|Age||Commit message (Collapse)||Author||Files||Lines|
(Brand new unit lasted one minute shy of 12 hours!)
* Don't stop clock before switching speeds
* Don't stop clock prior to transactions
* Stop clock at the end of transactions
Will result in slightly better performance and some power saving when
we're not actively using the SD peripheral.
* Check to see if clock is [not] running prior to [en|dis]abling it
* Stop clock _prior_ to resetting controller
* Stop clock after transaction is completed, not before initiating it
* Use controller's low power mode (disables clocks when idle)
* Fix, and enable, interrupt-driven DMA transfers
* Fixes for full interrupt-driven operation (WIP, still broken)
There's a code path that calls sd_init_device() while we hold sd_mtx, but
sd_init_device() tries to obtain the mutex while doing its work.
* Allows both SD interfaces to have requests in flight simultaneously
* Fixed a deadlock in the hotswap code
* Ensure TX DMA is idle before initiating a request (bug due to a typo)
(More specifically, use the SoC's "OS Timer", slaved to the main XTAL so
it doesn't matter how the main CPU is clocked)
PLL1 clock for those frequencies has been dropped from 508 to 169.5 MHz,
so it's still a respectable reduction.
(I'm not sure how/why it ever worked with the XTAL source, but it did,
and was off by an audible amount)
* for <= 48KHz, BCLK must be 256*freq (ie bdiv = 4)
* for <= 96KHz, BCLK must be 128*freq (ie bdiv = 2)
* for 11/22/44/88 KHz, disable PLL1 and run off XTAL
* cut PLL1 with 12/24/48/98 KHz audio from 516->86MHz
* cut PLL1 with 8/16/32/64 KHz audio from 426->106.5MHz
This should result in significant power savings for
common 44.1KHz audio playback, and pretty good savings
for everything else.
As an added bonus:
* enable de-emphasis filters at 32, 44.1, and 48 KHz
default/low speed is 192 MHz, Max is 576
Downclock PCLK/MCLK/etc to 96MHz to save a bit of juice
Honestly the high speed could be dialed down to, eg 384
as this thing is so bloody fast..
PLL0 Needs to be a multiple of 48MHz for sane USB operation!
(Indeed, "typical" clock for this part is 528, but that seems a
waste of power)
Also fixes a minor bugaboo in the jz4670 usb divisor calculation
that won't matter until we enable reclocking
Note: I left behind lcd_bitmap in features.txt, because removing it
would require considerable work in the manual and the translations.
Main/IRQ from: 7.5/0.75 to: 7.25/1.0
With the reduction of the opus codec stack usage, giving the IRQ stack
some additional breathing room is now possible.
This greatly increases the stability of SD card write operations.
(I suspect the underlying problem is not IRQ operation itself, instead
being exacerbated by it..)
This allows targets to automatically switch audio settings when the
line out is plugged/unplugged.
Only hooked up on the xDuoo X3, but there are other potential users.
7.50K main stack
0.75K irq stack
Prior values of 8K+1K overflowed IRAM by ~660 bytes on Onda 7x7 targets,
but worked on the xDuoo X3.
(The discrepancy is due to the Onda targets having more LCD code shoved
Should prevent the IRQ-related stack overflows seen on MIPS targets.
Also enable USB bootloader mode
It actually works for transferring data now!
Igor Poretsky gets full credit for this patch.
* Fully Interrupt-driven, with proper task yielding
* Much more robust error handling
* Eliminate duplicate code
* Pile of bugfixes
(Much of this adapted from Igor Poretsky's tree)
Silences some annoying clicks.
* Better multidrive support
* Common slot1/slot2 handling code
Cleaned up, rebased, and forward-ported from the xvortex fork.
(original credit to firstname.lastname@example.org)
Signed-off-by: Solomon Peachy <email@example.com>
* Editing a bunch of drivers' thread routines in order to
implement a new feature is tedious.
* No matter the number of storage drivers, they share one thread.
No extra threads needed for CONFIG_STORAGE_MULTI.
* Each has an event callback called by the storage thread.
* A default callback is provided to fake sleeping in order to
trigger idle callbacks. It could also do other default processing.
Changes to it will be part of driver code without editing each
* Drivers may sleep and wake as they please as long as they give
a low pulse on their storage bit to ask to go into sleep mode.
Idle callback is called on its behalf and driver immediately put
into sleep mode.
* Drivers may indicate they are to continue receiving events in
USB mode, otherwise they receve nothing until disconnect (they
do receive SYS_USB_DISCONNECTED no matter what).
* Rework a few things to keep the callback implementation sane
and maintainable. ata.c was dreadful with all those bools; make
it a state machine and easier to follow. Remove last_user_activity;
it has no purpose that isn't served by keeping the disk active
through last_disk_activity instead.
* Even-out stack sizes partly because of a lack of a decent place
to define them by driver or SoC or whatever; it doesn't seem too
critical to do that anyway. Many are simply too large while at
least one isn't really adequate. They may be individually
overridden if necessary (figure out where). The thread uses the
greatest size demanded. Newer file code is much more frugal with
stack space. I barely see use crack 50% after idle callbacks
(usually mid-40s). Card insert/eject doesn't demand much.
* No forcing of idle callbacks. If it isn't necessary for one or
more non-disk storage types, it really isn't any more necessary for
disk storage. Besides, it makes the whole thing easier to implement.
_remote_backlight_* and _buttonlight_* are cleaned as well
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Reviewed-by: Michael Sevakis <firstname.lastname@example.org>
Tested: Michael Sevakis <email@example.com>
Hopefyly this time is the final one.
I must have been blind and copied some intermediate file
instead of final version. No functional change, only fixes.
1) Avoid load/store delay slot by reorganizing instructions
in copy loops
2) Fix off-by-one error in cache initialization code.
This was harmless as it simply set line0 twice, now it sets
every cacheline only once.
3) Fix off-by-word error in .bss clearing loop.
The addiu in branch delay slot even if calculated is not seen
by the branch instruction itself, so the code did one word
too much in clearing.
4) Fix off-by-word error in deadbeefing stack.
Target that have a touchpad/touchscreen should disable it while
being locked (In order to avoid LCD to drain battery power due to
"key locked" constant reporting messages. If they a have a keylock
button this was already handled at driver level. If not (e.g. fuze+),
they will have to implement a switch at driver level that action.c
can operate on softlock.
This patch does the following for any target having a touchpad
or a touchscreen and no HAS_BUTTON_HOLD (ie any softlock target)
1) it implements the code to call button_enable_touch(bool en) in
2) button_enable_touch is implemented in button.c and call
either touchpad_enable or touchscreen_enable
3) those two function are implemented respectively in touchscreen.c
and a new touchpad.c file. They provide a generic way to silents touch's
device and call a function at driver level where target specific code
can be implemented if possible/needed (for power saving for instance).
Those function name are touchpad_enable_device and touchscreen_enable_device
4) we implement an empty function at driver level of targets that need it
to have them still being able to compiled.
Reviewed-by: Thomas Martitz <firstname.lastname@example.org>
Reviewed-by: Amaury Pouly <email@example.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
* SOUND_x enum can be generated by audiohw_settings.h along with settings
entries and sound_val2phys.
* VOLUME_MIN and VOLUME_MAX are no longer necessary within sound.c. If
you need them, they are for target-defined purposes.
* Fix up SDL volume implementation in sdl.c. Move sim volume calculation
code to pcm-sdl.c.
* Min trigger tresholds were based upon VOLUME_MIN for some reason.
These setting have nothing to do with playback volume. Since it is no
longer present, set these at -89dB which is the minimum peak meter
* Fix an oversight in wm8758.c. I forgot to add the dB->register
conversion to audiohw_set_volume.
This is going right in since it's long overdue. If anything is goofed,
drop me a line or just tweak it yourself if you know what's wrong. :-)
Make HW/SW codec interface more uniform when emulating HW functionality
on SWCODEC for functions such as "audiohw_set_pitch". The firmware-to-
DSP plumbing is in firmware/drivers/audiohw-swcodec.c. "sound_XXX"
APIs are all in sound.c with none in DSP code any longer.
Reduce number of settings definitions needed by each codec by providing
defaults for common ones like balance, channels and SW tone controls.
Remove need for separate SIM code and tables and add virtual codec header
for hosted targets.
Implements double-buffered volume, balance and prescaling control in
the main PCM driver when HAVE_SW_VOLUME_CONTROL is defined ensuring
that all PCM is volume controlled and level changes are low in latency.
Supports -73 to +6 dB using a 15-bit factor so that no large-integer
math is needed.
Low-level hardware drivers do not have to implement it themselves but
parameters can be changed (currently defined in pcm-internal.h) to work
best with a particular SoC or to provide different volume ranges.
Volume and prescale calls should be made in the codec driver. It should
appear as a normal hardware interface. PCM volume calls expect .1 dB
Reviewed-by: Michael Sevakis <firstname.lastname@example.org>
Tested-by: Michael Sevakis <email@example.com>
This was spotted while playing with qemu-jz:
1) rockbox reads TECR and TESR which are described as write-only
registers. Datasheet doesn't mention what happens if they are
readed. Apparently this doesn't have fatal side effects.
It comes down to two defines from jz4740.h
__tcu_stop_counter(n) and __tcu_start_counter(n) which use
2) rockbox accesses out of bound offset 0xd4 in DMA memspace.
It comes from dis_irq() in system-jz4740.c. NUM_DMA is 6 but
DMA channels are 0-5 so (irq <= IRQ_DMA_0 + NUM_DMA)) bound
check is wrong.
This are *NOT* tested on device.
Reviewed-by: Bertrik Sikken <firstname.lastname@example.org>
Tested-by: Purling Nayuki <email@example.com>
Reviewed-by: Marcin Bukat <firstname.lastname@example.org>
Additional status callback is added to pcm_play/rec_data instead of
using a special function to set it. Status includes DMA error
reporting to the status callback. Playback and recording callback
become more alike except playback uses "const void **addr" (because
the data should not be altered) and recording uses "void **addr".
"const" is put in place throughout where appropriate.
Most changes are fairly trivial. One that should be checked in
particular because it isn't so much is telechips, if anyone cares to
bother. PP5002 is not so trivial either but that tested as working.
Reviewed-by: Michael Sevakis <email@example.com>
Tested-by: Michael Sevakis <firstname.lastname@example.org>
Change all lcd drivers to using a pointer to the static framebuffer
instead of directly accessing the static array. This will let us
later do fun things like dynamic framebuffer sizes (RaaA) or
ability to use different buffers for different layers (dynamic