any architecture (using miniinterpreter)
-------------------------------------------------------------------------- */
-StgThreadReturnCode StgRun(StgFunPtr f, StgRegTable *basereg STG_UNUSED)
+StgRegTable * StgRun(StgFunPtr f, StgRegTable *basereg STG_UNUSED)
{
while (f) {
- if (RtsFlags[0].DebugFlags.interpreter) {
+ IF_DEBUG(interpreter,
debugBelch("Jumping to ");
printPtr((P_)f); fflush(stdout);
debugBelch("\n");
- }
+ );
f = (StgFunPtr) (f)();
}
- return (StgThreadReturnCode)R1.i;
+ return (StgRegTable *)R1.p;
}
StgFunPtr StgReturn(void)
#ifdef i386_HOST_ARCH
-StgThreadReturnCode
+#ifdef darwin_TARGET_OS
+#define STG_GLOBAL ".globl "
+#else
+#define STG_GLOBAL ".global "
+#endif
+
+StgRegTable *
StgRun(StgFunPtr f, StgRegTable *basereg) {
unsigned char space[ RESERVED_C_STACK_BYTES + 4*sizeof(void *) ];
- StgThreadReturnCode r;
+ StgRegTable * r;
__asm__ volatile (
/*
*/
"movl %3,%%ebx\n\t"
/*
- * grab the function argument from the stack, and jump to it.
+ * grab the function argument from the stack
*/
"movl %2,%%eax\n\t"
+
+#if darwin_TARGET_OS
+ /*
+ * Darwin: keep the stack aligned
+ */
+ "subl $12,%%esp\n\t"
+#endif
+
+ /*
+ * jump to it
+ */
"jmp *%%eax\n\t"
- ".global " STG_RETURN "\n"
+ STG_GLOBAL STG_RETURN "\n"
STG_RETURN ":\n\t"
+#if darwin_TARGET_OS
+ /*
+ * Darwin: keep the stack aligned
+ */
+ "addl $12,%%esp\n\t"
+#endif
+
"movl %%esi, %%eax\n\t" /* Return value in R1 */
/*
#ifdef x86_64_HOST_ARCH
-extern StgThreadReturnCode StgRun(StgFunPtr f, StgRegTable *basereg);
+extern StgRegTable * StgRun(StgFunPtr f, StgRegTable *basereg);
-static void StgRunIsImplementedInAssembler(void)
+static void GNUC3_ATTRIBUTE(used)
+StgRunIsImplementedInAssembler(void)
{
__asm__ volatile (
/*
/*
* Set BaseReg
*/
- "movq %%rsi,%%rbx\n\t"
+ "movq %%rsi,%%r13\n\t"
/*
* grab the function argument from the stack, and jump to it.
*/
".global " STG_RETURN "\n"
STG_RETURN ":\n\t"
- "movq %%r13, %%rax\n\t" /* Return value in R1 */
+ "movq %%rbx, %%rax\n\t" /* Return value in R1 */
/*
* restore callee-saves registers. (Don't stomp on %%rax!)
"addq %0, %%rsp\n\t"
"retq"
- : : "i"(RESERVED_C_STACK_BYTES+48 /*stack frame size*/));
+ : : "i"(RESERVED_C_STACK_BYTES+48+8 /*stack frame size*/));
+ /*
+ HACK alert!
+
+ The x86_64 ABI specifies that on a procedure call, %rsp is
+ aligned on a 16-byte boundary + 8. That is, the first
+ argument on the stack after the return address will be
+ 16-byte aligned.
+
+ Which should be fine: RESERVED_C_STACK_BYTES+48 is a multiple
+ of 16 bytes.
+
+ BUT... when we do a C-call from STG land, gcc likes to put the
+ stack alignment adjustment in the prolog. eg. if we're calling
+ a function with arguments in regs, gcc will insert 'subq $8,%rsp'
+ in the prolog, to keep %rsp aligned (the return address is 8
+ bytes, remember). The mangler throws away the prolog, so we
+ lose the stack alignment.
+
+ The hack is to add this extra 8 bytes to our %rsp adjustment
+ here, so that throughout STG code, %rsp is 16-byte aligned,
+ ready for a C-call.
+
+ A quick way to see if this is wrong is to compile this code:
+
+ main = System.Exit.exitWith ExitSuccess
+
+ And run it with +RTS -sstderr. The stats code in the RTS, in
+ particular statsPrintf(), relies on the stack alignment because
+ it saves the %xmm regs on the stack, so it'll fall over if the
+ stack isn't aligned, and calling exitWith from Haskell invokes
+ shutdownHaskellAndExit using a C call.
+
+ Future gcc releases will almost certainly break this hack...
+ */
}
#endif /* x86-64 */
#ifdef sparc_HOST_ARCH
-StgThreadReturnCode
+StgRegTable *
StgRun(StgFunPtr f, StgRegTable *basereg) {
unsigned char space[RESERVED_C_STACK_BYTES];
__asm__ volatile ("ld %1,%0"
: "=r" (i7) : "m" (((void **)(space))[100]));
#endif
- return (StgThreadReturnCode)R1.i;
+ return (StgRegTable *)R1.i;
}
#endif
#ifdef alpha_HOST_ARCH
-StgThreadReturnCode
+StgRegTable *
StgRun(StgFunPtr f, StgRegTable *basereg)
{
register long real_ra __asm__("$26"); volatile long save_ra;
register StgFunPtr real_pv __asm__("$27");
- StgThreadReturnCode ret;
+ StgRegTable * ret;
save_ra = real_ra;
save_gp = real_gp;
#ifdef hppa1_1_HOST_ARCH
-StgThreadReturnCode
+StgRegTable *
StgRun(StgFunPtr f, StgRegTable *basereg)
{
StgChar space[RESERVED_C_STACK_BYTES+16*sizeof(long)+10*sizeof(double)];
- StgThreadReturnCode ret;
+ StgRegTable * ret;
__asm__ volatile ("ldo %0(%%r30),%%r19\n"
"\tstw %%r3, 0(0,%%r19)\n"
#ifdef powerpc_HOST_ARCH
-extern StgThreadReturnCode StgRun(StgFunPtr f, StgRegTable *basereg);
+extern StgRegTable * StgRun(StgFunPtr f, StgRegTable *basereg);
#ifdef darwin_HOST_OS
-static void StgRunIsImplementedInAssembler(void)
+static void GNUC3_ATTRIBUTE(used)
+StgRunIsImplementedInAssembler(void)
{
#if HAVE_SUBSECTIONS_VIA_SYMBOLS
// if the toolchain supports deadstripping, we have to
// *) The Link Register is saved to a different offset in the caller's stack frame
// (Linux: 4(r1), Darwin 8(r1))
-static void StgRunIsImplementedInAssembler(void)
+static void GNUC3_ATTRIBUTE(used)
+StgRunIsImplementedInAssembler(void)
{
__asm__ volatile (
"\t.globl StgRun\n"
#ifdef powerpc64_HOST_ARCH
#ifdef linux_HOST_OS
-extern StgThreadReturnCode StgRun(StgFunPtr f, StgRegTable *basereg);
+extern StgRegTable * StgRun(StgFunPtr f, StgRegTable *basereg);
-static void StgRunIsImplementedInAssembler(void)
+static void GNUC3_ATTRIBUTE(used)
+StgRunIsImplementedInAssembler(void)
{
// r0 volatile
// r1 stack pointer
#define LOCALS 31
#endif
-static void StgRunIsImplementedInAssembler(void)
+static void GNUC3_ATTRIBUTE(used)
+StgRunIsImplementedInAssembler(void)
{
__asm__ volatile(
".global StgRun\n"
projects / ghc-hetmet.git / blobdiff