PPC64: Add ffi support

src/lj_ccall.c

@@ -380,6 +380,37 @@
 #define CCALL_HANDLE_COMPLEXARG \
   /* Pass complex by value in 2 or 4 GPRs. */
 
+#if LJ_ARCH_PPC64
+#define CCALL_HANDLE_REGARG \
+  if (isva) {  /* only GPRs will be used on C ellipsis operator */ \
+    goto gpr; \
+  } \
+  else { \
+    if (isfp) {  /* Try to pass argument in FPRs. */ \
+      if (nfpr + 1 <= CCALL_NARG_FPR) { \
+	dp = &cc->fpr[nfpr]; \
+	nfpr += 1; \
+	ngpr += 1;  /* align GPRs */ \
+	d = ctype_get(cts, CTID_DOUBLE);  /* FPRs always hold doubles. */ \
+	goto done; \
+      } \
+    } else {  /* Try to pass argument in GPRs. */ \
+  gpr: \
+      if (n > 1) { \
+	lua_assert(n == 2 || n == 4);  /* int64_t or complex (float). */ \
+	if (ctype_isinteger(d->info)) \
+	  ngpr = (ngpr + 1u) & ~1u;  /* Align int64_t to regpair. */ \
+	else if (ngpr + n > maxgpr) \
+	  ngpr = maxgpr;  /* Prevent reordering. */ \
+      } \
+      if (ngpr + n <= maxgpr) { \
+	dp = &cc->gpr[ngpr]; \
+	ngpr += n; \
+	goto done; \
+      } \
+    } \
+  }
+#else	/* 32 bits */
 #define CCALL_HANDLE_REGARG \
   if (isfp) {  /* Try to pass argument in FPRs. */ \
     if (nfpr + 1 <= CCALL_NARG_FPR) { \
@@ -402,6 +433,7 @@
       goto done; \
     } \
   }
+#endif
 
 #define CCALL_HANDLE_RET \
   if (ctype_isfp(ctr->info) && ctr->size == sizeof(float)) \

src/lj_ccall.h

@@ -85,12 +85,21 @@ typedef union FPRArg {
 
 #elif LJ_TARGET_PPC
 
+#if LJ_ARCH_PPC64
+#define CCALL_NARG_GPR		8
+#define CCALL_NARG_FPR		8
+#define CCALL_NRET_GPR		4	/* For complex double. */
+#define CCALL_NRET_FPR		1
+#define CCALL_SPS_EXTRA		14
+#define CCALL_SPS_FREE		0
+#else
 #define CCALL_NARG_GPR		8
 #define CCALL_NARG_FPR		8
 #define CCALL_NRET_GPR		4	/* For complex double. */
 #define CCALL_NRET_FPR		1
 #define CCALL_SPS_EXTRA		4
 #define CCALL_SPS_FREE		0
+#endif
 
 typedef intptr_t GPRArg;
 typedef double FPRArg;

src/lj_ccallback.c

@@ -61,7 +61,11 @@ static MSize CALLBACK_OFS2SLOT(MSize ofs)
 
 #elif LJ_TARGET_PPC
 
+#if LJ_ARCH_PPC64
+#define CALLBACK_MCODE_HEAD		40
+#else  /* PPC 32bits */
 #define CALLBACK_MCODE_HEAD		24
+#endif
 
 #elif LJ_TARGET_MIPS
 
@@ -189,10 +193,21 @@ static void callback_mcode_init(global_State *g, uint32_t *page)
   uint32_t *p = page;
   void *target = (void *)lj_vm_ffi_callback;
   MSize slot;
+#if LJ_ARCH_PPC64
+  *p++ = PPCI_LI | PPCF_T(RID_TMP) | ((((intptr_t)target) >> 32) & 0xffff);
+  *p++ = PPCI_LI | PPCF_T(RID_R12) | ((((intptr_t)g) >> 32) & 0xffff);
+  *p++ = PPCI_RLDICR | PPCF_T(RID_TMP) | PPCF_A(RID_TMP) | PPCF_SH(32) | PPCF_M6(63-32);  /* sldi */
+  *p++ = PPCI_RLDICR | PPCF_T(RID_R12) | PPCF_A(RID_R12) | PPCF_SH(32) | PPCF_M6(63-32);  /* sldi */
+  *p++ = PPCI_ORIS | PPCF_A(RID_TMP) | PPCF_T(RID_TMP) | ((((intptr_t)target) >> 16) & 0xffff);
+  *p++ = PPCI_ORIS | PPCF_A(RID_R12) | PPCF_T(RID_R12) | ((((intptr_t)g) >> 16) & 0xffff);
+  *p++ = PPCI_ORI | PPCF_A(RID_TMP) | PPCF_T(RID_TMP) | (((intptr_t)target) & 0xffff);
+  *p++ = PPCI_ORI | PPCF_A(RID_R12) | PPCF_T(RID_R12) | (((intptr_t)g) & 0xffff);
+#else  /* PPC 32bits */
   *p++ = PPCI_LIS | PPCF_T(RID_TMP) | (u32ptr(target) >> 16);
   *p++ = PPCI_LIS | PPCF_T(RID_R12) | (u32ptr(g) >> 16);
   *p++ = PPCI_ORI | PPCF_A(RID_TMP)|PPCF_T(RID_TMP) | (u32ptr(target) & 0xffff);
   *p++ = PPCI_ORI | PPCF_A(RID_R12)|PPCF_T(RID_R12) | (u32ptr(g) & 0xffff);
+#endif
   *p++ = PPCI_MTCTR | PPCF_T(RID_TMP);
   *p++ = PPCI_BCTR;
   for (slot = 0; slot < CALLBACK_MAX_SLOT; slot++) {

src/lj_target_ppc.h

@@ -1,5 +1,5 @@
 /*
-** Definitions for PPC CPUs.
+** Definitions for PPC/PPC64 CPUs.
 ** Copyright (C) 2005-2016 Mike Pall. See Copyright Notice in luajit.h
 */
 
@@ -131,6 +131,8 @@ static LJ_AINLINE uint32_t *exitstub_trace_addr_(uint32_t *p, uint32_t exitno)
 #define PPCF_C(r)	((r) << 6)
 #define PPCF_MB(n)	((n) << 6)
 #define PPCF_ME(n)	((n) << 1)
+#define PPCF_SH(n)	((((n) & 31) << (11+1)) | (((n) & 32) >> (5-1)))
+#define PPCF_M6(n)	((((n) & 31) << (5+1)) | (((n) & 32) << (11-5)))
 #define PPCF_Y		0x00200000
 #define PPCF_DOT	0x00000001
 
@@ -200,6 +202,13 @@ typedef enum PPCIns {
   PPCI_RLWINM = 0x54000000,
   PPCI_RLWIMI = 0x50000000,
 
+  PPCI_RLDICL = 0x78000000,
+  PPCI_RLDICR = 0x78000004,
+  PPCI_RLDIC = 0x78000008,
+  PPCI_RLDIMI = 0x7800000c,
+  PPCI_RLDCL = 0x78000010,
+  PPCI_RLDCR = 0x78000012,
+
   PPCI_B = 0x48000000,
   PPCI_BL = 0x48000001,
   PPCI_BC = 0x40800000,