fake pacga impl + tests build

.gitignore

@@ -1804,6 +1804,7 @@ configure~
 /none/tests/arm/vfpv4_fma
 
 # /none/tests/arm64/
+/none/tests/arm64/*.dSYM
 /none/tests/arm64/*.stderr.diff*
 /none/tests/arm64/*.stderr.out
 /none/tests/arm64/*.stdout.diff*

VEX/priv/guest_arm64_toIR.c

@@ -3819,7 +3819,7 @@ Bool dis_ARM64_data_processing_register(/*MB_OUT*/DisResult* dres,
       UInt dd = INSN(4,0);
       // see https://developer.arm.com/documentation/ddi0597/2023-12/Shared-Pseudocode/aarch64-functions-pac?lang=en#impl-aarch64.ComputePACIMPDEF.4
       // dd = compute_pac(dd, nn, mm | SP) & 0xFFFFFFFF00000000;
-      // FIXME: putIReg64orZR(dd, mkU64(0x101010FF00000000));
+      putIReg64orZR(dd, getIReg64orZR(nn));
       DIP("pacga %s, %s, %s (FAKED)\n", nameIRegOrZR(True, dd), nameIRegOrZR(True, nn), nameIRegOrSP(True, mm));
         return True;
     }

configure.ac

@@ -1480,7 +1480,8 @@ AM_CONDITIONAL(VGCONF_OS_IS_DARWIN_OR_FREEBSD,
                  -o x$VGCONF_PLATFORM_PRI_CAPS = xAMD64_FREEBSD \
                  -o x$VGCONF_PLATFORM_PRI_CAPS = xARM64_FREEBSD \
                  -o x$VGCONF_PLATFORM_PRI_CAPS = xX86_DARWIN \
-                 -o x$VGCONF_PLATFORM_PRI_CAPS = xAMD64_DARWIN)
+                 -o x$VGCONF_PLATFORM_PRI_CAPS = xAMD64_DARWIN \
+                 -o x$VGCONF_PLATFORM_PRI_CAPS = xARM64_DARWIN)
 
 
 # Sometimes, in the Makefile.am files, it's useful to know whether or not

helgrind/tests/tc07_hbl1.c

@@ -8,6 +8,7 @@
 
 #undef PLAT_x86_darwin
 #undef PLAT_amd64_darwin
+#undef PLAT_arm64_darwin
 #undef PLAT_x86_freebsd
 #undef PLAT_amd64_freebsd
 #undef PLAT_arm64_freebsd
@@ -26,6 +27,8 @@
 #  define PLAT_x86_darwin 1
 #elif defined(__APPLE__) && defined(__x86_64__)
 #  define PLAT_amd64_darwin 1
+#elif defined(__APPLE__) && defined(__aarch64__)
+#  define PLAT_arm64_darwin 1
 #elif defined(__FreeBSD__) && defined(__i386__)
 #  define PLAT_x86_freebsd 1
 #elif defined(__FreeBSD__) && defined(__amd64__)
@@ -86,7 +89,7 @@
       : /*out*/ : /*in*/ "r"(&(_lval))       \
       : /*trash*/ "r8", "r9", "cc", "memory" \
   );
-#elif defined(PLAT_arm64_linux) || defined(PLAT_arm64_freebsd)
+#elif defined(PLAT_arm64_linux) || defined(PLAT_arm64_freebsd) || defined(PLAT_arm64_darwin)
 #  define INC(_lval,_lqual) \
   __asm__ __volatile__( \
       "1:\n"                                 \

helgrind/tests/tc08_hbl2.c

@@ -18,12 +18,13 @@
 #include <unistd.h>
 
 /* Simple test program, no race.  Parent writes atomically to a counter
-   whilst child reads it.  When counter reaches a prearranged value, 
+   whilst child reads it.  When counter reaches a prearranged value,
    child joins back to parent.  Parent (writer) uses hardware bus lock;
    child is only reading and so does not need to use a bus lock. */
 
 #undef PLAT_x86_darwin
 #undef PLAT_amd64_darwin
+#undef PLAT_arm64_darwin
 #undef PLAT_x86_freebsd
 #undef PLAT_amd64_freebsd
 #undef PLAT_arm64_freebsd
@@ -43,6 +44,8 @@
 #  define PLAT_x86_darwin 1
 #elif defined(__APPLE__) && defined(__x86_64__)
 #  define PLAT_amd64_darwin 1
+#elif defined(__APPLE__) && defined(__aarch64__)
+#  define PLAT_arm64_darwin 1
 #elif defined(__FreeBSD__) && defined(__i386__)
 #  define PLAT_x86_freebsd 1
 #elif defined(__FreeBSD__) && defined(__amd64__)
@@ -108,7 +111,7 @@
       : /*out*/ : /*in*/ "r"(&(_lval))       \
       : /*trash*/ "r8", "r9", "cc", "memory" \
   );
-#elif defined(PLAT_arm64_linux) || defined(PLAT_arm64_freebsd)
+#elif defined(PLAT_arm64_linux) || defined(PLAT_arm64_freebsd) || defined(PLAT_arm64_darwin)
 #  define INC(_lval,_lqual) \
   __asm__ __volatile__( \
       "1:\n"                                 \

helgrind/tests/tc11_XCHG.c

@@ -11,6 +11,7 @@
 
 #undef PLAT_x86_darwin
 #undef PLAT_amd64_darwin
+#undef PLAT_arm64_darwin
 #undef PLAT_x86_freebsd
 #undef PLAT_amd64_freebsd
 #undef PLAT_arm64_freebsd
@@ -28,6 +29,8 @@
 #  define PLAT_x86_darwin 1
 #elif defined(__APPLE__) && defined(__x86_64__)
 #  define PLAT_amd64_darwin 1
+#elif defined(__APPLE__) && defined(__aarch64__)
+#  define PLAT_arm64_darwin 1
 #elif defined(__FreeBSD__) && defined(__i386__)
 #  define PLAT_x86_freebsd 1
 #elif defined(__FreeBSD__) && defined(__amd64__)
@@ -128,7 +131,7 @@
 
 #elif defined(PLAT_ppc32_linux) || defined(PLAT_ppc64_linux) \
       || defined(PLAT_arm_linux) || defined(PLAT_arm64_linux) \
-      || defined(PLAT_arm64_freebsd)
+      || defined(PLAT_arm64_freebsd) || defined(PLAT_arm64_darwin)
 #  if defined(HAVE_BUILTIN_ATOMIC)
 #    define XCHG_M_R(_addr,_lval)                                           \
         do {                                                                \

none/tests/arm64/atomics_v81.c

@@ -5,7 +5,7 @@
 */
 
 #include <stdio.h>
-#include <malloc.h>
+#include <stdlib.h>
 #include <stdint.h>
 #include <string.h>
 

none/tests/arm64/crc32.c

@@ -5,7 +5,6 @@
 */
 
 #include <stdio.h>
-#include <malloc.h>  // memalign
 #include <string.h>  // memset
 #include <assert.h>
 

none/tests/arm64/integer.c

@@ -10,7 +10,6 @@
 #endif
 
 #include <stdio.h>
-#include <malloc.h>  // memalign
 #include <string.h>  // memset
 #include <assert.h>
 
@@ -317,13 +316,13 @@ TESTINST2("bfm x2, x4, #0, #63", 0xaaaaaaaaaaaaaaaa, x2, x4, 0);
 
 TESTINST3("cmp x4, x5 ; cset x3, ne", 12345LL, 6789LL, x3, x4, x5, 0);
 TESTINST3("cmp x4, x5 ; cset x3, eq", 12345LL, 6789LL, x3, x4, x5, 0);
-  
+
 TESTINST3("cmp w4, w5 ; cset x3, ne",
           0xAB12345678, 0xCD12345678, x3, x4, x5, 0);
 
 TESTINST3("cmp w4, w5 ; cset x3, eq",
           0xCD12345678, 0xCD12345670, x3, x4, x5, 0);
-  
+
 TESTINST3("cmp x4, x5 ; mrs x3, nzcv", 0xdb432311d1e3a1d0, 0xa6325ae016fbd710, x3, x4, x5, 0);
 TESTINST3("cmp x4, x5 ; mrs x3, nzcv", 0x192e7217e1ab63b0, 0xfd370f11bfcd4a4a, x3, x4, x5, 0);
 TESTINST3("cmp x4, x5 ; mrs x3, nzcv", 0xfd79baaee550b488, 0x5bc94f0d3ee4863a, x3, x4, x5, 0);

none/tests/arm64/ldxp_stxp.c

@@ -4,7 +4,7 @@
 
 #include <stdio.h>
 #include <stdlib.h>
-#include <malloc.h>
+#include "tests/malloc.h"
 #include <assert.h>
 
 typedef  unsigned int            UInt;
@@ -36,7 +36,7 @@ void printBlock ( const char* who,
 
 int main ( void )
 {
-   ULong* block = memalign(16, 6 * sizeof(ULong));
+   ULong* block = memalign16(6 * sizeof(ULong));
    assert(block);
 
    ULong rt1in, rt2in, rt1out, rt2out;

none/tests/arm64/memory_test.c

@@ -4,7 +4,6 @@
 */
 
 #include <stdio.h>
-#include <malloc.h>  // memalign
 #include <string.h>  // memset
 #include "tests/malloc.h"
 #include <assert.h>
@@ -323,7 +322,7 @@ static void show_block_xor ( UChar* block1, UChar* block2, Int n )
 
 
 // In: rand:
-//       memory area, xferred vec regs, xferred int regs, 
+//       memory area, xferred vec regs, xferred int regs,
 //     caller spec:
 //       addr reg1, addr reg2
 //

none/tests/faultstatus.c

@@ -1,4 +1,4 @@
-/* 
+/*
    Check that a fault signal handler gets the expected info
  */
 #include <signal.h>
@@ -23,6 +23,8 @@
 #  define DIVISION_BY_ZERO_TRIGGERS_FPE 0
 #if defined(VGO_freebsd)
 #  define DIVISION_BY_ZERO_SI_CODE      SI_LWP
+#elif defined(VGO_darwin)
+#  define DIVISION_BY_ZERO_SI_CODE      FPE_INTDIV
 #else
 #  define DIVISION_BY_ZERO_SI_CODE      SI_TKILL
 #endif
@@ -71,7 +73,7 @@ static int testsig(int sig, int want)
 	if (sig != want) {
 		fprintf(stderr, "  FAIL: expected signal %d, not %d\n", want, sig);
 		return 0;
-	} 
+	}
 	return 1;
 }
 
@@ -150,7 +152,7 @@ int main()
 	sa.sa_sigaction = handler;
 	sa.sa_flags = SA_SIGINFO;
 	sigfillset(&sa.sa_mask);
-	
+
 	for(i = 0; i < sizeof(sigs)/sizeof(*sigs); i++)
 		sigaction(sigs[i], &sa, NULL);
 
@@ -182,7 +184,7 @@ int main()
 
 		for(i = 0; i < sizeof(tests)/sizeof(*tests); i++) {
 			cur_test = &tests[i];
-		
+
 			if (sigsetjmp(escape, 1) == 0) {
 				fprintf(stderr, "Test %d: ", i+1);
 				tests[i].test();

tests/arm64_features.c

@@ -181,6 +181,84 @@ typedef int    Bool;
 #define HWCAP2_FRINT        (1 << 8)
 #endif
 
+#if defined(VGO_darwin)
+#include <sys/types.h>
+#include <sys/sysctl.h>
+
+#define AT_HWCAP 0
+#define AT_HWCAP2 1
+
+#define CHECK_CAP(cap, name) \
+  (sysctlbyname((name), &hwcap_value, &size, NULL, 0) == 0 && hwcap_value == 1 ? (cap) : 0)
+
+// based on https://github.com/klauspost/cpuid/blob/master/os_darwin_arm64.go
+// and my computer `sysctl -a`
+unsigned long getauxval(unsigned long type)
+{
+  unsigned long hwcap = 0;
+  int hwcap_value = 0;
+  size_t size = 4;
+
+  switch (type) {
+    case AT_HWCAP:
+      hwcap |= CHECK_CAP(HWCAP_FP, "hw.optional.arm.FEAT_FP");
+      hwcap |= CHECK_CAP(HWCAP_FP, "hw.optional.floatingpoint"); // double
+      hwcap |= CHECK_CAP(HWCAP_FP, "hw.optional.arm.FEAT_FP");
+      hwcap |= CHECK_CAP(HWCAP_ASIMD, "hw.optional.AdvSIMD");
+      // ??? evtstrm
+      hwcap |= CHECK_CAP(HWCAP_AES, "hw.optional.arm.FEAT_AES");
+      hwcap |= CHECK_CAP(HWCAP_PMULL, "hw.optional.arm.FEAT_PMULL");
+      hwcap |= CHECK_CAP(HWCAP_SHA1, "hw.optional.arm.FEAT_SHA1");
+      hwcap |= CHECK_CAP(HWCAP_SHA2, "hw.optional.arm.FEAT_SHA256");
+      hwcap |= CHECK_CAP(HWCAP_CRC32, "hw.optional.FEAT_CRC32");
+      hwcap |= CHECK_CAP(HWCAP_CRC32, "hw.optional.armv8_crc32"); // double
+      hwcap |= CHECK_CAP(HWCAP_ATOMICS, "hw.optional.armv8_1_atomics");
+      hwcap |= CHECK_CAP(HWCAP_FPHP, "hw.optional.arm.FEAT_FP16");
+      hwcap |= CHECK_CAP(HWCAP_ASIMDHP, "hw.optional.AdvSIMD_HPFPCvt");
+      // ??? cpuid
+      hwcap |= CHECK_CAP(HWCAP_ASIMDRDM, "hw.optional.arm.FEAT_RDM");
+      hwcap |= CHECK_CAP(HWCAP_JSCVT, "hw.optional.arm.FEAT_JSCVT");
+      hwcap |= CHECK_CAP(HWCAP_FCMA, "hw.optional.arm.FEAT_FCMA");
+      hwcap |= CHECK_CAP(HWCAP_FCMA, "hw.optional.armv8_3_compnum"); // double
+      hwcap |= CHECK_CAP(HWCAP_LRCPC, "hw.optional.arm.FEAT_LRCPC");
+      hwcap |= CHECK_CAP(HWCAP_DCPOP, "hw.optional.arm.FEAT_DPB");
+      hwcap |= CHECK_CAP(HWCAP_SHA3, "hw.optional.arm.FEAT_SHA3");
+      hwcap |= CHECK_CAP(HWCAP_SHA3, "hw.optional.armv8_2_sha3"); // double
+      // ??? sm3
+      // ??? sm4
+      hwcap |= CHECK_CAP(HWCAP_ASIMDDP, "hw.optional.arm.FEAT_DotProd");
+      hwcap |= CHECK_CAP(HWCAP_SHA512, "hw.optional.arm.FEAT_SHA512");
+      hwcap |= CHECK_CAP(HWCAP_SHA512, "hw.optional.armv8_2_sha512"); // double
+      hwcap |= CHECK_CAP(HWCAP_SVE, "hw.optional.arm.FEAT_SVE");
+      // ??? asimdfhm
+      hwcap |= CHECK_CAP(HWCAP_DIT, "hw.optional.arm.FEAT_DIT");
+      // ??? uscat
+      // ??? ilrcpc
+      hwcap |= CHECK_CAP(HWCAP_FLAGM, "hw.optional.arm.FEAT_FlagM");
+      hwcap |= CHECK_CAP(HWCAP_SSBS, "hw.optional.arm.FEAT_SSBS");
+      hwcap |= CHECK_CAP(HWCAP_SB, "hw.optional.arm.FEAT_SB");
+      // ??? paca
+      // ??? pacg
+      return hwcap;
+    case AT_HWCAP2:
+      hwcap |= CHECK_CAP(HWCAP2_DCPODP, "hw.optional.arm.FEAT_DPB");
+      // ??? sve2
+      // ??? sveaes
+      // ??? svepmull
+      // ??? svebitperm
+      // ??? svesha3
+      // ??? svesm4
+      hwcap |= CHECK_CAP(HWCAP2_FLAGM2, "hw.optional.arm.FEAT_FlagM2");
+      // ??? frint
+      return hwcap;
+    default:
+      return 0UL;
+  }
+}
+
+#undef CHECK_CAP
+#endif
+
 unsigned long hwcaps[] = {
    HWCAP_FP,     HWCAP_ASIMD,  HWCAP_EVTSTRM, HWCAP_AES,     HWCAP_PMULL,
    HWCAP_SHA1,   HWCAP_SHA2,   HWCAP_CRC32,   HWCAP_ATOMICS, HWCAP_FPHP,