button_read_device() could be called before pca9555_read_thread
intializes pca9555_in_ports variable, and return incorrect value.
Change-Id: I960bff72fe230c9d0256b20e92d0a75e67266038
- Prevent the Changelog showing up when closing the configuration dialog if
"Show on startup" is enabled.
- Increase dialog size a bit.
- Detect and link Gerrit IDs.
Change-Id: I049be3c38a57b2559f2a78392b785fa51ec99dc1
When saving the system info from the progresslogger it shouldn't have knowledge
about the format of the data provided by getInfo(). Instead of relying on the
output being HTML formatted make getInfo() accept a parameter that indicates if
the data is to be formatted as HTML or text.
Change-Id: I733fe1a148e51b70ea1361d8feccffd7cbccd3d7
MSVC doesn't understand the gcc specific -Wno-unused-local-typedefs option.
Don't add it in this case.
Change-Id: I7ffb6c0c9c797eaad85975199c386020c464ca7b
We're compiling cutelogger directly, not as DLL. Therefore we must not add
__declspec attributes to the header.
Change-Id: I41b87fd4ba34dfbcd0d37245ff1c1f279139bb33
Change tracing from qDebug() to use cutelogger, which is available under the
LGPL2.1. This allows to automatically add filename and line number to the log,
and also provides multiple log levels.
Change-Id: I5dbdaf902ba54ea99f07ae10a07467c52fdac910
The build system used by rbutil is not very robust: all the files
are eventually compiled to the same directory so we cannot have
two files with the same name (system.c would conflicts with
system.cpp) so rename one file to avoid this. Also change all
include directives to local ones because we don't have to expose
the entire mspack source to inclusion.
Change-Id: I3fe0638d69fdc30566eb9425abfe33c807678b28
Reviewed-on: http://gerrit.rockbox.org/417
Reviewed-by: Dominik Riebeling <Dominik.Riebeling@gmail.com>
The previous code disabled the irq when there were fired. This seems
unreliable for some mysterious reason. Instead simply disable them
before enabling the irq.
Change-Id: I7ee1aa25dbc41c7dd53aa7c7744aa5e4d70ff2d8
Reviewed-on: http://gerrit.rockbox.org/640
Reviewed-by: benjamin brown <foolshperson@gmail.com>
Tested-by: benjamin brown <foolshperson@gmail.com>
Reviewed-by: Amaury Pouly <amaury.pouly@gmail.com>
The code was broken in two ways:
- it called storage_read_sectors with a wrong drive number
- calling storage_read_sectors too early at boot time will fail because
we are in the init function, so the drive is not yet registered.
To fix this, use a user provided read callback instead of storage read
functions.
Change-Id: I5ab79d48391fae619d5f8ae09c3d499fc43854c8
The old tools/mkzenboot has a number of problems: very hard to maintain,
poor integration with rbutil. Restart from scratch by recycling all the
crypto and descrambling code, rewrite the actual firmware modification part
to handle all scenarios in a much clearer way. The code is ready to be
integrated into Rockbox Utility, by using the very similar interface to
mkimxboot.
I copied all the keys from the old mkzenboot, so it can potentially support
the older Creative ports, but since this is untested, I prefer not do so
at the moment. However, I did add a "mixed" boot option to support the
dualboot style used in the older ports.
Change-Id: I80cfc48fa78187baa1b1692e8a30ec7137cea37b
Actually 0x20 is the right address, the 0x22 applies to the meizu which
probably use a Chinese clone with a non-spec compliant address.
Change-Id: I65b9ab6477c07a2293226536db3e696d34d3bc2c
This is mostly for consistency, this tool will be unused by the newer targets.
Also update the usage() display to mention -no-ciff
Change-Id: I4500f5fdce771ad3c53701a0bbaace916e88759d
The setting is currently unused but it provides some documentation and basic
support for a more comprehensive implementation of read-only support.
Change-Id: I353c33ef765ef6e6c09d59e943da6654a311ad84
The exact combination on which recording is needed is too complicated, so
just always cancel the unused warning, the compiler will optimise it away.
Change-Id: I0946586e4b01769c98babc4616f893c38f44c05f
The new code can select among several types of window (user, system, ...).
Furthermore, the type of partitions to use is selectable in config file.
Currently, two types are support: Freescale style MBR and Creative MBLK
Change-Id: I969d60a3d08f2c9448fb4b9c440051b7801b94cd
Currently we only support the BGA169 but if by chance Rockbox was to run on
a lqfp package for example, some pins may becomes unavailable or different.
Change-Id: I5c0d8d57ae31604572af37e0c2edd0bd7bda73a3
The old was doing register read/writes without calling tuner_power(), this
is broken for target with real tuner power management.
Change-Id: I5e60234697a8b91de6189daf6a10e871d5119b65
They share most of registers (except test and some extra bits), since our
driver is very basic, it is safe to relax the manufacurer & chip id check.
Change-Id: If54e8b9e0d852cd9128d5e28ff59bd2c7a55d98d
When bootloader is too intelligent, like on Creative devices, it might leave
some unexpected IRQs sources on, this is problematic when those sources share
the same IRQ line.
Change-Id: Ie9333918eb1048b1f2de4ea738cddf556aa75ee2
The STMP3700 has unreliable vbusvalid irq so we need to use
vdd5gtvddio. Therefore, using the vbusvalid bit produces
inconsistent results between events and detect. Fix that
by making power handle all the usb detect stuff.
Change-Id: Ic521b2f6677602e184fe46352c5359f4b4ba8d56
These functions are mostly used by the radio drivers and any blocking
call could potentially block the entire UI, which is pretty bad.
Since any request is expected to finish within a few us, having a 10ms
timeout doesn't seem unreasonable.
Change-Id: I03b19729511547e5bbdeb3476d020e5d87d0d7e1
On targets like stmp3700, resetting the dma channel on "soft" timeout
drives the whole block crazy unless it goes through complete reset.
Change-Id: I830b252279989bf9f9cd9c138240a6ea9b003527
The bootloader must call disk_init_subsystem() because it is multithread
(because of USB), otherwise strange things might happen. Calling disk_init()
is unnecessary since it is call when mounting partitions.
Change-Id: If7aff3dea0b96144e2a9b0f6179a9a0a632b93ed
Although Linux accepts several implicit definitions of SEEK_END found in
stdio.h, the compiler on FreeBSD won't. Rockbox compilation will fail
without stdio.h included.
There is a precedent for including this header, see
lib/rbcodec/codecs/libtremor/ivorbisfile.h.
Change-Id: I58510101b59a354cd6601cb3f323f385a824d2e8
Reviewed-on: http://gerrit.rockbox.org/639
Tested-by: Kevin Zheng <kevinz5000@gmail.com>
Reviewed-by: Frank Gevaerts <frank@gevaerts.be>
