arm64: 64bit iOS and SMP support (#7942)

Signed-off-by: Austin Seipp <austin@well-typed.com>

3 files changed, 25 insertions, 13 deletions

diff --git a/rts/StgCRun.c b/rts/StgCRun.c
index f649dbe9fb..55e0a48acf 100644
--- a/rts/StgCRun.c
+++ b/rts/StgCRun.c
@@ -756,22 +756,28 @@ StgRun(StgFunPtr f, StgRegTable *basereg) {
     __asm__ volatile (
         /*
          * save callee-saves registers on behalf of the STG code.
+         * floating point registers only need the bottom 64 bits preserved.
+         * x16 and x17 are ip0 and ip1, but we can't refer to them by that name with clang.
          */
+        "stp fp,  lr,  [sp, #-16]!\n\t"
+        "mov fp, sp\n\t"
+        "stp x16, x17, [sp, #-16]!\n\t"
         "stp x19, x20, [sp, #-16]!\n\t"
         "stp x21, x22, [sp, #-16]!\n\t"
         "stp x23, x24, [sp, #-16]!\n\t"
         "stp x25, x26, [sp, #-16]!\n\t"
         "stp x27, x28, [sp, #-16]!\n\t" 
-        "stp ip0, ip1, [sp, #-16]!\n\t"
-        "str lr, [sp, #-8]!\n\t"
+        "stp d8,  d9,  [sp, #-16]!\n\t"
+        "stp d10, d11, [sp, #-16]!\n\t"
+        "stp d12, d13, [sp, #-16]!\n\t"
+        "stp d14, d15, [sp, #-16]!\n\t"
 
         /*
          * allocate some space for Stg machine's temporary storage.
          * Note: RESERVER_C_STACK_BYTES has to be a round number here or
          * the assembler can't assemble it.
          */
-        "str lr, [sp, %3]"
-        /* "sub sp, sp, %3\n\t" */
+        "sub sp, sp, %3\n\t"
         /*
          * Set BaseReg
          */
@@ -779,16 +785,17 @@ StgRun(StgFunPtr f, StgRegTable *basereg) {
         /*
          * Jump to function argument.
          */
-        "bx %1\n\t"
+        "br %1\n\t"
 
         ".globl " STG_RETURN "\n\t"
+#if !defined(ios_HOST_OS)
         ".type " STG_RETURN ", %%function\n"
+#endif
         STG_RETURN ":\n\t"
         /*
          * Free the space we allocated
          */
-        "ldr lr, [sp], %3\n\t"
-        /* "add sp, sp, %3\n\t" */
+        "add sp, sp, %3\n\t"
         /*
          * Return the new register table, taking it from Stg's R1 (ARM64's R22).
          */
@@ -796,18 +803,23 @@ StgRun(StgFunPtr f, StgRegTable *basereg) {
         /*
          * restore callee-saves registers.
          */
-        "ldr lr, [sp], #8\n\t"
-        "ldp ip0, ip1, [sp], #16\n\t"
+
+        "ldp d14, d15, [sp], #16\n\t"
+        "ldp d12, d13, [sp], #16\n\t"
+        "ldp d10, d11, [sp], #16\n\t"
+        "ldp d8,  d9,  [sp], #16\n\t"
         "ldp x27, x28, [sp], #16\n\t"
         "ldp x25, x26, [sp], #16\n\t"
         "ldp x23, x24, [sp], #16\n\t"
         "ldp x21, x22, [sp], #16\n\t"
         "ldp x19, x20, [sp], #16\n\t"
+        "ldp x16, x17, [sp], #16\n\t"
+        "ldp fp,  lr,  [sp], #16\n\t"
 
       : "=r" (r)
       : "r" (f), "r" (basereg), "i" (RESERVED_C_STACK_BYTES)
         : "%x19", "%x20", "%x21", "%x22", "%x23", "%x24", "%x25", "%x26", "%x27", "%x28",
-          "%ip0", "%ip1", "%lr"
+          "%x16", "%x17", "%lr"
     );
     return r;
 }
diff --git a/rts/posix/OSMem.c b/rts/posix/OSMem.c
index 8a6a85e3ad..359df7022b 100644
--- a/rts/posix/OSMem.c
+++ b/rts/posix/OSMem.c
@@ -32,7 +32,7 @@
 
 #include <errno.h>
 
-#if darwin_HOST_OS
+#if darwin_HOST_OS || ios_HOST_OS
 #include <mach/mach.h>
 #include <mach/vm_map.h>
 #include <sys/sysctl.h>
diff --git a/rts/sm/Storage.c b/rts/sm/Storage.c
index afb171b568..7174425e04 100644
--- a/rts/sm/Storage.c
+++ b/rts/sm/Storage.c
@@ -1168,7 +1168,7 @@ calcNeeded (rtsBool force_major, memcount *blocks_needed)
          should be modified to use allocateExec instead of VirtualAlloc.
    ------------------------------------------------------------------------- */
 
-#if defined(arm_HOST_ARCH) && defined(ios_HOST_OS)
+#if (defined(arm_HOST_ARCH) || defined(aarch64_HOST_ARCH)) && defined(ios_HOST_OS)
 void sys_icache_invalidate(void *start, size_t len);
 #endif
 
@@ -1180,7 +1180,7 @@ void flushExec (W_ len, AdjustorExecutable exec_addr)
   /* x86 doesn't need to do anything, so just suppress some warnings. */
   (void)len;
   (void)exec_addr;
-#elif defined(arm_HOST_ARCH) && defined(ios_HOST_OS)
+#elif (defined(arm_HOST_ARCH) || defined(aarch64_HOST_ARCH)) && defined(ios_HOST_OS)
   /* On iOS we need to use the special 'sys_icache_invalidate' call. */
   sys_icache_invalidate(exec_addr, ((unsigned char*)exec_addr)+len);
 #elif defined(__GNUC__)