Sync opus codec to upstream commit
02fed471a4568852d6618e041c4f2af0d7730ee2 (August 30 2013)
This brings in a lot of optimizations but also makes the diff
between our codec and the upstream much smaller as most of our
optimizations have been upstreamed or supeceded.
Speedups across the board for CELT mode files:
64kbps 128kbps
H300 9.82MHz 15.48MHz
c200 4.86MHz 9.63MHz
fuze v1 10.32MHz 15.92MHz
For the silk mode test file (16kbps) arm targets get a speedup
of about 2MHz while the H300 is 7.8MHz slower, likely because it's
now using the pseudostack more rather than the real stack which
is in iram. Patches to get around that are upcomming.
Change-Id: Ifecf963e461c51ac42e09dac1e91bc4bc3b12fa3
Instead of providing yet another memory allocator implementation
use tlsf and simply link tlsf library.
Another small improvement is to *grow* memory pool by grabbing
audiobuffer instead of just switching to use audiobuf exclusively.
Tested with simple lua 'memory eater' script.
This patch extends tlsf lib slightly. You can provide
void *get_new_area(size_t * size) function which will override
weak dummy implementation provided in lib itself. This allows to
automaticaly initialize memory pool as well as grow memory
pool if needed (for example grab audiobuffer when pluginbuffer
is exhaused).
Change-Id: I841af6b6b5bbbf546c14cbf139a7723fbb982f1b
codec makefiles larger freedom in what they can do to it.
Use this in libopus to prepend the libopus searchpaths to
CODECFLAGS so that its internal config.h will be picked up before
our global one. This avoids having to do a s/config.h/opus_config.h/
when syncing which will be handy soon.
Change-Id: I018d729aa0c8300fa3149f22a5a8c5668b339dfa
Reviewed-on: http://gerrit.rockbox.org/496
Reviewed-by: Nils Wallménius <nils@rockbox.org>
Drop most of the cases: only keep 64 MHz and 133 MHz. Pick values
from the manual which seem to match real life values.
Change-Id: I912752fbe372f9f44207db6853d0ff92fd619bed
I was too aggressive when converting reads to standard reads, this
one is memory based and not file based
Change-Id: Ibe7162894cc44cbd79f56cafe7136bda0f76bcdc
Now always generate a "make.db" file which aims at being the exact
representation of the file, ie running sbtoelf and elftosb using
the generated command file should produce the exact same file
(except for the random paddings). We still miss the support
for some option parsing to achieve that though.
Change-Id: Ib7d6b241f7855fd35225df8ab8e0711f69d9ee5a
This dump has been obtained with hwpatcher + hwstub, the fuze+
was running OF 2.36.8 in normal mode while playing some mp3.
Change-Id: Ieecaa760fe8ccade0858db929b9d7c175a3eaddf
This tool allows one to explore any register map. Register dumps
(like produced by hwstub tools) can be loaded and decoded by the
tool. Finally some analysers are provided for specific soc analysis
like clock tree and emi on imx233 for example.
Change-Id: Iaf81bd52d15f3e44ab4fe9bc039153fcf60cf92a
Now that the drive strength problem has been fixed, we can safely
drive sd cards at 48MHz in HS mode to get the best possible
transfer speed at 3.3V.
Change-Id: I0291589c399fb4880deba97895ff578451a32f99
Instead of going back and forth between TRAN and STBY modes,
stay in TRAN mode all the time, this avoid two commands on each
read/write and a potential delay to wait for the card.
Change-Id: Iafd456ab9a581d870331b622eeb48dcc254eda7f
This allows tells the card the number of blocks that will be
transfered. This is usually faster than continuous read/write.
It is mandatory for MMC and on SD cards, support is probed.
Change-Id: Ide3f97c26c2b714390884c69e05b00c2caa552f8
The old max dma transfer size was set to 2^15 because allowing
values up to 2^16 would result in overflow of the transfer count
in many places. This was a problem with the old register headers
because the macros didn't make sure the result is contained in
the bitmask of the field. The new macro do so 2^16 will actually
end up being 0 which means 2^16 for the hardware. This is kind of
hacky but it works and I prefer that this stays a power of two
because setting it to 2^16-1 would be extremely weird.
Change-Id: I9ce5661c477a79ab52efd5c573948e2353117804
- take advantage of the new rmi power function implemented to:
1) lower usual power_mode to low_power as it seems to be enough and might save
some battery
2) implement a system that lower that state to very_low_power
after 1 minute of inactivity.
3) implement touchdev_enable(bool) that can be use later to disable the
touchpad when needed
4) improve the debug screen report of the current power state and
changing the power state using volume keys
Change-Id: I0b372696d4b2bef5360c778d0500870fd9badee1
Reviewed-on: http://gerrit.rockbox.org/525
Reviewed-by: Amaury Pouly <amaury.pouly@gmail.com>
When using variadic macros there's no need for IF_MD2/IF_MV2 to deal
with function parameters. IF_MD/IF_MV are enough.
Throw in IF_MD_DRV/ID_MV_VOL that return the parameter if MD/MV, or 0
if not.
Change-Id: I7605e6039f3be19cb47110c84dcb3c5516f2c3eb
The code was broken in a subtle but crucial way: storage idle
callbacks are oneshot so after the first flush everything would
stay in the buffer forever and would never be written to the disk
thus resulting into many events being lost. This changed correctly
registers the idle callback each time the buffer is not empty.
Note that the idle storage code checks if a callback
has is in the queue already so we don't register twice.
Change-Id: Ifdf331d4b757e05b8a6902bf5926cbc7689f5109
The quickscreen calls settings_apply() and the crossfeed code wasn't
checking that the right crossfeed was set before updating the filter
for the custom setting, which was overwriting the Meier crossfeed
data (custom and Meier share the same data space).
Change-Id: Ifaa2f46fe062d4497681a2dd0d5068ec906c96a3
This small patch aims to fix a trivial
problem: when exiting the "cube" plugin
in a hosted environment (at least on YP-R0)
this caused the whole program, and consequently
also the player, to be turned off.
I don't know why the exit() function was
used here...
Change-Id: I7b58d4ca5c01c230f2c53204b51ec553c5fe1287
Reviewed-on: http://gerrit.rockbox.org/321
Reviewed-by: Bertrik Sikken <bertrik@sikken.nl>
Remove the hackish elf_translate_addresses which should not have
existed in the first place, on write always compute the physical
address of a section using elf_translate_virtual_address which
makes it possible to specify any virtual to physical mapping and
fail nicely if there is none.
Change-Id: I4f436945e90280a6fd9430de6c642dbeb8e23d40
