| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
zhashfs was writing the eblocks: count line based on the file size of
the input. However, this assumes that the last block of the input file
had data in it, which is not necessarily true. Resultant .zd images
flashed OK but caused OFW to produce a warning.
WARNING: The file said highest block 15911 but wrote only as high as block 15903
Adjust the calculation to look at the last occupied data block to fix
this.
Test case, last block used is in the middle of the file:
rm -f testfile
dd if=/dev/zero of=testfile count=0 seek=128M bs=1
dd if=/dev/zero of=testfile count=1 seek=64M bs=1 conv=notrunc
./bin/zhashfs 0x20000 sha256 testfile testfile.zsp testfile.zd
The output used to trigger the OFW warning, now it does not.
Test case, last block used is at the end of the file:
rm -f testfile
dd if=/dev/zero of=testfile count=0 seek=128M bs=1
dd if=/dev/zero of=testfile count=1 seek=134217727 bs=1
./bin/zhashfs 0x20000 sha256 testfile testfile.zsp testfile.zd
The output here is unchanged by this patch - good.
|
|
|
|
| |
For compatibility with livecd-tools-18.13
|
|
|
|
|
|
|
|
| |
imgcreate now needs firewalld instead of system-config-firewall.
Provide it to make it happy.
Hopefully can go away soon.
https://bugzilla.redhat.com/show_bug.cgi?id=884878
|
| |
|
|
|
|
| |
Also enable usb olpc-update image generation.
|
|
|
|
|
|
|
|
|
| |
Reduce the impacts of having two system loggers active by limiting how
much memory journald will use.
In a later cycle we should remove rsyslogd and just rely on the journal.
Thanks to Shawn Landden.
|
|
|
|
| |
Requested by wad for C1 build.
|
|
|
|
|
| |
wad requests the next build reserve 400mb instead of 200mb of
RAM as part of #12285 diagnosis. Hack this in place for now.
|
|
|
|
|
|
|
|
| |
This reverts commit e41aaec60d874cbd32ab4a4f39891ae4c18b37b7.
The firmware bug behind this commit has been solved a while ago.
It is not present in A2 firmware releases, but enough time has passed
now that everyone has a B1.
|
|
|
|
|
| |
This is no longer pulled in by default, and is needed for 3G modem
support.
|
|
|
|
|
|
| |
The bluetooth service starts by default, but you don't need the service
just to do a bluetooth scan, which is what we believe the factory wants
to do.
|
|
|
|
|
|
|
|
| |
For XO-4 production, a method to run a simple bluetooth test is needed.
Add the bluez package so that it can be tested from the command line.
This does bring in an irritating collection of desktop packages and
depdendencies too, which we're not really supporting just yet.
|
|
|
|
|
|
|
|
|
| |
See
http://lists.fedoraproject.org/pipermail/test/2012-October/111230.html
Accompanied by a change to olpc-configure.
Thanks to Sam Greenfeld for pointing this out.
|
|
|
|
| |
It's not included in default dependency chains any more.
|
|
|
|
|
|
|
| |
This reverts commit 32d7015ac1b6287f3e1dae4b84c056e625643edd.
This design/implementation was found to be incompatible with
secure fs-update with old (shipped) firmware versions.
|
|
|
|
|
| |
Adobe Flash may be available on ARM as a bare libflashplayer.so file.
Support installation of this format via the adobe_flash module.
|
|
|
|
|
|
|
| |
Break out of the loop once we find the first good group URL.
Removes indenting, puts the "break" in a more natural place
and allows error handling.
|
| |
|
|
|
|
| |
This has now been ported to GTK3.
|
|
|
|
|
|
| |
osk (on-screen keyboard) is a better name for this module: touch support
will be shipped on all platforms, this module is only really for the
keyboard addition, and will soon be growing OSK-specific configuration.
|
|
|
|
| |
We now use libabiword introspection.
|
|
|
|
| |
Now using the OLPC wiki as the Sugar activity source.
|
|
|
|
|
|
|
|
| |
The keyboard control panel has been removed from the last few
releases since its buggy and doesn't suit the OLPC user-base well.
The software update is replaced with our own updater component,
to be reinstated soon.
|
|
|
|
| |
Found by Paul Fox.
|
|
|
|
|
|
| |
This reverts commit c68dcf2b8ca9816d2aa92306b6bbe23e6c15eeda.
gst-browser-plugin was not ready in time for this release;
reinstate totem once more.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Use arch considerations in yumcfg like the repos module.
This involved promoting fedora_arch to be a global option.
|
|
|
|
|
| |
We're working exclusively with F18 at the moment, and having these
files around is just a maintenance hassle.
|
|
|
|
|
|
|
|
|
| |
The XO-4 at the present time requires model-suffixes zip file names
e.g. runos4.zip.
This also seems to make sense in the general case, and is implemented
here as being fully backwards-compatible with firmware that don't
support it.
|
|
|
|
|
|
|
|
|
|
|
| |
Generate the environment just once during the build process, and
additionally dump it in shell format to a file.
Load this file when executing nochroot scripts. Such scripts now have
access to config settings, etc.
The default interpreter was changed from sh to bash because shlib uses
bashisms.
|
| |
|
|
|
|
|
|
|
| |
And add API to make this contract stronger.
Some chrooted kspost scripts are now interested in knowing the laptop
model number.
|
|
|
|
|
| |
Promote the shared directory to a core directory and create it early,
meaning that it can be used from all stages, including prepare.
|
|
|
|
|
|
| |
The XO-4 has some problems and unresolved questions around booting
from zip files. Leave the unzipped boot files in place until that
is resolved.
|
| |
|
|
|
|
|
| |
Enable it on XO-4, and also XO-1.75 until the early touchscreen
prototypes have been phased out.
|
| |
|
|
|
|
| |
This is no longer a dependency of the kernel package.
|
| |
|
|
|
|
|
|
| |
This package was previously pulled in via sugar-* deps, but not any
more. However, some activities still use it, as stated in the
sugar-devel mail: Port all activities from cjson to json
|
|
|
|
|
| |
Based on the fact that Sugar localization is very incomplete,
Reuben suggests removing this.
|
|
|
|
| |
Requested by Reuben.
|
|
|
|
|
|
|
|
| |
On some systems, we seem particularly prone to a race condition where
losetup cannot probe loop partitions because udev is probing the newly
appeared loop device.
Add a repeated fallback to partx to work around this situation.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We now ship the kernel/initramfs in zip files only.
This reduces duplication for when images get signed, and it allows
us to ship a simplified olpc.fth.
This implements the plan here:
http://lists.laptop.org/pipermail/devel/2012-March/034439.html
This didn't work on first attempt because we hit a firmware bug
on XO-1.75 (#12107). Now that we have a fix, and a workaround for
old firmware versions, give this change another spin.
|
|
|
|
|
| |
This reverts commit 41b5c8982e70ea61fde8aac6d2ed4bbe90458dc0.
It breaks boot on XO-1.75; the kernel boots but can't find dtree/atb.
|
| |
|
| |
|