x86 uses append-metadata, but only for signing and not for the
metadata itself.
Since recently SUPPORTED_DEVICES was assigned with a global value
and is not empty anymore, append-metadata will now actually put
metadata into x86 images. This breaks sysupgrade on x86.
To fix it for the moment, let's just empty SUPPORTED_DEVICES for
this target again.
In the long term, one should either not add metadata to the images
if it is not desired, and/or remove the unintended fwtool check.
Fixes: f52081bcf9 ("treewide: provide global default for SUPPORTED_DEVICES")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Many target use a repetitive if-include scheme for their subtarget
image files, though their names are consistent with the subtarget
names.
This patch removes these redundant conditions and just uses the
variable for the include where the target setup allows it.
For sunxi, this includes a trivial rename of the subtarget image
Makefiles.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
It was reported, that after image generation rework
x86-64-generic-squashfs-rootfs.img image won't boot on XenServer x86_64
anymore:
F2FS-fs (xvda): Magic Mismatch, valid(0xf2f52010) - read(0x84289960)
F2FS-fs (xvda): Can't find valid F2FS filesystem in 1th superblock
F2FS-fs (xvda): Magic Mismatch, valid(0xf2f52010) - read(0x4e8ee223)
F2FS-fs (xvda): Can't find valid F2FS filesystem in 2th superblock
List of all partitions:
ca00 4207 xvda
driver: vbd
No filesystem could mount root, tried:
ext3
ext2
ext4
squashfs
iso9660
f2fs
Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(202,0)
So lets fix this by adding back padding which was introduced in commit
a17d9482f5 ("x86: image: fix small disk space in squashfs overlay").
Ref: FS#3036
Fixes: 258f070d1a ("x86: fix missing squashfs and ext4 rootfs images")
Fixes: cb007a7bf6 ("x86: switch image generation to new code")
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Now that the x86 target uses the new image generation code we can also
attach metadata to the created images.
As currently the `SUPPORTED_DEVICES` list is empty, no JSON metadata is
attached, however the signing happens in the same step.
This results in signature verification for x86 images.
Signed-off-by: Paul Spooren <mail@aparcar.org>
grub-bios-setup requires two images (boot.img and core.img),
but they are missing. This make an error during sysupgrade:
Upgrading bootloader on /dev/sda...
grub-bios-setup: error: cannot open `/tmp/boot/boot/grub/boot.img': No
such file or directory.
Signed-off-by: 李国 <uxgood.org@gmail.com>
Add EFI platform bootable images for x86 platforms. These images can
also boot from legacy BIOS platform.
EFI System Partition need to be fat12/fat16/fat32 (not need to load
filesystem drivers), so the first partition of EFI images are not ext4
filesystem any more.
GPT partition table has an alternate partition table, we did not
generate it. This may cause problems when use these images as qemu disk
(kernel can not find rootfs), we pad enough sectors will be ok.
Signed-off-by: 李国 <uxgood.org@gmail.com>
[part_magic_* refactoring, removed genisoimage checks]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
The previous rework of x86 image creation done in commit cb007a7bf6
("x86: switch image generation to new code") removed images of separate
squashfs and ext4 filesystems which are handy for example in testing
under QEMU.
So this patch adds back creation of those missing rootfs images for ext4
and squashfs based filesystems.
Fixes: cb007a7bf6 ("x86: switch image generation to new code")
Signed-off-by: Paul Spooren <mail@aparcar.org>
[commit subject and description tweaks]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
The previous rework of x86 image creation done in commit cb007a7bf6
("x86: switch image generation to new code") broke the padding in
images.
sda: p2 size 212992 extends beyond EOD, enabling native capacity
sda: p2 size 212992 extends beyond EOD, truncated
sd 0:0:0:0: [sda] Attached SCSI disk
SQUASHFS error: squashfs_read_data failed to read block 0x2cc556
unable to read id index table
VFS: Cannot open root device "PARTUUID=ac5c9cd8-02" or unknown-block(8,2): error -5
Please append a correct "root=" boot option; here are the available partitions:
0800 19761 sda
driver: sd
0801 16384 sda1 ac5c9cd8-01
0802 2865 sda2 ac5c9cd8-02
Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(8,2)
Tested with x86/64 with Docker (squashfs), qemustart (ext4/squashfs) and
virtualbox (ext4/squashfs).
Ref: FS#2935
Fixes: cb007a7bf6 ("x86: switch image generation to new code")
Suggested-by: Russell Senior <russell@personaltelco.net>
Signed-off-by: Paul Spooren <mail@aparcar.org>
[commit subject and description tweaks]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
The x86 image generation was refacted via cb007a7bf6 and accidently not
included `geode.mk` when selected as subtarget.
Now the file is included and image compilation for x86/geode works
again.
Thanks to Russell Senior <russell@personaltelco.net> for reporting the
problem and suggesting a patch!
Signed-off-by: Paul Spooren <mail@aparcar.org>
The previous image generation code would always gzipped images.
This patch changes the behaviour and only compresses images when
selected in menuconfig.
Signed-off-by: Paul Spooren <mail@aparcar.org>
This commit introduces few related changes which need to be done in
single commit to keep images buildable between git revisions. In result
it retains all previous image creation possibilities with slight name
change of generated images. Brief summary of the commit:
* Split up image generation recipe to smaller chunks to make it more
generic and reusable.
* Make iso images x86 specific and drop their definition as root
filesystem.
* Convert image creation process to generic code specified in image.mk.
* Make geode subtarget inherit features from the main target instead of
redefining them.
* For subtargets create device definitions with basic packages set.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
[rebased]
Signed-off-by: Paul Spooren <mail@aparcar.org>
Let the grub2 package take care of creating installable grub2 images,
this will allow creating grub2 images without first calling x86 image
generation recipe. Also as side effect, since those images are now
shared, it'll reduce the number of calling grub-mkimage.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
[rebase, adjusted commit title]
Signed-off-by: Paul Spooren <mail@aparcar.org>
Currently bootloader always stays on the same version as when first
written to boot medium (not true if partition layout changed, which will
trigger sysupgrade process to write full disk image). That creates
inconveniences as it always stays with same features or/and bugs. Users
wishing to add support to additional modules or new version, would need
to write the whole image, potentially destroying previous system
configuration. To fix these, this commit adds additional routine to
sysupgrade which upgrades unconditionally the bootloader to the latest
state provided by OpenWrt.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
There can be some leftovers from other image recipes, if the same
directory names are used and multiply image types are selected.
Therefore remove directories used in the recipe, before contents for the
image are prepared.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
VMWare ESXI 6.5 and above is not compatible with
subformat=monolithicSparse (The default qemu-img convert -O VMDK option).
Monolithic Sparse vmdk can be imported, but issues occur when running
sysupgrade with new images and other tasks that modify the file system
(issues like Kernel panics, reboot loops, sometimes crashing the Host ESXI
box).
This change creates an additional VMDK output file for ESXI that sets the
subformat to monlithicFlat, and the adapter_type to the SCSI lsilogic
controller.
This change existed back on:
25e36d379e
But it looks like the change was removed when refactoring occurred with:
5f6a2732f892b6229473576d89cc963ae9c97d5d
Signed-off-by: John Sommerville <jsommerville@untangle.com>
These targets are currently using more or less same SIGNATURE variable
which provides unique partition ID/signature, so it makes sense to
refactor it out into common IMG_PART_SIGNATURE variable which could be
reused by all targets.
This is another step in the direction of reproducible OpenWrt images.
Signed-off-by: Paul Spooren <mail@aparcar.org>
[split into separate commit, renamed to IMG_PART_SIGNATURE]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
The host tool package qemu is build for the target x86. The installed tool
qemu-img is needed to build vdi or vmdk images. In the image Makefile we
use however the host installed qemu-img command and not the tool from the
buildsystem. This commit force to use the already build qemu-img command
from the openwrt toolchain.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Current squashfs snapshot images are improperly padded by just 128k,
which doesn't correspond with TARGET_ROOTFS_PARTSIZE=128 default config
option, leading to following:
Filesystem Size Used Available Use% Mounted on
/dev/root 2.5M 2.5M 0 100% /rom
/dev/loop0 85.0K 9.0K 71.0K 11% /overlay
overlayfs:/overlay 85.0K 9.0K 71.0K 11% /
Signed-off-by: Petr Štetiar <ynezz@true.cz>
In `build: image: Add IMG_ROOTFS and IMG_COMBINED variables` we've added
this new variables so switch to them now.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Currently we're producing unusable x86 combined squashfs images
(18.06.1, 18.06.2 and snapshots) as we don't enable TARGET_IMAGES_PAD,
thus providing very small space for the overlay filesystem, leading to
the following with OpenWrt 18.06.1 r7258-5eb055306f images on x86 QEMU:
root@(none):/# mount | egrep 'root|overlay'
/dev/root on /rom type squashfs
/dev/loop0 on /overlay type ext4
overlayfs:/overlay on / type overlay
root@(none):/# df -h | egrep 'root|overlay|Size'
Filesystem Size Used Available Use% Mounted on
/dev/root 2.5M 2.5M 0 100% /rom
/dev/loop0 113.0K 8.0K 97.0K 8% /overlay
overlayfs:/overlay 113.0K 8.0K 97.0K 8% /
We should rather ensure proper image padding in image generation code
and we shouldn't rely on config options in order to generate usable
images.
So this patch removes TARGET_IMAGES_PAD config option and enforces
padding on combined images.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
It was previously added in 546fced, which was part of "intel-microcode:
load as early as possible" series.
Unfortunately the conditionals added to GRUB config caused error on boot,
because on sysupgrade, bootloader is not updated and is left with old
features/modules. Since this module is needed for early microcode load
and transition to this needs to be done step by step, enable the test
module now, so that every newly created image has it already embedded.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
VBoxManage is not used and the image is created with proper permisions:
0f5d0f6 image: use internal qemu-img for vmdk and vdi images drop host
dependencies on qemu-utils and VirtualBox
Unreachable config symbols:
9e0759e x86: merge all geode based subtargets into one
No need to define those symbols since x86_64 is subtarget of x86:
196fb76 x86: make x86_64 a subtarget instead of a standalone target
Unreachable config symbols, so remove GRUB_ROOT:
371b382 x86: remove the xen_domu subtarget
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
Create initrd enries for x86 images, that'll load intel microcode as
early as possible. To achieve that the test module for grub is enabled
which provides shell-like conditionals. Also restrict the late load of
microcode to AMD processors.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
Add files to bootfs image from selected as built-in packages, which want
to install files to targets boot file system.
Signed-off-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
On the more sophisticated (i.e. deeper FIFO) serial controllers,
flow-control might be needed to avoid dropping output.
Signed-off-by: Philip Prindeville <philipp@redfish-solutions.com>
It's unused since commit 7427007193 ("x86: remove the olpc subtarget,
it has been unmaintained for a long time").
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
In the latest version of grub-mkimage, the prefix option is mandatory.
Not supplying it fails with:
```
Prefix not specified (use the -p option).
```
In grub-2.02-beta2 a DEFAULT_DIRECTORY was defined
in `include/grub/osdep/hostfile_unix.h` as:
```
#if defined (__NetBSD__)
/* NetBSD uses /boot for its boot block. */
# define DEFAULT_DIRECTORY "/"GRUB_DIR_NAME
#else
# define DEFAULT_DIRECTORY "/"GRUB_BOOT_DIR_NAME"/"GRUB_DIR_NAME
#endif
```
Where:
* GRUB_BOOT_DIR_NAME == boot
* GRUB_DIR_NAME == grub
This was used if the -p option was omitted.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Change grub's root device to xen/xvda,msdos1 for the x86_xen_domu
target so that it will boot without further changes.
Signed-off-by: Wilfried Klaebe <w+lede-project@chaos.in-kiel.de>
[Jo-Philipp Wich: fixed and rebased patch from FS#264, added subject]
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
To improve reproducibility, prevent the inclusion of timestamps
in the gzip header.
Signed-off-by: Reiner Herrmann <reiner@reiner-h.de>
SVN-Revision: 46361
This changes the x86 image generation to match x86_64, using the PARTUUID for
the rootfs instead of explicitly configuring the device.
It unbreaks KVM with VirtIO, which uses /dev/vda2 instead of /dev/sda2.
Tested in QEMU/KVM with VirtIO, VirtualBox and VMware.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
SVN-Revision: 44966
