Add dma_retain() and dma_release() to reference count the users
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19344 a1c6a512-1295-4272-9138-f99709370657
Note that on low memory targets (Clip/m200v4 tested) you will encounter
random crashes. Applying FS#9332 seems to help a lot.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19342 a1c6a512-1295-4272-9138-f99709370657
Correct the logic when enabling interrupt line (multivolume)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19337 a1c6a512-1295-4272-9138-f99709370657
* Adds a callback to be called on end of transfer
* Add a function to disable a channel
* Services the 2 channels if both are active in the isr
SD driver: panics on error
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19333 a1c6a512-1295-4272-9138-f99709370657
Reorder system_init() to initialize peripherals not only in bootloader
Use a 65MHz PCLK (and memclk) which will be needed for realtime decoding
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19330 a1c6a512-1295-4272-9138-f99709370657
It is read separately, and so backlight/buttonlight can be disabled
while hold switch is on
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19302 a1c6a512-1295-4272-9138-f99709370657
We set directly the needed bits to write into DBOP_DOUT (15:13 and 3:0)
Since we can't set the mask 0xf00f with one mov instruction, revert the
logic and use orrne instead of biceq (in the whole function for consistency)
Fix suggested by Jens Arnold
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19279 a1c6a512-1295-4272-9138-f99709370657
Fix lcd drivers which stopped working after changing the PLL.
Move set_cpu_frequency() to a place where it is actually used.
Remove enable_irq() call already done by the bootloader
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19276 a1c6a512-1295-4272-9138-f99709370657
Use a 24MHz peripheral clock instead of 64MHz for battery life
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19234 a1c6a512-1295-4272-9138-f99709370657
We don't need to mess with CCU_IO register (it's only needed for targets
with a SD slot) so don't keep its value when changing buttonlight
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19139 a1c6a512-1295-4272-9138-f99709370657
A button press of ~10s will trigger the hardware power off instead of ~1s
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19123 a1c6a512-1295-4272-9138-f99709370657
- Make write_sectors() function fail.
- Disable interrupts while transferring data and do not yield(), so we are sure the FIFO is not overflowed
Note this solution is only temporary since it's not friendly to other threads and confuse kernel tick precision
This will be reverted when we will be using DMA to access the SD card, but for now it permits further development
- PL180: Rename the MMC_* registers into MCI_*, to not make people believe it is a MMC only controller
- Supports non aligned destination buffers when reading
- Correct the timeout units which were lamely copied from ata-sd-pp.c and were in milliseconds (note that the timeouts are disabled now)
- Higher a bit the stack size
- Use the full initialization procedure in the bootloader and the loaded firmware
- Use the CCU_IO register only when a SD slot is present
- Put some panicf() around to catch problems
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19122 a1c6a512-1295-4272-9138-f99709370657
Move the ADC defines to as3514.h, and make adc-target.h only include as3514.h
Implement the missing API (ascodec_readbytes, ascodec_(un)lock)
Revert the changes to the PP-specific arm/ascodec-target.h in r19073,
use a AS3525 specific ascodec-target.h while moving the AS3514 specific
code in export/as3514.h
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@19076 a1c6a512-1295-4272-9138-f99709370657
