Files
tinygo/src/internal/task/task_stack_avr.S
T
Ayke van Laethem 907d90105a runtime: use the main (startup) stack for the main goroutine
Instead of always starting a new goroutine for the main goroutine, run
the main goroutine on the system stack.
The system stack is not occupied with scheduling, instead each goroutine
that wants to pause itself calls into the scheduler which will switch to
the next task (goroutine) to run, or sleeps.

There are various advantages of this over the previous system:

  * When the program doesn't start a goroutine, the code size and RAM
    consumption is close to what you'd get with `-scheduler=none`.
  * When the program does start a goroutine, there is still a reduction
    in RAM consumption because only one extra stack is needed.
  * Because tasks directly switch to the next task to run, only a single
    task switch is needed instead of two (goroutine -> scheduler ->
    goroutine). This should improve task switching performance.

I kept the current behavior for WebAssembly/Asyncify. I looked into how
the same benefits can be realized for WebAssembly but couldn't easily
find how to do that. Maybe this can be done separately, or maybe we'll
just wait for the stack switching proposal to finish.

The code for Cortex-M is currently more complicated than I'd like, and
therefore can sometimes result in a slight increase in code size. I'd
like to fix this eventually but am still looking into good ways to do
this. I still think this change is generally beneficial because many
programs see big reductions in code size when compiling for Cortex-M.
2022-07-30 01:57:45 +02:00

97 lines
2.5 KiB
ArmAsm

.section .text.tinygo_startTask
.global tinygo_startTask
.type tinygo_startTask, %function
tinygo_startTask:
// Small assembly stub for starting a goroutine. This is already run on the
// new stack, with the callee-saved registers already loaded.
// Most importantly, r2r3 contain the pc of the to-be-started function and
// r4r5 contain the only argument it is given. Multiple arguments are packed
// into one by storing them in a new allocation.
// Set the first argument of the goroutine start wrapper, which contains all
// the arguments.
movw r24, r4
// Branch to the "goroutine start" function. Note that the Z register is
// call-clobbered, so does not need to be restored after use.
movw Z, r2
icall
// After return, exit this goroutine. This is a tail call.
#if __AVR_ARCH__ == 2 || __AVR_ARCH__ == 25
// Small memory devices (8kB flash) that do not have the long call
// instruction availble will need to use rcall instead.
// Note that they will probably not be able to run more than the main
// goroutine anyway, but this file is compiled for all AVRs so it needs to
// compile at least.
rcall tinygo_pause
#else
// Other devices can (and must) use the regular call instruction.
call tinygo_pause
#endif
.global tinygo_swapTask
.type tinygo_swapTask, %function
tinygo_swapTask:
// This function gets the following parameters:
// r24:r25 = newStack uintptr
// r22:r23 = oldStack *uintptr
// Save all call-saved registers:
// https://gcc.gnu.org/wiki/avr-gcc#Call-Saved_Registers
push r29 // Y
push r28 // Y
push r17
push r16
push r15
push r14
push r13
push r12
push r11
push r10
push r9
push r8
push r7
push r6
push r5
push r4
push r3
push r2
// Save the current stack pointer in oldStack.
in r2, 0x3d; SPL
in r3, 0x3e; SPH
movw Y, r22
std Y+0, r2
std Y+1, r3
// Switch to the new stack pointer.
in r0, 0x3f ; SREG
cli
out 0x3d, r24; SPL
out 0x3f, r0 ; SREG, restore interrupts (after the next instruction)
out 0x3e, r25; SPH
// Load saved register from the new stack.
pop r2
pop r3
pop r4
pop r5
pop r6
pop r7
pop r8
pop r9
pop r10
pop r11
pop r12
pop r13
pop r14
pop r15
pop r16
pop r17
pop r28 // Y
pop r29 // Y
// Return into the new task, as if tinygo_swapTask was a regular call.
ret