro, but not remounted rw later.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
This was the first implementation that downloads the complete ISO,
but now we have the posibility of download only needed *.sfs files.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
This was useful to me during development of dm-snapshot support
to create very quickly isos without wating for SquashFS compression.
Is time to remove this, I think the is no practical usage,
and make the code a bit more simple.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
We are only using these files only for SYSLINUX bootloader that supports
".." as directories.
ISOLINUX jumps to archiso.cfg via isolinux.cfg
PXELINUX goes directly to archiso.cfg via DHCP configuration.
Suggested by Thomas.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
* In that way we can have two differents syslinux menus depending if booting from
PXE or from non-PXE.
Some file renames are made to keep things ordered:
archiso_pxe* all things related to PXELINUX
archiso_sys* all things related to SYSLINUX
*_inc.cfg only contains INCLUDE directive
* By default NBD PXE option is selected for booting on timeout.
* As (NDB/NFS/HTTP/FTP) server is used the same IP from pxeserver.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
* Now we have more than one PXE boot methods.
* archiso_ndb_srv= accepts special ${pxeserver} parameter like in others PXE hooks.
* This allows to setup network from initramfs without forcing using a special PXE hook.
* Update README
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
* UI is more correct than DEFAULT for simple menu system.
* Remove TIMEOUT
* Remove PROMPT 0, this is the default value.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
On machines with common graphics chipsets (radeon, intel, nvidia), enable
KMS early. Downloading images from a remote server can take time, better
view the status with a nice resolution.
Otherwise does not work.
22100 Nov 29 09:29 ./lib/libnss_dns-2.14.1.so
20 Nov 29 09:29 ./lib/libnss_dns.so.2 -> libnss_dns-2.14.1.so
46732 Nov 29 09:29 ./lib/libnss_files-2.14.1.so
22 Nov 29 09:29 ./lib/libnss_files.so.2 -> libnss_files-2.14.1.so
Looks like my test was made before introducing symlinks
and using directly libnss_dns-$VERSION.so.2 as libnss_dns.so.2.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
We are currently checksuming all files inside $archisobasedir
(aitab, *sfs, kernel/initramfs and boot/syslinux related files)
this is stored in one file only. Its works good for now,
but I think we only need to checksum only aitab and *.sfs if they are needed.
Currently if you boot a dual-iso in i686, also checksum is done
for *.sfs of x86_64. Not a big issue, this just take a bit more time.
The real issue is when booting via PXE with HTTP/FTP methods,
since they download only aitab and needed *.sfs files,
can not use directly the only one file where checksums are stored.
This patch does:
(1) Do not checksum syslinux related files, anyway you are already reach initramfs stage.
(2) Do not checksum kernel/initramfs files, for the same as (1)
(3) Two checksum.${arch}.md5 for each i686 and x86_64 for only aitab and *.sfs.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
* In that way, we have unified directory scheme among {pxe,sys,ext,iso}linux,
sharing the same directory structure and config files.
* Also enable IPAPPEND (will show a warning message about unknown directive in non-pxelinux before menu)
NEW:
* Rename syslinux*.cfg -> archiso*.cfg
* Now syslinux.cfg is only a "loadconfig" trick like we are doing for isolinux
* All paths are now relative to %INSTALL_DIR% ($archisobasedir) (Default: /arch/)
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
[ramfs /]# cat /etc/resolv.conf
nameserver 192.168.0.77
[ramfs /]# curl -O http://releng.archlinux.org/pxeboot/README
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 1129 100 1129 0 0 2191 0 --:--:-- --:--:-- --:--:-- 4589
[ramfs /]#
Do not "add_binary /lib/libnss_dns.so.2", because will add libnss_dns-VERSION.so.2 as libnss_dns.so.2
Is not an issue will work (nobody links directly to VERSION), but is not nice.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
It add support for using *.sfs images via NFS.
It must be located after archiso_pxe_nbd in mkinitcpio.conf, in that way
NBD is the default.
New boot parameters:
archiso_nfs_srv=IP:/path
Set the NFS IP and path. Accepts ${pxeserver} as IP.
archiso_nfs_opt= Sets NFS mount opts (comma separated)
Defaults to:
port = as given by server portmap daemon
rsize = 1024
wsize = 1024
timeo = 7
retrans = 3
acregmin = 3
acregmax = 60
acdirmin = 30
acdirmax = 60
flags = hard, nointr, noposix, cto, ac
:: Running Hook [archiso_pxe_common]
IP-Config: eth0 hardware address 52:54:00:12:34:57 mtu 1500
IP-Config: eth0 guessed broadcast address 192.168.0.255
IP-Config: eth0 guessed nameserver address 192.168.0.218
IP-Config: eth0 complete (from 192.168.0.218):
address: 192.168.0.90 broadcast: 192.168.0.255 netmask: 255.255.255.0
gateway: 192.168.0.218 dns0 : 192.168.0.218 dns1 : 0.0.0.0
rootserver: 192.168.0.218 rootpath:
filename :
:: Running Hook [archiso_pxe_nbd]
:: Running Hook [archiso_pxe_curl]
:: Running Hook [archiso_pxe_nfs]
:: Mounting '192.168.0.77:/mnt/iso'
...
192.168.0.77:/mnt/iso/ on /run/archiso/bootmnt type nfs (rw,relatime,vers=3,rsize=524288,wsize=524288,namlen=255,hard,nolock,proto=tcp,port=65535,timeo=7,retrans=3,sec=sys,local_lock=all,addr=192.168.0.77)
...
My /etc/exports
/mnt/iso 192.168.0.90(ro,no_subtree_check,no_root_squash)
Its add support for downloading the ISO image via HTTP/FTP with curl or download *.sfs files.
The files are downloaded on /run/archiso/curlspace (tmpfs) and setup a loop dev from it
(like in archiso_loop_mnt) in the ISO case, or just do a bind mount to
/run/archiso/bootmnt in the other case.
This hook must be located after archiso_pxe_nbd in mkinitcpio.conf, in that way
NBD is the default.
New bootparm:
* archiso_curl_url=(first form)
archiso_curl_url=http://192.168.0.7/archlinux/iso/archlinux.iso (absolute form)
Use an absolute URL. Fetch the entire ISO
archiso_curl_url=ftp://${pxeserver}/archlinux/iso/archlinux.iso (relative form)
Use the as IP the same server where PXE reside. Fetch the entire ISO
* archiso_curl_url=(second form)
archiso_curl_url=http://192.168.0.7/archlinux/iso/unpacked (absolute form)
Use an absolute URL. Where unpacked (an example name) is a directory where
${archisobasedir} ("arch" by default) is located with all *.sfs.
This fetch only needed files.
archiso_curl_url=ftp://${pxeserver}/archlinux/iso/unpacked (relative form)
Use the as IP the same server where PXE reside. Where unpacked (an example name)
is a directory where ${archisobasedir} ("arch" by default)
is located with all *.sfs. This fetch only needed files.
* curlspace_size= Set a size for a tmpfs filesystem where files are downloaded.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
One hook only setup the network device (archiso_pxe_common) the other
setup the NBD client (archiso_pxe_nbd).
New bootparam:
archiso_pxe_srv=IP
Allow to set an IP different from the PXE server.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
In this way allow to mount /run/archiso/bootmnt from another hook,
but still use the logic from the main mount_hook, skipping only this mount.
Signed-off-by: Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar>
