it's only needed if we hotswap the µSD or if we build the fuzev2
scrollwheel code without irq (which we don't do anyway)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26165 a1c6a512-1295-4272-9138-f99709370657
test_disk doesn't freeze anymore when closing the file between writing & reading
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26164 a1c6a512-1295-4272-9138-f99709370657
The list is ordered by priority, put the DMA and SD interrupts at
the top
This seems to fix the random freezes on heavy storage operation like
building database or pictureflow cache
The recently added audio interrupt had a priority higher than SD slot on
as3525v1, give it a lower priority
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26160 a1c6a512-1295-4272-9138-f99709370657
Show all the adc channels in the IO ports debug menu (not for as3525v2
since the channels are different)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26149 a1c6a512-1295-4272-9138-f99709370657
build it on fuzev2 and clipv2 too, it should be the same controller
as3525v1 file is only build for CONFIG_USBOTG == USBOTG_AS3525, no need
to check for it in the .c file
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26142 a1c6a512-1295-4272-9138-f99709370657
Note to self: Next time don't forget to check a AMSv2 build before committing...
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26130 a1c6a512-1295-4272-9138-f99709370657
- Enable end of charge monitoring once, it doesn't need to be disabled
- Acknowledge the first (wrong) end of charge interrupt on charger enable
(this had been broken in r25299)
- Centralize reads to ENRD* registers and cache the results when needed
- on PP it is not needed because reads are atomic, we only check for
end of charge when the charging, and for charger presence when
discharging
as3525v2 (using as3543) specifics
- I got the datasheet today from AMS, thanks to them for being so fast
and not require me to sign tons of papers!
- USB detection now works on as3525v2 using the as3543. Clip+ won't
reboot to OF yet, it needs mkamsboot support first (usbstack disabled)
- Charging should work, the CHARGER register is at a different place, it
is an extended PMU register -> use ascodec_read/write_charger() to
access it
- real interrupts are not used yet for ENRD, we get thousands of
interrupts per second, apparently only limited by the i2c clock.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26116 a1c6a512-1295-4272-9138-f99709370657
Better be safe than sorry, don't try to read/write outside our storage,
because we might overwrite the OF on the internal storage
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26077 a1c6a512-1295-4272-9138-f99709370657
-> Change references to as3517 into as3515
there is a reference to the as3515 in the public as3525 datasheet to
support this fact
as3517 is used in the as3527 SoC, the (public) as3527 datasheet has
detailed information on its registers
the audio codec/PMU used in as3525v2 (thought to be as3543) appears to be
compatible (identical?) with as3517
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26075 a1c6a512-1295-4272-9138-f99709370657
it would still flash when the light was off
it turns out that if B5 is set to input the light can't be turned on
but we can still select between µSD slot and internal storage
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26059 a1c6a512-1295-4272-9138-f99709370657
adjust logical starting sector to physical starting sector after the check
number of blocks is already adjusted from physical to logical
don't panic anymore when accessing the last part of storage
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26050 a1c6a512-1295-4272-9138-f99709370657
Indent the asm constraints at the same level than instructions
Also add a trick to reduce the number of instructions outputted by gcc
in the commented C version of the loop
The difference between C and asm is now 1 instruction
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26027 a1c6a512-1295-4272-9138-f99709370657
This fixes the error introduced in r25989, where it would release the
DMA engine when uninitializing recording.
The real bug has been fixed instead (in pitch_detector)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@26003 a1c6a512-1295-4272-9138-f99709370657
also avoid disabling completely the DMA engine in pitch_detector
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25989 a1c6a512-1295-4272-9138-f99709370657
The OF does this and while we don't have the documentation to be sure I think it's highly likely this is a low power mode referred to in one of the product "briefs".
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25936 a1c6a512-1295-4272-9138-f99709370657
This seems to make switching the uSD card to high speed timing much more reliable.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25935 a1c6a512-1295-4272-9138-f99709370657
When the rest of the code is confirmed to work properly without CPU
frequency changes, it will be possible to reliabily test set_cpu_frequency()
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25924 a1c6a512-1295-4272-9138-f99709370657
Get a bit more battery life (between 30 minutes and 2 hours)
as3525v2: only enabled in bootloader, we need a new binary release anwyway
as3525v1: enabled in rockbox.sansa as well, so we can continue using the
current bootloaders
Flyspray: FS#11246
Author: myself
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25900 a1c6a512-1295-4272-9138-f99709370657
This is to a) to cleanup firmware/common and firmware/include a bit, but also b) for Rockbox as an application which should use the host system's c library and headers, separating makes it easy to exclude our files from the build.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25850 a1c6a512-1295-4272-9138-f99709370657
Some SD cards were having problems switching to 4 Bit widebus and this solution appears to remedy that.
Thanks to Luca_S!
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25846 a1c6a512-1295-4272-9138-f99709370657
These INT's shouldn't be unmasked until the transfer is just about to happen.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25836 a1c6a512-1295-4272-9138-f99709370657
all as3525v2 use ADC_CHG_IN because ADC_BVDD is way too high when charging
Fuzev2 now displays battery icon correctly when charging
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25815 a1c6a512-1295-4272-9138-f99709370657
Change SD_SELECT/DESELECT_CARD commands to MCI_NO_RESP as we don't need a response here.
Renumber some command errors.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25811 a1c6a512-1295-4272-9138-f99709370657
Also enable write support, test_disk passes on all targets
Flyspray: FS#11140
Authors: Jack Halpin and myself
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25799 a1c6a512-1295-4272-9138-f99709370657
delays after modifying the registers seems not to be needed
moving RAM operation (cpu_frequency variable) before modifying the
register also seems to help
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25754 a1c6a512-1295-4272-9138-f99709370657
We do not need any delay after modifying it in system_init(), so the
same applies in set_cpu_frequency()
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25753 a1c6a512-1295-4272-9138-f99709370657
- remove udelay(1) and use the previous busy loop : the delay was too
long for proper hold detection
- remove the 2nd delay (unneeded)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25741 a1c6a512-1295-4272-9138-f99709370657
detect wraps so we don't miss the value we were waiting for
fix audio stuttering on fuzev2
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25740 a1c6a512-1295-4272-9138-f99709370657
Use it in fuzev2 to improve some big delays (correct the biggest one to actually wait for the fifo to empty), and use it in the sd drivers.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25734 a1c6a512-1295-4272-9138-f99709370657
Revive the delay removed by r25636 (and make it longer so it works when
boosted)
Also read GPIOA pins 7 & 6 in one load
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25722 a1c6a512-1295-4272-9138-f99709370657
Before the clock settings were fixed this setting actually caused the problem it now seems to solve.
Random freezups seem to be gone for clip+ at least.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25531 a1c6a512-1295-4272-9138-f99709370657
We only use DMA transfers so we should not need to change/reset the FIFO configuration.
Also add comment showing what we set as the TX/RX watermarks.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25507 a1c6a512-1295-4272-9138-f99709370657
XPD was being switched between SD-MMC interface and general i/o functions in the sd_enable() function similar to as3525v1.
Since the as3525v2's do not seem to use XPD as general i/o there's no need to switch it back.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25497 a1c6a512-1295-4272-9138-f99709370657
It is equally fast and doesn't need an additional mask
Thanks to amiconn for pointing to it
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25477 a1c6a512-1295-4272-9138-f99709370657
PCLK doesn't use PLLA as a source but FCLK, so when changing FCLK with
CGU_PROC register, we must change PCLK as well with CGU_PERI register
Operate with 24MHz PCLK (and unboosted FCLK) for Clipv2/Clip+
Use 60MHz on Fuzev2 to keep the display fast enough (still slower than
Fuzev1 though)
µSD seems to function correctly now
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25475 a1c6a512-1295-4272-9138-f99709370657
We can write to DBOP_DOUT in 8 bits mode : bits 7..0 are mapped to the
LCD, not bits 15..12 and 3..0 like on Clipv1
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25455 a1c6a512-1295-4272-9138-f99709370657
1 more hour of battery life measured on Clip+ : 16h30 with mp3 @192kbps
Fuzev2 frequency isn't changed because the scrollwheel is less
responsive below 60MHz
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25448 a1c6a512-1295-4272-9138-f99709370657
Brightness changes on the Fuzev2 now work flawlessly
This might also explain why my Clip+ would sometimes "not boot", that
could have been because the backlight was not enabled at all.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25445 a1c6a512-1295-4272-9138-f99709370657
Move scrollwheel parsing function into separate file as it's reused.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25425 a1c6a512-1295-4272-9138-f99709370657
the arm926-ejs doesn't have synchronous/asynchronous/fastbus modes, so
just change CGU_PROC directly
Note: we could use a lower unboosted frequency now
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25417 a1c6a512-1295-4272-9138-f99709370657
Write the setting before enabling the PLL
Fix booting problem (black screen) with Clip+
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25415 a1c6a512-1295-4272-9138-f99709370657
Instead of modifying CGU_PROC to get 24MHz pclk, just switch to fastbus else Clip+ wouldn't boot
Tested on fuzev1/Clip+
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25413 a1c6a512-1295-4272-9138-f99709370657
They're all the same when charger is unplugged, but when it's plugged:
- BVDD is way too high
- RTCSUP oscillates between 2 values
- CHG_IN acts like RTCSUP on some models but seems to works fine on my Clip+
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25405 a1c6a512-1295-4272-9138-f99709370657
I have also made the CMD_CHECK_CRC_BIT unused for now since we do not check any response crc values yet.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25343 a1c6a512-1295-4272-9138-f99709370657
The internal card does not appear to be HS capable, at least not in 2GB clip+
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25316 a1c6a512-1295-4272-9138-f99709370657
The controller only needs to be reset if we had an error to clean up any leftover trash...
Move comment pertaining to retry variable so it's actually nearby.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25315 a1c6a512-1295-4272-9138-f99709370657
Adjust the initial MCI_MASK value to also mask the MCI_INT_RXDR and MCI_INT_TXDR bits as it seems we don't use them for dma transfers.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25314 a1c6a512-1295-4272-9138-f99709370657
Move CLKDIV macros into clock-target.h.
Only enable the necessary interfaces for the 3 clock registers used for SD.
Add MEMSTICK and SDSLOT registers to bottom of register display in View HW info debug page.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25309 a1c6a512-1295-4272-9138-f99709370657
The registers value does not change so we don't need to read them
This avoids dividing by 0
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25303 a1c6a512-1295-4272-9138-f99709370657
We don't need the post transfer call this way. We check on TRAN state before each partial transfer.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25282 a1c6a512-1295-4272-9138-f99709370657
as3525 comes with a as3517, as3525v2 with a as3543, rename specific
registers accordingly
Existing problems: FM (line out) doesn't work, volume can't go below a
certain point
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25260 a1c6a512-1295-4272-9138-f99709370657
We do not need a stack pointer at this step, and we are sure to not use
memory not initialized yet (like bss)
Fixes FS#11114 (tested on Fuze)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25229 a1c6a512-1295-4272-9138-f99709370657
- buttonlight works
- backlight turns on (quite weak though and no brightness adjustment/backlight off functional)
- don't share drivers with e200v2/fuzev1 for now as it's not entirely clear how dbop plays into this
- deactivte scrollwheel as it's messing up the timer setup in kernel-as3525.c indicating the dbop input reading doesn't work well
- still no working bootloader/no lcd or buttons
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25225 a1c6a512-1295-4272-9138-f99709370657
Fixes some buttons not being read (hold is still buggy)
FlySpray: FS#11111
Author: Pascal Below
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25209 a1c6a512-1295-4272-9138-f99709370657
Especially when using caches we might read the response too fast and get
the response for the previous command instead.
Now Clip+ boots fine with both instruction & data caches enabled, the
delay might need to be lowered though: boot time is a bit longer.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25183 a1c6a512-1295-4272-9138-f99709370657
calc_freq(CLK_PLLA) for as3525v2 simply returns 240MHz instead of calculating until we understand how it is calculated.
Now displays 922T or 926ejs depending on version.
Remove SD and uSD info from as3525v2 as it is not useful due to different SD controller.
Increased the lines used for each page of display on clip screen size.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25120 a1c6a512-1295-4272-9138-f99709370657
Introduce a new .init section for initialisation code, so that it can be copied to an area which is later overwritten before calling. The stack/bss can then overwrite that code, effectively freeing the code size that the initialisation routines need. Gives a few kB ram usage back.
Only implemented for PP and as3525 so far. More targets could be added, as well as more functions.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25013 a1c6a512-1295-4272-9138-f99709370657
SD_ALL_SEND_CID was using cardinfo.cid to store the response instead of the temp array used for long responses.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24944 a1c6a512-1295-4272-9138-f99709370657
Compilation shows a warning for the first argument of the lcd_bitmap() call in show_logo().
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24911 a1c6a512-1295-4272-9138-f99709370657
Remove MCI_HCON read from init_controller() as it now appears unneccesary.
Make sd-init_card() use similar init sequence to the one in sd-as3525.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24903 a1c6a512-1295-4272-9138-f99709370657
I've left in some commented out code for now as this is still a bit experimental.....
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24891 a1c6a512-1295-4272-9138-f99709370657
Differences remaining:
- list of peripherals reset
- CGU_PROC isn't modified on as3525v2
- CGU_PLLA bits aren't known, but we use a known setting for 240MHz
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24868 a1c6a512-1295-4272-9138-f99709370657
Do not use a static variable for buttons, else they're never reset
Remove unneeded code
Move GPIO_DIR setting to init function
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24855 a1c6a512-1295-4272-9138-f99709370657
check all error bits
only signal wakeup on data transfers, not on commands
trim down send_cmd
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24851 a1c6a512-1295-4272-9138-f99709370657
not touching MCI_CTYPE (leaving bus width to 1) gives data transfers
ignore the hardware locked up while error bit for now
remove printf() helper & debug code
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24838 a1c6a512-1295-4272-9138-f99709370657
Disable errors on response timeout since it can happen on SD_SEND_IF_COND
Disable errors on start bit error : it's ignored by the linux driver
No panic on my side with those 2 bits unchecked, but no transfer
completion either.
Note: the Linux driver doesn't implement DMA
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24837 a1c6a512-1295-4272-9138-f99709370657
Reset DMA before transfers and check data transfer over bit in isr
Still no error or data transfer over conditions
Read the correct status register in isr : there is a masked interrupt
status register and a general status register
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24836 a1c6a512-1295-4272-9138-f99709370657
Initial button driver for clip+. I've followed the steps the OF uses to read the buttons.
There is no hold button on the clip+ so for now calling button_hold() siply returns false.
The OF actually sets an additional flag for the 2nd read from D6 and not BUTTON_POWER as I have done here.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24669 a1c6a512-1295-4272-9138-f99709370657
The code assumed LCD pixels were packed on 16 bits values but for the
Clip we use 8 bits values.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24563 a1c6a512-1295-4272-9138-f99709370657
Now making the Fuzev2 bootloader build should be pretty easy
TODO:
- write button driver (FlynDice found all buttons already)
- find button light
- decide if lcd-ssd1303.c must be modified for Clip+ using SSP or forked
- check if backlight code works (I copied Clipv2 code)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24520 a1c6a512-1295-4272-9138-f99709370657
Select it based on as3535/as3525v2 so Clip+/Fuzev2 will be able to use
it too
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24222 a1c6a512-1295-4272-9138-f99709370657
Voltage scaling seems to be causing various problems mostly related to issues with the uSD cards.
The increased runtime benefit only amounts to ~30 minutes as currently implemented so it seems prudent to disable it once again at this time.
We still don't understand why the core voltage being lowered would impact the uSD card but in fact it does. The internal cards do not seem to have problems.
I have simply #ifdef'd the voltage scaling code with HAVE_ADJUSTABLE_CPU_VOLTAGE so if you want to use voltage scaling simply define that and the voltage scaling code should run.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24217 a1c6a512-1295-4272-9138-f99709370657
Unify this optimized dbop transfer function and re-use it more often (it still handles 16bit transfers).
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24198 a1c6a512-1295-4272-9138-f99709370657
This gets rid of LCD glitches on Sansa Fuze, and now LCD transfers can
get interrupted by button reading
Flyspray: FS #10603
Author: Bertrik Sikken
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24192 a1c6a512-1295-4272-9138-f99709370657
- add lcd_set_invert()
- add lcd_set_flip() commented out since not working fine yet:
mpegplayer isn't flipped, and statusbar display is buggy
- use the registers list from e200v2
- cosmetics (indentation fix, comments changes, function names, casts,
function moves, ...) to make the diff with lcd-e200v2.c shorter
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24181 a1c6a512-1295-4272-9138-f99709370657
Reuse some code from Clip (LCD) and a lot of code from AS3525
Add a new CPU type : AS3525v2, identical to AS3525 except it's an ARMv5 (arm926-ejs)
SD code still not working
For an unknown reason LCD doesn't work anymore (to be investigated)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24131 a1c6a512-1295-4272-9138-f99709370657
For some reason 4 bit widebus is creating issues when writing to the .rockbox directory so revert 4 bit widebus and the revision to the write delay that was added as a fix.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24054 a1c6a512-1295-4272-9138-f99709370657
Making this delay apply to all non-HS SD cards seems to fix some data corruption issues that came up with the switch to 4-bit widebus.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@24019 a1c6a512-1295-4272-9138-f99709370657
