Skip to content

Add aarch64 support #12

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Add aarch64 support #12

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
2fc402d
Added aarch64 support
rickvantooster Feb 3, 2025
e7bd9b1
Fixed wrong SM in sleep_write
rickvantooster Feb 3, 2025
15983ff
Added guard around fd cast and added code to save the v8-v15 registers
rickvantooster Feb 5, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
127 changes: 127 additions & 0 deletions coroutine.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -84,10 +84,32 @@ typedef enum {

// Linux x86_64 call convention
// %rdi, %rsi, %rdx, %rcx, %r8, and %r9
// Linux aarch64 call convention
// r0, r1, r2, r3, r4, r5, r6 and r7

void __attribute__((naked)) coroutine_yield(void)
{
// @arch
#ifdef __aarch64__

asm(

" sub sp, sp, #240\n"
" stp q8, q9, [sp,#0]\n"
" stp q10, q11, [sp,#32]\n"
" stp q12, q13, [sp,#64]\n"
" stp q14, q15, [sp,#96]\n"
" stp x19, x20, [sp,#128]\n"
" stp x21, x22, [sp,#144]\n"
" stp x23, x24, [sp,#160]\n"
" stp x25, x26, [sp,#176]\n"
" stp x27, x28, [sp,#192]\n"
" stp x29, x30, [sp,#208]\n"
" mov x0, sp\n"
" mov x1, #0\n" // sm = SM_NONE
" b coroutine_switch_context\n");

#else
asm(
" pushq %rdi\n"
" pushq %rbp\n"
Expand All @@ -99,12 +121,35 @@ void __attribute__((naked)) coroutine_yield(void)
" movq %rsp, %rdi\n" // rsp
" movq $0, %rsi\n" // sm = SM_NONE
" jmp coroutine_switch_context\n");
#endif
}

void __attribute__((naked)) coroutine_sleep_read(int fd)
{
#ifndef __ANDROID__
(void) fd;
#endif

// @arch

#ifdef __aarch64__
asm(
" sub sp, sp, #240\n"
" stp q8, q9, [sp,#0]\n"
" stp q10, q11, [sp,#32]\n"
" stp q12, q13, [sp,#64]\n"
" stp q14, q15, [sp,#96]\n"
" stp x19, x20, [sp,#128]\n"
" stp x21, x22, [sp,#144]\n"
" stp x23, x24, [sp,#160]\n"
" stp x25, x26, [sp,#176]\n"
" stp x27, x28, [sp,#192]\n"
" stp x29, x30, [sp,#208]\n"
" mov x2, x0\n"
" mov x0, sp\n"
" mov x1, #1\n" // sm = SM_READ
" b coroutine_switch_context\n");
#else
asm(
" pushq %rdi\n"
" pushq %rbp\n"
Expand All @@ -117,12 +162,33 @@ void __attribute__((naked)) coroutine_sleep_read(int fd)
" movq %rsp, %rdi\n" // rsp
" movq $1, %rsi\n" // sm = SM_READ
" jmp coroutine_switch_context\n");
#endif
}

void __attribute__((naked)) coroutine_sleep_write(int fd)
{
#ifndef __ANDROID__
(void) fd;
#endif
// @arch
#ifdef __aarch64__
asm(
" sub sp, sp, #240\n"
" stp q8, q9, [sp,#0]\n"
" stp q10, q11, [sp,#32]\n"
" stp q12, q13, [sp,#64]\n"
" stp q14, q15, [sp,#96]\n"
" stp x19, x20, [sp,#128]\n"
" stp x21, x22, [sp,#144]\n"
" stp x23, x24, [sp,#160]\n"
" stp x25, x26, [sp,#176]\n"
" stp x27, x28, [sp,#192]\n"
" stp x29, x30, [sp,#208]\n"
" mov x2, x0\n"
" mov x0, sp\n"
" mov x1, #2\n" // sm = SM_WRITE
" b coroutine_switch_context\n");
#else
asm(
" pushq %rdi\n"
" pushq %rbp\n"
Expand All @@ -135,12 +201,33 @@ void __attribute__((naked)) coroutine_sleep_write(int fd)
" movq %rsp, %rdi\n" // rsp
" movq $2, %rsi\n" // sm = SM_WRITE
" jmp coroutine_switch_context\n");
#endif
}

void __attribute__((naked)) coroutine_restore_context(void *rsp)
{
// @arch
(void)rsp;
#ifdef __aarch64__

asm(
" mov sp, x0\n"
" ldp q8, q9, [sp,#0]\n"
" ldp q10, q11, [sp,#32]\n"
" ldp q12, q13, [sp,#64]\n"
" ldp q14, q15, [sp,#96]\n"
" ldp x19, x20, [sp,#128]\n"
" ldp x21, x22, [sp,#144]\n"
" ldp x23, x24, [sp,#160]\n"
" ldp x25, x26, [sp,#176]\n"
" ldp x27, x28, [sp,#192]\n"
" ldp x29, x30, [sp,#208]\n"
" mov x1, x30\n"
" ldr x30, [sp, #224]\n"
" ldr x0, [sp, #232]\n"
" add sp, sp, #240\n"
" ret x1\n");
#else
asm(
" movq %rdi, %rsp\n"
" popq %r15\n"
Expand All @@ -151,6 +238,7 @@ void __attribute__((naked)) coroutine_restore_context(void *rsp)
" popq %rbp\n"
" popq %rdi\n"
" ret\n");
#endif
}

void coroutine_switch_context(void *rsp, Sleep_Mode sm, int fd)
Expand Down Expand Up @@ -251,6 +339,44 @@ void coroutine_go(void (*f)(void*), void *arg)

void **rsp = (void**)((char*)contexts.items[id].stack_base + STACK_CAPACITY);
// @arch
#ifdef __aarch64__
*(--rsp) = arg;
*(--rsp) = coroutine__finish_current;
*(--rsp) = f; // push r0



*(--rsp) = 0; // push r29
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't you also need to reserve space for r30 here, since the load will read both r29 and r30?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Don't you also need to reserve space for r30 here, since the load will read both r29 and r30?

r30 is the same as the link register so the line

*(--rsp) = coroutine__finish_current;

will already reserve space for it.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok, yeah, we don't need to preserve it's value, so we can just read some garbage into it, instead of the previous value. Now but what about storing to the stack, wouldn't it override *(--rsp) = f; // push r0?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I am not sure if I am following you. So far everything runs perfectly fine on the raspberry pi 4 model B.

*(--rsp) = 0; // push r28
*(--rsp) = 0; // push r27
*(--rsp) = 0; // push r26
*(--rsp) = 0; // push r25
*(--rsp) = 0; // push r24
*(--rsp) = 0; // push r23
*(--rsp) = 0; // push r22
*(--rsp) = 0; // push r21
*(--rsp) = 0; // push r20
*(--rsp) = 0; // push r19

*(--rsp) = 0; // push v15
*(--rsp) = 0;
*(--rsp) = 0; // push v14
*(--rsp) = 0;
*(--rsp) = 0; // push v13
*(--rsp) = 0;
*(--rsp) = 0; // push v12
*(--rsp) = 0;
*(--rsp) = 0; // push v11
*(--rsp) = 0;
*(--rsp) = 0; // push v10
*(--rsp) = 0;
*(--rsp) = 0; // push v09
*(--rsp) = 0;
*(--rsp) = 0; // push v08
*(--rsp) = 0;

#else

*(--rsp) = coroutine__finish_current;
*(--rsp) = f;
*(--rsp) = arg; // push rdi
Expand All @@ -260,6 +386,7 @@ void coroutine_go(void (*f)(void*), void *arg)
*(--rsp) = 0; // push r13
*(--rsp) = 0; // push r14
*(--rsp) = 0; // push r15
#endif
contexts.items[id].rsp = rsp;

da_append(&active, id);
Expand Down