| Commit message (Collapse) | Author | Age |
|---|
| |
|
|
|
|
| |
This makes doubles on the stack (as for var-args functions) work
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
| |
Support -Wcast-align and -Wpointer-arith while still allowing
architectures to pick whether they want an 8-bit or 32-bit stack.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
| |
This makes sure that doubles are aligned properly when passed on the stack.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
This driver uses the generic GPIO functions and allows per-LED port
and pin configuration. It supports up to 32 LEDs.
Rename SoC-specific LED drivers.
Remove enabled parameter to ao_led_init
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
| |
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
| |
Redefines some stdio bits so that we can build with either pdclib or
newlib + avr stdio.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Gcc 5.4.1 tracks alignment of data through assignments, so that a
uint32_t pointer which comes from byte-aligned uint8_t data:
extern uint8_t foo[];
uint32_t *q = (void *) foo;
Fetches and stores through this pointer are done bytewise. This is
slow (meh), but if q references a device register, things to bad very
quickly.
This patch works around this bug in the compiler by adding
__attribute__((aligned(4))) tags to some variables, or changing them
from uint8_t to uint32_t. Places doing this will now be caught as I've
added -Wcast-align to the compiler flags. That required adding (void
*) casts, after the relevant code was checked to make sure the
compiler could tell that the addresses were aligned.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
| |
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
|
|
| |
We use a separate stack pointer for task code, which means we can
verify that it is in range in any interrupt handler. This adds checks
for the task stack (under #ifdef DEBUG) that run in ao_wakeup as well
as at every timer interrupt.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Having arbitrary alarms firing in the middle of complicated device
logic makes no sense at all. Therefore only correct use of ao_alarm
and ao_clear_alarm was around a specific ao_sleep call, with correct
recovery in case the alarm fires.
This patch replaces all uses of ao_alarm/ao_sleep/ao_clear_alarm with
ao_sleep_for, a new function which takes the alarm timeout directly.
A few cases which weren't simply calling ao_sleep have been reworked
to pass the timeout value down to the place where sleep *is* being
called, and having that code deal with the return correctly.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
|
|
core remains a bad name to use -- dirvish skips files (and
directories, it seems) with that name.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
