Merge branch 'development'

Change-Id: Icb5432f157ceb2182d93e2d327cfa63ad02a8c0e

.gitignore

@@ -1,3 +1,4 @@
+*~
 *.o
 *.elf
 *.bin

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "ihk"]
+	path = ihk
+	url = https://github.com/RIKEN-SysSoft/ihk.git

Makefile.in

@@ -4,13 +4,21 @@ INCDIR =  @INCDIR@
 ETCDIR = @ETCDIR@
 MANDIR = @MANDIR@
 
-all::
+all: executer-mcctrl executer-mcoverlayfs executer-user mckernel mck-tools
-	@(cd executer/kernel/mcctrl; make modules)
+
-	@(cd executer/kernel/mcoverlayfs; make modules)
+executer-mcctrl:
-	@(cd executer/user; make)
+	+@(cd executer/kernel/mcctrl; $(MAKE) modules)
-	@case "$(TARGET)" in \
+
+executer-mcoverlayfs:
+	+@(cd executer/kernel/mcoverlayfs; $(MAKE) modules)
+
+executer-user:
+	+@(cd executer/user; $(MAKE))
+
+mckernel:
+	+@case "$(TARGET)" in \
 	    attached-mic | builtin-x86 | builtin-mic | smp-x86 | smp-arm64) \
-		(cd kernel; make) \
+		(cd kernel; $(MAKE)) \
 		;; \
 	    *) \
 		echo "unknown target $(TARGET)" >&2 \
@@ -18,13 +26,16 @@ all::
 		;; \
 	esac
 
-install::
+mck-tools:
-	@(cd executer/kernel/mcctrl; make install)
+	+@(cd tools/mcstat; $(MAKE))
-	@(cd executer/kernel/mcoverlayfs; make install)
+
-	@(cd executer/user; make install)
+install:
+	@(cd executer/kernel/mcctrl; $(MAKE) install)
+	@(cd executer/kernel/mcoverlayfs; $(MAKE) install)
+	@(cd executer/user; $(MAKE) install)
 	@case "$(TARGET)" in \
 	    attached-mic | builtin-x86 | builtin-mic | smp-x86 | smp-arm64) \
-		(cd kernel; make install) \
+		(cd kernel; $(MAKE) install) \
 		;; \
 	    *) \
 		echo "unknown target $(TARGET)" >&2 \
@@ -32,51 +43,39 @@ install::
 		;; \
 	esac
 	@case "$(TARGET)" in \
-	    attached-mic) \
-		mkdir -p -m 755 $(SBINDIR); \
-		install -m 755 arch/x86/tools/mcreboot-attached-mic.sh $(SBINDIR)/mcreboot; \
-		install -m 755 arch/x86/tools/mcshutdown-attached-mic.sh $(SBINDIR)/mcshutdown; \
-		mkdir -p -m 755 $(MANDIR)/man1; \
-		install -m 644 arch/x86/tools/mcreboot.1 $(MANDIR)/man1/mcreboot.1; \
-		;; \
-	    builtin-x86) \
-		mkdir -p -m 755 $(SBINDIR); \
-		install -m 755 arch/x86/tools/mcreboot-builtin-x86.sh $(SBINDIR)/mcreboot; \
-		install -m 755 arch/x86/tools/mcshutdown-builtin-x86.sh $(SBINDIR)/mcshutdown; \
-		mkdir -p -m 755 $(MANDIR)/man1; \
-		install -m 644 arch/x86/tools/mcreboot.1 $(MANDIR)/man1/mcreboot.1; \
-		;; \
 		smp-x86 | smp-arm64) \
 		mkdir -p -m 755 $(SBINDIR); \
-		install -m 755 arch/x86/tools/mcreboot-smp-x86.sh $(SBINDIR)/mcreboot.sh; \
+		install -m 755 arch/x86_64/tools/mcreboot-smp-x86.sh $(SBINDIR)/mcreboot.sh; \
-		install -m 755 arch/x86/tools/mcstop+release-smp-x86.sh $(SBINDIR)/mcstop+release.sh; \
+		install -m 755 arch/x86_64/tools/mcstop+release-smp-x86.sh $(SBINDIR)/mcstop+release.sh; \
-		install -m 600 arch/x86/tools/mcoverlay-destroy-smp-x86.sh $(SBINDIR)/mcoverlay-destroy.sh; \
+		install -m 755 arch/x86_64/tools/mcoverlay-destroy-smp-x86.sh $(SBINDIR)/mcoverlay-destroy.sh; \
-		install -m 600 arch/x86/tools/mcoverlay-create-smp-x86.sh $(SBINDIR)/mcoverlay-create.sh; \
+		install -m 755 arch/x86_64/tools/mcoverlay-create-smp-x86.sh $(SBINDIR)/mcoverlay-create.sh; \
-		install -m 755 arch/x86/tools/eclair-dump-backtrace.exp $(SBINDIR)/eclair-dump-backtrace.exp;\
+		install -m 755 arch/x86_64/tools/eclair-dump-backtrace.exp $(SBINDIR)/eclair-dump-backtrace.exp;\
 		mkdir -p -m 755 $(ETCDIR); \
-		install -m 644 arch/x86/tools/irqbalance_mck.service $(ETCDIR)/irqbalance_mck.service; \
+		install -m 644 arch/x86_64/tools/irqbalance_mck.service $(ETCDIR)/irqbalance_mck.service; \
-		install -m 644 arch/x86/tools/irqbalance_mck.in $(ETCDIR)/irqbalance_mck.in; \
+		install -m 644 arch/x86_64/tools/irqbalance_mck.in $(ETCDIR)/irqbalance_mck.in; \
 		mkdir -p -m 755 $(INCDIR); \
 		install -m 644 kernel/include/swapfmt.h $(INCDIR); \
 		mkdir -p -m 755 $(MANDIR)/man1; \
-		install -m 644 arch/x86/tools/mcreboot.1 $(MANDIR)/man1/mcreboot.1; \
+		install -m 644 arch/x86_64/tools/mcreboot.1 $(MANDIR)/man1/mcreboot.1; \
 		;; \
 	    *) \
 		echo "unknown target $(TARGET)" >&2 \
 		exit 1 \
 		;; \
 	esac
+	@(cd tools/mcstat/; $(MAKE) install)
 
-clean::
+clean:
-	@(cd executer/kernel/mcctrl; make clean)
+	@(cd executer/kernel/mcctrl; $(MAKE) clean)
-	@(cd executer/kernel/mcoverlayfs; make clean)
+	@(cd executer/kernel/mcoverlayfs; $(MAKE) clean)
-	@(cd executer/user; make clean)
+	@(cd executer/user; $(MAKE) clean)
 	@case "$(TARGET)" in \
 	    attached-mic | builtin-x86 | builtin-mic | smp-x86 | smp-arm64) \
-		(cd kernel; make clean) \
+		(cd kernel; $(MAKE) clean) \
 		;; \
 	    *) \
 		echo "unknown target $(TARGET)" >&2 \
 		exit 1 \
 		;; \
 	esac
+	@(cd tools/mcstat; $(MAKE) clean)

arch/arm64/kernel/cpu.c

@@ -30,6 +30,7 @@
 #include <debug-monitors.h>
 #include <sysreg.h>
 #include <cpufeature.h>
+#include <debug.h>
 #ifdef POSTK_DEBUG_ARCH_DEP_65
 #include <hwcap.h>
 #endif /* POSTK_DEBUG_ARCH_DEP_65 */
@@ -39,16 +40,10 @@
 #include "postk_print_sysreg.c"
 
 #ifdef DEBUG_PRINT_CPU
-#define dkprintf kprintf
+#undef DDEBUG_DEFAULT
-#define ekprintf kprintf
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define ekprintf kprintf
 #endif
 
-#define BUG_ON(condition) do { if (condition) { kprintf("PANIC: %s: %s(line:%d)\n",\
-				__FILE__, __FUNCTION__, __LINE__); panic(""); } } while(0)
-
 struct cpuinfo_arm64 cpuinfo_data[NR_CPUS];	/* index is logical cpuid */
 static unsigned int per_cpu_timer_val[NR_CPUS] = { 0 };
 
@@ -590,6 +585,8 @@ static void show_context_stack(struct pt_regs *regs)
 		return;
 	}
 
+	ihk_mc_debug_show_interrupt_context(regs);
+
 	sp = (uintptr_t)regs + sizeof(*regs);
 	stack_top = ALIGN_UP(sp, (uintptr_t)KERNEL_STACK_SIZE);
 	max_loop = (stack_top - sp) / min_stack_frame_size;
@@ -1170,8 +1167,6 @@ void arch_clone_thread(struct thread *othread, unsigned long pc,
 	asm("mrs %0, tpidr_el0" : "=r" (tls));
 	othread->tlsblock_base = nthread->tlsblock_base = tls;
 
-	/* copy fp_regs values from parent. */
-	save_fp_regs(othread);
 	if ((othread->fp_regs != NULL) && (check_and_allocate_fp_regs(nthread) == 0)) {
 		memcpy(nthread->fp_regs, othread->fp_regs, sizeof(fp_regs_struct));
 	}
@@ -1205,6 +1200,10 @@ void ihk_mc_delay_us(int us)
 	arch_delay(us);
 }
 
+void arch_print_stack()
+{
+}
+
 void arch_show_interrupt_context(const void *reg)
 {
 	const struct pt_regs *regs = (struct pt_regs *)reg;
@@ -1279,7 +1278,6 @@ int ihk_mc_interrupt_cpu(int cpu, int vector)
 	return 0;
 }
 
-#ifdef POSTK_DEBUG_ARCH_DEP_22
 /*
  * @ref.impl linux-linaro/arch/arm64/kernel/process.c::tls_thread_switch()
  */
@@ -1305,14 +1303,13 @@ struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 	extern void perf_start(struct mc_perf_event *event);
 	extern void perf_reset(struct mc_perf_event *event);
 	struct thread *last;
-#ifdef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
 	struct mcs_rwlock_node_irqsave lock;
-#endif /* POSTK_DEBUG_TEMP_FIX_41 */
 
 	/* Set up new TLS.. */
 	dkprintf("[%d] arch_switch_context: tlsblock_base: 0x%lX\n", 
 		 ihk_mc_get_processor_id(), next->tlsblock_base);
 
+#ifdef ENABLE_PERF
 	/* Performance monitoring inherit */
 	if(next->proc->monitoring_event) {
 		if(next->proc->perf_status == PP_RESET)
@@ -1322,10 +1319,10 @@ struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 			perf_start(next->proc->monitoring_event);
 		}
 	}
+#endif /*ENABLE_PERF*/
 	if (likely(prev)) {
 		tls_thread_switch(prev, next);
 
-#ifdef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
 		mcs_rwlock_writer_lock(&prev->proc->update_lock, &lock);
 		if (prev->proc->status & (PS_DELAY_STOPPED | PS_DELAY_TRACED)) {
 			switch (prev->proc->status) {
@@ -1339,11 +1336,12 @@ struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 				break;
 			}
 			mcs_rwlock_writer_unlock(&prev->proc->update_lock, &lock);
+
+			/* Wake up the parent who tried wait4 and sleeping */
 			waitq_wakeup(&prev->proc->parent->waitpid_q);
 		} else {
 			mcs_rwlock_writer_unlock(&prev->proc->update_lock, &lock);
 		}
-#endif /* POSTK_DEBUG_TEMP_FIX_41 */
 
 		last = ihk_mc_switch_context(&prev->ctx, &next->ctx, prev);
 	}
@@ -1353,7 +1351,6 @@ struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 
 	return last;
 }
-#endif /* POSTK_DEBUG_ARCH_DEP_22 */
 
 /*@
   @ requires \valid(thread);
@@ -1428,8 +1425,14 @@ save_fp_regs(struct thread *thread)
 	}
 }
 
-void
+void copy_fp_regs(struct thread *from, struct thread *to)
-clear_fp_regs(struct thread *thread)
+{
+	if ((from->fp_regs != NULL) && (check_and_allocate_fp_regs(to) == 0)) {
+		memcpy(to->fp_regs, from->fp_regs, sizeof(fp_regs_struct));
+	}
+}
+
+void clear_fp_regs(void)
 {
 	if (likely(elf_hwcap & (HWCAP_FP | HWCAP_ASIMD))) {
 #ifdef CONFIG_ARM64_SVE
@@ -1466,7 +1469,7 @@ restore_fp_regs(struct thread *thread)
 	if (likely(elf_hwcap & (HWCAP_FP | HWCAP_ASIMD))) {
 		if (!thread->fp_regs) {
 			// only clear fpregs.
-			clear_fp_regs(thread);
+			clear_fp_regs();
 			return;
 		}
 		thread_fpsimd_load(thread);
@@ -1499,7 +1502,6 @@ unhandled_page_fault(struct thread *thread, void *fault_addr, void *regs)
 	const uintptr_t address = (uintptr_t)fault_addr;
 	struct process_vm *vm = thread->vm;
 	struct vm_range *range;
-	char found;
 	unsigned long irqflags;
 	unsigned long error = 0;
 
@@ -1513,17 +1515,12 @@ unhandled_page_fault(struct thread *thread, void *fault_addr, void *regs)
 			(error & PF_RSVD ? "was" : "wasn't"),
 			(error & PF_INSTR ? "was" : "wasn't"));
 
-	found = 0;
+	range = lookup_process_memory_range(vm, address, address+1);
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	if (range) {
-		if (range->start <= address && range->end > address) {
+		__kprintf("address is in range, flag: 0x%lx\n",
-			found = 1;
+				range->flag);
-			__kprintf("address is in range, flag: 0x%lx\n",
+		ihk_mc_pt_print_pte(vm->address_space->page_table, (void*)address);
-					range->flag);
+	} else {
-			ihk_mc_pt_print_pte(vm->address_space->page_table, (void*)address);
-			break;
-		}
-	}
-	if (!found) {
 		__kprintf("address is out of range! \n");
 	}
 

arch/arm64/kernel/fpsimd.c

@@ -9,20 +9,16 @@
 #include <prctl.h>
 #include <cpufeature.h>
 #include <kmalloc.h>
+#include <debug.h>
+#include <process.h>
 
 //#define DEBUG_PRINT_FPSIMD
 
 #ifdef DEBUG_PRINT_FPSIMD
-#define dkprintf kprintf
+#undef DDEBUG_DEFAULT
-#define ekprintf kprintf
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define ekprintf kprintf
 #endif
 
-#define BUG_ON(condition) do { if (condition) { kprintf("PANIC: %s: %s(line:%d)\n",\
-				__FILE__, __FUNCTION__, __LINE__); panic(""); } } while(0)
-
 #ifdef CONFIG_ARM64_SVE
 
 /* Maximum supported vector length across all CPUs (initially poisoned) */
@@ -73,9 +69,6 @@ static int get_nr_threads(struct process *proc)
 	return nr_threads;
 }
 
-extern void save_fp_regs(struct thread *thread);
-extern void clear_fp_regs(struct thread *thread);
-extern void restore_fp_regs(struct thread *thread);
 /* @ref.impl arch/arm64/kernel/fpsimd.c::sve_set_vector_length */
 int sve_set_vector_length(struct thread *thread,
 			unsigned long vl, unsigned long flags)
@@ -129,7 +122,7 @@ int sve_set_vector_length(struct thread *thread,
 			/* for self at prctl syscall */
 			if (thread == cpu_local_var(current)) {
 				save_fp_regs(thread);
-				clear_fp_regs(thread);
+				clear_fp_regs();
 				thread_sve_to_fpsimd(thread, &fp_regs);
 				sve_free(thread);
 

arch/arm64/kernel/gencore.c

@@ -7,6 +7,7 @@
 #include <process.h>
 #include <string.h>
 #include <elfcore.h>
+#include <debug.h>
 
 #define	align32(x) ((((x) + 3) / 4) * 4)
 #define	alignpage(x) ((((x) + (PAGE_SIZE) - 1) / (PAGE_SIZE)) * (PAGE_SIZE))
@@ -14,11 +15,8 @@
 //#define DEBUG_PRINT_GENCORE
 
 #ifdef DEBUG_PRINT_GENCORE
-#define	dkprintf(...)	kprintf(__VA_ARGS__)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...)	do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
 #endif
 
 /*
@@ -221,7 +219,7 @@ int gencore(struct thread *thread, void *regs,
 	Elf64_Ehdr eh;
 	Elf64_Phdr *ph = NULL;
 	void *note = NULL;
-	struct vm_range *range;
+	struct vm_range *range, *next;
 	struct process_vm *vm = thread->vm;
 	int segs = 1;	/* the first one is for NOTE */
 	int notesize, phsize, alignednotesize;
@@ -235,7 +233,10 @@ int gencore(struct thread *thread, void *regs,
 		return -1;
 	}
 
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	next = lookup_process_memory_range(vm, 0, -1);
+	while ((range = next)) {
+		next = next_process_memory_range(vm, range);
+
 		dkprintf("start:%lx end:%lx flag:%lx objoff:%lx\n", 
 			 range->start, range->end, range->flag, range->objoff);
 		/* We omit reserved areas because they are only for
@@ -323,7 +324,10 @@ int gencore(struct thread *thread, void *regs,
 
 	/* program header for each memory chunk */
 	i = 1;
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	next = lookup_process_memory_range(vm, 0, -1);
+	while ((range = next)) {
+		next = next_process_memory_range(vm, range);
+
 		unsigned long flag = range->flag;
 		unsigned long size = range->end - range->start;
 
@@ -364,7 +368,10 @@ int gencore(struct thread *thread, void *regs,
 	dkprintf("coretable[2]: %lx@%lx(%lx)\n", ct[2].len, ct[2].addr, note);
 
 	i = 3;	/* memory segments */
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	next = lookup_process_memory_range(vm, 0, -1);
+	while ((range = next)) {
+		next = next_process_memory_range(vm, range);
+
 		unsigned long phys;
 
 		if (range->flag & VR_RESERVED)

arch/arm64/kernel/include/arch-futex.h

@@ -134,4 +134,12 @@ futex_atomic_cmpxchg_inatomic(int __user *uaddr, int oldval, int newval)
 	return ret;
 }
 
+static inline int get_futex_value_locked(uint32_t *dest, uint32_t *from)
+{
+
+	*dest = *(volatile uint32_t *)from;
+
+	return 0;
+}
+
 #endif /* !__HEADER_ARM64_COMMON_ARCH_FUTEX_H */

arch/arm64/kernel/include/arch-lock.h

@@ -6,6 +6,8 @@
 
 #include <ihk/cpu.h>
 #include <ihk/atomic.h>
+#include "affinity.h"
+#include <lwk/compiler.h>
 
 //#define DEBUG_SPINLOCK
 //#define DEBUG_MCS_RWLOCK
@@ -152,6 +154,8 @@ typedef struct mcs_lock_node {
 	unsigned long irqsave;
 } __attribute__((aligned(64))) mcs_lock_node_t;
 
+typedef mcs_lock_node_t mcs_lock_t;
+
 static void mcs_lock_init(struct mcs_lock_node *node)
 {
 	node->locked = 0;
@@ -602,4 +606,16 @@ __mcs_rwlock_reader_unlock(struct mcs_rwlock_lock *lock, struct mcs_rwlock_node_
 #endif
 }
 
+static inline int irqflags_can_interrupt(unsigned long flags)
+{
+#ifdef CONFIG_HAS_NMI
+#warning irqflags_can_interrupt needs testing/fixing on such a target
+	return flags > ICC_PMR_EL1_MASKED;
+#else
+	// PSTATE.DAIF I bit clear means interrupt is possible
+	return !(flags & (1 << 7));
+#endif
+}
+
+
 #endif /* !__HEADER_ARM64_COMMON_ARCH_LOCK_H */

arch/arm64/kernel/include/arch-memory.h

@@ -215,7 +215,7 @@ static const unsigned int  PTL1_ENTRIES = __PTL1_ENTRIES;
 #define PAGE_P2ALIGN        0
 #define page_offset(addr)   __page_offset(addr, PAGE_SIZE)
 #define page_align(addr)    __page_align(addr, PAGE_SIZE)
-#define page_align_up(addr) __page_align_up((addr, PAGE_SIZE)
+#define page_align_up(addr) __page_align_up(addr, PAGE_SIZE)
 
 /*
  * large page
@@ -263,6 +263,8 @@ static const unsigned int  PTL1_ENTRIES = __PTL1_ENTRIES;
 
 #define PTE_FILEOFF		PTE_SPECIAL
 
+#define PT_ENTRIES		(PAGE_SIZE >> 3)
+
 #ifndef __ASSEMBLY__
 
 #include <ihk/types.h>

arch/arm64/kernel/include/arch-perfctr.h

@@ -35,38 +35,4 @@ void arm64_disable_pmu(void);
 int armv8pmu_init(struct arm_pmu* cpu_pmu);
 
 /* TODO[PMU]: 共通部に定義があっても良い。今後の動向を見てここの定義を削除する */
-/*
- * Generalized hardware cache events:
- *
- *       { L1-D, L1-I, LLC, ITLB, DTLB, BPU, NODE } x
- *       { read, write, prefetch } x
- *       { accesses, misses }
- */
-enum perf_hw_cache_id {
-	PERF_COUNT_HW_CACHE_L1D			= 0,
-	PERF_COUNT_HW_CACHE_L1I			= 1,
-	PERF_COUNT_HW_CACHE_LL			= 2,
-	PERF_COUNT_HW_CACHE_DTLB		= 3,
-	PERF_COUNT_HW_CACHE_ITLB		= 4,
-	PERF_COUNT_HW_CACHE_BPU			= 5,
-	PERF_COUNT_HW_CACHE_NODE		= 6,
-
-	PERF_COUNT_HW_CACHE_MAX,		/* non-ABI */
-};
-
-enum perf_hw_cache_op_id {
-	PERF_COUNT_HW_CACHE_OP_READ		= 0,
-	PERF_COUNT_HW_CACHE_OP_WRITE		= 1,
-	PERF_COUNT_HW_CACHE_OP_PREFETCH		= 2,
-
-	PERF_COUNT_HW_CACHE_OP_MAX,		/* non-ABI */
-};
-
-enum perf_hw_cache_op_result_id {
-	PERF_COUNT_HW_CACHE_RESULT_ACCESS	= 0,
-	PERF_COUNT_HW_CACHE_RESULT_MISS		= 1,
-
-	PERF_COUNT_HW_CACHE_RESULT_MAX,		/* non-ABI */
-};
-
 #endif

arch/arm64/kernel/include/arch/cpu.h

@@ -25,6 +25,8 @@
 #define smp_rmb()	dmb(ishld)
 #define smp_wmb()	dmb(ishst)
 
+#define arch_barrier()	smp_mb()
+
 #define smp_store_release(p, v)						\
 do {									\
 	compiletime_assert_atomic_type(*p);				\

arch/arm64/kernel/include/arch_rusage.h

@@ -0,0 +1,34 @@
+#ifndef ARCH_RUSAGE_H_INCLUDED
+#define ARCH_RUSAGE_H_INCLUDED
+
+#include <arch-memory.h>
+
+//#define DEBUG_RUSAGE
+
+extern struct rusage_global *rusage;
+
+#define IHK_OS_PGSIZE_4KB  0
+#define IHK_OS_PGSIZE_16KB 1
+#define IHK_OS_PGSIZE_64KB 2
+
+static inline int rusage_pgsize_to_pgtype(size_t pgsize)
+{
+	int ret = IHK_OS_PGSIZE_4KB;
+	switch (pgsize) {
+	case __PTL1_SIZE:
+		ret = IHK_OS_PGSIZE_4KB;
+		break;
+	case __PTL2_SIZE:
+		ret = IHK_OS_PGSIZE_16KB;
+		break;
+	case __PTL3_SIZE:
+		ret = IHK_OS_PGSIZE_64KB;
+		break;
+	default:
+		kprintf("%s: Error: Unknown pgsize=%ld\n", __FUNCTION__, pgsize);
+		break;
+	}
+	return ret;
+}
+
+#endif /* !defined(ARCH_RUSAGE_H_INCLUDED) */

arch/arm64/kernel/include/assembler.h

@@ -6,6 +6,8 @@
 
 #if defined(CONFIG_HAS_NMI)
 #include <arm-gic-v3.h>
+#else /* defined(CONFIG_HAS_NMI) */
+#include <sysreg.h>
 #endif /* defined(CONFIG_HAS_NMI) */
 
 #if defined(CONFIG_HAS_NMI)

arch/arm64/kernel/include/cputype.h

@@ -35,6 +35,8 @@
 #define MIDR_IMPLEMENTOR(midr)	\
 	(((midr) & MIDR_IMPLEMENTOR_MASK) >> MIDR_IMPLEMENTOR_SHIFT)
 
+#define ARM_CPU_IMP_CAVIUM	0x43
+
 #ifndef __ASSEMBLY__
 
 static unsigned int read_cpuid_id(void)

arch/arm64/kernel/include/ihk/context.h

@@ -73,6 +73,7 @@ static inline void pt_regs_write_reg(struct pt_regs *regs, int r,
 #define ihk_mc_syscall_arg5(uc) (uc)->regs[5]
 
 #define ihk_mc_syscall_ret(uc)  (uc)->regs[0]
+#define ihk_mc_syscall_number(uc)  (uc)->regs[8]
 
 #define ihk_mc_syscall_pc(uc)   (uc)->pc
 #define ihk_mc_syscall_sp(uc)   (uc)->sp

arch/arm64/kernel/include/signal.h

@@ -9,6 +9,11 @@
 #define _NSIG_BPW	64
 #define _NSIG_WORDS (_NSIG / _NSIG_BPW)
 
+static inline int valid_signal(unsigned long sig)
+{
+	return sig <= _NSIG ? 1 : 0;
+}
+
 typedef unsigned long int __sigset_t;
 #define __sigmask(sig)  (((__sigset_t) 1) << ((sig) - 1))
 

arch/arm64/kernel/include/syscall_list.h

@@ -114,14 +114,18 @@ SYSCALL_HANDLED(236, get_mempolicy)
 SYSCALL_HANDLED(237, set_mempolicy)
 SYSCALL_HANDLED(238, migrate_pages)
 SYSCALL_HANDLED(239, move_pages)
+#ifdef PERF_ENABLE
 SYSCALL_HANDLED(241, perf_event_open)
+#endif // PERF_ENABLE
 SYSCALL_HANDLED(260, wait4)
 SYSCALL_HANDLED(270, process_vm_readv)
 SYSCALL_HANDLED(271, process_vm_writev)
+#ifdef PERF_ENABLE
 SYSCALL_HANDLED(601, pmc_init)
 SYSCALL_HANDLED(602, pmc_start)
 SYSCALL_HANDLED(603, pmc_stop)
 SYSCALL_HANDLED(604, pmc_reset)
+#endif // PERF_ENABLE
 SYSCALL_HANDLED(700, get_cpu_id)
 #ifdef PROFILE_ENABLE
 SYSCALL_HANDLED(__NR_profile, profile)
@@ -144,5 +148,3 @@ SYSCALL_HANDLED(1045, signalfd)
 SYSCALL_DELEGATED(1049, stat)
 SYSCALL_DELEGATED(1060, getpgrp)
 SYSCALL_DELEGATED(1062, time)
-SYSCALL_HANDLED(1071, vfork)
-SYSCALL_DELEGATED(1079, fork)

arch/arm64/kernel/irq-gic-v2.c

@@ -7,15 +7,13 @@
 #include <arch/cpu.h>
 #include <memory.h>
 #include <syscall.h>
+#include <debug.h>
 
 // #define DEBUG_GICV2
 
 #ifdef DEBUG_GICV2
-#define	dkprintf(...)	kprintf(__VA_ARGS__)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...)
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
 #endif
 
 void *dist_base;

arch/arm64/kernel/irq-gic-v3.c

@@ -7,15 +7,19 @@
 #include <cputype.h>
 #include <process.h>
 #include <syscall.h>
+#include <debug.h>
 
 //#define DEBUG_GICV3
 
+#define USE_CAVIUM_THUNDER_X
+
 #ifdef DEBUG_GICV3
-#define	dkprintf(...)	kprintf(__VA_ARGS__)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
+#endif
-#define dkprintf(...)
+
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
+#ifdef USE_CAVIUM_THUNDER_X
+static char is_cavium_thunderx = 0;
 #endif
 
 void *dist_base;
@@ -108,8 +112,8 @@ static uint64_t gic_read_iar_cavium_thunderx(void)
 	asm volatile("nop;nop;nop;nop;");
 	asm volatile("mrs_s %0, " __stringify(ICC_IAR1_EL1) : "=r" (irqstat));
 	asm volatile("nop;nop;nop;nop;");
-	mb();
 #endif /* CONFIG_HAS_NMI */
+	mb();
 
 	return irqstat;
 }
@@ -118,7 +122,7 @@ static uint64_t gic_read_iar_cavium_thunderx(void)
 static uint64_t gic_read_iar(void)
 {
 #ifdef USE_CAVIUM_THUNDER_X
-	if (static_key_false(&is_cavium_thunderx))
+	if (is_cavium_thunderx)
 		return gic_read_iar_cavium_thunderx();
 	else
 #endif
@@ -266,6 +270,7 @@ void arm64_issue_ipi_gicv3(uint32_t cpuid, uint32_t vector)
 {
 	dkprintf("Send irq#%d to cpuid=%d\n", vector, cpuid);
 
+	barrier();
 	if(vector < 16){
 		// send SGI
 		arm64_raise_sgi_gicv3(cpuid, vector);
@@ -304,7 +309,9 @@ void gic_dist_init_gicv3(unsigned long dist_base_pa, unsigned long size)
 
 #ifdef USE_CAVIUM_THUNDER_X
 	/* Cavium ThunderX erratum 23154 */
-	gicv3_check_capabilities();
+	if (MIDR_IMPLEMENTOR(read_cpuid_id()) == ARM_CPU_IMP_CAVIUM) {
+		is_cavium_thunderx = 1;
+	}
 #endif
 }
 
@@ -399,6 +406,7 @@ void gic_enable_gicv3(void)
 	/* Set specific IPI to NMI */
 	writeb_relaxed(GICD_INT_NMI_PRI, rd_sgi_base + GIC_DIST_PRI + INTRID_CPU_STOP);
 	writeb_relaxed(GICD_INT_NMI_PRI, rd_sgi_base + GIC_DIST_PRI + INTRID_MEMDUMP);
+	writeb_relaxed(GICD_INT_NMI_PRI, rd_sgi_base + GIC_DIST_PRI + INTRID_STACK_TRACE);
 
 	/* sync wait */
 	gic_do_wait_for_rwp(rbase);

arch/arm64/kernel/local.c

@@ -20,10 +20,11 @@ size_t arm64_cpu_local_variables_span = LOCALS_SPAN;	/* for debugger */
 void init_processors_local(int max_id)
 {
 	int i = 0;
+	const int sz = (max_id + 1) * KERNEL_STACK_SIZE;
 	union arm64_cpu_local_variables *tmp;
 
 	/* allocate one more for alignment */
-	locals = ihk_mc_alloc_pages((max_id + 1) * (KERNEL_STACK_SIZE / PAGE_SIZE), IHK_MC_AP_CRITICAL);
+	locals = ihk_mc_alloc_pages(((sz + PAGE_SIZE - 1) / PAGE_SIZE), IHK_MC_AP_CRITICAL);
 	locals = (union arm64_cpu_local_variables *)ALIGN_UP((unsigned long)locals, KERNEL_STACK_SIZE);
 
 	/* clear struct process, struct process_vm, struct thread_info area */

arch/arm64/kernel/memory.c

@@ -14,9 +14,7 @@
 #include <context.h>
 #include <kmalloc.h>
 #include <vdso.h>
-
+#include <debug.h>
-#define	dkprintf(...)	do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
 
 #define NOT_IMPLEMENTED()  do { kprintf("%s is not implemented\n", __func__); while(1);} while(0)
 
@@ -1760,6 +1758,12 @@ int visit_pte_range(page_table_t pt, void *start0, void *end0, int pgshift,
 	return initial_lookup.walk(tt, 0, start, end, initial_lookup.callback, &args);
 }
 
+int visit_pte_range_safe(page_table_t pt, void *start0, void *end0, int pgshift,
+                        enum visit_pte_flag flags, pte_visitor_t *funcp, void *arg)
+{
+	return 0;
+}
+
 struct clear_range_args {
 	int free_physical;
 	struct memobj *memobj;
@@ -1801,7 +1805,6 @@ static int clear_range_l1(void *args0, pte_t *ptep, uint64_t base,
 			ihk_mc_free_pages_user(phys_to_virt(phys), npages);
 			dkprintf("%s: freeing regular page at 0x%lx\n", __FUNCTION__, base);
 		}
-		args->vm->currss -= PTL1_SIZE;
 	}
 	
 	return 0;
@@ -1881,7 +1884,6 @@ static int clear_range_middle(void *args0, pte_t *ptep, uint64_t base,
 				ihk_mc_free_pages_user(phys_to_virt(phys), npages);
 				dkprintf("%s(level=%d): freeing large page at 0x%lx\n", __FUNCTION__, level, base);
 			}
-			args->vm->currss -= tbl.pgsize;
 		}
 
 		return 0;
@@ -2920,17 +2922,12 @@ int read_process_vm(struct process_vm *vm, void *kdst, const void *usrc, size_t
 			return error;
 		}
 
-#ifdef POSTK_DEBUG_TEMP_FIX_52 /* NUMA support(memory area determination) */
+		if (!is_mckernel_memory(pa, pa + cpsize)) {
-		if (!is_mckernel_memory(pa)) {
-#else
-		if (pa < ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0) ||
-			pa >= ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0)) {
-#endif /* POSTK_DEBUG_TEMP_FIX_52 */
 			dkprintf("%s: pa is outside of LWK memory, to: %p, pa: %p,"
 				"cpsize: %d\n", __FUNCTION__, to, pa, cpsize);
 			va = ihk_mc_map_virtual(pa, 1, PTATTR_ACTIVE);
 			memcpy(to, va, cpsize);
-			ihk_mc_unmap_virtual(va, 1, 1);
+			ihk_mc_unmap_virtual(va, 1);
 		}
 		else {
 			va = phys_to_virt(pa);
@@ -3003,17 +3000,12 @@ int write_process_vm(struct process_vm *vm, void *udst, const void *ksrc, size_t
 			return error;
 		}
 
-#ifdef POSTK_DEBUG_TEMP_FIX_52 /* NUMA support(memory area determination) */
+		if (!is_mckernel_memory(pa, pa + cpsize)) {
-		if (!is_mckernel_memory(pa)) {
-#else
-		if (pa < ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0) ||
-			pa >= ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0)) {
-#endif /* POSTK_DEBUG_TEMP_FIX_52 */
 			dkprintf("%s: pa is outside of LWK memory, from: %p,"
 				"pa: %p, cpsize: %d\n", __FUNCTION__, from, pa, cpsize);
 			va = ihk_mc_map_virtual(pa, 1, PTATTR_WRITABLE|PTATTR_ACTIVE);
 			memcpy(va, from, cpsize);
-			ihk_mc_unmap_virtual(va, 1, 1);
+			ihk_mc_unmap_virtual(va, 1);
 		}
 		else {
 			va = phys_to_virt(pa);
@@ -3074,17 +3066,12 @@ int patch_process_vm(struct process_vm *vm, void *udst, const void *ksrc, size_t
 			return error;
 		}
 
-#ifdef POSTK_DEBUG_TEMP_FIX_52 /* NUMA support(memory area determination) */
+		if (!is_mckernel_memory(pa, pa + cpsize)) {
-		if (!is_mckernel_memory(pa)) {
-#else
-		if (pa < ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0) ||
-			pa >= ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0)) {
-#endif /* POSTK_DEBUG_TEMP_FIX_52 */
 			dkprintf("%s: pa is outside of LWK memory, from: %p,"
 				"pa: %p, cpsize: %d\n", __FUNCTION__, from, pa, cpsize);
 			va = ihk_mc_map_virtual(pa, 1, PTATTR_WRITABLE|PTATTR_ACTIVE);
 			memcpy(va, from, cpsize);
-			ihk_mc_unmap_virtual(va, 1, 1);
+			ihk_mc_unmap_virtual(va, 1);
 		}
 		else {
 			va = phys_to_virt(pa);

arch/arm64/kernel/perfctr.c

@@ -93,21 +93,50 @@ int ihk_mc_perfctr_init(int counter, uint64_t config, int mode)
 	return ret;
 }
 
-int ihk_mc_perfctr_start(int counter)
+int ihk_mc_perfctr_start(unsigned long counter_mask)
 {
-	int ret;
+	int ret = 0;
-	ret = cpu_pmu.enable_counter(counter);
+	int counter;
-	return ret;
+	unsigned long counter_bit;
+
+	for (counter = 0, counter_bit = 1;
+	     counter_bit < counter_mask;
+	     counter++, counter_bit <<= 1) {
+		if (!(counter_mask & counter_bit))
+			continue;
+
+		ret = cpu_pmu.enable_counter(counter_mask);
+		if (ret < 0)
+			break;
+	}
+
+	return ret < 0 ? ret : 0;
 }
 
-int ihk_mc_perfctr_stop(int counter)
+int ihk_mc_perfctr_stop(unsigned long counter_mask)
 {
-	cpu_pmu.disable_counter(counter);
+	int ret = 0;
+	int counter;
+	unsigned long counter_bit;
 
-	// ihk_mc_perfctr_startが呼ばれるときには、
+	for (counter = 0, counter_bit = 1;
-	// init系関数が呼ばれるのでdisableにする。
+	     counter_bit < counter_mask;
-	cpu_pmu.disable_intens(counter);
+	     counter++, counter_bit <<= 1) {
-	return 0;
+		if (!(counter_mask & counter_bit))
+			continue;
+
+		ret = cpu_pmu.disable_counter(counter);
+		if (ret < 0)
+			break;
+
+		// ihk_mc_perfctr_startが呼ばれるときには、
+		// init系関数が呼ばれるのでdisableにする。
+		ret = cpu_pmu.disable_intens(counter);
+		if (ret < 0)
+			break;
+	}
+
+	return ret < 0 ? ret : 0;
 }
 
 int ihk_mc_perfctr_reset(int counter)

arch/arm64/kernel/perfctr_armv8pmu.c

@@ -4,16 +4,14 @@
 #include <ihk/perfctr.h>
 #include <errno.h>
 #include <ihk/debug.h>
+#include <debug.h>
 
 #define BIT(nr) (1UL << (nr))
 
 //#define DEBUG_PRINT_PMU
 #ifdef DEBUG_PRINT_PMU
-#define	dkprintf(...) kprintf(__VA_ARGS__)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...) kprintf(__VA_ARGS__)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define	ekprintf(...) kprintf(__VA_ARGS__)
 #endif
 
 

arch/arm64/kernel/psci.c

@@ -21,15 +21,13 @@
 #include <ihk/debug.h>
 #include <compiler.h>
 #include <lwk/compiler.h>
+#include <debug.h>
 
 //#define DEBUG_PRINT_PSCI
 
 #ifdef DEBUG_PRINT_PSCI
-#define dkprintf(...) kprintf(__VA_ARGS__)
+#undef DDEBUG_DEFAULT
-#define ekprintf(...) kprintf(__VA_ARGS__)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define ekprintf(...) kprintf(__VA_ARGS__)
 #endif
 
 #define PSCI_POWER_STATE_TYPE_POWER_DOWN	1

arch/arm64/kernel/ptrace.c

@@ -11,22 +11,17 @@
 #include <hwcap.h>
 #include <string.h>
 #include <thread_info.h>
+#include <debug.h>
 
 //#define DEBUG_PRINT_SC
 
 #ifdef DEBUG_PRINT_SC
-#define dkprintf kprintf
+#undef DDEBUG_DEFAULT
-#define ekprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define ekprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
 #endif
 
 #define NOT_IMPLEMENTED()  do { kprintf("%s is not implemented\n", __func__); while(1);} while(0)
 
-#define BUG_ON(condition) do { if (condition) { kprintf("PANIC: %s: %s(line:%d)\n",\
-				__FILE__, __FUNCTION__, __LINE__); panic(""); } } while(0)
-
 extern void save_debugreg(unsigned long *debugreg);
 extern unsigned long do_kill(struct thread *thread, int pid, int tid, int sig, struct siginfo *info, int ptracecont);
 extern int interrupt_from_user(void *);
@@ -959,13 +954,9 @@ void ptrace_report_signal(struct thread *thread, int sig)
 	}
 	thread->exit_status = sig;
 	/* Transition thread state */
-#ifdef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
 	proc->status = PS_DELAY_TRACED;
-#else /* POSTK_DEBUG_TEMP_FIX_41 */
-	proc->status = PS_TRACED;
-#endif /* POSTK_DEBUG_TEMP_FIX_41 */
 	thread->status = PS_TRACED;
-	proc->ptrace &= ~PT_TRACE_SYSCALL_MASK;
+	proc->ptrace &= ~PT_TRACE_SYSCALL;
 	if (sig == SIGSTOP || sig == SIGTSTP ||
 			sig == SIGTTIN || sig == SIGTTOU) {
 		proc->signal_flags |= SIGNAL_STOP_STOPPED;
@@ -982,10 +973,6 @@ void ptrace_report_signal(struct thread *thread, int sig)
 	info._sifields._sigchld.si_pid = thread->tid;
 	info._sifields._sigchld.si_status = thread->exit_status;
 	do_kill(cpu_local_var(current), parent_pid, -1, SIGCHLD, &info, 0);
-#ifndef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
-	/* Wake parent (if sleeping in wait4()) */
-	waitq_wakeup(&proc->parent->waitpid_q);
-#endif /* !POSTK_DEBUG_TEMP_FIX_41 */
 
 	dkprintf("ptrace_report_signal,sleeping\n");
 	/* Sleep */

arch/arm64/kernel/syscall.c

@@ -12,6 +12,10 @@
 #include <lwk/compiler.h>
 #include <hwcap.h>
 #include <prctl.h>
+#include <limits.h>
+#include <syscall.h>
+#include <uio.h>
+#include <debug.h>
 
 extern void ptrace_report_signal(struct thread *thread, int sig);
 extern void clear_single_step(struct thread *thread);
@@ -25,18 +29,12 @@ static void __check_signal(unsigned long rc, void *regs, int num, int irq_disabl
 //#define DEBUG_PRINT_SC
 
 #ifdef DEBUG_PRINT_SC
-#define dkprintf kprintf
+#undef DDEBUG_DEFAULT
-#define ekprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define ekprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
 #endif
 
 #define NOT_IMPLEMENTED()  do { kprintf("%s is not implemented\n", __func__); while(1);} while(0)
 
-#define BUG_ON(condition) do { if (condition) { kprintf("PANIC: %s: %s(line:%d)\n",\
-				__FILE__, __FUNCTION__, __LINE__); panic(""); } } while(0)
-
 uintptr_t debug_constants[] = {
 	sizeof(struct cpu_local_var),
 	offsetof(struct cpu_local_var, current),
@@ -57,7 +55,7 @@ static int cpuid_head = 1;
 
 extern int num_processors;
 
-int obtain_clone_cpuid(cpu_set_t *cpu_set) {
+int obtain_clone_cpuid(cpu_set_t *cpu_set, int use_last) {
 	int min_queue_len = -1;
 	int i, min_cpu = -1;
 
@@ -1175,19 +1173,10 @@ do_signal(unsigned long rc, void *regs0, struct thread *thread, struct sig_pendi
 				/* Reap and set new signal_flags */
 				proc->signal_flags = SIGNAL_STOP_STOPPED;
 
-#ifdef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
 				proc->status = PS_DELAY_STOPPED;
-#else /* POSTK_DEBUG_TEMP_FIX_41 */
-				proc->status = PS_STOPPED;
-#endif /* POSTK_DEBUG_TEMP_FIX_41 */
 				thread->status = PS_STOPPED;
 				mcs_rwlock_writer_unlock(&proc->update_lock, &lock);	
 
-#ifndef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
-				/* Wake up the parent who tried wait4 and sleeping */
-				waitq_wakeup(&proc->parent->waitpid_q);
-#endif /* !POSTK_DEBUG_TEMP_FIX_41 */
-
 				dkprintf("do_signal(): pid: %d, tid: %d SIGSTOP, sleeping\n", 
 					proc->pid, thread->tid);
 				/* Sleep */
@@ -1204,19 +1193,10 @@ do_signal(unsigned long rc, void *regs0, struct thread *thread, struct sig_pendi
 			/* Update thread state in fork tree */
 			mcs_rwlock_writer_lock(&proc->update_lock, &lock);	
 			thread->exit_status = SIGTRAP;
-#ifdef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
 			proc->status = PS_DELAY_TRACED;
-#else /* POSTK_DEBUG_TEMP_FIX_41 */
-			proc->status = PS_TRACED;
-#endif /* POSTK_DEBUG_TEMP_FIX_41 */
 			thread->status = PS_TRACED;
 			mcs_rwlock_writer_unlock(&proc->update_lock, &lock);	
 
-#ifndef POSTK_DEBUG_TEMP_FIX_41 /* early to wait4() wakeup for ptrace, fix. */
-			/* Wake up the parent who tried wait4 and sleeping */
-			waitq_wakeup(&thread->proc->parent->waitpid_q);
-#endif /* !POSTK_DEBUG_TEMP_FIX_41 */
-
 			/* Sleep */
 			dkprintf("do_signal,SIGTRAP,sleeping\n");
 
@@ -1321,6 +1301,17 @@ interrupt_from_user(void *regs0)
 	return((regs->pstate & PSR_MODE_MASK) == PSR_MODE_EL0t);
 }
 
+void save_syscall_return_value(int num, unsigned long rc)
+{
+	/*
+	 * Save syscall return value.
+	 */
+	if (cpu_local_var(current) && cpu_local_var(current)->uctx &&
+			num != __NR_rt_sigsuspend) {
+		ihk_mc_syscall_arg0(cpu_local_var(current)->uctx) = rc;
+	}
+}
+
 void
 check_signal(unsigned long rc, void *regs0, int num)
 {
@@ -1345,16 +1336,6 @@ __check_signal(unsigned long rc, void *regs0, int num, int irq_disabled)
 		return;
 	thread = cpu_local_var(current);
 
-	/**
-	 * If check_signal is called from syscall(), 
-	 * then save syscall return value.
-	 */
-	if((regs == NULL)&&(num != __NR_rt_sigsuspend)){ /* It's call from syscall! */
-	 	// Get user context through current thread
-	 	// and update syscall return.
-	 	ihk_mc_syscall_arg0(thread->uctx) = rc;
-	}
-
 	if(thread == NULL || thread->proc->pid == 0){
 		struct thread *t;
 		irqstate = ihk_mc_spinlock_lock(&(cpu_local_var(runq_lock)));
@@ -1591,7 +1572,7 @@ done:
 		return 0;
 	}
 
-	if (tthread->thread_offloaded) {
+	if (tthread->uti_state == UTI_STATE_RUNNING_IN_LINUX) {
 		interrupt_syscall(tthread, sig);
 		release_thread(tthread);
 		return 0;
@@ -1726,7 +1707,7 @@ SYSCALL_DECLARE(mmap)
 		| MAP_NONBLOCK		// 0x10000
 		;
 
-	const intptr_t addr0 = ihk_mc_syscall_arg0(ctx);
+	const uintptr_t addr0 = ihk_mc_syscall_arg0(ctx);
 	const size_t len0 = ihk_mc_syscall_arg1(ctx);
 	const int prot = ihk_mc_syscall_arg2(ctx);
 	const int flags0 = ihk_mc_syscall_arg3(ctx);
@@ -1735,7 +1716,7 @@ SYSCALL_DECLARE(mmap)
 	struct thread *thread = cpu_local_var(current);
 	struct vm_regions *region = &thread->vm->region;
 	int error;
-	intptr_t addr = 0;
+	uintptr_t addr = 0;
 	size_t len;
 	int flags = flags0;
 	size_t pgsize;
@@ -1798,8 +1779,9 @@ SYSCALL_DECLARE(mmap)
 		goto out;
 	}
 
-	if ((flags & MAP_FIXED) && ((addr < region->user_start)
+	if (addr < region->user_start
-			|| (region->user_end <= addr))) {
+			|| region->user_end <= addr
+			|| len > (region->user_end - region->user_start)) {
 		ekprintf("sys_mmap(%lx,%lx,%x,%x,%x,%lx):ENOMEM\n",
 				addr0, len0, prot, flags0, fd, off0);
 		error = -ENOMEM;
@@ -1866,4 +1848,645 @@ save_uctx(void *uctx, struct pt_regs *regs)
 	/* TODO: skeleton for UTI */
 }
 
+int do_process_vm_read_writev(int pid, 
+		const struct iovec *local_iov,
+		unsigned long liovcnt,
+		const struct iovec *remote_iov,
+		unsigned long riovcnt,
+		unsigned long flags,
+		int op)
+{
+	int ret = -EINVAL;	
+	int li, ri;
+	int pli, pri;
+	off_t loff, roff;
+	size_t llen = 0, rlen = 0;
+	size_t copied = 0;
+	size_t to_copy;
+	struct thread *lthread = cpu_local_var(current);
+	struct process *rproc;
+	struct process *lproc = lthread->proc;
+	struct process_vm *rvm = NULL;
+	unsigned long rphys;
+	unsigned long rpage_left;
+	unsigned long psize;
+	void *rva;
+	struct vm_range *range;
+	struct mcs_rwlock_node_irqsave lock;
+	struct mcs_rwlock_node update_lock;
+
+	/* Sanity checks */
+	if (flags) {
+		return -EINVAL;
+	}
+	
+	if (liovcnt > IOV_MAX || riovcnt > IOV_MAX) {
+		return -EINVAL;
+	}
+
+	/* Check if parameters are okay */
+	ihk_mc_spinlock_lock_noirq(&lthread->vm->memory_range_lock);
+
+	range = lookup_process_memory_range(lthread->vm, 
+			(uintptr_t)local_iov, 
+			(uintptr_t)(local_iov + liovcnt * sizeof(struct iovec)));
+
+	if (!range) {
+		ret = -EFAULT; 
+		goto arg_out;
+	}
+
+	range = lookup_process_memory_range(lthread->vm, 
+			(uintptr_t)remote_iov, 
+			(uintptr_t)(remote_iov + riovcnt * sizeof(struct iovec)));
+
+	if (!range) {
+		ret = -EFAULT; 
+		goto arg_out;
+	}
+
+	ret = 0;
+arg_out:
+	ihk_mc_spinlock_unlock_noirq(&lthread->vm->memory_range_lock);
+
+	if (ret != 0) {
+		goto out;
+	}
+
+	for (li = 0; li < liovcnt; ++li) {
+		llen += local_iov[li].iov_len;
+		dkprintf("local_iov[%d].iov_base: 0x%lx, len: %lu\n",
+			li, local_iov[li].iov_base, local_iov[li].iov_len);
+	}
+
+	for (ri = 0; ri < riovcnt; ++ri) {
+		rlen += remote_iov[ri].iov_len;
+		dkprintf("remote_iov[%d].iov_base: 0x%lx, len: %lu\n",
+			ri, remote_iov[ri].iov_base, remote_iov[ri].iov_len);
+	}
+
+	if (llen != rlen) {
+		return -EINVAL;
+	}
+	
+	/* Find remote process */
+	rproc = find_process(pid, &lock);
+	if (!rproc) {
+		ret = -ESRCH;
+		goto out;
+	}
+
+	mcs_rwlock_reader_lock_noirq(&rproc->update_lock, &update_lock);
+	if(rproc->status == PS_EXITED ||
+	   rproc->status == PS_ZOMBIE){
+		mcs_rwlock_reader_unlock_noirq(&rproc->update_lock, &update_lock);
+		process_unlock(rproc, &lock);
+		ret = -ESRCH;
+		goto out;
+	}
+	rvm = rproc->vm;
+	hold_process_vm(rvm);
+	mcs_rwlock_reader_unlock_noirq(&rproc->update_lock, &update_lock);
+	process_unlock(rproc, &lock);
+
+	if (lproc->euid != 0 &&
+	    (lproc->ruid != rproc->ruid ||
+	     lproc->ruid != rproc->euid ||
+	     lproc->ruid != rproc->suid ||
+	     lproc->rgid != rproc->rgid ||
+	     lproc->rgid != rproc->egid ||
+	     lproc->rgid != rproc->sgid)) {
+		ret = -EPERM;
+		goto out;
+	}
+
+	dkprintf("pid %d found, doing %s: liovcnt: %d, riovcnt: %d \n", pid,
+		(op == PROCESS_VM_READ) ? "PROCESS_VM_READ" : "PROCESS_VM_WRITE",
+		liovcnt, riovcnt);
+
+	pli = pri = -1; /* Previous indeces in iovecs */
+	li = ri = 0; /* Current indeces in iovecs */
+	loff = roff = 0; /* Offsets in current iovec */
+
+	/* Now iterate and do the copy */
+	while (copied < llen) {
+		int faulted = 0;
+
+		/* New local vector? */
+		if (pli != li) {
+			struct vm_range *range;
+
+			ihk_mc_spinlock_lock_noirq(&lthread->vm->memory_range_lock);
+
+			/* Is base valid? */
+			range = lookup_process_memory_range(lthread->vm,
+					(uintptr_t)local_iov[li].iov_base,
+					(uintptr_t)(local_iov[li].iov_base + 1));
+
+			if (!range) {
+				ret = -EFAULT;
+				goto pli_out;
+			}
+
+			/* Is range valid? */
+			range = lookup_process_memory_range(lthread->vm,
+					(uintptr_t)local_iov[li].iov_base,
+					(uintptr_t)(local_iov[li].iov_base + local_iov[li].iov_len));
+
+			if (range == NULL) {
+				ret = -EINVAL;
+				goto pli_out;
+			}
+
+			if (!(range->flag & ((op == PROCESS_VM_READ) ?
+				VR_PROT_WRITE : VR_PROT_READ))) {
+				ret = -EFAULT;
+				goto pli_out;
+			}
+
+			ret = 0;
+pli_out:
+			ihk_mc_spinlock_unlock_noirq(&lthread->vm->memory_range_lock);
+
+			if (ret != 0) {
+				goto out;
+			}
+
+			pli = li;
+		}
+
+		/* New remote vector? */
+		if (pri != ri) {
+			struct vm_range *range;
+
+			ihk_mc_spinlock_lock_noirq(&rvm->memory_range_lock);
+
+			/* Is base valid? */
+			range = lookup_process_memory_range(rvm,
+					(uintptr_t)remote_iov[li].iov_base,
+					(uintptr_t)(remote_iov[li].iov_base + 1));
+
+			if (range == NULL) {
+				ret = -EFAULT;
+				goto pri_out;
+			}
+
+			/* Is range valid? */
+			range = lookup_process_memory_range(rvm,
+					(uintptr_t)remote_iov[li].iov_base,
+					(uintptr_t)(remote_iov[li].iov_base + remote_iov[li].iov_len));
+
+			if (range == NULL) {
+				ret = -EINVAL;
+				goto pri_out;
+			}
+
+			if (!(range->flag & ((op == PROCESS_VM_READ) ?
+				VR_PROT_READ : VR_PROT_WRITE))) {
+				ret = -EFAULT;
+				goto pri_out;
+			}
+
+			ret = 0;
+pri_out:
+			ihk_mc_spinlock_unlock_noirq(&rvm->memory_range_lock);
+
+			if (ret != 0) {
+				goto out;
+			}
+
+			pri = ri;
+		}
+
+		/* Figure out how much we can copy at most in this iteration */
+		to_copy = (local_iov[li].iov_len - loff);	
+		if ((remote_iov[ri].iov_len - roff) < to_copy) {
+			to_copy = remote_iov[ri].iov_len - roff;
+		}
+
+retry_lookup:
+		/* TODO: remember page and do this only if necessary */
+		ret = ihk_mc_pt_virt_to_phys_size(rvm->address_space->page_table,
+				remote_iov[ri].iov_base + roff, &rphys, &psize);
+
+		if (ret) {
+			uint64_t reason = PF_POPULATE | PF_WRITE | PF_USER;
+			void *addr;
+
+			if (faulted) {
+				ret = -EFAULT;
+				goto out;
+			}
+
+			/* Fault in pages */
+			for (addr = (void *)
+					(((unsigned long)remote_iov[ri].iov_base + roff)
+					& PAGE_MASK);
+					addr < (remote_iov[ri].iov_base + roff + to_copy);
+					addr += PAGE_SIZE) {
+
+				ret = page_fault_process_vm(rvm, addr, reason);
+				if (ret) {
+					ret = -EFAULT;
+					goto out;
+				}
+			}
+
+			faulted = 1;
+			goto retry_lookup;
+		}
+
+		rpage_left = ((((unsigned long)remote_iov[ri].iov_base + roff +
+			psize) & ~(psize - 1)) -
+			((unsigned long)remote_iov[ri].iov_base + roff));
+		if (rpage_left < to_copy) {
+			to_copy = rpage_left;
+		}
+
+		rva = phys_to_virt(rphys);
+
+		fast_memcpy(
+				(op == PROCESS_VM_READ) ? local_iov[li].iov_base + loff : rva,
+				(op == PROCESS_VM_READ) ? rva : local_iov[li].iov_base + loff,
+				to_copy);
+
+		copied += to_copy;
+		dkprintf("local_iov[%d]: 0x%lx %s remote_iov[%d]: 0x%lx, %lu copied, psize: %lu, rpage_left: %lu\n",
+			li, local_iov[li].iov_base + loff, 
+			(op == PROCESS_VM_READ) ? "<-" : "->", 
+			ri, remote_iov[ri].iov_base + roff, to_copy,
+			psize, rpage_left);
+
+		loff += to_copy;
+		roff += to_copy;
+
+		if (loff == local_iov[li].iov_len) {
+			li++;
+			loff = 0;
+		}
+
+		if (roff == remote_iov[ri].iov_len) {
+			ri++;
+			roff = 0;
+		}
+	}
+
+	release_process_vm(rvm);
+
+	return copied;
+
+out:
+	if(rvm)
+		release_process_vm(rvm);
+	return ret;
+}
+
+int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
+{
+	int i, i_s, i_e, phase = 1;
+	struct move_pages_smp_req *mpsr =
+		(struct move_pages_smp_req *)arg;
+	struct process_vm *vm = mpsr->proc->vm;
+	int count = mpsr->count;
+	struct page_table *save_pt;
+	extern struct page_table *get_init_page_table(void);
+
+	i_s = (count / nr_cpus) * cpu_index;
+	i_e = i_s + (count / nr_cpus);
+	if (cpu_index == (nr_cpus - 1)) {
+		i_e = count;
+	}
+
+	/* Load target process' PT so that we can access user-space */
+	save_pt = cpu_local_var(current) == &cpu_local_var(idle) ?
+		get_init_page_table() :
+		cpu_local_var(current)->vm->address_space->page_table;
+
+	if (save_pt != vm->address_space->page_table) {
+		ihk_mc_load_page_table(vm->address_space->page_table);
+	}
+	else {
+		save_pt = NULL;
+	}
+
+	if (nr_cpus == 1) {
+		switch (cpu_index) {
+			case 0:
+				memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
+						sizeof(void *) * count);
+				memcpy(mpsr->status, mpsr->user_status,
+						sizeof(int) * count);
+				memcpy(mpsr->nodes, mpsr->user_nodes,
+						sizeof(int) * count);
+				memset(mpsr->ptep, 0, sizeof(pte_t) * count);
+				memset(mpsr->status, 0, sizeof(int) * count);
+				memset(mpsr->nr_pages, 0, sizeof(int) * count);
+				memset(mpsr->dst_phys, 0,
+						sizeof(unsigned long) * count);
+				mpsr->nodes_ready = 1;
+				break;
+
+			default:
+				break;
+		}
+	}
+	else if (nr_cpus > 1 && nr_cpus < 4) {
+		switch (cpu_index) {
+			case 0:
+				memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
+						sizeof(void *) * count);
+				memcpy(mpsr->status, mpsr->user_status,
+						sizeof(int) * count);
+			case 1:
+				memcpy(mpsr->nodes, mpsr->user_nodes,
+						sizeof(int) * count);
+				memset(mpsr->ptep, 0, sizeof(pte_t) * count);
+				memset(mpsr->status, 0, sizeof(int) * count);
+				memset(mpsr->nr_pages, 0, sizeof(int) * count);
+				memset(mpsr->dst_phys, 0,
+						sizeof(unsigned long) * count);
+				mpsr->nodes_ready = 1;
+				break;
+
+			default:
+				break;
+		}
+	}
+	else if (nr_cpus >= 4 && nr_cpus < 8) {
+		switch (cpu_index) {
+			case 0:
+				memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
+						sizeof(void *) * count);
+				break;
+			case 1:
+				memcpy(mpsr->status, mpsr->user_status,
+						sizeof(int) * count);
+				break;
+			case 2:
+				memcpy(mpsr->nodes, mpsr->user_nodes,
+						sizeof(int) * count);
+				mpsr->nodes_ready = 1;
+				break;
+			case 3:
+				memset(mpsr->ptep, 0, sizeof(pte_t) * count);
+				memset(mpsr->status, 0, sizeof(int) * count);
+				memset(mpsr->nr_pages, 0, sizeof(int) * count);
+				memset(mpsr->dst_phys, 0,
+						sizeof(unsigned long) * count);
+				break;
+
+			default:
+				break;
+		}
+	}
+	else if (nr_cpus >= 8) {
+		switch (cpu_index) {
+			case 0:
+				memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
+						sizeof(void *) * (count / 2));
+				break;
+			case 1:
+				memcpy(mpsr->virt_addr + (count / 2),
+						mpsr->user_virt_addr + (count / 2),
+						sizeof(void *) * (count / 2));
+				break;
+			case 2:
+				memcpy(mpsr->status, mpsr->user_status,
+						sizeof(int) * count);
+				break;
+			case 3:
+				memcpy(mpsr->nodes, mpsr->user_nodes,
+						sizeof(int) * count);
+				mpsr->nodes_ready = 1;
+				break;
+			case 4:
+				memset(mpsr->ptep, 0, sizeof(pte_t) * count);
+				break;
+			case 5:
+				memset(mpsr->status, 0, sizeof(int) * count);
+				break;
+			case 6:
+				memset(mpsr->nr_pages, 0, sizeof(int) * count);
+				break;
+			case 7:
+				memset(mpsr->dst_phys, 0,
+						sizeof(unsigned long) * count);
+				break;
+
+			default:
+				break;
+		}
+	}
+
+	while (!(volatile int)mpsr->nodes_ready) {
+		cpu_pause();
+	}
+
+	/* NUMA verification in parallel */
+	for (i = i_s; i < i_e; i++) {
+		if (mpsr->nodes[i] < 0 ||
+				mpsr->nodes[i] >= ihk_mc_get_nr_numa_nodes() ||
+				!test_bit(mpsr->nodes[i],
+					mpsr->proc->vm->numa_mask)) {
+			mpsr->phase_ret = -EINVAL;
+			break;
+		}
+	}
+
+	/* Barrier */
+	ihk_atomic_inc(&mpsr->phase_done);
+	while (ihk_atomic_read(&mpsr->phase_done) <
+			(phase * nr_cpus)) {
+		cpu_pause();
+	}
+
+	if (mpsr->phase_ret != 0) {
+		goto out;
+	}
+
+	dkprintf("%s: phase %d done\n", __FUNCTION__, phase);
+	++phase;
+
+	/* PTE lookup in parallel */
+	for (i = i_s; i < i_e; i++) {
+		void *phys;
+		size_t pgsize;
+		int p2align;
+		/*
+		 * XXX: No page structures for anonymous mappings.
+		 * Look up physical addresses by scanning page tables.
+		 */
+		mpsr->ptep[i] = ihk_mc_pt_lookup_pte(vm->address_space->page_table,
+				(void *)mpsr->virt_addr[i], 0, &phys, &pgsize, &p2align);
+
+		/* PTE valid? */
+		if (!mpsr->ptep[i] || !pte_is_present(mpsr->ptep[i])) {
+			mpsr->status[i] = -ENOENT;
+			mpsr->ptep[i] = NULL;
+			continue;
+		}
+
+		/* PTE is file? */
+		if (pte_is_fileoff(mpsr->ptep[i], PAGE_SIZE)) {
+			mpsr->status[i] = -EINVAL;
+			mpsr->ptep[i] = NULL;
+			continue;
+		}
+
+		dkprintf("%s: virt 0x%lx:%lu requested to be moved to node %d\n",
+			__FUNCTION__, mpsr->virt_addr[i], pgsize, mpsr->nodes[i]);
+
+		/* Large page? */
+		if (pgsize > PAGE_SIZE) {
+			int nr_sub_pages = (pgsize / PAGE_SIZE);
+			int j;
+
+			if (i + nr_sub_pages > count) {
+				kprintf("%s: ERROR: page at index %d exceeds the region\n",
+						__FUNCTION__, i);
+				mpsr->status[i] = -EINVAL;
+				break;
+			}
+
+			/* Is it contiguous across nr_sub_pages and all
+			 * requested to be moved to the same target node? */
+			for (j = 0; j < nr_sub_pages; ++j) {
+				if (mpsr->virt_addr[i + j] !=
+				(mpsr->virt_addr[i] + (j * PAGE_SIZE)) ||
+						mpsr->nodes[i] != mpsr->nodes[i + j]) {
+					kprintf("%s: ERROR: virt address or node at index %d"
+							" is inconsistent\n",
+							__FUNCTION__, i + j);
+					mpsr->phase_ret = -EINVAL;
+					goto pte_out;
+				}
+			}
+
+			mpsr->nr_pages[i] = nr_sub_pages;
+			i += (nr_sub_pages - 1);
+		}
+		else {
+			mpsr->nr_pages[i] = 1;
+		}
+	}
+
+pte_out:
+	/* Barrier */
+	ihk_atomic_inc(&mpsr->phase_done);
+	while (ihk_atomic_read(&mpsr->phase_done) <
+			(phase * nr_cpus)) {
+		cpu_pause();
+	}
+
+	if (mpsr->phase_ret != 0) {
+		goto out;
+	}
+
+	dkprintf("%s: phase %d done\n", __FUNCTION__, phase);
+	++phase;
+
+	if (cpu_index == 0) {
+		/* Allocate new pages on target NUMA nodes */
+		for (i = 0; i < count; i++) {
+			int pgalign = 0;
+			int j;
+			void *dst;
+
+			if (!mpsr->ptep[i] || mpsr->status[i] < 0 || !mpsr->nr_pages[i])
+				continue;
+
+			/* TODO: store pgalign info in an array as well? */
+			if (mpsr->nr_pages[i] > 1) {
+				if (mpsr->nr_pages[i] * PAGE_SIZE == PTL2_SIZE)
+					pgalign = PTL2_SHIFT - PTL1_SHIFT;
+			}
+
+			dst = ihk_mc_alloc_aligned_pages_node(mpsr->nr_pages[i],
+					pgalign, IHK_MC_AP_USER, mpsr->nodes[i]);
+
+			if (!dst) {
+				mpsr->status[i] = -ENOMEM;
+				continue;
+			}
+
+			for (j = i; j < (i + mpsr->nr_pages[i]); ++j) {
+				mpsr->status[j] = mpsr->nodes[i];
+			}
+
+			mpsr->dst_phys[i] = virt_to_phys(dst);
+
+			dkprintf("%s: virt 0x%lx:%lu to node %d, pgalign: %d,"
+					" allocated phys: 0x%lx\n",
+					__FUNCTION__, mpsr->virt_addr[i],
+					mpsr->nr_pages[i] * PAGE_SIZE,
+					mpsr->nodes[i], pgalign, mpsr->dst_phys[i]);
+		}
+	}
+
+	/* Barrier */
+	ihk_atomic_inc(&mpsr->phase_done);
+	while (ihk_atomic_read(&mpsr->phase_done) <
+			(phase * nr_cpus)) {
+		cpu_pause();
+	}
+
+	if (mpsr->phase_ret != 0) {
+		goto out;
+	}
+
+	dkprintf("%s: phase %d done\n", __FUNCTION__, phase);
+	++phase;
+
+	/* Copy, PTE update, memfree in parallel */
+	for (i = i_s; i < i_e; ++i) {
+		if (!mpsr->dst_phys[i])
+			continue;
+
+		fast_memcpy(phys_to_virt(mpsr->dst_phys[i]),
+				phys_to_virt(pte_get_phys(mpsr->ptep[i])),
+				mpsr->nr_pages[i] * PAGE_SIZE);
+
+		ihk_mc_free_pages(
+				phys_to_virt(pte_get_phys(mpsr->ptep[i])),
+				mpsr->nr_pages[i]);
+
+		pte_update_phys(mpsr->ptep[i], mpsr->dst_phys[i]);
+
+		dkprintf("%s: virt 0x%lx:%lu copied and remapped to phys: 0x%lu\n",
+				__FUNCTION__, mpsr->virt_addr[i],
+				mpsr->nr_pages[i] * PAGE_SIZE,
+				mpsr->dst_phys[i]);
+	}
+
+	/* XXX: do a separate SMP call with only CPUs running threads
+	 * of this process? */
+	if (cpu_local_var(current)->proc == mpsr->proc) {
+		/* Invalidate all TLBs */
+		for (i = 0; i < mpsr->count; i++) {
+			if (!mpsr->dst_phys[i])
+				continue;
+
+			flush_tlb_single((unsigned long)mpsr->virt_addr[i]);
+		}
+	}
+
+out:
+	if (save_pt) {
+		ihk_mc_load_page_table(save_pt);
+	}
+
+	return mpsr->phase_ret;
+}
+
+time_t time(void) {
+	struct timespec ats;
+
+	if (gettime_local_support) {
+		calculate_time_from_tsc(&ats);
+		return ats.tv_sec;
+	}
+
+	return (time_t)0;
+}
+
 /*** End of File ***/

arch/arm64/kernel/vdso.c

@@ -14,15 +14,13 @@
 #include <ihk/debug.h>
 #include <ikc/queue.h>
 #include <vdso.h>
+#include <debug.h>
 
 //#define DEBUG_PRINT_VDSO
 
 #ifdef DEBUG_PRINT_VDSO
-#define	dkprintf(...)	kprintf(__VA_ARGS__)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...)	do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
 #endif
 
 #ifdef POSTK_DEBUG_ARCH_DEP_52

arch/arm64/kernel/vdso/gettimeofday.c

@@ -1,5 +1,7 @@
 /* gettimeofday.c COPYRIGHT FUJITSU LIMITED 2016 */
 
+#include <affinity.h>
+#include <arch-memory.h>
 #include <time.h>
 #include <syscall.h>
 #include <registers.h>

arch/x86/elfboot/raw.lds

@@ -1,37 +0,0 @@
-ENTRY(_start)
-
-PHDRS
-{
-	text PT_LOAD;
-	data PT_LOAD;
-}
-
-SECTIONS
-{
-	. = SIZEOF_HEADERS;
-    . = ALIGN(4096);
-	.text :	{
-	      *(.text)
-	} :text
-	.data : {
-	      *(.data)
-	      *(.data.*)
-	} :data
-	.rodata : {
-	      *(.rodata .rodata.*)
-	} :data
-
-	. = ALIGN(8);
-	.bss : {
-	     _bss_start = .;
-	     *(.bss .bss.*)
-	     _bss_end = .;
-         . = ALIGN(4096);
-         _stack_end = .;
-    } :data
-
-	/DISCARD/ : {
-	     *(.eh_frame)
-         *(.note.gnu.build-id)
-	}
-}

arch/x86/kernel/Makefile.arch

@@ -1,7 +0,0 @@
-IHK_OBJS += cpu.o interrupt.o memory.o trampoline.o local.o context.o
-IHK_OBJS += perfctr.o syscall.o vsyscall.o
-# POSTK_DEBUG_ARCH_DEP_18 coredump arch separation.
-# IHK_OBJS added coredump.o
-ifeq ($(ARCH), arm64)
-IHK_OBJS += coredump.o
-endif

arch/x86_64/elfboot/raw.lds

@@ -0,0 +1,37 @@
+ENTRY(_start)
+
+PHDRS
+{
+	text PT_LOAD;
+	data PT_LOAD;
+}
+
+SECTIONS
+{
+	. = SIZEOF_HEADERS;
+	. = ALIGN(4096);
+	.text :	{
+		*(.text)
+	} :text
+	.data : {
+		*(.data)
+		*(.data.*)
+	} :data
+	.rodata : {
+		*(.rodata .rodata.*)
+	} :data
+
+	. = ALIGN(8);
+	.bss : {
+		_bss_start = .;
+		*(.bss .bss.*)
+		_bss_end = .;
+		. = ALIGN(4096);
+		_stack_end = .;
+	} :data
+
+	/DISCARD/ : {
+		*(.eh_frame)
+		*(.note.gnu.build-id)
+	}
+}

arch/x86_64/kernel/Makefile.arch.in

@@ -0,0 +1,2 @@
+IHK_OBJS += cpu.o interrupt.o memory.o trampoline.o local.o context.o
+IHK_OBJS += perfctr.o syscall.o vsyscall.o

arch/x86_64/kernel/cpu.c

@@ -31,6 +31,7 @@
 #include <prctl.h>
 #include <page.h>
 #include <kmalloc.h>
+#include <debug.h>
 
 #define LAPIC_ID            0x020
 #define LAPIC_TIMER         0x320
@@ -69,11 +70,8 @@
 //#define DEBUG_PRINT_CPU
 
 #ifdef DEBUG_PRINT_CPU
-#define dkprintf kprintf
+#undef DDEBUG_DEFAULT
-#define ekprintf kprintf
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define ekprintf kprintf
 #endif
 
 static void *lapic_vp;
@@ -96,6 +94,8 @@ int gettime_local_support = 0;
 extern int ihk_mc_pt_print_pte(struct page_table *pt, void *virt);
 extern int kprintf(const char *format, ...);
 extern int interrupt_from_user(void *);
+extern void perf_start(struct mc_perf_event *event);
+extern void perf_reset(struct mc_perf_event *event);
 
 static struct idt_entry{
 	uint32_t desc[4];
@@ -847,8 +847,6 @@ void setup_x86_ap(void (*next_func)(void))
 }
 
 void arch_show_interrupt_context(const void *reg);
-void set_signal(int sig, void *regs, struct siginfo *info);
-void check_signal(unsigned long, void *, int);
 extern void tlb_flush_handler(int vector);
 
 void __show_stack(uintptr_t *sp) {
@@ -870,6 +868,19 @@ void show_context_stack(uintptr_t *rbp) {
 	return;
 }
 
+void interrupt_exit(struct x86_user_context *regs)
+{
+	if (interrupt_from_user(regs)) {
+		cpu_enable_interrupt();
+		check_sig_pending();
+		check_need_resched();
+		check_signal(0, regs, -1);
+	}
+	else {
+		check_sig_pending();
+	}
+}
+
 void handle_interrupt(int vector, struct x86_user_context *regs)
 {
 	struct ihk_mc_interrupt_handler *h;
@@ -992,14 +1003,16 @@ void handle_interrupt(int vector, struct x86_user_context *regs)
 		}
 	}
 
-	if(interrupt_from_user(regs)){
+	interrupt_exit(regs);
-		cpu_enable_interrupt();
+	set_cputime(interrupt_from_user(regs)? 0: 1);
-		check_signal(0, regs, 0);
-		check_need_resched();
-	}
-	set_cputime(0);
 
 	--v->in_interrupt;
+
+	/* for migration by IPI */
+	if (v->flags & CPU_FLAG_NEED_MIGRATE) {
+		schedule();
+		check_signal(0, regs, 0);
+	}
 }
 
 void gpe_handler(struct x86_user_context *regs)
@@ -1012,13 +1025,9 @@ void gpe_handler(struct x86_user_context *regs)
 		panic("gpe_handler");
 	}
 	set_signal(SIGSEGV, regs, NULL);
-	if(interrupt_from_user(regs)){
+	interrupt_exit(regs);
-		cpu_enable_interrupt();
+	set_cputime(interrupt_from_user(regs)? 0: 1);
-		check_signal(0, regs, 0);
+	panic("GPF");
-		check_need_resched();
-	}
-	set_cputime(0);
-	// panic("GPF");
 }
 
 void debug_handler(struct x86_user_context *regs)
@@ -1045,12 +1054,8 @@ void debug_handler(struct x86_user_context *regs)
 	memset(&info, '\0', sizeof info);
 	info.si_code = si_code;
 	set_signal(SIGTRAP, regs, &info);
-	if(interrupt_from_user(regs)){
+	interrupt_exit(regs);
-		cpu_enable_interrupt();
+	set_cputime(interrupt_from_user(regs)? 0: 1);
-		check_signal(0, regs, 0);
-		check_need_resched();
-	}
-	set_cputime(0);
 }
 
 void int3_handler(struct x86_user_context *regs)
@@ -1067,12 +1072,8 @@ void int3_handler(struct x86_user_context *regs)
 	memset(&info, '\0', sizeof info);
 	info.si_code = TRAP_BRKPT;
 	set_signal(SIGTRAP, regs, &info);
-	if(interrupt_from_user(regs)){
+	interrupt_exit(regs);
-		cpu_enable_interrupt();
+	set_cputime(interrupt_from_user(regs)? 0: 1);
-		check_signal(0, regs, 0);
-		check_need_resched();
-	}
-	set_cputime(0);
 }
 
 void
@@ -1081,7 +1082,6 @@ unhandled_page_fault(struct thread *thread, void *fault_addr, void *regs)
 	const uintptr_t address = (uintptr_t)fault_addr;
 	struct process_vm *vm = thread->vm;
 	struct vm_range *range;
-	char found;
 	unsigned long irqflags;
 	unsigned long error = ((struct x86_user_context *)regs)->gpr.error;
 
@@ -1095,17 +1095,12 @@ unhandled_page_fault(struct thread *thread, void *fault_addr, void *regs)
 			(error & PF_RSVD ? "was" : "wasn't"),
 			(error & PF_INSTR ? "was" : "wasn't"));
 
-	found = 0;
+	range = lookup_process_memory_range(vm, address, address+1);
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	if (range) {
-		if (range->start <= address && range->end > address) {
+		__kprintf("address is in range, flag: 0x%lx\n",
-			found = 1;
+				range->flag);
-			__kprintf("address is in range, flag: 0x%lx\n",
+		ihk_mc_pt_print_pte(vm->address_space->page_table, (void*)address);
-					range->flag);
+	} else {
-			ihk_mc_pt_print_pte(vm->address_space->page_table, (void*)address);
-			break;
-		}
-	}
-	if (!found) {
 		__kprintf("address is out of range! \n");
 	}
 
@@ -1473,29 +1468,91 @@ void ihk_mc_delay_us(int us)
 	arch_delay(us);
 }
 
-#define EXTENDED_ARCH_SHOW_CONTEXT
-#ifdef EXTENDED_ARCH_SHOW_CONTEXT
 void arch_show_extended_context(void)
 {
-	unsigned long cr0, cr4, msr, xcr0;
+	unsigned long cr0, cr4, msr, xcr0 = 0;
 
 	/*  Read and print CRs, MSR_EFER, XCR0  */
 	asm volatile("movq %%cr0, %0" : "=r"(cr0));
 	asm volatile("movq %%cr4, %0" : "=r"(cr4));
 	msr = rdmsr(MSR_EFER);
-	xcr0 = xgetbv(0);
+	if (xsave_available) {
-
+		xcr0 = xgetbv(0);
+	}
 	__kprintf("\n             CR0              CR4\n");
 	__kprintf("%016lX %016lX\n", cr0, cr4);
 
 	__kprintf("             MSR_EFER\n");
 	__kprintf("%016lX\n", msr);
 
-	__kprintf("             XCR0\n");
+	if (xsave_available) {
-	__kprintf("%016lX\n", xcr0);
+		__kprintf("             XCR0\n");
-
+		__kprintf("%016lX\n", xcr0);
+	}
+}
+
+struct stack {
+	struct stack *rbp;
+	unsigned long eip;
+};
+
+/* KPRINTF_LOCAL_BUF_LEN is 1024, useless to go further */
+#define STACK_BUF_LEN (1024-sizeof("[  0]: "))
+static void __print_stack(struct stack *rbp, unsigned long first) {
+	char buf[STACK_BUF_LEN];
+	size_t len;
+
+	/* Build string in buffer to output a single line */
+	len = snprintf(buf, STACK_BUF_LEN,
+		       "addr2line -e smp-x86/kernel/mckernel.img -fpia");
+
+	if (first)
+		len += snprintf(buf + len, STACK_BUF_LEN - len,
+				" %#16lx", first);
+
+	while ((unsigned long)rbp > 0xffff880000000000 &&
+			STACK_BUF_LEN - len > sizeof(" 0x0123456789abcdef")) {
+		len += snprintf(buf + len, STACK_BUF_LEN - len,
+				" %#16lx", rbp->eip);
+		rbp = rbp->rbp;
+	}
+	__kprintf("%s\n", buf);
+}
+
+void arch_print_pre_interrupt_stack(const struct x86_basic_regs *regs) {
+	struct stack *rbp;
+
+	/* only for kernel stack */
+	if (regs->error & PF_USER)
+		return;
+
+	__kprintf("Pre-interrupt stack trace:\n");
+
+	/* interrupt stack heuristics:
+	 * - the first entry looks like it is always garbage, so skip.
+	 * (that is done by taking regs->rsp instead of &regs->rsp)
+	 * - that still looks sometimes wrong. For now, if it is not
+	 * within 64k of itself, look for the next entry that matches.
+	 */
+
+	rbp = (struct stack*)regs->rsp;
+
+	while ((uintptr_t)rbp > (uintptr_t)rbp->rbp
+			|| (uintptr_t)rbp + 0x10000 < (uintptr_t)rbp->rbp)
+		rbp = (struct stack *)(((uintptr_t *)rbp) + 1);
+
+	__print_stack(rbp, regs->rip);
+}
+
+void arch_print_stack() {
+	struct stack *rbp;
+
+	__kprintf("Approximative stack trace:\n");
+
+	asm("mov %%rbp, %0" : "=r"(rbp) );
+
+	__print_stack(rbp, 0);
 }
-#endif
 
 /*@
   @ requires \valid(reg);
@@ -1526,9 +1583,11 @@ void arch_show_interrupt_context(const void *reg)
 	__kprintf("%16lx %16lx %16lx %16lx\n",
 	        regs->cs, regs->ss, regs->rflags, regs->error);
 
-#ifdef EXTENDED_ARCH_SHOW_CONTEXT
+kprintf_unlock(irqflags);
-        arch_show_extended_context();
+return;
-#endif	
+	arch_show_extended_context();
+
+	arch_print_pre_interrupt_stack(regs);
 
 	kprintf_unlock(irqflags);
 }
@@ -1588,12 +1647,10 @@ int ihk_mc_interrupt_cpu(int cpu, int vector)
 	return 0;
 }
 
-#ifdef POSTK_DEBUG_ARCH_DEP_22
-extern void perf_start(struct mc_perf_event *event);
-extern void perf_reset(struct mc_perf_event *event);
 struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 {
 	struct thread *last;
+	struct mcs_rwlock_node_irqsave lock;
 
 	dkprintf("[%d] schedule: tlsblock_base: 0x%lX\n",
 	         ihk_mc_get_processor_id(), next->tlsblock_base);
@@ -1612,7 +1669,7 @@ struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 	}
 
 #ifdef PROFILE_ENABLE
-	if (prev->profile && prev->profile_start_ts != 0) {
+	if (prev && prev->profile && prev->profile_start_ts != 0) {
 		prev->profile_elapsed_ts +=
 			(rdtsc() - prev->profile_start_ts);
 		prev->profile_start_ts = 0;
@@ -1624,6 +1681,28 @@ struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 #endif
 
 	if (prev) {
+		mcs_rwlock_writer_lock(&prev->proc->update_lock, &lock);
+		if (prev->proc->status & (PS_DELAY_STOPPED | PS_DELAY_TRACED)) {
+			switch (prev->proc->status) {
+			case PS_DELAY_STOPPED:
+				prev->proc->status = PS_STOPPED;
+				break;
+			case PS_DELAY_TRACED:
+				prev->proc->status = PS_TRACED;
+				break;
+			default:
+				break;
+			}
+			mcs_rwlock_writer_unlock(&prev->proc->update_lock,
+						&lock);
+
+			/* Wake up the parent who tried wait4 and sleeping */
+			waitq_wakeup(&prev->proc->parent->waitpid_q);
+		} else {
+			mcs_rwlock_writer_unlock(&prev->proc->update_lock,
+						&lock);
+		}
+
 		last = ihk_mc_switch_context(&prev->ctx, &next->ctx, prev);
 	}
 	else {
@@ -1631,7 +1710,6 @@ struct thread *arch_switch_context(struct thread *prev, struct thread *next)
 	}
 	return last;
 }
-#endif
 
 /*@
   @ requires \valid(thread);
@@ -1651,13 +1729,11 @@ release_fp_regs(struct thread *thread)
 	thread->fp_regs = NULL;
 }
 
-/*@
+static int
-  @ requires \valid(thread);
+check_and_allocate_fp_regs(struct thread *thread)
-  @*/
-void
-save_fp_regs(struct thread *thread)
 {
-	int	pages;
+	int pages;
+	int result = 0;
 
 	if (!thread->fp_regs) {
 		pages = (xsave_size + (PAGE_SIZE - 1)) >> PAGE_SHIFT;
@@ -1666,12 +1742,26 @@ save_fp_regs(struct thread *thread)
 
 		if (!thread->fp_regs) {
 			kprintf("error: allocating fp_regs pages\n");
-			return;
+			result = 1;
+			goto out;
 		}
 
-		memset(thread->fp_regs, 0, sizeof(fp_regs_struct));
 		memset(thread->fp_regs, 0, pages * PAGE_SIZE);
 	}
+out:
+	return result;
+}
+
+/*@
+  @ requires \valid(thread);
+  @*/
+void
+save_fp_regs(struct thread *thread)
+{
+	if (check_and_allocate_fp_regs(thread) != 0) {
+		// alloc error
+		return;
+	}
 
 	if (xsave_available) {
 		unsigned int low, high;
@@ -1687,13 +1777,12 @@ save_fp_regs(struct thread *thread)
 	}
 }
 
-#ifdef POSTK_DEBUG_TEMP_FIX_19
+void copy_fp_regs(struct thread *from, struct thread *to)
-void
-clear_fp_regs(struct thread *thread)
 {
-	return;
+	if ((from->fp_regs != NULL) && (check_and_allocate_fp_regs(to) == 0)) {
+		memcpy(to->fp_regs, from->fp_regs, sizeof(fp_regs_struct));
+	}
 }
-#endif /* POSTK_DEBUG_TEMP_FIX_19 */
 
 /*@
   @ requires \valid(thread);
@@ -1702,8 +1791,11 @@ clear_fp_regs(struct thread *thread)
 void
 restore_fp_regs(struct thread *thread)
 {
-	if (!thread->fp_regs)
+	if (!thread->fp_regs) {
+		// only clear fpregs.
+		clear_fp_regs();
 		return;
+	}
 
 	if (xsave_available) {
 		unsigned int low, high;
@@ -1722,6 +1814,13 @@ restore_fp_regs(struct thread *thread)
 	//release_fp_regs(thread);
 }
 
+void clear_fp_regs(void)
+{
+	struct cpu_local_var *v = get_this_cpu_local_var();
+
+	restore_fp_regs(&v->idle);
+}
+
 ihk_mc_user_context_t *lookup_user_context(struct thread *thread)
 {
 	ihk_mc_user_context_t *uctx = thread->uctx;

arch/x86_64/kernel/gencore.c

@@ -6,6 +6,7 @@
 #include <process.h>
 #include <string.h>
 #include <elfcore.h>
+#include <debug.h>
 
 #define	align32(x) ((((x) + 3) / 4) * 4)
 #define	alignpage(x) ((((x) + (PAGE_SIZE) - 1) / (PAGE_SIZE)) * (PAGE_SIZE))
@@ -13,13 +14,16 @@
 //#define DEBUG_PRINT_GENCORE
 
 #ifdef DEBUG_PRINT_GENCORE
-#define	dkprintf(...)	kprintf(__VA_ARGS__)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...)	do { if (0) kprintf(__VA_ARGS__); } while (0)
-#define	ekprintf(...)	kprintf(__VA_ARGS__)
 #endif
 
+/* Exclude reserved (mckernel's internal use), device file,
+ * hole created by mprotect
+ */
+#define GENCORE_RANGE_IS_INACCESSIBLE(range) \
+	    ((range->flag & (VR_RESERVED | VR_MEMTYPE_UC | VR_DONTDUMP)))
+
 /*
  * Generate a core file image, which consists of many chunks.
  * Returns an allocated table, an etnry of which is a pair of the address 
@@ -289,7 +293,7 @@ int gencore(struct thread *thread, void *regs,
 	Elf64_Ehdr eh;
 	Elf64_Phdr *ph = NULL;
 	void *note = NULL;
-	struct vm_range *range;
+	struct vm_range *range, *next;
 	struct process_vm *vm = thread->vm;
 	int segs = 1;	/* the first one is for NOTE */
 	int notesize, phsize, alignednotesize;
@@ -303,13 +307,16 @@ int gencore(struct thread *thread, void *regs,
 		return -1;
 	}
 
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	next = lookup_process_memory_range(vm, 0, -1);
+	while ((range = next)) {
+		next = next_process_memory_range(vm, range);
+
 		dkprintf("start:%lx end:%lx flag:%lx objoff:%lx\n", 
 			 range->start, range->end, range->flag, range->objoff);
-		/* We omit reserved areas because they are only for
+
-		   mckernel's internal use. */		   
+		if (GENCORE_RANGE_IS_INACCESSIBLE(range)) {
-		if (range->flag & VR_RESERVED)
 			continue;
+		}
 		/* We need a chunk for each page for a demand paging area.
 		   This can be optimized for spacial complexity but we would
 		   lose simplicity instead. */
@@ -391,12 +398,16 @@ int gencore(struct thread *thread, void *regs,
 
 	/* program header for each memory chunk */
 	i = 1;
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	next = lookup_process_memory_range(vm, 0, -1);
+	while ((range = next)) {
+		next = next_process_memory_range(vm, range);
+
 		unsigned long flag = range->flag;
 		unsigned long size = range->end - range->start;
 
-		if (range->flag & VR_RESERVED)
+		if (GENCORE_RANGE_IS_INACCESSIBLE(range)) {
 			continue;
+		}
 
 		ph[i].p_type = PT_LOAD;
 		ph[i].p_flags = ((flag & VR_PROT_READ) ? PF_R : 0)
@@ -432,11 +443,15 @@ int gencore(struct thread *thread, void *regs,
 	dkprintf("coretable[2]: %lx@%lx(%lx)\n", ct[2].len, ct[2].addr, note);
 
 	i = 3;	/* memory segments */
-	list_for_each_entry(range, &vm->vm_range_list, list) {
+	next = lookup_process_memory_range(vm, 0, -1);
+	while ((range = next)) {
+		next = next_process_memory_range(vm, range);
+
 		unsigned long phys;
 
-		if (range->flag & VR_RESERVED)
+		if (GENCORE_RANGE_IS_INACCESSIBLE(range)) {
 			continue;
+		}
 		if (range->flag & VR_DEMAND_PAGING) {
 			/* Just an ad hoc kluge. */
 			unsigned long p, start, phys;

arch/x86_64/kernel/include/arch-futex.h

@@ -64,13 +64,13 @@ static inline int futex_atomic_cmpxchg_inatomic(int __user *uaddr, int oldval,
 	return oldval;
 }
 
-#ifdef POSTK_DEBUG_ARCH_DEP_8 /* arch depend hide */
+static inline int futex_atomic_op_inuser(int encoded_op,
-static inline int futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
+					 int __user *uaddr)
 {
 	int op = (encoded_op >> 28) & 7;
 	int cmp = (encoded_op >> 24) & 15;
-	int oparg = (encoded_op << 8) >> 20;
+	int oparg = (encoded_op & 0x00fff000) >> 12;
-	int cmparg = (encoded_op << 20) >> 20;
+	int cmparg = encoded_op & 0xfff;
 	int oldval = 0, ret, tem;
 
 	if (encoded_op & (FUTEX_OP_OPARG_SHIFT << 28))
@@ -128,6 +128,13 @@ static inline int futex_atomic_op_inuser(int encoded_op, int __user *uaddr)
 	}
 	return ret;
 }
-#endif	/* !POSTK_DEBUG_ARCH_DEP_8 */
+
+static inline int get_futex_value_locked(uint32_t *dest, uint32_t *from)
+{
+
+	*dest = *(volatile uint32_t *)from;
+
+	return 0;
+}
 
 #endif

arch/x86_64/kernel/include/arch-lock.h

@@ -6,6 +6,7 @@
 
 #include <ihk/cpu.h>
 #include <ihk/atomic.h>
+#include <lwk/compiler.h>
 
 //#define DEBUG_SPINLOCK
 //#define DEBUG_MCS_RWLOCK
@@ -14,7 +15,17 @@
 int __kprintf(const char *format, ...);
 #endif
 
-typedef int ihk_spinlock_t;
+typedef unsigned short __ticket_t;
+typedef unsigned int __ticketpair_t;
+
+typedef struct ihk_spinlock {
+	union {
+		__ticketpair_t head_tail;
+		struct __raw_tickets {
+			__ticket_t head, tail;
+		} tickets;
+	};
+} ihk_spinlock_t;
 
 extern void preempt_enable(void);
 extern void preempt_disable(void);
@@ -23,9 +34,61 @@ extern void preempt_disable(void);
 
 static void ihk_mc_spinlock_init(ihk_spinlock_t *lock)
 {
-	*lock = 0;
+	lock->head_tail = 0;
+}
+#define SPIN_LOCK_UNLOCKED { .head_tail = 0 }
+
+
+#ifdef DEBUG_SPINLOCK
+#define ihk_mc_spinlock_trylock_noirq(l) { int rc;						\
+__kprintf("[%d] call ihk_mc_spinlock_trylock_noirq %p %s:%d\n", ihk_mc_get_processor_id(), (l), __FILE__, __LINE__); \
+rc = __ihk_mc_spinlock_trylock_noirq(l); \
+ __kprintf("[%d] ret ihk_mc_spinlock_trylock_noirq\n", ihk_mc_get_processor_id()); rc; \
+}
+#else
+#define ihk_mc_spinlock_trylock_noirq __ihk_mc_spinlock_trylock_noirq
+#endif
+
+static int __ihk_mc_spinlock_trylock_noirq(ihk_spinlock_t *lock)
+{
+	ihk_spinlock_t cur = { .head_tail = lock->head_tail };
+	ihk_spinlock_t next = { .tickets.head = cur.tickets.head, .tickets.tail = cur.tickets.tail + 2 };
+	int success;
+
+	if (cur.tickets.head != cur.tickets.tail) {
+		return 0;
+	}
+
+	preempt_disable();
+
+	/* Use the same increment amount as other functions! */
+	success = __sync_bool_compare_and_swap((__ticketpair_t*)lock, cur.head_tail, next.head_tail);
+
+	if (!success) {
+		preempt_enable();
+	}
+	return success;
+}
+
+#ifdef DEBUG_SPINLOCK
+#define ihk_mc_spinlock_trylock(l, result) ({ unsigned long rc;		\
+__kprintf("[%d] call ihk_mc_spinlock_trylock %p %s:%d\n", ihk_mc_get_processor_id(), (l), __FILE__, __LINE__); \
+ rc = __ihk_mc_spinlock_trylock(l, result);									\
+__kprintf("[%d] ret ihk_mc_spinlock_trylock\n", ihk_mc_get_processor_id()); rc;\
+})
+#else
+#define ihk_mc_spinlock_trylock __ihk_mc_spinlock_trylock
+#endif
+static unsigned long __ihk_mc_spinlock_trylock(ihk_spinlock_t *lock, int *result)
+{
+	unsigned long flags;
+	
+	flags = cpu_disable_interrupt_save();
+
+	*result = __ihk_mc_spinlock_trylock_noirq(lock);
+
+	return flags;
 }
-#define SPIN_LOCK_UNLOCKED 0
 
 #ifdef DEBUG_SPINLOCK
 #define ihk_mc_spinlock_lock_noirq(l) { \
@@ -39,40 +102,24 @@ __kprintf("[%d] ret ihk_mc_spinlock_lock_noirq\n", ihk_mc_get_processor_id()); \
 
 static void __ihk_mc_spinlock_lock_noirq(ihk_spinlock_t *lock)
 {
-	int inc = 0x00010000;
+	register struct __raw_tickets inc = { .tail = 0x0002 };
-	int tmp;
-
-#if 0
-	asm volatile("lock ; xaddl %0, %1\n"
-	             "movzwl %w0, %2\n\t"
-	             "shrl $16, %0\n\t"
-	             "1:\t"
-	             "cmpl %0, %2\n\t"
-	             "je 2f\n\t"
-	             "rep ; nop\n\t"
-	             "movzwl %1, %2\n\t"
-	             "jmp 1b\n"
-	             "2:"
-	             : "+Q" (inc), "+m" (*lock), "=r" (tmp) : : "memory", "cc");
-#endif
 
 	preempt_disable();
 
-	asm volatile("lock; xaddl %0, %1\n"
+	asm volatile ("lock xaddl %0, %1\n"
-			"movzwl %w0, %2\n\t"
+			: "+r" (inc), "+m" (*(lock)) : : "memory", "cc");
-			"shrl $16, %0\n\t"
-			"1:\t"
-			"cmpl %0, %2\n\t"
-			"je 2f\n\t"
-			"rep ; nop\n\t"
-			"movzwl %1, %2\n\t"
-			/* don't need lfence here, because loads are in-order */
-			"jmp 1b\n"
-			"2:"
-			: "+r" (inc), "+m" (*lock), "=&r" (tmp)
-			:
-			: "memory", "cc");
 
+	if (inc.head == inc.tail)
+		goto out;
+
+	for (;;) {
+		if (*((volatile __ticket_t *)&lock->tickets.head) == inc.tail)
+			goto out;
+		cpu_pause();
+	}
+
+out:
+	barrier();	/* make sure nothing creeps before the lock is taken */
 }
 
 #ifdef DEBUG_SPINLOCK
@@ -106,7 +153,10 @@ __kprintf("[%d] ret ihk_mc_spinlock_unlock_noirq\n", ihk_mc_get_processor_id());
 #endif
 static void __ihk_mc_spinlock_unlock_noirq(ihk_spinlock_t *lock)
 {
-	asm volatile ("lock incw %0" : "+m"(*lock) : : "memory", "cc");
+	__ticket_t inc = 0x0002;
+
+	asm volatile ("lock addw %1, %0\n"
+			: "+m" (lock->tickets.head) : "ri" (inc) : "memory", "cc");
 
 	preempt_enable();
 }
@@ -134,6 +184,8 @@ typedef struct mcs_lock_node {
 	unsigned long irqsave;
 } __attribute__((aligned(64))) mcs_lock_node_t;
 
+typedef mcs_lock_node_t mcs_lock_t;
+
 static void mcs_lock_init(struct mcs_lock_node *node)
 {
 	node->locked = 0;
@@ -600,4 +652,9 @@ __mcs_rwlock_reader_unlock(struct mcs_rwlock_lock *lock, struct mcs_rwlock_node_
 #endif
 }
 
+static inline int irqflags_can_interrupt(unsigned long flags)
+{
+	return !!(flags & 0x200);
+}
+
 #endif

arch/x86_64/kernel/include/arch-memory.h

@@ -42,16 +42,34 @@
 
 #define USER_END           0x0000800000000000UL
 #define TASK_UNMAPPED_BASE 0x00002AAAAAA00000UL
+
+/*
+ * Canonical negative addresses (i.e., the smallest kernel virtual address)
+ * on x86 64 bit mode (in its most restricted 48 bit format) starts from
+ * 0xffff800000000000, but Linux starts mapping physical memory at 0xffff880000000000.
+ * The 0x80000000000 long gap (8TBs, i.e., 16 PGD level entries in the page tables)
+ * is used for Xen hyervisor (see arch/x86/include/asm/page.h) and that is
+ * what we utilize for McKernel.
+ * This gives us the benefit of being able to use Linux kernel virtual
+ * addresses identically as in Linux.
+ *
+ * NOTE: update these also in eclair.c when modified!
+ */
 #define MAP_ST_START       0xffff800000000000UL
-#define MAP_VMAP_START     0xfffff00000000000UL
+#define MAP_VMAP_START     0xffff850000000000UL
-#define MAP_FIXED_START    0xffffffff70000000UL
+#define MAP_FIXED_START    0xffff860000000000UL
-#define MAP_KERNEL_START   0xffffffff80000000UL
+#define LINUX_PAGE_OFFSET  0xffff880000000000UL
+/*
+ * MAP_KERNEL_START is 8MB below MODULES_END in Linux.
+ * Placing the LWK image in the virtual address space at the end of
+ * the Linux modules section enables us to map the LWK TEXT in Linux
+ * as well, so that Linux can also call into LWK text.
+ */
+#define MAP_KERNEL_START   0xFFFFFFFFFE800000UL
 #define STACK_TOP(region)  ((region)->user_end)
 
 #define MAP_VMAP_SIZE      0x0000000100000000UL
 
-#define KERNEL_PHYS_OFFSET MAP_ST_START
-
 #define PTL4_SHIFT         39
 #define PTL4_SIZE          (1UL << PTL4_SHIFT)
 #define PTL3_SHIFT         30

arch/x86_64/kernel/include/arch/cpu.h

@@ -13,16 +13,16 @@
 #ifndef ARCH_CPU_H
 #define ARCH_CPU_H
 
-#include <ihk/cpu.h>
+#define arch_barrier()	asm volatile("" : : : "memory")
 
 static inline void rmb(void)
 {
-	barrier();
+	arch_barrier();
 }
 
 static inline void wmb(void)
 {
-	barrier();
+	arch_barrier();
 }
 
 static unsigned long read_tsc(void)

arch/x86_64/kernel/include/ihk/context.h

@@ -49,6 +49,7 @@ typedef struct x86_user_context ihk_mc_user_context_t;
 #define ihk_mc_syscall_arg5(uc) (uc)->gpr.r9
 
 #define ihk_mc_syscall_ret(uc)  (uc)->gpr.rax
+#define ihk_mc_syscall_number(uc)  (uc)->gpr.orig_rax
 
 #define ihk_mc_syscall_pc(uc)   (uc)->gpr.rip
 #define ihk_mc_syscall_sp(uc)   (uc)->gpr.rsp

arch/x86_64/kernel/include/registers.h

@@ -189,9 +189,30 @@ struct tss64 {
 } __attribute__((packed));
 
 struct x86_basic_regs {
-	unsigned long r15, r14, r13, r12, rbp, rbx, r11, r10;
+	unsigned long r15;
-	unsigned long r9, r8, rax, rcx, rdx, rsi, rdi, error;
+	unsigned long r14;
-	unsigned long rip, cs, rflags, rsp, ss;
+	unsigned long r13;
+	unsigned long r12;
+	unsigned long rbp;
+	unsigned long rbx;
+	unsigned long r11;
+	unsigned long r10;
+	unsigned long r9;
+	unsigned long r8;
+	unsigned long rax;
+	unsigned long rcx;
+	unsigned long rdx;
+	unsigned long rsi;
+	unsigned long rdi;
+	union {
+		unsigned long orig_rax; /* syscall */
+		unsigned long error;    /* interrupts */
+	};
+	unsigned long rip;
+	unsigned long cs;
+	unsigned long rflags;
+	unsigned long rsp;
+	unsigned long ss;
 };
 
 struct x86_sregs {

arch/x86_64/kernel/include/signal.h

@@ -18,6 +18,11 @@
 #define _NSIG_BPW	64
 #define _NSIG_WORDS (_NSIG / _NSIG_BPW)
 
+static inline int valid_signal(unsigned long sig)
+{
+	return sig <= _NSIG ? 1 : 0;
+}
+
 typedef unsigned long int __sigset_t;
 #define __sigmask(sig)  (((__sigset_t) 1) << ((sig) - 1))
 

arch/x86_64/kernel/include/syscall_list.h

@@ -56,7 +56,7 @@ SYSCALL_HANDLED(36, getitimer)
 SYSCALL_HANDLED(38, setitimer)
 SYSCALL_HANDLED(39, getpid)
 SYSCALL_HANDLED(56, clone)
-SYSCALL_DELEGATED(57, fork)
+SYSCALL_HANDLED(57, fork)
 SYSCALL_HANDLED(58, vfork)
 SYSCALL_HANDLED(59, execve)
 SYSCALL_HANDLED(60, exit)
@@ -114,7 +114,7 @@ SYSCALL_HANDLED(160, setrlimit)
 SYSCALL_HANDLED(164, settimeofday)
 SYSCALL_HANDLED(186, gettid)
 SYSCALL_HANDLED(200, tkill)
-SYSCALL_DELEGATED(201, time)
+SYSCALL_HANDLED(201, time)
 SYSCALL_HANDLED(202, futex)
 SYSCALL_HANDLED(203, sched_setaffinity)
 SYSCALL_HANDLED(204, sched_getaffinity)
@@ -161,6 +161,7 @@ SYSCALL_HANDLED(__NR_profile, profile)
 SYSCALL_HANDLED(730, util_migrate_inter_kernel)
 SYSCALL_HANDLED(731, util_indicate_clone)
 SYSCALL_HANDLED(732, get_system)
+SYSCALL_HANDLED(733, util_register_desc)
 
 /* McKernel Specific */
 SYSCALL_HANDLED(801, swapout)

arch/x86_64/kernel/interrupt.S

@@ -145,6 +145,8 @@ nmi:
 		movq	%rsp,%gs:PANIC_REGS+0x08
 
 		movl	nmi_mode(%rip),%eax
+		cmp $3,%rax
+		je  4f
 		cmp	$1,%rax
 		je	1f
 		cmp	$2,%rax
@@ -199,9 +201,9 @@ nmi:
 		movl	%eax,%gs:PANIC_REGS+0xA0
 		movq	$1,%gs:PANICED
 		call	ihk_mc_query_mem_areas
-1:
+4:
 		hlt
-		jmp	1b
+		jmp	4b
 
 .globl x86_syscall
 x86_syscall:

arch/x86_64/kernel/local.c

@@ -107,9 +107,17 @@ void init_boot_processor_local(void)
   @ ensures \result == %gs;
   @ assigns \nothing;
   */
+extern int num_processors;
 int ihk_mc_get_processor_id(void)
 {
 	int id;
+	void *gs;
+
+	gs = (void *)rdmsr(MSR_GS_BASE);
+	if (gs < (void *)locals ||
+			gs > ((void *)locals + LOCALS_SPAN * num_processors)) {
+		return -1;
+	}
 
 	asm volatile("movl %%gs:0, %0" : "=r"(id));
 

arch/x86_64/kernel/memory.c

@@ -25,15 +25,13 @@
 #include <cls.h>
 #include <kmalloc.h>
 #include <rusage_private.h>
+#include <debug.h>
 
 //#define DEBUG
 
 #ifdef DEBUG
-#define	dkprintf(...)	do { kprintf(__VA_ARGS__); } while (0)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...)	do { kprintf(__VA_ARGS__); } while (0)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define dkprintf(...) do { } while (0)
-#define ekprintf(...) do { kprintf(__VA_ARGS__); } while (0)
 #endif
 
 static char *last_page;
@@ -41,6 +39,8 @@ extern char _head[], _end[];
 
 extern unsigned long x86_kernel_phys_base;
 
+int safe_kernel_map = 0;
+
 /* Arch specific early allocation routine */
 void *early_alloc_pages(int nr_pages)
 {
@@ -109,6 +109,7 @@ struct page_table {
 };
 
 static struct page_table *init_pt;
+static int init_pt_loaded = 0;
 static ihk_spinlock_t init_pt_lock;
 
 static int use_1gb_page = 0;
@@ -167,30 +168,6 @@ static unsigned long setup_l3(struct page_table *pt,
 	return virt_to_phys(pt);
 }
 
-static void init_normal_area(struct page_table *pt)
-{
-	unsigned long map_start, map_end, phys, pt_phys;
-	int ident_index, virt_index;
-
-	map_start = ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0);
-	map_end = ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0);
-
-	kprintf("map_start = %lx, map_end = %lx\n", map_start, map_end);
-	ident_index = map_start >> PTL4_SHIFT;
-	virt_index = (MAP_ST_START >> PTL4_SHIFT) & (PT_ENTRIES - 1);
-
-	memset(pt, 0, sizeof(struct page_table));
-
-	for (phys = (map_start & ~(PTL4_SIZE - 1)); phys < map_end;
-	     phys += PTL4_SIZE) {
-		pt_phys = setup_l3(ihk_mc_alloc_pages(1, IHK_MC_AP_CRITICAL), phys,
-		                   map_start, map_end);
-
-		pt->entry[ident_index++] = pt_phys | PFL4_PDIR_ATTR;
-		pt->entry[virt_index++] = pt_phys | PFL4_PDIR_ATTR;
-	}
-}
-
 static struct page_table *__alloc_new_pt(ihk_mc_ap_flag ap_flag)
 {
 	struct page_table *newpt = ihk_mc_alloc_pages(1, ap_flag);
@@ -258,6 +235,11 @@ static unsigned long attr_to_l1attr(enum ihk_mc_pt_attribute attr)
 	}
 }
 
+#define PTLX_SHIFT(index) PTL ## index ## _SHIFT
+
+#define GET_VIRT_INDEX(virt, index, dest) \
+	dest = ((virt) >> PTLX_SHIFT(index)) & (PT_ENTRIES - 1)
+
 #define GET_VIRT_INDICES(virt, l4i, l3i, l2i, l1i) \
 	l4i = ((virt) >> PTL4_SHIFT) & (PT_ENTRIES - 1); \
 	l3i = ((virt) >> PTL3_SHIFT) & (PT_ENTRIES - 1); \
@@ -493,7 +475,7 @@ uint64_t ihk_mc_pt_virt_to_pagemap(struct page_table *pt, unsigned long virt)
 
 	error = ihk_mc_pt_virt_to_phys(pt, (void *)virt, &phys);
 	if (error) {
-		return 0;
+		return PM_PSHIFT(PAGE_SHIFT);
 	}
 
 	pagemap = PM_PFRAME(phys >> PAGE_SHIFT);
@@ -1518,12 +1500,12 @@ static int clear_range_l1(void *args0, pte_t *ptep, uint64_t base,
 	if (page) {
 		dkprintf("%s: page=%p,is_in_memobj=%d,(old & PFL1_DIRTY)=%lx,memobj=%p,args->memobj->flags=%x\n", __FUNCTION__, page, page_is_in_memobj(page), (old & PFL1_DIRTY), args->memobj, args->memobj ? args->memobj->flags : -1);
 	}
-	if (page && page_is_in_memobj(page) && (old & PFL1_DIRTY) && (args->memobj) &&
+	if (page && page_is_in_memobj(page) && pte_is_dirty(&old, PTL1_SIZE) &&
-			!(args->memobj->flags & MF_ZEROFILL)) {
+			args->memobj && !(args->memobj->flags & MF_ZEROFILL)) {
 		memobj_flush_page(args->memobj, phys, PTL1_SIZE);
 	}
 
-	if (!(old & PFL1_FILEOFF)) {
+	if (!pte_is_fileoff(&old, PTL1_SIZE)) {
 		if(args->free_physical) {
 			if (!page) {
 				/* Anonymous || !XPMEM attach */
@@ -1542,7 +1524,6 @@ static int clear_range_l1(void *args0, pte_t *ptep, uint64_t base,
 				dkprintf("%lx-,%s: calling memory_stat_rss_sub(),phys=%lx,size=%ld,pgsize=%ld\n", pte_get_phys(&old), __FUNCTION__, pte_get_phys(&old), PTL1_SIZE, PTL1_SIZE);
 				rusage_memory_stat_sub(args->memobj, PTL1_SIZE, PTL1_SIZE); 
 			}
-			args->vm->currss -= PTL1_SIZE;
 		}  else {
 			dkprintf("%s: !calling memory_stat_rss_sub(),virt=%lx,phys=%lx\n", __FUNCTION__, base, pte_get_phys(&old));
 		}
@@ -1586,11 +1567,11 @@ static int clear_range_l2(void *args0, pte_t *ptep, uint64_t base,
 			page = phys_to_page(phys);
 		}
 
-		if (page && page_is_in_memobj(page) && (old & PFL2_DIRTY)) {
+		if (page && page_is_in_memobj(page) && pte_is_dirty(&old, PTL2_SIZE)) {
 			memobj_flush_page(args->memobj, phys, PTL2_SIZE);
 		}
 
-		if (!(old & PFL2_FILEOFF)) {
+		if (!pte_is_fileoff(&old, PTL2_SIZE)) {
 			if(args->free_physical) {
 				if (!page) {
 					/* Anonymous || !XPMEM attach */
@@ -1611,7 +1592,6 @@ static int clear_range_l2(void *args0, pte_t *ptep, uint64_t base,
 					dkprintf("%lx-,%s: calling memory_stat_rss_sub(),phys=%lx,size=%ld,pgsize=%ld\n", pte_get_phys(&old), __FUNCTION__, pte_get_phys(&old), PTL2_SIZE, PTL2_SIZE);
 					rusage_memory_stat_sub(args->memobj, PTL2_SIZE, PTL2_SIZE); 
 				}
-				args->vm->currss -= PTL2_SIZE;
 			}
 		}
 
@@ -1668,13 +1648,13 @@ static int clear_range_l3(void *args0, pte_t *ptep, uint64_t base,
 			page = phys_to_page(phys);
 		}
 
-		if (page && page_is_in_memobj(page) && (old & PFL3_DIRTY)) {
+		if (page && page_is_in_memobj(page) && pte_is_dirty(&old, PTL3_SIZE)) {
 			memobj_flush_page(args->memobj, phys, PTL3_SIZE);
 		}
 
 		dkprintf("%s: phys=%ld, pte_get_phys(&old),PTL3_SIZE\n", __FUNCTION__, pte_get_phys(&old));
 
-		if (!(old & PFL3_FILEOFF)) {
+		if (!pte_is_fileoff(&old, PTL3_SIZE)) {
 			if(args->free_physical) {
 				if (!page) {
 					/* Anonymous || !XPMEM attach */
@@ -1693,7 +1673,6 @@ static int clear_range_l3(void *args0, pte_t *ptep, uint64_t base,
 					dkprintf("%lx-,%s: calling memory_stat_rss_sub(),phys=%lx,size=%ld,pgsize=%ld\n", pte_get_phys(&old), __FUNCTION__, pte_get_phys(&old), PTL3_SIZE, PTL3_SIZE);
 					rusage_memory_stat_sub(args->memobj, PTL3_SIZE, PTL3_SIZE); 
 				}
-				args->vm->currss -= PTL3_SIZE;
 			}
 		}
 
@@ -2543,6 +2522,82 @@ static void init_fixed_area(struct page_table *pt)
 	return;
 }
 
+static void init_normal_area(struct page_table *pt)
+{
+	unsigned long map_start, map_end, phys;
+	void *virt;
+
+	map_start = ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0);
+	map_end = ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0);
+	virt = (void *)MAP_ST_START + map_start;
+
+	kprintf("map_start = %lx, map_end = %lx, virt %lx\n",
+		map_start, map_end, virt);
+
+	for (phys = map_start; phys < map_end; phys += LARGE_PAGE_SIZE) {
+		if (set_pt_large_page(pt, virt, phys, PTATTR_WRITABLE) != 0) {
+			kprintf("%s: error setting mapping for 0x%lx\n",
+					__func__, virt);
+		}
+		virt += LARGE_PAGE_SIZE;
+	}
+}
+
+static void init_linux_kernel_mapping(struct page_table *pt)
+{
+	unsigned long map_start, map_end, phys;
+	void *virt;
+	int nr_memory_chunks, chunk_id, numa_id;
+
+	/* In case of safe_kernel_map option (safe_kernel_map == 1),
+	 * processing to prevent destruction of the memory area on Linux side
+	 * is executed */
+	if (safe_kernel_map == 0) {
+		kprintf("Straight-map entire physical memory\n");
+
+		/* Map 2 TB for now */
+		map_start = 0;
+		map_end = 0x20000000000;
+
+		virt = (void *)LINUX_PAGE_OFFSET;
+
+		kprintf("Linux kernel virtual: 0x%lx - 0x%lx -> 0x%lx - 0x%lx\n",
+			LINUX_PAGE_OFFSET, LINUX_PAGE_OFFSET + map_end, 0, map_end);
+
+		for (phys = map_start; phys < map_end; phys += LARGE_PAGE_SIZE) {
+			if (set_pt_large_page(pt, virt, phys, PTATTR_WRITABLE) != 0) {
+				kprintf("%s: error setting mapping for 0x%lx\n", __FUNCTION__, virt);
+			}
+			virt += LARGE_PAGE_SIZE;
+		}
+	} else {
+		kprintf("Straight-map physical memory areas allocated to McKernel\n");
+
+		nr_memory_chunks = ihk_mc_get_nr_memory_chunks();
+		if (nr_memory_chunks == 0) {
+			kprintf("%s: ERROR: No memory chunk available.\n", __FUNCTION__);
+			return;
+		}
+
+		for (chunk_id = 0; chunk_id < nr_memory_chunks; chunk_id++) {
+			if (ihk_mc_get_memory_chunk(chunk_id, &map_start, &map_end, &numa_id)) {
+				kprintf("%s: ERROR: Memory chunk id (%d) out of range.\n", __FUNCTION__, chunk_id);
+				continue;
+			}
+
+			dkprintf("Linux kernel virtual: 0x%lx - 0x%lx -> 0x%lx - 0x%lx\n",
+					 LINUX_PAGE_OFFSET + map_start, LINUX_PAGE_OFFSET + map_end, map_start, map_end);
+
+			virt = (void *)(LINUX_PAGE_OFFSET + map_start);
+			for (phys = map_start; phys < map_end; phys += LARGE_PAGE_SIZE, virt += LARGE_PAGE_SIZE) {
+				if (set_pt_large_page(pt, virt, phys, PTATTR_WRITABLE) != 0) {
+					kprintf("%s: set_pt_large_page() failed for 0x%lx\n", __FUNCTION__, virt);
+				}
+			}
+		}
+	}
+}
+
 void init_text_area(struct page_table *pt)
 {
 	unsigned long __end, phys, virt;
@@ -2627,17 +2682,19 @@ void init_page_table(void)
 	init_pt = ihk_mc_alloc_pages(1, IHK_MC_AP_CRITICAL);
 	ihk_mc_spinlock_init(&init_pt_lock);
 	
-	memset(init_pt, 0, sizeof(PAGE_SIZE));
+	memset(init_pt, 0, sizeof(*init_pt));
 
 	/* Normal memory area */
 	init_normal_area(init_pt);
+	init_linux_kernel_mapping(init_pt);
 	init_fixed_area(init_pt);
 	init_low_area(init_pt);
 	init_text_area(init_pt);
 	init_vsyscall_area(init_pt);
 
 	load_page_table(init_pt);
-	kprintf("Page table is now at %p\n", init_pt);
+	init_pt_loaded = 1;
+	kprintf("Page table is now at 0x%lx\n", init_pt);
 }
 
 extern void __reserve_arch_pages(unsigned long, unsigned long,
@@ -2667,15 +2724,31 @@ unsigned long virt_to_phys(void *v)
 	unsigned long va = (unsigned long)v;
 
 	if (va >= MAP_KERNEL_START) {
+		dkprintf("%s: MAP_KERNEL_START <= 0x%lx <= LINUX_PAGE_OFFSET\n",
+				__FUNCTION__, va);
 		return va - MAP_KERNEL_START + x86_kernel_phys_base;
-	} else {
+	}
+	else if (va >= LINUX_PAGE_OFFSET) {
+		return va - LINUX_PAGE_OFFSET;
+	}
+	else if (va >= MAP_FIXED_START) {
+		return va - MAP_FIXED_START;
+	}
+	else {
+		dkprintf("%s: MAP_ST_START <= 0x%lx <= MAP_FIXED_START\n",
+				__FUNCTION__, va);
 		return va - MAP_ST_START;
 	}
 }
 
 void *phys_to_virt(unsigned long p)
 {
-	return (void *)(p + MAP_ST_START);
+	/* Before loading our own PT use straight mapping */
+	if (!init_pt_loaded) {
+		return (void *)(p + MAP_ST_START);
+	}
+
+	return (void *)(p + LINUX_PAGE_OFFSET);
 }
 
 int copy_from_user(void *dst, const void *src, size_t siz)
@@ -2843,17 +2916,12 @@ int read_process_vm(struct process_vm *vm, void *kdst, const void *usrc, size_t
 			return error;
 		}
 
-#ifdef POSTK_DEBUG_TEMP_FIX_52 /* NUMA support(memory area determination) */
+		if (!is_mckernel_memory(pa, pa + cpsize)) {
-		if (!is_mckernel_memory(pa)) {
-#else
-		if (pa < ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0) ||
-			pa >= ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0)) {
-#endif /* POSTK_DEBUG_TEMP_FIX_52 */
 			dkprintf("%s: pa is outside of LWK memory, to: %p, pa: %p,"
 				"cpsize: %d\n", __FUNCTION__, to, pa, cpsize);
 			va = ihk_mc_map_virtual(pa, 1, PTATTR_ACTIVE);
 			memcpy(to, va, cpsize);
-			ihk_mc_unmap_virtual(va, 1, 1);
+			ihk_mc_unmap_virtual(va, 1);
 		}
 		else {
 			va = phys_to_virt(pa);
@@ -2927,17 +2995,12 @@ int write_process_vm(struct process_vm *vm, void *udst, const void *ksrc, size_t
 			return error;
 		}
 
-#ifdef POSTK_DEBUG_TEMP_FIX_52 /* NUMA support(memory area determination) */
+		if (!is_mckernel_memory(pa, pa + cpsize)) {
-		if (!is_mckernel_memory(pa)) {
-#else
-		if (pa < ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0) ||
-			pa >= ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0)) {
-#endif /* POSTK_DEBUG_TEMP_FIX_52 */
 			dkprintf("%s: pa is outside of LWK memory, from: %p,"
 				"pa: %p, cpsize: %d\n", __FUNCTION__, from, pa, cpsize);
 			va = ihk_mc_map_virtual(pa, 1, PTATTR_ACTIVE);
 			memcpy(va, from, cpsize);
-			ihk_mc_unmap_virtual(va, 1, 1);
+			ihk_mc_unmap_virtual(va, 1);
 		}
 		else {
 			va = phys_to_virt(pa);
@@ -2998,17 +3061,12 @@ int patch_process_vm(struct process_vm *vm, void *udst, const void *ksrc, size_t
 			return error;
 		}
 
-#ifdef POSTK_DEBUG_TEMP_FIX_52 /* NUMA support(memory area determination) */
+		if (!is_mckernel_memory(pa, pa + cpsize)) {
-		if (!is_mckernel_memory(pa)) {
-#else
-		if (pa < ihk_mc_get_memory_address(IHK_MC_GMA_MAP_START, 0) ||
-			pa >= ihk_mc_get_memory_address(IHK_MC_GMA_MAP_END, 0)) {
-#endif /* POSTK_DEBUG_TEMP_FIX_52 */
 			dkprintf("%s: pa is outside of LWK memory, from: %p,"
 				"pa: %p, cpsize: %d\n", __FUNCTION__, from, pa, cpsize);
 			va = ihk_mc_map_virtual(pa, 1, PTATTR_ACTIVE);
 			memcpy(va, from, cpsize);
-			ihk_mc_unmap_virtual(va, 1, 1);
+			ihk_mc_unmap_virtual(va, 1);
 		}
 		else {
 			va = phys_to_virt(pa);

arch/x86_64/kernel/mikc.c

@@ -30,7 +30,7 @@ int ihk_mc_ikc_init_first_local(struct ihk_ikc_channel_desc *channel,
 
 	memset(channel, 0, sizeof(struct ihk_ikc_channel_desc));
 
-	mikc_queue_pages = ((2 * num_processors * MASTER_IKCQ_PKTSIZE)
+	mikc_queue_pages = ((4 * num_processors * MASTER_IKCQ_PKTSIZE)
 			+ (PAGE_SIZE - 1)) / PAGE_SIZE;
 
 	/* Place both sides in this side */

arch/x86_64/kernel/perfctr.c

@@ -10,23 +10,22 @@
 #include <ihk/perfctr.h>
 #include <march.h>
 #include <errno.h>
+#include <cls.h>
 #include <ihk/debug.h>
+#include <ihk/cpu.h>
 #include <registers.h>
 #include <mc_perf_event.h>
+#include <config.h>
+#include <debug.h>
 
 extern unsigned int *x86_march_perfmap;
 extern int running_on_kvm(void);
-#ifdef POSTK_DEBUG_TEMP_FIX_31
 int ihk_mc_perfctr_fixed_init(int counter, int mode);
-#endif/*POSTK_DEBUG_TEMP_FIX_31*/
 
 //#define PERFCTR_DEBUG
 #ifdef PERFCTR_DEBUG
-#define	dkprintf(...)	do { kprintf(__VA_ARGS__); } while (0)
+#undef DDEBUG_DEFAULT
-#define	ekprintf(...)	do { kprintf(__VA_ARGS__); } while (0)
+#define DDEBUG_DEFAULT DDEBUG_PRINT
-#else
-#define	dkprintf(...)	do { } while (0)
-#define	ekprintf(...)	do { kprintf(__VA_ARGS__); } while (0)
 #endif
 
 #define X86_CR4_PCE     0x00000100
@@ -40,11 +39,11 @@ int ihk_mc_perfctr_fixed_init(int counter, int mode);
 		}                                                               \
     } while(0)
 
-int perf_counters_discovered = 0;
+int perf_counters_discovered;
-int X86_IA32_NUM_PERF_COUNTERS = 0;
+int NUM_PERF_COUNTERS;
-unsigned long X86_IA32_PERF_COUNTERS_MASK = 0;
+unsigned long PERF_COUNTERS_MASK;
-int X86_IA32_NUM_FIXED_PERF_COUNTERS = 0;
+int NUM_FIXED_PERF_COUNTERS;
-unsigned long X86_IA32_FIXED_PERF_COUNTERS_MASK = 0;
+unsigned long FIXED_PERF_COUNTERS_MASK;
 
 void x86_init_perfctr(void)
 {
@@ -57,6 +56,10 @@ void x86_init_perfctr(void)
 	uint64_t ecx;
 	uint64_t edx;
 
+#ifndef ENABLE_PERF
+	return;
+#endif //ENABLE_PERF
+
 	/* Do not do it on KVM */
 	if (running_on_kvm()) return;
 
@@ -71,17 +74,17 @@ void x86_init_perfctr(void)
 		op = 0x0a;
 		asm volatile("cpuid" : "=a"(eax),"=b"(ebx),"=c"(ecx),"=d"(edx):"a"(op));
 
-		X86_IA32_NUM_PERF_COUNTERS = ((eax & 0xFF00) >> 8);
+		NUM_PERF_COUNTERS = ((eax & 0xFF00) >> 8);
-		X86_IA32_PERF_COUNTERS_MASK = (1 << X86_IA32_NUM_PERF_COUNTERS) - 1;
+		PERF_COUNTERS_MASK = (1 << NUM_PERF_COUNTERS) - 1;
 		
-		X86_IA32_NUM_FIXED_PERF_COUNTERS = (edx & 0x0F);
+		NUM_FIXED_PERF_COUNTERS = (edx & 0x0F);
-		X86_IA32_FIXED_PERF_COUNTERS_MASK = 
+		FIXED_PERF_COUNTERS_MASK =
-			((1UL << X86_IA32_NUM_FIXED_PERF_COUNTERS) - 1) <<
+			((1UL << NUM_FIXED_PERF_COUNTERS) - 1) <<
-			X86_IA32_BASE_FIXED_PERF_COUNTERS;
+			BASE_FIXED_PERF_COUNTERS;
 	
 		perf_counters_discovered = 1;
-		kprintf("X86_IA32_NUM_PERF_COUNTERS: %d, X86_IA32_NUM_FIXED_PERF_COUNTERS: %d\n",
+		kprintf("NUM_PERF_COUNTERS: %d, NUM_FIXED_PERF_COUNTERS: %d\n",
-				X86_IA32_NUM_PERF_COUNTERS, X86_IA32_NUM_FIXED_PERF_COUNTERS);
+				NUM_PERF_COUNTERS, NUM_FIXED_PERF_COUNTERS);
 	}
 
 	/* Clear Fixed Counter Control */
@@ -90,20 +93,20 @@ void x86_init_perfctr(void)
 	wrmsr(MSR_PERF_FIXED_CTRL, value);
 
 	/* Clear Generic Counter Control */
-	for(i = 0; i < X86_IA32_NUM_PERF_COUNTERS; i++) {
+	for (i = 0; i < NUM_PERF_COUNTERS; i++) {
 		wrmsr(MSR_IA32_PERFEVTSEL0 + i, 0);
 	}
 
 	/* Enable PMC Control */
 	value = rdmsr(MSR_PERF_GLOBAL_CTRL);
-	value |= X86_IA32_PERF_COUNTERS_MASK;
+	value |= PERF_COUNTERS_MASK;
-	value |= X86_IA32_FIXED_PERF_COUNTERS_MASK;
+	value |= FIXED_PERF_COUNTERS_MASK;
 	wrmsr(MSR_PERF_GLOBAL_CTRL, value);
 }
 
 static int set_perfctr_x86_direct(int counter, int mode, unsigned int value)
 {
-	if (counter < 0 || counter >= X86_IA32_NUM_PERF_COUNTERS) {
+	if (counter < 0 || counter >= NUM_PERF_COUNTERS) {
 		return -EINVAL;
 	}
 
@@ -142,13 +145,14 @@ static int set_pmc_x86_direct(int counter, long val)
 	val &= 0x000000ffffffffff; // 40bit Mask
 
 	cnt_bit = 1UL << counter;
-	if ( cnt_bit & X86_IA32_PERF_COUNTERS_MASK ) {
+	if (cnt_bit & PERF_COUNTERS_MASK) {
 		// set generic pmc
 		wrmsr(MSR_IA32_PMC0 + counter, val);
 	}
-	else if ( cnt_bit & X86_IA32_FIXED_PERF_COUNTERS_MASK ) {
+	else if (cnt_bit & FIXED_PERF_COUNTERS_MASK) {
 		// set fixed pmc
-		wrmsr(MSR_IA32_FIXED_CTR0 + counter - X86_IA32_BASE_FIXED_PERF_COUNTERS, val);
+		wrmsr(MSR_IA32_FIXED_CTR0 +
+			counter - BASE_FIXED_PERF_COUNTERS, val);
 	}
 	else {
 		return -EINVAL;
@@ -168,10 +172,10 @@ static int set_fixed_counter(int counter, int mode)
 {
 	unsigned long value = 0;
 	unsigned int  ctr_mask = 0xf;
-	int counter_idx = counter - X86_IA32_BASE_FIXED_PERF_COUNTERS ;
+	int counter_idx = counter - BASE_FIXED_PERF_COUNTERS;
 	unsigned int  set_val = 0;
 
-	if (counter_idx < 0 || counter_idx >= X86_IA32_NUM_FIXED_PERF_COUNTERS) {
+	if (counter_idx < 0 || counter_idx >= NUM_FIXED_PERF_COUNTERS) {
 		return -EINVAL;
 	}
 
@@ -201,14 +205,13 @@ int ihk_mc_perfctr_init_raw(int counter, uint64_t config, int mode)
 int ihk_mc_perfctr_init_raw(int counter, unsigned int code, int mode)
 #endif /*POSTK_DEBUG_TEMP_FIX_29*/
 {
-#ifdef POSTK_DEBUG_TEMP_FIX_31
 	// PAPI_REF_CYC counted by fixed counter
-	if (counter >= X86_IA32_BASE_FIXED_PERF_COUNTERS) {
+	if (counter >= BASE_FIXED_PERF_COUNTERS &&
+		counter < BASE_FIXED_PERF_COUNTERS + NUM_FIXED_PERF_COUNTERS) {
 		return ihk_mc_perfctr_fixed_init(counter, mode);
 	}
-#endif /*POSTK_DEBUG_TEMP_FIX_31*/
 
-	if (counter < 0 || counter >= X86_IA32_NUM_PERF_COUNTERS) {
+	if (counter < 0 || counter >= NUM_PERF_COUNTERS) {
 		return -EINVAL;
 	}
 
@@ -241,7 +244,7 @@ int ihk_mc_perfctr_init(int counter, enum ihk_perfctr_type type, int mode)
 	}
 #endif /*POSTK_DEBUG_TEMP_FIX_29*/
 
-	if (counter < 0 || counter >= X86_IA32_NUM_PERF_COUNTERS) {
+	if (counter < 0 || counter >= NUM_PERF_COUNTERS) {
 		return -EINVAL;
 	}
 	if (type < 0 || type >= PERFCTR_MAX_TYPE) {
@@ -254,22 +257,50 @@ int ihk_mc_perfctr_init(int counter, enum ihk_perfctr_type type, int mode)
 	return set_perfctr_x86_direct(counter, mode, x86_march_perfmap[type]);
 }
 
+int ihk_mc_perfctr_set_extra(struct mc_perf_event *event)
+{
+	struct thread *thread = cpu_local_var(current);
+
+	// allocate extra_reg
+	if (thread->extra_reg_alloc_map & (1UL << event->extra_reg.idx)) {
+		if (event->extra_reg.idx == EXTRA_REG_RSP_0) {
+			event->extra_reg.idx = EXTRA_REG_RSP_1;
+		}
+		else if (event->extra_reg.idx == EXTRA_REG_RSP_1) {
+			event->extra_reg.idx = EXTRA_REG_RSP_0;
+		}
+
+		if (thread->extra_reg_alloc_map & (1UL << event->extra_reg.idx)) {
+			// extra_regs are full
+			return -1;
+		}
+	}
+
+	if (event->extra_reg.idx == EXTRA_REG_RSP_0) {
+		event->hw_config &= ~0xffUL;
+		event->hw_config |= ihk_mc_get_extra_reg_event(EXTRA_REG_RSP_0);
+		event->extra_reg.reg = MSR_OFFCORE_RSP_0;
+	}
+	else if (event->extra_reg.idx == EXTRA_REG_RSP_1) {
+		event->hw_config &= ~0xffUL;
+		event->hw_config |= ihk_mc_get_extra_reg_event(EXTRA_REG_RSP_1);
+		event->extra_reg.reg = MSR_OFFCORE_RSP_1;
+	}
+		
+	thread->extra_reg_alloc_map |= (1UL << event->extra_reg.idx);
+	wrmsr(event->extra_reg.reg, event->extra_reg.config);
+	return 0;
+}
+
 #ifdef HAVE_MARCH_PERFCTR_START
 extern void x86_march_perfctr_start(unsigned long counter_mask);
 #endif
 
-#ifdef POSTK_DEBUG_TEMP_FIX_30
-int ihk_mc_perfctr_start(int counter)
-#else
 int ihk_mc_perfctr_start(unsigned long counter_mask)
-#endif /*POSTK_DEBUG_TEMP_FIX_30*/
 {
 	int ret = 0;
 	unsigned long value = 0;
-	unsigned long mask = X86_IA32_PERF_COUNTERS_MASK | X86_IA32_FIXED_PERF_COUNTERS_MASK;
+	unsigned long mask = PERF_COUNTERS_MASK | FIXED_PERF_COUNTERS_MASK;
-#ifdef POSTK_DEBUG_TEMP_FIX_30
-	unsigned long counter_mask = 1UL << counter;
-#endif /*POSTK_DEBUG_TEMP_FIX_30*/
 
 	PERFCTR_CHKANDJUMP(counter_mask & ~mask, "counter_mask out of range", -EINVAL);
 
@@ -286,18 +317,11 @@ int ihk_mc_perfctr_start(unsigned long counter_mask)
 	goto fn_exit;
 }
 
-#ifdef POSTK_DEBUG_TEMP_FIX_30
-int ihk_mc_perfctr_stop(int counter)
-#else
 int ihk_mc_perfctr_stop(unsigned long counter_mask)
-#endif/*POSTK_DEBUG_TEMP_FIX_30*/
 {
 	int ret = 0;
 	unsigned long value;
-	unsigned long mask = X86_IA32_PERF_COUNTERS_MASK | X86_IA32_FIXED_PERF_COUNTERS_MASK;
+	unsigned long mask = PERF_COUNTERS_MASK | FIXED_PERF_COUNTERS_MASK;
-#ifdef POSTK_DEBUG_TEMP_FIX_30
-	unsigned long counter_mask = 1UL << counter;
-#endif/*POSTK_DEBUG_TEMP_FIX_30*/
 
 	PERFCTR_CHKANDJUMP(counter_mask & ~mask, "counter_mask out of range", -EINVAL);
 
@@ -334,10 +358,10 @@ int ihk_mc_perfctr_fixed_init(int counter, int mode)
 {
 	unsigned long value = 0;
 	unsigned int  ctr_mask = 0xf;
-	int counter_idx = counter - X86_IA32_BASE_FIXED_PERF_COUNTERS ;
+	int counter_idx = counter - BASE_FIXED_PERF_COUNTERS;
 	unsigned int  set_val = 0;
 
-	if (counter_idx < 0 || counter_idx >= X86_IA32_NUM_FIXED_PERF_COUNTERS) {
+	if (counter_idx < 0 || counter_idx >= NUM_FIXED_PERF_COUNTERS) {
 		return -EINVAL;
 	}
 
@@ -378,7 +402,7 @@ int ihk_mc_perfctr_read_mask(unsigned long counter_mask, unsigned long *value)
 {
 	int i, j;
 
-	for (i = 0, j = 0; i < X86_IA32_NUM_PERF_COUNTERS && counter_mask;
+	for (i = 0, j = 0; i < NUM_PERF_COUNTERS && counter_mask;
 	     i++, counter_mask >>= 1) {
 		if (counter_mask & 1) {
 			value[j++] = rdpmc(i);
@@ -398,13 +422,14 @@ unsigned long ihk_mc_perfctr_read(int counter)
 
 	cnt_bit = 1UL << counter;
 
-	if ( cnt_bit & X86_IA32_PERF_COUNTERS_MASK ) {
+	if (cnt_bit & PERF_COUNTERS_MASK) {
 		// read generic pmc
 		retval = rdpmc(counter);
 	}
-	else if ( cnt_bit & X86_IA32_FIXED_PERF_COUNTERS_MASK ) {
+	else if (cnt_bit & FIXED_PERF_COUNTERS_MASK) {
 		// read fixed pmc
-		retval = rdpmc((1 << 30) + (counter - X86_IA32_BASE_FIXED_PERF_COUNTERS));
+		retval = rdpmc((1 << 30) +
+			(counter - BASE_FIXED_PERF_COUNTERS));
 	}
 	else {
 		retval = -EINVAL;
@@ -426,12 +451,12 @@ unsigned long ihk_mc_perfctr_read_msr(int counter)
 
 	cnt_bit = 1UL << counter;
 
-	if ( cnt_bit & X86_IA32_PERF_COUNTERS_MASK ) {
+	if (cnt_bit & PERF_COUNTERS_MASK) {
 		// read generic pmc
 		idx = MSR_IA32_PMC0 + counter;
 		retval = (unsigned long) rdmsr(idx);
 	}
-	else if ( cnt_bit & X86_IA32_FIXED_PERF_COUNTERS_MASK ) {
+	else if (cnt_bit & FIXED_PERF_COUNTERS_MASK) {
 		// read fixed pmc
 		idx = MSR_IA32_FIXED_CTR0 + counter;
 		retval = (unsigned long) rdmsr(idx);
@@ -464,8 +489,8 @@ int ihk_mc_perfctr_alloc_counter(unsigned int *type, unsigned long *config, unsi
 	}
 
 	// find avail generic counter
-       	for(i = 0; i < X86_IA32_NUM_PERF_COUNTERS; i++) {
+	for (i = 0; i < NUM_PERF_COUNTERS; i++) {
-		if(!(pmc_status & (1 << i))) {
+		if (!(pmc_status & (1 << i))) {
 			ret = i;
 			break;
 		}

arch/x86_64/kernel/vsyscall.c

@@ -31,51 +31,6 @@ struct tod_data_s tod_data
 	.version =	IHK_ATOMIC64_INIT(0),
 };
 
-static inline void cpu_pause_for_vsyscall(void)
-{
-	asm volatile ("pause" ::: "memory");
-	return;
-} /* cpu_pause_for_vsyscall() */
-
-static inline void calculate_time_from_tsc(struct timespec *ts)
-{
-	long ver;
-	unsigned long current_tsc;
-	__time_t sec_delta;
-	long ns_delta;
-
-	for (;;) {
-		while ((ver = ihk_atomic64_read(&tod_data.version)) & 1) {
-			/* settimeofday() is in progress */
-			cpu_pause_for_vsyscall();
-		}
-		rmb();
-		*ts = tod_data.origin;
-		rmb();
-		if (ver == ihk_atomic64_read(&tod_data.version)) {
-			break;
-		}
-
-		/* settimeofday() has intervened */
-		cpu_pause_for_vsyscall();
-	}
-
-	current_tsc = rdtsc();
-	sec_delta = current_tsc / tod_data.clocks_per_sec;
-	ns_delta = NS_PER_SEC * (current_tsc % tod_data.clocks_per_sec)
-		/ tod_data.clocks_per_sec;
-	/* calc. of ns_delta overflows if clocks_per_sec exceeds 18.44 GHz */
-
-	ts->tv_sec += sec_delta;
-	ts->tv_nsec += ns_delta;
-	if (ts->tv_nsec >= NS_PER_SEC) {
-		ts->tv_nsec -= NS_PER_SEC;
-		++ts->tv_sec;
-	}
-
-	return;
-} /* calculate_time_from_tsc() */
-
 int vsyscall_gettimeofday(struct timeval *tv, void *tz)
 {
 	int error;
@@ -102,7 +57,7 @@ int vsyscall_gettimeofday(struct timeval *tv, void *tz)
 			: "%rcx", "%r11", "memory");
 
 	if (error) {
-		*(int *)0 = 0;	/* i.e. raise(SIGSEGV) */
+		*(volatile int *)0 = 0;	/* i.e. raise(SIGSEGV) */
 	}
 	return error;
 } /* vsyscall_gettimeofday() */

arch/x86_64/tools/mcoverlay-create-smp-x86.sh.in

@@ -43,9 +43,9 @@ error_exit() {
 		;;
 	esac
 
-	exit 1
+	# Retun -EINVAL
+	exit -22
 }
-
 fi
 
 if [ ! -e /tmp/mcos ]; then 
@@ -145,3 +145,5 @@ for cpuid in `find /sys/bus/cpu/devices/* -maxdepth 0 -name "cpu[0123456789]*" -
 	rm -rf /tmp/mcos/mcos0_sys/bus/cpu/devices/$cpuid
     fi
 done
+
+exit 0

arch/x86_64/tools/mcoverlay-destroy-smp-x86.sh.in

@@ -8,6 +8,9 @@ if grep mcoverlay /proc/modules &>/dev/null; then
     if [ -e /tmp/mcos ]; then rm -rf /tmp/mcos; fi
     if ! rmmod mcoverlay 2>/dev/null; then
 	echo "error: removing mcoverlay" >&2
-	exit 1
+	# Return -EINVAL
+	exit -22
     fi
 fi
+
+exit 0

arch/x86_64/tools/mcreboot-smp-x86.sh.in

@@ -12,6 +12,7 @@
 # the same set of resources as it used previously.
 # Note that the script does not output anything unless an error occurs.
 
+ret=1
 prefix="@prefix@"
 BINDIR="${prefix}/bin"
 SBINDIR="${prefix}/sbin"
@@ -19,6 +20,7 @@ ETCDIR=@ETCDIR@
 KMODDIR="${prefix}/kmod"
 KERNDIR="${prefix}/@TARGET@/kernel"
 ENABLE_MCOVERLAYFS="@ENABLE_MCOVERLAYFS@"
+MCK_BUILDID=@BUILDID@
 
 mem="512M@0"
 cpus=""
@@ -43,8 +45,12 @@ fi
 
 turbo=""
 ihk_irq=""
+safe_kernel_map=""
+umask_old=`umask`
+idle_halt=""
+allow_oversubscribe=""
 
-while getopts :tk:c:m:o:f:r:q:i:d: OPT
+while getopts stk:c:m:o:f:r:q:i:d:e:hO OPT
 do
 	case ${OPT} in
 	f)	facility=${OPTARG}
@@ -57,18 +63,26 @@ do
 		;;
 	m) mem=${OPTARG}
 		;;
+	s) safe_kernel_map="safe_kernel_map"
+		;;
 	r) ikc_map=${OPTARG}
 		;;
 	q) ihk_irq=${OPTARG}
 		;;
 	t) turbo="turbo"
 		;;
+	e) extra_kopts=${OPTARG}
+		;;
 	d) DUMP_LEVEL=${OPTARG}
 		;;
 	i) mon_interval=${OPTARG}
 		;;
-	*)  echo "invalid option -${OPT}" >&2
+	h) idle_halt="idle_halt"
-		exit 1
+		;;
+	O) allow_oversubscribe="allow_oversubscribe"
+		;;
+	\?) exit 1
+		;;
 	esac
 done
 
@@ -80,6 +94,16 @@ fi
 if [ "${redirect_kmsg}" != "0" -o "${mon_interval}" != "-1" ]; then
     ${SBINDIR}/ihkmond -f ${facility} -k ${redirect_kmsg} -i ${mon_interval}
 fi
+
+disable_irqbalance_mck() {
+	if [ -f /etc/systemd/system/irqbalance_mck.service ]; then
+		systemctl disable irqbalance_mck.service >/dev/null 2>/dev/null
+
+		# Invalid .service file persists so remove it
+		rm -f /etc/systemd/system/irqbalance_mck.service
+	fi
+}
+
 #
 # Revert any state that has been initialized before the error occured.
 #
@@ -87,6 +111,16 @@ error_exit() {
 	local status=$1
 
 	case $status in
+	irqbalance_mck_started)
+		if [ "${irqbalance_used}" == "yes" ]; then
+		    if [ "`systemctl status irqbalance_mck.service 2> /dev/null |grep -E 'Active: active'`" != "" ]; then
+			if ! systemctl stop irqbalance_mck.service 2>/dev/null; then
+			    echo "warning: failed to stop irqbalance_mck" >&2
+			fi
+			disable_irqbalance_mck
+		    fi
+		fi
+		;&
 	mcos_sys_mounted)
 		if [ "$enable_mcoverlay" == "yes" ]; then
 			umount /tmp/mcos/mcos0_sys
@@ -148,25 +182,25 @@ error_exit() {
 		fi
 		;&
 	ihk_smp_loaded)
-		rmmod ihk_smp_x86 2>/dev/null || echo "warning: failed to remove ihk_smp_x86" >&2
+		rmmod ihk_smp_@ARCH@ 2>/dev/null || echo "warning: failed to remove ihk_smp_@ARCH@" >&2
 		;&
 	ihk_loaded)
 		rmmod ihk 2>/dev/null || echo "warning: failed to remove ihk" >&2
 		;&
+	smp_affinity_modified)
+		umask $umask_old
+		if [ "${irqbalance_used}" == "yes" ]; then
+		    if ! perl -e '$tmpdir="/tmp/mcreboot"; @files = grep { -f } glob "$tmpdir/proc/irq/*/smp_affinity"; foreach $file (@files) { $dest = substr($file, length($tmpdir)); if (0) {print "cp $file $dest\n";} system("cp $file $dest 2>/dev/null"); }'; then
+			echo "warning: failed to restore /proc/irq/*/smp_affinity" >&2
+		    fi
+		    if [ -e /tmp/mcreboot ]; then rm -rf /tmp/mcreboot; fi
+		fi
+		;&
 	irqbalance_stopped)
-		if [ "`systemctl status irqbalance_mck.service 2> /dev/null |grep -E 'Active: active'`" != "" ]; then
+		if [ "${irqbalance_used}" == "yes" ]; then
-			if ! systemctl stop irqbalance_mck.service 2>/dev/null; then
+		    if ! systemctl start irqbalance.service; then
-				echo "warning: failed to stop irqbalance_mck" >&2
+			echo "warning: failed to start irqbalance" >&2;
-			fi
+		    fi
-			if ! systemctl disable irqbalance_mck.service >/dev/null 2>/dev/null; then
-				echo "warning: failed to disable irqbalance_mck" >&2
-			fi
-			if ! etcdir=@ETCDIR@ perl -e '$etcdir=$ENV{'etcdir'}; @files = grep { -f } glob "$etcdir/proc/irq/*/smp_affinity"; foreach $file (@files) { $dest = substr($file, length($etcdir)); if(0) {print "cp $file $dest\n";} system("cp $file $dest 2>/dev/null"); }'; then
-				echo "warning: failed to restore /proc/irq/*/smp_affinity" >&2
-			fi
-			if ! systemctl start irqbalance.service; then
-				echo "warning: failed to start irqbalance" >&2;
-			fi
 		fi
 		;&
 	initial)
@@ -174,7 +208,8 @@ error_exit() {
 		;;
 	esac
 
-	exit 1
+	# Propagate exit status if any
+	exit $ret
 }
 
 ihk_ikc_irq_core=0
@@ -200,7 +235,7 @@ if [ "${ENABLE_MCOVERLAYFS}" == "yes" ]; then
 			enable_mcoverlay="yes"
 		fi
 	else
-		if [ ${linux_version_code} -eq 199168 -a ${rhel_release} -ge 327 ]; then
+		if [ ${linux_version_code} -eq 199168 -a ${rhel_release} -ge 327 -a ${rhel_release} -le 862 ]; then
 			enable_mcoverlay="yes"
 		fi
 		if [ ${linux_version_code} -ge 262144 -a ${linux_version_code} -lt 262400 ]; then
@@ -225,7 +260,11 @@ fi
 
 # Remove mcoverlay if loaded
 if [ "$enable_mcoverlay" == "yes" ]; then
-    . ${SBINDIR}/mcoverlay-destroy.sh
+    ${SBINDIR}/mcoverlay-destroy.sh
+    ret=$?
+    if [ $ret -ne 0 ]; then
+	error_exit "initial"
+    fi
 fi
 
 # Stop irqbalance
@@ -236,9 +275,9 @@ if [ "${irqbalance_used}" == "yes" ]; then
 		exit 1
     fi;
 
-    if ! etcdir=@ETCDIR@ perl -e 'use File::Copy qw(copy); $etcdir=$ENV{'etcdir'}; @files = grep { -f } glob "/proc/irq/*/smp_affinity"; foreach $file (@files) { $rel = substr($file, 1); $dir=substr($rel, 0, length($rel)-length("/smp_affinity")); if(0) { print "cp $file $etcdir/$rel\n";} if(system("mkdir -p $etcdir/$dir")){ exit 1;} if(!copy($file,"$etcdir/$rel")){ exit 1;} }'; then
+    if ! perl -e 'use File::Copy qw(copy); $tmpdir="/tmp/mcreboot"; @files = grep { -f } glob "/proc/irq/*/smp_affinity"; foreach $file (@files) { $rel = substr($file, 1); $dir = substr($rel, 0, length($rel) - length("/smp_affinity")); if (system("mkdir -p $tmpdir/$dir")) { exit 1; } if (0) { print "cp $file $tmpdir/$rel\n"; } if (!copy($file,"$tmpdir/$rel")) { exit 1; } }'; then
 		echo "error: saving /proc/irq/*/smp_affinity" >&2
-		error_exit "mcos_sys_mounted"
+		error_exit "irqbalance_stopped"
 	fi;
 
 # Prevent /proc/irq/*/smp_affinity from getting zero after offlining
@@ -252,16 +291,20 @@ if [ "${irqbalance_used}" == "yes" ]; then
 
     if ! ncpus=$ncpus smp_affinity_mask=$smp_affinity_mask perl -e '@dirs = grep { -d } glob "/proc/irq/*"; foreach $dir (@dirs) { $hit = 0; $affinity_str = `cat $dir/smp_affinity`; chomp $affinity_str; @int32strs = split /,/, $affinity_str; @int32strs_mask=split /,/, $ENV{'smp_affinity_mask'}; for($i=0;$i <= $#int32strs_mask; $i++) { $int32strs_inv[$i] = sprintf("%08x",hex($int32strs_mask[$i])^0xffffffff); if($i == 0) { $len = int((($ENV{'ncpus'}%32)+3)/4); if($len != 0) { $int32strs_inv[$i] = substr($int32strs_inv[$i], -$len, $len); } } } $inv = join(",", @int32strs_inv); $nint32s = int(($ENV{'ncpus'}+31)/32); for($j = $nint32s - 1; $j >= 0; $j--) { if(hex($int32strs[$nint32s - 1 - $j]) & hex($int32strs_mask[$nint32s - 1 - $j])) { $hit = 1; }} if($hit == 1) { $cmd = "echo $inv > $dir/smp_affinity 2>/dev/null"; system $cmd;}}'; then
 		echo "error: modifying /proc/irq/*/smp_affinity" >&2
-		error_exit "mcos_sys_mounted"
+		error_exit "irqbalance_stopped"
 	fi
-
 fi
 
+# Set umask so that proc/sys files/directories created by
+# mcctrl.ko and mcreboot.sh have appropriate permission bits
+umask_dec=$(( 8#${umask_old} & 8#0002 ))
+umask 0`printf "%o" ${umask_dec}`
+
 # Load IHK if not loaded
 if ! grep -E 'ihk\s' /proc/modules &>/dev/null; then
 	if ! taskset -c 0 insmod ${KMODDIR}/ihk.ko 2>/dev/null; then
 		echo "error: loading ihk" >&2
-		error_exit "irqbalance_stopped"
+		error_exit "smp_affinity_modified"
 	fi
 fi
 
@@ -277,7 +320,7 @@ echo 1 > /proc/sys/vm/compact_memory 2>/dev/null
 sync
 
 # Load IHK-SMP if not loaded and reserve CPUs and memory
-if ! grep ihk_smp_x86 /proc/modules &>/dev/null; then
+if ! grep ihk_smp_@ARCH@ /proc/modules &>/dev/null; then
 	if [ "$ihk_irq" == "" ]; then
 		for i in `seq 64 255`; do
 			if [ ! -d /proc/irq/$i ] && [ "`cat /proc/interrupts | grep ":" | awk '{print $1}' | grep -o '[0-9]*' | grep -e '^$i$'`" == "" ]; then
@@ -290,12 +333,12 @@ if ! grep ihk_smp_x86 /proc/modules &>/dev/null; then
 			error_exit "ihk_loaded"
 		fi
 	fi
-	if ! taskset -c 0 insmod ${KMODDIR}/ihk-smp-x86.ko ihk_start_irq=$ihk_irq ihk_ikc_irq_core=$ihk_ikc_irq_core 2>/dev/null; then
+	if ! taskset -c 0 insmod ${KMODDIR}/ihk-smp-@ARCH@.ko ihk_start_irq=$ihk_irq ihk_ikc_irq_core=$ihk_ikc_irq_core 2>/dev/null; then
-		echo "error: loading ihk-smp-x86" >&2
+		echo "error: loading ihk-smp-@ARCH@" >&2
 		error_exit "ihk_loaded"
 	fi
 
-	# Offline-reonline RAM (special case for OFP SNC-4 mode)
+	# Offline-reonline RAM (special case for OFP SNC-4 flat mode)
 	if [ "`hostname | grep "c[0-9][0-9][0-9][0-9].ofp"`" != "" ] && [ "`cat /sys/devices/system/node/online`" == "0-7" ]; then
 		for i in  0 1 2 3; do
 			find /sys/devices/system/node/node$i/memory*/ -name "online" | while read f; do
@@ -309,6 +352,22 @@ if ! grep ihk_smp_x86 /proc/modules &>/dev/null; then
 			find /sys/devices/system/node/node$i/memory*/ -name "online" | while read f; do
 				echo 0 > $f 2>&1 > /dev/null;
 			done
+		done
+		for i in 4 5 6 7; do
+			find /sys/devices/system/node/node$i/memory*/ -name "online" | while read f; do
+				echo 1 > $f 2>&1 > /dev/null;
+			done
+		done
+	fi
+
+	# Offline-reonline RAM (special case for OFP Quadrant flat mode)
+	if [ "`hostname | grep "c[0-9][0-9][0-9][0-9].ofp"`" != "" ] && [ "`cat /sys/devices/system/node/online`" == "0-1" ]; then
+		for i in 1; do
+			find /sys/devices/system/node/node$i/memory*/ -name "online" | while read f; do
+				echo 0 > $f 2>&1 > /dev/null;
+			done
+		done
+		for i in 1; do
 			find /sys/devices/system/node/node$i/memory*/ -name "online" | while read f; do
 				echo 1 > $f 2>&1 > /dev/null;
 			done
@@ -333,13 +392,20 @@ if ! grep mcctrl /proc/modules &>/dev/null; then
 	fi
 fi
 
+# Check that different versions of binaries/scripts are not mixed
+IHK_BUILDID=`${SBINDIR}/ihkconfig 0 get buildid`
+if [  "${IHK_BUILDID}" != "${MCK_BUILDID}" ]; then
+	echo "IHK build-id (${IHK_BUILDID}) didn't match McKernel build-id (${MCK_BUILDID})." >&2
+	exit 1
+fi
+
 # Destroy all LWK instances
 if ls /dev/mcos* 1>/dev/null 2>&1; then
 	for i in /dev/mcos*; do
 		ind=`echo $i|cut -c10-`;
 		# Retry when conflicting with ihkmond
 		nretry=0
-		until ${SBINDIR}/ihkconfig 0 destroy $ind || [ $nretry -lt 4 ]; do
+		until ${SBINDIR}/ihkconfig 0 destroy $ind || [ $nretry -ge 4 ]; do
 		    sleep 0.25
 		    nretry=$[ $nretry + 1 ]
 		done
@@ -383,7 +449,7 @@ if ! ${SBINDIR}/ihkosctl 0 load ${KERNDIR}/mckernel.img; then
 fi
 
 # Set kernel arguments
-if ! ${SBINDIR}/ihkosctl 0 kargs "hidos $turbo dump_level=${DUMP_LEVEL}"; then
+if ! ${SBINDIR}/ihkosctl 0 kargs "hidos $turbo $safe_kernel_map $idle_halt dump_level=${DUMP_LEVEL} $extra_kopts $allow_oversubscribe"; then
 	echo "error: setting kernel arguments" >&2
 	error_exit "os_created"
 fi
@@ -401,7 +467,11 @@ fi
 
 # Overlay /proc, /sys with McKernel specific contents
 if [ "$enable_mcoverlay" == "yes" ]; then
-    . ${SBINDIR}/mcoverlay-create.sh
+    ${SBINDIR}/mcoverlay-create.sh
+    ret=$?
+    if [ $ret -ne 0 ]; then
+	error_exit "os_created"
+    fi
 fi
 
 # Start irqbalance with CPUs and IRQ for McKernel banned
@@ -409,7 +479,9 @@ if [ "${irqbalance_used}" == "yes" ]; then
     banirq=`cat /proc/interrupts| perl -e 'while(<>) { if(/^\s*(\d+).*IHK\-SMP\s*$/) {print $1;}}'`
 
     sed "s/%mask%/$smp_affinity_mask/g" $ETCDIR/irqbalance_mck.in | sed "s/%banirq%/$banirq/g" > /tmp/irqbalance_mck
-	systemctl disable irqbalance_mck.service >/dev/null 2>/dev/null
+
+	disable_irqbalance_mck
+
 	if ! systemctl link $ETCDIR/irqbalance_mck.service >/dev/null 2>/dev/null; then
 		echo "error: linking irqbalance_mck" >&2
 		error_exit "mcos_sys_mounted"
@@ -421,3 +493,8 @@ if [ "${irqbalance_used}" == "yes" ]; then
 	fi
 #    echo cpus=$cpus ncpus=$ncpus banirq=$banirq
 fi
+
+# Restore umask
+umask ${umask_old}
+
+exit 0

arch/x86_64/tools/mcstop+release-smp-x86.sh.in

@@ -18,17 +18,24 @@ mem=""
 cpus=""
 irqbalance_used=""
 
+disable_irqbalance_mck() {
+	if [ -f /etc/systemd/system/irqbalance_mck.service ]; then
+		systemctl disable irqbalance_mck.service >/dev/null 2>/dev/null
+
+		# Invalid .service file persists so remove it
+		rm -f /etc/systemd/system/irqbalance_mck.service
+	fi
+}
+
 # No SMP module? Exit.
-if ! grep ihk_smp_x86 /proc/modules &>/dev/null; then exit 0; fi
+if ! grep ihk_smp_@ARCH@ /proc/modules &>/dev/null; then exit 0; fi
 
 if [ "`systemctl status irqbalance_mck.service 2> /dev/null |grep -E 'Active: active'`" != "" ]; then
 	irqbalance_used="yes"
 	if ! systemctl stop irqbalance_mck.service 2>/dev/null; then
 		echo "warning: failed to stop irqbalance_mck" >&2
 	fi
-	if ! systemctl disable irqbalance_mck.service >/dev/null 2>/dev/null; then
+	disable_irqbalance_mck
-		echo "warning: failed to disable irqbalance_mck" >&2
-	fi
 fi
 
 # Destroy all LWK instances
@@ -37,7 +44,7 @@ if ls /dev/mcos* 1>/dev/null 2>&1; then
 		ind=`echo $i|cut -c10-`;
 		# Retry when conflicting with ihkmond
 		nretry=0
-		until ${SBINDIR}/ihkconfig 0 destroy $ind || [ $nretry -lt 4 ]; do
+		until ${SBINDIR}/ihkconfig 0 destroy $ind || [ $nretry -ge 4 ]; do
 		    sleep 0.25
 		    nretry=$[ $nretry + 1 ]
 		done
@@ -48,6 +55,9 @@ if ls /dev/mcos* 1>/dev/null 2>&1; then
 	done
 fi
 
+# Allow ihkmond to flush kmsg buffer
+sleep 2.0
+
 # Query IHK-SMP resources and release them
 if ! ${SBINDIR}/ihkconfig 0 query cpu > /dev/null; then
 	echo "error: querying cpus" >&2
@@ -62,17 +72,23 @@ if [ "${cpus}" != "" ]; then
 	fi
 fi
 
-if ! ${SBINDIR}/ihkconfig 0 query mem > /dev/null; then
+#if ! ${SBINDIR}/ihkconfig 0 query mem > /dev/null; then
-	echo "error: querying memory" >&2
+#	echo "error: querying memory" >&2
-	exit 1
+#	exit 1
-fi
+#fi
+#
+#mem=`${SBINDIR}/ihkconfig 0 query mem`
+#if [ "${mem}" != "" ]; then
+#	if ! ${SBINDIR}/ihkconfig 0 release mem $mem > /dev/null; then
+#		echo "error: releasing memory" >&2
+#		exit 1
+#	fi
+#fi
 
-mem=`${SBINDIR}/ihkconfig 0 query mem`
+# Release all memory
-if [ "${mem}" != "" ]; then
+if ! ${SBINDIR}/ihkconfig 0 release mem "all" > /dev/null; then
-	if ! ${SBINDIR}/ihkconfig 0 release mem $mem > /dev/null; then
+	echo "error: releasing memory" >&2
-		echo "error: releasing memory" >&2
+	exit 1
-		exit 1
-	fi
 fi
 
 # Remove delegator if loaded
@@ -84,12 +100,18 @@ if grep mcctrl /proc/modules &>/dev/null; then
 fi
 
 # Remove mcoverlay if loaded
-. ${SBINDIR}/mcoverlay-destroy.sh
+${SBINDIR}/mcoverlay-destroy.sh
+ret=$?
+if [ $ret -ne 0 ]; then
+    echo "error: mcoverlay-destroy.sh" >&2
+    exit $ret
+fi
+
 
 # Remove SMP module
-if grep ihk_smp_x86 /proc/modules &>/dev/null; then
+if grep ihk_smp_@ARCH@ /proc/modules &>/dev/null; then
-	if ! rmmod ihk_smp_x86 2>/dev/null; then
+	if ! rmmod ihk_smp_@ARCH@ 2>/dev/null; then
-		echo "error: removing ihk_smp_x86" >&2
+		echo "error: removing ihk_smp_@ARCH@" >&2
 		exit 1
 	fi
 fi
@@ -110,9 +132,10 @@ fi
 
 # Start irqbalance with the original settings
 if [ "${irqbalance_used}" != "" ]; then
-	if ! etcdir=@ETCDIR@ perl -e '$etcdir=$ENV{'etcdir'}; @files = grep { -f } glob "$etcdir/proc/irq/*/smp_affinity"; foreach $file (@files) { $dest = substr($file, length($etcdir)); if(0) {print "cp $file $dest\n";} system("cp $file $dest 2>/dev/null"); }'; then
+	if ! perl -e '$tmpdir="/tmp/mcreboot"; @files = grep { -f } glob "$tmpdir/proc/irq/*/smp_affinity"; foreach $file (@files) { $dest = substr($file, length($tmpdir)); if (0) {print "cp $file $dest\n";} system("cp $file $dest 2>/dev/null"); }'; then
 		echo "warning: failed to restore /proc/irq/*/smp_affinity" >&2
 	fi
+	if [ -e /tmp/mcreboot ]; then rm -rf /tmp/mcreboot; fi
 	if ! systemctl start irqbalance.service; then
 		echo "warning: failed to start irqbalance" >&2;
 	fi

config.h.in

@@ -3,13 +3,19 @@
 /* Path of install directory for binary */
 #undef BINDIR
 
+/* IHK build-id to confirm IHK and McKernel built at the same time are used */
+#undef BUILDID
+
 /* whether mcoverlayfs is enabled */
 #undef ENABLE_MCOVERLAYFS
 
 /* whether memdump feature is enabled */
 #undef ENABLE_MEMDUMP
 
-/* whether mcoverlayfs is enabled */
+/* whether perf is enabled */
+#undef ENABLE_PERF
+
+/* whether qlmpi is enabled */
 #undef ENABLE_QLMPI
 
 /* whether rusage is enabled */
@@ -48,48 +54,6 @@
 /* Define to 1 if you have the <unistd.h> header file. */
 #undef HAVE_UNISTD_H
 
-/* Define to address of kernel symbol __vvar_page, or 0 if exported */
-#undef MCCTRL_KSYM___vvar_page
-
-/* Define to address of kernel symbol hpet_address, or 0 if exported */
-#undef MCCTRL_KSYM_hpet_address
-
-/* Define to address of kernel symbol hv_clock, or 0 if exported */
-#undef MCCTRL_KSYM_hv_clock
-
-/* Define to address of kernel symbol sys_mount, or 0 if exported */
-#undef MCCTRL_KSYM_sys_mount
-
-/* Define to address of kernel symbol sys_readlink, or 0 if exported */
-#undef MCCTRL_KSYM_sys_readlink
-
-/* Define to address of kernel symbol sys_umount, or 0 if exported */
-#undef MCCTRL_KSYM_sys_umount
-
-/* Define to address of kernel symbol sys_unshare, or 0 if exported */
-#undef MCCTRL_KSYM_sys_unshare
-
-/* Define to address of kernel symbol vdso_end, or 0 if exported */
-#undef MCCTRL_KSYM_vdso_end
-
-/* Define to address of kernel symbol vdso_image_64, or 0 if exported */
-#undef MCCTRL_KSYM_vdso_image_64
-
-/* Define to address of kernel symbol vdso_pages, or 0 if exported */
-#undef MCCTRL_KSYM_vdso_pages
-
-/* Define to address of kernel symbol vdso_spec, or 0 if exported */
-#undef MCCTRL_KSYM_vdso_spec
-
-/* Define to address of kernel symbol vdso_start, or 0 if exported */
-#undef MCCTRL_KSYM_vdso_start
-
-/* Define to address of kernel symbol walk_page_range, or 0 if exported */
-#undef MCCTRL_KSYM_walk_page_range
-
-/* Define to address of kernel symbol zap_page_range, or 0 if exported */
-#undef MCCTRL_KSYM_zap_page_range
-
 /* McKernel specific headers */
 #undef MCKERNEL_INCDIR
 
@@ -122,3 +86,6 @@
 
 /* Define to 1 if you have the ANSI C header files. */
 #undef STDC_HEADERS
+
+/* whether or not syscall_intercept library is linked */
+#undef WITH_SYSCALL_INTERCEPT

configure

@@ -1,6 +1,6 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for mckernel 0.9.0.
+# Generated by GNU Autoconf 2.69 for mckernel 1.6.0.
 #
 #
 # Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
@@ -577,8 +577,8 @@ MAKEFLAGS=
 # Identity of this package.
 PACKAGE_NAME='mckernel'
 PACKAGE_TARNAME='mckernel'
-PACKAGE_VERSION='0.9.0'
+PACKAGE_VERSION='1.6.0'
-PACKAGE_STRING='mckernel 0.9.0'
+PACKAGE_STRING='mckernel 1.6.0'
 PACKAGE_BUGREPORT=''
 PACKAGE_URL=''
 
@@ -628,9 +628,12 @@ IHK_RELEASE_DATE
 DCFA_VERSION
 MCKERNEL_VERSION
 IHK_VERSION
+WITH_SYSCALL_INTERCEPT
 ENABLE_QLMPI
 ENABLE_RUSAGE
 ENABLE_MCOVERLAYFS
+LDFLAGS_SYSCALL_INTERCEPT
+CPPFLAGS_SYSCALL_INTERCEPT
 MANDIR
 KERNDIR
 KMODDIR
@@ -645,6 +648,7 @@ TARGET
 UNAME_R
 KDIR
 ARCH
+BUILDID
 XCC
 FGREP
 EGREP
@@ -701,6 +705,9 @@ enable_option_checking
 with_mpi
 with_mpi_include
 with_mpi_lib
+with_syscall_intercept
+with_syscall_intercept_include
+with_syscall_intercept_lib
 with_kernelsrc
 with_target
 with_system_map
@@ -708,6 +715,7 @@ enable_dcfa
 enable_memdump
 enable_mcoverlayfs
 enable_rusage
+enable_perf
 enable_qlmpi
 with_uname_r
 '
@@ -1260,7 +1268,7 @@ if test "$ac_init_help" = "long"; then
   # Omit some internal or obsolete options to make the list less imposing.
   # This message is too long to be a string in the A/UX 3.1 sh.
   cat <<_ACEOF
-\`configure' configures mckernel 0.9.0 to adapt to many kinds of systems.
+\`configure' configures mckernel 1.6.0 to adapt to many kinds of systems.
 
 Usage: $0 [OPTION]... [VAR=VALUE]...
 
@@ -1321,7 +1329,7 @@ fi
 
 if test -n "$ac_init_help"; then
   case $ac_init_help in
-     short | recursive ) echo "Configuration of mckernel 0.9.0:";;
+     short | recursive ) echo "Configuration of mckernel 1.6.0:";;
    esac
   cat <<\_ACEOF
 
@@ -1333,6 +1341,7 @@ Optional Features:
   --enable-memdump        enable dumping memory and analyzing a dump
   --enable-mcoverlayfs    enable mcoverlayfs implementation
   --enable-rusage         enable rusage implementation
+  --enable-perf           enable perf_event implementation
   --enable-qlmpi          enable qlmpi implementation
 
 Optional Packages:
@@ -1343,6 +1352,15 @@ Optional Packages:
   --with-mpi-include=PATH specify path where mpi include directory can be
                           found
   --with-mpi-lib=PATH     specify path where mpi lib directory can be found
+  --with-syscall_intercept=PATH
+                          specify path where syscall_intercept include
+                          directory and lib directory can be found
+  --with-syscall_intercept-include=PATH
+                          specify path where syscall_intercept include
+                          directory can be found
+  --with-syscall_intercept-lib=PATH
+                          specify path where syscall_intercept lib directory
+                          can be found
   --with-kernelsrc=path   Path to 'kernel src', default is
                           /lib/modules/uname_r/build
   --with-target={attached-mic | builtin-mic | builtin-x86 | smp-x86}
@@ -1428,7 +1446,7 @@ fi
 test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
-mckernel configure 0.9.0
+mckernel configure 1.6.0
 generated by GNU Autoconf 2.69
 
 Copyright (C) 2012 Free Software Foundation, Inc.
@@ -1726,7 +1744,7 @@ cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
-It was created by mckernel $as_me 0.9.0, which was
+It was created by mckernel $as_me 1.6.0, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
@@ -2079,12 +2097,14 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
 
-IHK_VERSION=0.9.0
+
-MCKERNEL_VERSION=0.9.0
+
-DCFA_VERSION=0.9.0
+IHK_VERSION=1.6.0
-IHK_RELEASE_DATE=2013-11-18
+MCKERNEL_VERSION=1.6.0
-MCKERNEL_RELEASE_DATE=2013-11-18
+DCFA_VERSION=DCFA_VERSION_m4
-DCFA_RELEASE_DATE=2013-11-18
+IHK_RELEASE_DATE=2018-11-11
+MCKERNEL_RELEASE_DATE=2018-11-11
+DCFA_RELEASE_DATE=DCFA_RELEASE_DATE_m4
 
 
 
@@ -3510,6 +3530,195 @@ fi
 
 
 
+
+# Check whether --with-syscall_intercept was given.
+if test "${with_syscall_intercept+set}" = set; then :
+  withval=$with_syscall_intercept; case "$withval" in #(
+  yes|no|'') :
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: --without-syscall_intercept=PATH expects a valid PATH" >&5
+$as_echo "$as_me: WARNING: --without-syscall_intercept=PATH expects a valid PATH" >&2;}
+                          with_syscall_intercept="" ;; #(
+  *) :
+     ;;
+esac
+else
+  with_syscall_intercept=
+fi
+
+
+# Check whether --with-syscall_intercept-include was given.
+if test "${with_syscall_intercept_include+set}" = set; then :
+  withval=$with_syscall_intercept_include; case "$withval" in #(
+  yes|no|'') :
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: --without-syscall_intercept-include=PATH expects a valid PATH" >&5
+$as_echo "$as_me: WARNING: --without-syscall_intercept-include=PATH expects a valid PATH" >&2;}
+                          with_syscall_intercept_include="" ;; #(
+  *) :
+     ;;
+esac
+fi
+
+
+# Check whether --with-syscall_intercept-lib was given.
+if test "${with_syscall_intercept_lib+set}" = set; then :
+  withval=$with_syscall_intercept_lib; case "$withval" in #(
+  yes|no|'') :
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: --without-syscall_intercept-lib=PATH expects a valid PATH" >&5
+$as_echo "$as_me: WARNING: --without-syscall_intercept-lib=PATH expects a valid PATH" >&2;}
+                          with_syscall_intercept_lib="" ;; #(
+  *) :
+     ;;
+esac
+fi
+
+
+    # The args have been sanitized into empty/non-empty values above.
+    # Now append -I/-L args to CPPFLAGS/LDFLAGS, with more specific options
+    # taking priority
+
+    if test -n "${with_syscall_intercept_include}"; then :
+
+
+	if echo "$CPPFLAGS_SYSCALL_INTERCEPT" | $FGREP -e "\<-I${with_syscall_intercept_include}\>" >/dev/null 2>&1; then :
+  echo "CPPFLAGS_SYSCALL_INTERCEPT(='$CPPFLAGS_SYSCALL_INTERCEPT') contains '-I${with_syscall_intercept_include}', not appending" >&5
+else
+  echo "CPPFLAGS_SYSCALL_INTERCEPT(='$CPPFLAGS_SYSCALL_INTERCEPT') does not contain '-I${with_syscall_intercept_include}', appending" >&5
+		CPPFLAGS_SYSCALL_INTERCEPT="$CPPFLAGS_SYSCALL_INTERCEPT -I${with_syscall_intercept_include}"
+
+fi
+
+else
+  if test -n "${with_syscall_intercept}"; then :
+
+
+	if echo "$CPPFLAGS_SYSCALL_INTERCEPT" | $FGREP -e "\<-I${with_syscall_intercept}/include\>" >/dev/null 2>&1; then :
+  echo "CPPFLAGS_SYSCALL_INTERCEPT(='$CPPFLAGS_SYSCALL_INTERCEPT') contains '-I${with_syscall_intercept}/include', not appending" >&5
+else
+  echo "CPPFLAGS_SYSCALL_INTERCEPT(='$CPPFLAGS_SYSCALL_INTERCEPT') does not contain '-I${with_syscall_intercept}/include', appending" >&5
+		CPPFLAGS_SYSCALL_INTERCEPT="$CPPFLAGS_SYSCALL_INTERCEPT -I${with_syscall_intercept}/include"
+
+fi
+
+fi
+fi
+
+    if test -n "${with_syscall_intercept_lib}"; then :
+
+
+	if echo "$LDFLAGS_SYSCALL_INTERCEPT" | $FGREP -e "\<-L${with_syscall_intercept_lib} -Wl,-rpath,${with_syscall_intercept_lib}\>" >/dev/null 2>&1; then :
+  echo "LDFLAGS_SYSCALL_INTERCEPT(='$LDFLAGS_SYSCALL_INTERCEPT') contains '-L${with_syscall_intercept_lib} -Wl,-rpath,${with_syscall_intercept_lib}', not appending" >&5
+else
+  echo "LDFLAGS_SYSCALL_INTERCEPT(='$LDFLAGS_SYSCALL_INTERCEPT') does not contain '-L${with_syscall_intercept_lib} -Wl,-rpath,${with_syscall_intercept_lib}', appending" >&5
+		LDFLAGS_SYSCALL_INTERCEPT="$LDFLAGS_SYSCALL_INTERCEPT -L${with_syscall_intercept_lib} -Wl,-rpath,${with_syscall_intercept_lib}"
+
+fi
+
+else
+  if test -n "${with_syscall_intercept}"; then :
+
+
+	if echo "$LDFLAGS_SYSCALL_INTERCEPT" | $FGREP -e "\<-L${with_syscall_intercept}/lib -Wl,-rpath,${with_syscall_intercept}/lib\>" >/dev/null 2>&1; then :
+  echo "LDFLAGS_SYSCALL_INTERCEPT(='$LDFLAGS_SYSCALL_INTERCEPT') contains '-L${with_syscall_intercept}/lib -Wl,-rpath,${with_syscall_intercept}/lib', not appending" >&5
+else
+  echo "LDFLAGS_SYSCALL_INTERCEPT(='$LDFLAGS_SYSCALL_INTERCEPT') does not contain '-L${with_syscall_intercept}/lib -Wl,-rpath,${with_syscall_intercept}/lib', appending" >&5
+		LDFLAGS_SYSCALL_INTERCEPT="$LDFLAGS_SYSCALL_INTERCEPT -L${with_syscall_intercept}/lib -Wl,-rpath,${with_syscall_intercept}/lib"
+
+fi
+
+                  if test -d "${with_syscall_intercept}/lib64"; then :
+
+
+	if echo "$LDFLAGS_SYSCALL_INTERCEPT" | $FGREP -e "\<-L${with_syscall_intercept}/lib64 -Wl,-rpath,${with_syscall_intercept}/lib64\>" >/dev/null 2>&1; then :
+  echo "LDFLAGS_SYSCALL_INTERCEPT(='$LDFLAGS_SYSCALL_INTERCEPT') contains '-L${with_syscall_intercept}/lib64 -Wl,-rpath,${with_syscall_intercept}/lib64', not appending" >&5
+else
+  echo "LDFLAGS_SYSCALL_INTERCEPT(='$LDFLAGS_SYSCALL_INTERCEPT') does not contain '-L${with_syscall_intercept}/lib64 -Wl,-rpath,${with_syscall_intercept}/lib64', appending" >&5
+		LDFLAGS_SYSCALL_INTERCEPT="$LDFLAGS_SYSCALL_INTERCEPT -L${with_syscall_intercept}/lib64 -Wl,-rpath,${with_syscall_intercept}/lib64"
+
+fi
+
+fi
+
+fi
+
+fi
+
+    if test -n "${with_syscall_intercept}" || test -n "${with_syscall_intercept_include}" || test -n "${with_syscall_intercept_lib}"; then :
+  WITH_SYSCALL_INTERCEPT=yes
+else
+  WITH_SYSCALL_INTERCEPT=no
+fi
+
+
+if test "x$WITH_SYSCALL_INTERCEPT" == "xno" ; then
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for syscall_no_intercept in -lsyscall_intercept" >&5
+$as_echo_n "checking for syscall_no_intercept in -lsyscall_intercept... " >&6; }
+if ${ac_cv_lib_syscall_intercept_syscall_no_intercept+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lsyscall_intercept -lcapstone -ldl $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char syscall_no_intercept ();
+int
+main ()
+{
+return syscall_no_intercept ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_lib_syscall_intercept_syscall_no_intercept=yes
+else
+  ac_cv_lib_syscall_intercept_syscall_no_intercept=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_syscall_intercept_syscall_no_intercept" >&5
+$as_echo "$ac_cv_lib_syscall_intercept_syscall_no_intercept" >&6; }
+if test "x$ac_cv_lib_syscall_intercept_syscall_no_intercept" = xyes; then :
+  syscall_intercept_lib_found=yes
+else
+  syscall_intercept_lib_found=no
+fi
+
+	if test "x$syscall_intercept_lib_found" != "xyes"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: libsyscall_intercept.so not found" >&5
+$as_echo "$as_me: libsyscall_intercept.so not found" >&6;}
+fi
+
+	ac_fn_c_check_header_mongrel "$LINENO" "libsyscall_intercept_hook_point.h" "ac_cv_header_libsyscall_intercept_hook_point_h" "$ac_includes_default"
+if test "x$ac_cv_header_libsyscall_intercept_hook_point_h" = xyes; then :
+  syscall_intercept_header_found=yes
+else
+  syscall_intercept_header_found=no
+fi
+
+
+	if test "x$syscall_intercept_header_found" != "xyes"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: libsyscall_intercept_hook_point.h not found" >&5
+$as_echo "$as_me: libsyscall_intercept_hook_point.h not found" >&6;}
+fi
+
+	if test "x$syscall_intercept_lib_found" == "xyes" && test "x$syscall_intercept_header_found" == "xyes"; then :
+  WITH_SYSCALL_INTERCEPT=yes
+else
+  WITH_SYSCALL_INTERCEPT=no
+fi
+fi
+
+
+
 # Check whether --with-kernelsrc was given.
 if test "${with_kernelsrc+set}" = set; then :
   withval=$with_kernelsrc; WITH_KERNELSRC=$withval
@@ -3568,6 +3777,14 @@ else
 fi
 
 
+# Check whether --enable-perf was given.
+if test "${enable_perf+set}" = set; then :
+  enableval=$enable_perf; ENABLE_PERF=$enableval
+else
+  ENABLE_PERF=yes
+fi
+
+
 # Check whether --enable-qlmpi was given.
 if test "${enable_qlmpi+set}" = set; then :
   enableval=$enable_qlmpi; ENABLE_QLMPI=$enableval
@@ -4275,7 +4492,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     builtin-mic)
@@ -4292,7 +4509,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/attached/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/attached/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     builtin-x86)
@@ -4309,7 +4526,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/attached/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     smp-x86)
@@ -4341,7 +4558,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/smp-x86/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     smp-arm64)
@@ -4366,11 +4583,14 @@ case $WITH_TARGET in
 	if test "X$ETCDIR" = X; then
 		ETCDIR="$prefix/etc"
 	fi
+	if test "X$INCLUDEDIR" = X; then
+		INCLUDEDIR="$prefix/include"
+	fi
 	if test "X$KMODDIR" = X; then
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/smp-arm64/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     *)
@@ -4382,399 +4602,6 @@ KDIR="$WITH_KERNELSRC"
 UNAME_R="$WITH_UNAME_R"
 TARGET="$WITH_TARGET"
 
-MCCTRL_LINUX_SYMTAB=""
-case "X$WITH_SYSTEM_MAP" in
-    Xyes | Xno | X)
-	MCCTRL_LINUX_SYMTAB=""
-	;;
-    *)
-	MCCTRL_LINUX_SYMTAB="$WITH_SYSTEM_MAP"
-	;;
-esac
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for System.map" >&5
-$as_echo_n "checking for System.map... " >&6; }
-if test -r "$MCCTRL_LINUX_SYMTAB"; then
-	MCCTRL_LINUX_SYMTAB="$MCCTRL_LINUX_SYMTAB"
-elif test -r "/boot/System.map-`uname -r`"; then
-	MCCTRL_LINUX_SYMTAB="/boot/System.map-`uname -r`"
-elif test -r "$KDIR/System.map"; then
-	MCCTRL_LINUX_SYMTAB="$KDIR/System.map"
-fi
-
-if test "$MCCTRL_LINUX_SYMTAB" == ""; then
-	as_fn_error $? "could not find" "$LINENO" 5
-fi
-
-if test -z "`eval cat $MCCTRL_LINUX_SYMTAB`"; then
-	as_fn_error $? "could not read System.map file, no read permission?" "$LINENO" 5
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MCCTRL_LINUX_SYMTAB" >&5
-$as_echo "$MCCTRL_LINUX_SYMTAB" >&6; }
-
-MCCTRL_LINUX_SYMTAB_CMD="cat $MCCTRL_LINUX_SYMTAB"
-
-# MCCTRL_FIND_KSYM(SYMBOL)
-# ------------------------------------------------------
-# Search System.map for address of the given symbol and
-# do one of three things in config.h:
-# If not found, leave MCCTRL_KSYM_foo undefined
-# If found to be exported, "#define MCCTRL_KSYM_foo 0"
-# If found not to be exported, "#define MCCTRL_KSYM_foo 0x<value>"
-
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol sys_mount" >&5
-$as_echo_n "checking System.map for symbol sys_mount... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " sys_mount\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_sys_mount\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_sys_mount $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol sys_umount" >&5
-$as_echo_n "checking System.map for symbol sys_umount... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " sys_umount\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_sys_umount\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_sys_umount $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol sys_unshare" >&5
-$as_echo_n "checking System.map for symbol sys_unshare... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " sys_unshare\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_sys_unshare\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_sys_unshare $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol zap_page_range" >&5
-$as_echo_n "checking System.map for symbol zap_page_range... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " zap_page_range\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_zap_page_range\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_zap_page_range $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol vdso_image_64" >&5
-$as_echo_n "checking System.map for symbol vdso_image_64... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " vdso_image_64\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_vdso_image_64\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_vdso_image_64 $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol vdso_start" >&5
-$as_echo_n "checking System.map for symbol vdso_start... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " vdso_start\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_vdso_start\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_vdso_start $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol vdso_end" >&5
-$as_echo_n "checking System.map for symbol vdso_end... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " vdso_end\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_vdso_end\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_vdso_end $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol vdso_pages" >&5
-$as_echo_n "checking System.map for symbol vdso_pages... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " vdso_pages\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_vdso_pages\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_vdso_pages $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol __vvar_page" >&5
-$as_echo_n "checking System.map for symbol __vvar_page... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __vvar_page\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab___vvar_page\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM___vvar_page $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol hpet_address" >&5
-$as_echo_n "checking System.map for symbol hpet_address... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " hpet_address\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_hpet_address\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_hpet_address $mcctrl_addr
-_ACEOF
-
-  fi
-
-# POSTK_DEBUG_ARCH_DEP_50, add:find kernel symbol.
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol vdso_spec" >&5
-$as_echo_n "checking System.map for symbol vdso_spec... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " vdso_spec\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_vdso_spec\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_vdso_spec $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol hv_clock" >&5
-$as_echo_n "checking System.map for symbol hv_clock... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " hv_clock\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_hv_clock\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_hv_clock $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol sys_readlink" >&5
-$as_echo_n "checking System.map for symbol sys_readlink... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " sys_readlink\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_sys_readlink\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_sys_readlink $mcctrl_addr
-_ACEOF
-
-  fi
-
-
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking System.map for symbol walk_page_range" >&5
-$as_echo_n "checking System.map for symbol walk_page_range... " >&6; }
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " walk_page_range\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: not found" >&5
-$as_echo "not found" >&6; }
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_walk_page_range\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $mcctrl_result" >&5
-$as_echo "$mcctrl_result" >&6; }
-
-cat >>confdefs.h <<_ACEOF
-#define MCCTRL_KSYM_walk_page_range $mcctrl_addr
-_ACEOF
-
-  fi
-
-
 case $ENABLE_MEMDUMP in
 	yes|no|auto)
 		;;
@@ -4961,6 +4788,28 @@ else
 $as_echo "$as_me: rusage is disabled" >&6;}
 fi
 
+if test "x$ENABLE_PERF" = "xyes" ; then
+
+$as_echo "#define ENABLE_PERF 1" >>confdefs.h
+
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: perf is enabled" >&5
+$as_echo "$as_me: perf is enabled" >&6;}
+else
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: perf is disabled" >&5
+$as_echo "$as_me: perf is disabled" >&6;}
+fi
+
+if test "x$WITH_SYSCALL_INTERCEPT" = "xyes" ; then
+
+$as_echo "#define WITH_SYSCALL_INTERCEPT 1" >>confdefs.h
+
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: syscall_intercept library is linked" >&5
+$as_echo "$as_me: syscall_intercept library is linked" >&6;}
+else
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: syscall_intercept library isn't linked" >&5
+$as_echo "$as_me: syscall_intercept library isn't linked" >&6;}
+fi
+
 if test "x$MCKERNEL_INCDIR" != "x" ; then
 
 cat >>confdefs.h <<_ACEOF
@@ -4988,6 +4837,23 @@ cat >>confdefs.h <<_ACEOF
 _ACEOF
 
 
+ABS_SRCDIR=$( cd $( dirname $0 ); pwd )
+IHK_ABS_SRCDIR=${ABS_SRCDIR}/../ihk
+BUILDID=$( cd $IHK_ABS_SRCDIR; if [ ! -d .git ]; then echo $IHK_VERSION; else bash -c 'git rev-list -1 HEAD | cut -c1-8'; fi )
+{ $as_echo "$as_me:${as_lineno-$LINENO}: BUILDID=$BUILDID" >&5
+$as_echo "$as_me: BUILDID=$BUILDID" >&6;}
+if test "x$BUILDID" != "x" ; then
+
+cat >>confdefs.h <<_ACEOF
+#define BUILDID "$BUILDID"
+_ACEOF
+
+fi
+
+
+
+
+
 
 
 
@@ -5021,9 +4887,19 @@ ac_config_headers="$ac_config_headers config.h"
 
 # POSTK_DEBUG_ARCH_DEP_37
 # AC_CONFIG_FILES arch dependfiles separate
-ac_config_files="$ac_config_files Makefile executer/user/Makefile executer/user/mcexec.1:executer/user/mcexec.1in executer/user/vmcore2mckdump executer/user/arch/$ARCH/Makefile executer/user/arch/x86_64/Makefile executer/kernel/mcctrl/Makefile executer/kernel/mcctrl/arch/$ARCH/Makefile executer/kernel/mcoverlayfs/Makefile executer/kernel/mcoverlayfs/linux-3.10.0-327.36.1.el7/Makefile executer/kernel/mcoverlayfs/linux-4.0.9/Makefile executer/kernel/mcoverlayfs/linux-4.6.7/Makefile executer/include/qlmpilib.h kernel/Makefile kernel/Makefile.build kernel/include/swapfmt.h arch/x86/tools/mcreboot-attached-mic.sh arch/x86/tools/mcshutdown-attached-mic.sh arch/x86/tools/mcreboot-builtin-x86.sh arch/x86/tools/mcreboot-smp-x86.sh arch/x86/tools/mcstop+release-smp-x86.sh arch/x86/tools/mcoverlay-destroy-smp-x86.sh arch/x86/tools/mcoverlay-create-smp-x86.sh arch/x86/tools/eclair-dump-backtrace.exp arch/x86/tools/mcshutdown-builtin-x86.sh arch/x86/tools/mcreboot.1:arch/x86/tools/mcreboot.1in arch/x86/tools/irqbalance_mck.service arch/x86/tools/irqbalance_mck.in"
+ac_config_files="$ac_config_files Makefile executer/user/Makefile executer/user/mcexec.1:executer/user/mcexec.1in executer/user/vmcore2mckdump executer/user/arch/$ARCH/Makefile executer/user/arch/x86_64/Makefile executer/kernel/mcctrl/Makefile executer/kernel/mcctrl/arch/$ARCH/Makefile executer/kernel/mcoverlayfs/Makefile executer/kernel/mcoverlayfs/linux-3.10.0-327.36.1.el7/Makefile executer/kernel/mcoverlayfs/linux-4.0.9/Makefile executer/kernel/mcoverlayfs/linux-4.6.7/Makefile executer/include/qlmpilib.h kernel/Makefile kernel/Makefile.build kernel/include/swapfmt.h arch/x86_64/tools/mcreboot-attached-mic.sh arch/x86_64/tools/mcshutdown-attached-mic.sh arch/x86_64/tools/mcreboot-builtin-x86.sh arch/x86_64/tools/mcreboot-smp-x86.sh arch/x86_64/tools/mcstop+release-smp-x86.sh arch/x86_64/tools/mcoverlay-destroy-smp-x86.sh arch/x86_64/tools/mcoverlay-create-smp-x86.sh arch/x86_64/tools/eclair-dump-backtrace.exp arch/x86_64/tools/mcshutdown-builtin-x86.sh arch/x86_64/tools/mcreboot.1:arch/x86_64/tools/mcreboot.1in arch/x86_64/tools/irqbalance_mck.service arch/x86_64/tools/irqbalance_mck.in tools/mcstat/mcstat.1:tools/mcstat/mcstat.1in tools/mcstat/Makefile"
 
 
+if test -e "${ABS_SRCDIR}/test"; then
+ac_config_files="$ac_config_files mck_test_config.sample:test/mck_test_config.sample.in"
+
+fi
+
+if test "$TARGET" = "smp-x86"; then
+ac_config_files="$ac_config_files arch/x86_64/kernel/Makefile.arch"
+
+fi
+
 if test "$TARGET" = "smp-arm64"; then
 ac_config_files="$ac_config_files kernel/config/config.smp-arm64 arch/arm64/kernel/vdso/Makefile arch/arm64/kernel/Makefile.arch"
 
@@ -5541,7 +5417,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
-This file was extended by mckernel $as_me 0.9.0, which was
+This file was extended by mckernel $as_me 1.6.0, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
@@ -5603,7 +5479,7 @@ _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
-mckernel config.status 0.9.0
+mckernel config.status 1.6.0
 configured by $0, generated by GNU Autoconf 2.69,
   with options \\"\$ac_cs_config\\"
 
@@ -5741,18 +5617,22 @@ do
     "kernel/Makefile") CONFIG_FILES="$CONFIG_FILES kernel/Makefile" ;;
     "kernel/Makefile.build") CONFIG_FILES="$CONFIG_FILES kernel/Makefile.build" ;;
     "kernel/include/swapfmt.h") CONFIG_FILES="$CONFIG_FILES kernel/include/swapfmt.h" ;;
-    "arch/x86/tools/mcreboot-attached-mic.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcreboot-attached-mic.sh" ;;
+    "arch/x86_64/tools/mcreboot-attached-mic.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcreboot-attached-mic.sh" ;;
-    "arch/x86/tools/mcshutdown-attached-mic.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcshutdown-attached-mic.sh" ;;
+    "arch/x86_64/tools/mcshutdown-attached-mic.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcshutdown-attached-mic.sh" ;;
-    "arch/x86/tools/mcreboot-builtin-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcreboot-builtin-x86.sh" ;;
+    "arch/x86_64/tools/mcreboot-builtin-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcreboot-builtin-x86.sh" ;;
-    "arch/x86/tools/mcreboot-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcreboot-smp-x86.sh" ;;
+    "arch/x86_64/tools/mcreboot-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcreboot-smp-x86.sh" ;;
-    "arch/x86/tools/mcstop+release-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcstop+release-smp-x86.sh" ;;
+    "arch/x86_64/tools/mcstop+release-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcstop+release-smp-x86.sh" ;;
-    "arch/x86/tools/mcoverlay-destroy-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcoverlay-destroy-smp-x86.sh" ;;
+    "arch/x86_64/tools/mcoverlay-destroy-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcoverlay-destroy-smp-x86.sh" ;;
-    "arch/x86/tools/mcoverlay-create-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcoverlay-create-smp-x86.sh" ;;
+    "arch/x86_64/tools/mcoverlay-create-smp-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcoverlay-create-smp-x86.sh" ;;
-    "arch/x86/tools/eclair-dump-backtrace.exp") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/eclair-dump-backtrace.exp" ;;
+    "arch/x86_64/tools/eclair-dump-backtrace.exp") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/eclair-dump-backtrace.exp" ;;
-    "arch/x86/tools/mcshutdown-builtin-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcshutdown-builtin-x86.sh" ;;
+    "arch/x86_64/tools/mcshutdown-builtin-x86.sh") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcshutdown-builtin-x86.sh" ;;
-    "arch/x86/tools/mcreboot.1") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/mcreboot.1:arch/x86/tools/mcreboot.1in" ;;
+    "arch/x86_64/tools/mcreboot.1") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/mcreboot.1:arch/x86_64/tools/mcreboot.1in" ;;
-    "arch/x86/tools/irqbalance_mck.service") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/irqbalance_mck.service" ;;
+    "arch/x86_64/tools/irqbalance_mck.service") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/irqbalance_mck.service" ;;
-    "arch/x86/tools/irqbalance_mck.in") CONFIG_FILES="$CONFIG_FILES arch/x86/tools/irqbalance_mck.in" ;;
+    "arch/x86_64/tools/irqbalance_mck.in") CONFIG_FILES="$CONFIG_FILES arch/x86_64/tools/irqbalance_mck.in" ;;
+    "tools/mcstat/mcstat.1") CONFIG_FILES="$CONFIG_FILES tools/mcstat/mcstat.1:tools/mcstat/mcstat.1in" ;;
+    "tools/mcstat/Makefile") CONFIG_FILES="$CONFIG_FILES tools/mcstat/Makefile" ;;
+    "mck_test_config.sample") CONFIG_FILES="$CONFIG_FILES mck_test_config.sample:test/mck_test_config.sample.in" ;;
+    "arch/x86_64/kernel/Makefile.arch") CONFIG_FILES="$CONFIG_FILES arch/x86_64/kernel/Makefile.arch" ;;
     "kernel/config/config.smp-arm64") CONFIG_FILES="$CONFIG_FILES kernel/config/config.smp-arm64" ;;
     "arch/arm64/kernel/vdso/Makefile") CONFIG_FILES="$CONFIG_FILES arch/arm64/kernel/vdso/Makefile" ;;
     "arch/arm64/kernel/Makefile.arch") CONFIG_FILES="$CONFIG_FILES arch/arm64/kernel/Makefile.arch" ;;

configure.ac

@@ -1,11 +1,9 @@
 # configure.ac COPYRIGHT FUJITSU LIMITED 2015-2016
 AC_PREREQ(2.63)
-m4_define([IHK_VERSION_m4],[0.9.0])dnl
+m4_define([IHK_VERSION_m4],[1.6.0])dnl
-m4_define([MCKERNEL_VERSION_m4],[0.9.0])dnl
+m4_define([MCKERNEL_VERSION_m4],[1.6.0])dnl
-m4_define([DCFA_VERSION_m4],[0.9.0])dnl
+m4_define([IHK_RELEASE_DATE_m4],[2018-11-11])dnl
-m4_define([IHK_RELEASE_DATE_m4],[2013-11-18])dnl
+m4_define([MCKERNEL_RELEASE_DATE_m4],[2018-11-11])dnl
-m4_define([MCKERNEL_RELEASE_DATE_m4],[2013-11-18])dnl
-m4_define([DCFA_RELEASE_DATE_m4],[2013-11-18])dnl
 
 AC_INIT([mckernel], MCKERNEL_VERSION_m4)
 
@@ -79,6 +77,58 @@ AC_DEFUN([PAC_SET_HEADER_LIB_PATH],[
           ])
 ])
 
+AC_DEFUN([PAC_SET_HEADER_LIB_PATH_SYSCALL_INTERCEPT],[
+    AC_ARG_WITH([$1],
+                [AC_HELP_STRING([--with-$1=PATH],
+                                [specify path where $1 include directory and lib directory can be found])],
+
+                [AS_CASE(["$withval"],
+                         [yes|no|''],
+                         [AC_MSG_WARN([--with[out]-$1=PATH expects a valid PATH])
+                          with_$1=""])],
+                [with_$1=$2])
+    AC_ARG_WITH([$1-include],
+                [AC_HELP_STRING([--with-$1-include=PATH],
+                                [specify path where $1 include directory can be found])],
+                [AS_CASE(["$withval"],
+                         [yes|no|''],
+                         [AC_MSG_WARN([--with[out]-$1-include=PATH expects a valid PATH])
+                          with_$1_include=""])],
+                [])
+    AC_ARG_WITH([$1-lib],
+                [AC_HELP_STRING([--with-$1-lib=PATH],
+                                [specify path where $1 lib directory can be found])],
+                [AS_CASE(["$withval"],
+                         [yes|no|''],
+                         [AC_MSG_WARN([--with[out]-$1-lib=PATH expects a valid PATH])
+                          with_$1_lib=""])],
+                [])
+
+    # The args have been sanitized into empty/non-empty values above.
+    # Now append -I/-L args to CPPFLAGS/LDFLAGS, with more specific options
+    # taking priority
+
+    AS_IF([test -n "${with_$1_include}"],
+          [PAC_APPEND_FLAG([-I${with_$1_include}],[CPPFLAGS_SYSCALL_INTERCEPT])],
+          [AS_IF([test -n "${with_$1}"],
+                 [PAC_APPEND_FLAG([-I${with_$1}/include],[CPPFLAGS_SYSCALL_INTERCEPT])])])
+
+    AS_IF([test -n "${with_$1_lib}"],
+          [PAC_APPEND_FLAG([-L${with_$1_lib} -Wl,-rpath,${with_$1_lib}],[LDFLAGS_SYSCALL_INTERCEPT])],
+          [AS_IF([test -n "${with_$1}"],
+                 dnl is adding lib64 by default really the right thing to do?  What if
+                 dnl we are on a 32-bit host that happens to have both lib dirs available?
+                 [PAC_APPEND_FLAG([-L${with_$1}/lib -Wl,-rpath,${with_$1}/lib],[LDFLAGS_SYSCALL_INTERCEPT])
+                  AS_IF([test -d "${with_$1}/lib64"],
+		        [PAC_APPEND_FLAG([-L${with_$1}/lib64 -Wl,-rpath,${with_$1}/lib64],[LDFLAGS_SYSCALL_INTERCEPT])])
+                 ])
+          ])
+
+    AS_IF([test -n "${with_$1}" || test -n "${with_$1_include}" || test -n "${with_$1_lib}"],
+	      [WITH_SYSCALL_INTERCEPT=yes],
+	      [WITH_SYSCALL_INTERCEPT=no])
+])
+
 IHK_VERSION=IHK_VERSION_m4
 MCKERNEL_VERSION=MCKERNEL_VERSION_m4
 DCFA_VERSION=DCFA_VERSION_m4
@@ -97,6 +147,23 @@ AS_IF([test "x$numa_lib_found" != "xyes"],
 
 PAC_SET_HEADER_LIB_PATH([mpi])
 
+PAC_SET_HEADER_LIB_PATH_SYSCALL_INTERCEPT([syscall_intercept])
+
+if test "x$WITH_SYSCALL_INTERCEPT" == "xno" ; then
+	AC_CHECK_LIB([syscall_intercept],[syscall_no_intercept],[syscall_intercept_lib_found=yes],[syscall_intercept_lib_found=no],[-lcapstone -ldl])
+	AS_IF([test "x$syscall_intercept_lib_found" != "xyes"],
+		[AC_MSG_NOTICE([libsyscall_intercept.so not found])])
+
+	AC_CHECK_HEADER([libsyscall_intercept_hook_point.h],[syscall_intercept_header_found=yes],[syscall_intercept_header_found=no])
+	AS_IF([test "x$syscall_intercept_header_found" != "xyes"],
+		[AC_MSG_NOTICE([libsyscall_intercept_hook_point.h not found])])
+
+	AS_IF([test "x$syscall_intercept_lib_found" == "xyes" && test "x$syscall_intercept_header_found" == "xyes"],
+		[WITH_SYSCALL_INTERCEPT=yes],
+		[WITH_SYSCALL_INTERCEPT=no])
+fi
+
+
 AC_ARG_WITH([kernelsrc],
   AC_HELP_STRING(
     [--with-kernelsrc=path],[Path to 'kernel src', default is /lib/modules/uname_r/build]),
@@ -134,6 +201,12 @@ AC_ARG_ENABLE([rusage],
 	      [ENABLE_RUSAGE=$enableval],
 	      [ENABLE_RUSAGE=yes])
 
+AC_ARG_ENABLE([perf],
+	      AC_HELP_STRING([--enable-perf],
+			     [enable perf_event implementation]),
+	      [ENABLE_PERF=$enableval],
+	      [ENABLE_PERF=yes])
+
 AC_ARG_ENABLE([qlmpi],
 	      AC_HELP_STRING([--enable-qlmpi],
 			     [enable qlmpi implementation]),
@@ -225,7 +298,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     builtin-mic)
@@ -242,7 +315,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/attached/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/attached/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     builtin-x86)
@@ -259,7 +332,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/attached/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     smp-x86)
@@ -291,7 +364,7 @@ case $WITH_TARGET in
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/smp-x86/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     smp-arm64)
@@ -316,11 +389,14 @@ case $WITH_TARGET in
 	if test "X$ETCDIR" = X; then
 		ETCDIR="$prefix/etc"
 	fi
+	if test "X$INCLUDEDIR" = X; then
+		INCLUDEDIR="$prefix/include"
+	fi
 	if test "X$KMODDIR" = X; then
 		KMODDIR="$prefix/kmod"
 	fi
 	if test "X$MANDIR" = X; then
-		MANDIR="$prefix/smp-arm64/man"
+		MANDIR="$prefix/share/man"
 	fi
 	;;
     *)
@@ -332,78 +408,6 @@ KDIR="$WITH_KERNELSRC"
 UNAME_R="$WITH_UNAME_R"
 TARGET="$WITH_TARGET"
 
-MCCTRL_LINUX_SYMTAB=""
-case "X$WITH_SYSTEM_MAP" in
-    Xyes | Xno | X)
-	MCCTRL_LINUX_SYMTAB=""
-	;;
-    *)
-	MCCTRL_LINUX_SYMTAB="$WITH_SYSTEM_MAP"
-	;;
-esac
-
-AC_MSG_CHECKING([[for System.map]])
-if test -r "$MCCTRL_LINUX_SYMTAB"; then
-	MCCTRL_LINUX_SYMTAB="$MCCTRL_LINUX_SYMTAB"
-elif test -r "/boot/System.map-`uname -r`"; then
-	MCCTRL_LINUX_SYMTAB="/boot/System.map-`uname -r`"
-elif test -r "$KDIR/System.map"; then
-	MCCTRL_LINUX_SYMTAB="$KDIR/System.map"
-fi
-
-if test "$MCCTRL_LINUX_SYMTAB" == ""; then
-	AC_MSG_ERROR([could not find])
-fi
-
-if test -z "`eval cat $MCCTRL_LINUX_SYMTAB`"; then
-	AC_MSG_ERROR([could not read System.map file, no read permission?])
-fi
-AC_MSG_RESULT([$MCCTRL_LINUX_SYMTAB])
-
-MCCTRL_LINUX_SYMTAB_CMD="cat $MCCTRL_LINUX_SYMTAB"
-
-# MCCTRL_FIND_KSYM(SYMBOL)
-# ------------------------------------------------------
-# Search System.map for address of the given symbol and
-# do one of three things in config.h:
-# If not found, leave MCCTRL_KSYM_foo undefined
-# If found to be exported, "#define MCCTRL_KSYM_foo 0"
-# If found not to be exported, "#define MCCTRL_KSYM_foo 0x<value>"
-AC_DEFUN([MCCTRL_FIND_KSYM],[
-  AC_MSG_CHECKING([[System.map for symbol $1]])
-  mcctrl_addr=`eval $MCCTRL_LINUX_SYMTAB_CMD | grep " $1\$" | cut -d\  -f1`
-  if test -z $mcctrl_addr; then
-    AC_MSG_RESULT([not found])
-  else
-    mcctrl_result=$mcctrl_addr
-    mcctrl_addr="0x$mcctrl_addr"
-    m4_ifval([$2],[],[
-      if `eval $MCCTRL_LINUX_SYMTAB_CMD | grep " __ksymtab_$1\$" >/dev/null`; then
-        mcctrl_result="exported"
-        mcctrl_addr="0"
-      fi
-    ])
-    AC_MSG_RESULT([$mcctrl_result])
-    AC_DEFINE_UNQUOTED(MCCTRL_KSYM_[]$1,$mcctrl_addr,[Define to address of kernel symbol $1, or 0 if exported])
-  fi
-])
-
-MCCTRL_FIND_KSYM([sys_mount])
-MCCTRL_FIND_KSYM([sys_umount])
-MCCTRL_FIND_KSYM([sys_unshare])
-MCCTRL_FIND_KSYM([zap_page_range])
-MCCTRL_FIND_KSYM([vdso_image_64])
-MCCTRL_FIND_KSYM([vdso_start])
-MCCTRL_FIND_KSYM([vdso_end])
-MCCTRL_FIND_KSYM([vdso_pages])
-MCCTRL_FIND_KSYM([__vvar_page])
-MCCTRL_FIND_KSYM([hpet_address])
-# POSTK_DEBUG_ARCH_DEP_50, add:find kernel symbol.
-MCCTRL_FIND_KSYM([vdso_spec])
-MCCTRL_FIND_KSYM([hv_clock])
-MCCTRL_FIND_KSYM([sys_readlink])
-MCCTRL_FIND_KSYM([walk_page_range])
-
 case $ENABLE_MEMDUMP in
 	yes|no|auto)
 		;;
@@ -451,7 +455,7 @@ else
 fi
 
 if test "x$ENABLE_QLMPI" = "xyes" ; then
-	AC_DEFINE([ENABLE_QLMPI],[1],[whether mcoverlayfs is enabled])
+	AC_DEFINE([ENABLE_QLMPI],[1],[whether qlmpi is enabled])
 	AC_MSG_NOTICE([qlmpi is enabled])
 else
 	AC_MSG_NOTICE([qlmpi is disabled])
@@ -475,6 +479,20 @@ else
 	AC_MSG_NOTICE([rusage is disabled])
 fi
 
+if test "x$ENABLE_PERF" = "xyes" ; then
+	AC_DEFINE([ENABLE_PERF],[1],[whether perf is enabled])
+	AC_MSG_NOTICE([perf is enabled])
+else
+	AC_MSG_NOTICE([perf is disabled])
+fi
+
+if test "x$WITH_SYSCALL_INTERCEPT" = "xyes" ; then
+	AC_DEFINE([WITH_SYSCALL_INTERCEPT],[1],[whether or not syscall_intercept library is linked])
+	AC_MSG_NOTICE([syscall_intercept library is linked])
+else
+	AC_MSG_NOTICE([syscall_intercept library isn't linked])
+fi
+
 if test "x$MCKERNEL_INCDIR" != "x" ; then
    AC_DEFINE_UNQUOTED(MCKERNEL_INCDIR,"$MCKERNEL_INCDIR",[McKernel specific headers])
 fi
@@ -486,6 +504,15 @@ fi
 AC_DEFINE_UNQUOTED(BINDIR,"$BINDIR",[Path of install directory for binary])
 AC_DEFINE_UNQUOTED(SBINDIR,"$SBINDIR",[Path of install directory for system binary])
 
+ABS_SRCDIR=$( cd $( dirname $0 ); pwd )
+IHK_ABS_SRCDIR=${ABS_SRCDIR}/../ihk
+BUILDID=$( cd $IHK_ABS_SRCDIR; if @<:@ ! -d .git @:>@; then echo $IHK_VERSION; else bash -c 'git rev-list -1 HEAD | cut -c1-8'; fi )
+AC_MSG_NOTICE([BUILDID=$BUILDID])
+if test "x$BUILDID" != "x" ; then
+   AC_DEFINE_UNQUOTED(BUILDID,"$BUILDID",[IHK build-id to confirm IHK and McKernel built at the same time are used])
+fi
+AC_SUBST(BUILDID)
+
 AC_SUBST(CC)
 AC_SUBST(XCC)
 AC_SUBST(ARCH)
@@ -503,9 +530,12 @@ AC_SUBST(KMODDIR)
 AC_SUBST(KERNDIR)
 AC_SUBST(MANDIR)
 AC_SUBST(CFLAGS)
+AC_SUBST(CPPFLAGS_SYSCALL_INTERCEPT)
+AC_SUBST(LDFLAGS_SYSCALL_INTERCEPT)
 AC_SUBST(ENABLE_MCOVERLAYFS)
 AC_SUBST(ENABLE_RUSAGE)
 AC_SUBST(ENABLE_QLMPI)
+AC_SUBST(WITH_SYSCALL_INTERCEPT)
 
 AC_SUBST(IHK_VERSION)
 AC_SUBST(MCKERNEL_VERSION)
@@ -535,20 +565,34 @@ AC_CONFIG_FILES([
 	kernel/Makefile
 	kernel/Makefile.build
 	kernel/include/swapfmt.h
-	arch/x86/tools/mcreboot-attached-mic.sh
+	arch/x86_64/tools/mcreboot-attached-mic.sh
-	arch/x86/tools/mcshutdown-attached-mic.sh
+	arch/x86_64/tools/mcshutdown-attached-mic.sh
-	arch/x86/tools/mcreboot-builtin-x86.sh
+	arch/x86_64/tools/mcreboot-builtin-x86.sh
-	arch/x86/tools/mcreboot-smp-x86.sh
+	arch/x86_64/tools/mcreboot-smp-x86.sh
-	arch/x86/tools/mcstop+release-smp-x86.sh
+	arch/x86_64/tools/mcstop+release-smp-x86.sh
-	arch/x86/tools/mcoverlay-destroy-smp-x86.sh
+	arch/x86_64/tools/mcoverlay-destroy-smp-x86.sh
-	arch/x86/tools/mcoverlay-create-smp-x86.sh
+	arch/x86_64/tools/mcoverlay-create-smp-x86.sh
-	arch/x86/tools/eclair-dump-backtrace.exp
+	arch/x86_64/tools/eclair-dump-backtrace.exp
-	arch/x86/tools/mcshutdown-builtin-x86.sh
+	arch/x86_64/tools/mcshutdown-builtin-x86.sh
-	arch/x86/tools/mcreboot.1:arch/x86/tools/mcreboot.1in
+	arch/x86_64/tools/mcreboot.1:arch/x86_64/tools/mcreboot.1in
-	arch/x86/tools/irqbalance_mck.service
+	arch/x86_64/tools/irqbalance_mck.service
-	arch/x86/tools/irqbalance_mck.in
+	arch/x86_64/tools/irqbalance_mck.in
+	tools/mcstat/mcstat.1:tools/mcstat/mcstat.1in
+	tools/mcstat/Makefile
 ])
 
+if test -e "${ABS_SRCDIR}/test"; then
+AC_CONFIG_FILES([
+mck_test_config.sample:test/mck_test_config.sample.in
+])
+fi
+
+if test "$TARGET" = "smp-x86"; then
+AC_CONFIG_FILES([
+	arch/x86_64/kernel/Makefile.arch
+])
+fi
+
 if test "$TARGET" = "smp-arm64"; then
 AC_CONFIG_FILES([
 	kernel/config/config.smp-arm64

executer/include/ihklib_rusage.h

@@ -5,6 +5,10 @@
 #define IHK_MAX_NUM_NUMA_NODES 1024
 #define IHK_MAX_NUM_CPUS 1024
 
+#define IHK_OS_PGSIZE_4KB 0
+#define IHK_OS_PGSIZE_2MB 1
+#define IHK_OS_PGSIZE_1GB 2
+
 struct mckernel_rusage {
 	unsigned long memory_stat_rss[IHK_MAX_NUM_PGSIZES];
 	unsigned long memory_stat_mapped_file[IHK_MAX_NUM_PGSIZES];

executer/include/uprotocol.h

@@ -55,13 +55,14 @@
 #define MCEXEC_UP_SYS_UMOUNT     0x30a02915
 #define MCEXEC_UP_SYS_UNSHARE    0x30a02916
 
-#define MCEXEC_UP_UTIL_THREAD1   0x30a02920
+#define MCEXEC_UP_UTI_GET_CTX    0x30a02920
-#define MCEXEC_UP_UTIL_THREAD2   0x30a02921
+#define MCEXEC_UP_UTI_SAVE_FS    0x30a02921
 #define MCEXEC_UP_SIG_THREAD     0x30a02922
 #define MCEXEC_UP_SYSCALL_THREAD 0x30a02924
 #define MCEXEC_UP_TERMINATE_THREAD 0x30a02925
 #define MCEXEC_UP_GET_NUM_POOL_THREADS  0x30a02926
 #define MCEXEC_UP_UTI_ATTR       0x30a02927
+#define MCEXEC_UP_RELEASE_USER_SPACE 0x30a02928
 
 #define MCEXEC_UP_DEBUG_LOG     0x40000000
 
@@ -91,6 +92,7 @@ struct program_image_section {
 
 struct get_cpu_set_arg {
 	int nr_processes;
+	int *process_rank;
 	void *cpu_set;
 	size_t cpu_set_size;	// Size in bytes
 	int *target_core;
@@ -111,10 +113,8 @@ typedef unsigned long __cpu_set_unit;
 
 struct program_load_desc {
 	int num_sections;
-	int status;
 	int cpu;
 	int pid;
-	int err;
 	int stack_prot;
 	int pgid;
 	int cred[8];
@@ -140,8 +140,12 @@ struct program_load_desc {
 	unsigned long mpol_flags;
 	unsigned long mpol_threshold;
 	unsigned long heap_extension;
+	long stack_premap;
+	unsigned long mpol_bind_mask;
+	int uti_thread_rank; /* N-th clone() spawns a thread on Linux CPU */
+	int uti_use_last_cpu; /* Work-around not to share CPU with OpenMP thread */
 	int nr_processes;
-	char shell_path[SHELL_PATH_MAX_LEN];
+	int process_rank;
 	__cpu_set_unit cpu_set[PLD_CPU_SET_SIZE];
 	int profile;
 	struct program_image_section sections[0];
@@ -242,6 +246,28 @@ struct sys_unshare_desc {
 	unsigned long unshare_flags;
 };
 
+struct release_user_space_desc {
+	unsigned long user_start;
+	unsigned long user_end;
+};
+
+struct terminate_thread_desc {
+	int pid;
+	int tid;
+
+	long code; 
+	/* 32------32  31--16  15--------8  7----0
+       exit_group          exit-status  signal */
+
+	unsigned long tsk; /* struct task_struct * */
+};
+
+struct rpgtable_desc {
+	uintptr_t rpgtable;
+	uintptr_t start;
+	uintptr_t len;
+};
+
 enum perf_ctrl_type {
 	PERF_CTRL_SET,
 	PERF_CTRL_GET,
@@ -251,7 +277,7 @@ enum perf_ctrl_type {
 
 struct perf_ctrl_desc {
 	enum perf_ctrl_type ctrl_type;
-	int status;
+	int err;
 	union {
 		/* for SET, GET */
 		struct {
@@ -291,6 +317,10 @@ struct perf_ctrl_desc {
 #define UTI_FLAG_HIGH_PRIORITY (1ULL<<12)
 #define UTI_FLAG_NON_COOPERATIVE (1ULL<<13)
 
+#define UTI_FLAG_PREFER_LWK (1ULL << 14)
+#define UTI_FLAG_PREFER_FWK (1ULL << 15)
+#define UTI_FLAG_FABRIC_INTR_AFFINITY (1ULL << 16)
+
 /* Linux default value is used */
 #define UTI_MAX_NUMA_DOMAINS (1024)
 
@@ -309,6 +339,30 @@ struct kuti_attr {
 
 struct uti_attr_desc {
 	unsigned long phys_attr;
+	char *uti_cpu_set_str; /* UTI_CPU_SET environmental variable */
+	size_t uti_cpu_set_len;
+};
+
+struct uti_ctx {
+	union {
+		char ctx[4096]; /* TODO: Get the size from config.h */
+		struct {
+			int uti_refill_tid;
+		};
+	};
+}; 
+
+struct uti_get_ctx_desc {
+	unsigned long rp_rctx; /* Remote physical address of remote context */
+	void *rctx; /* Remote context */
+	void *lctx; /* Local context */
+	int uti_refill_tid;
+	unsigned long key; /* OUT: struct task_struct* of mcexec thread, used to search struct host_thread */
+};
+
+struct uti_save_fs_desc {
+	void *rctx; /* Remote context */
+	void *lctx; /* Local context */
 };
 
 #endif

executer/include/uti.h

@@ -0,0 +1,31 @@
+#ifndef UTI_H_INCLUDED
+#define UTI_H_INCLUDED
+
+struct syscall_struct {
+	int number;
+	unsigned long args[6];
+	unsigned long ret;
+	unsigned long uti_clv; /* copy of a clv in McKernel */
+};
+
+#define UTI_SZ_SYSCALL_STACK 16
+
+/* Variables accessed by mcexec.c and syscall_intercept.c */
+struct uti_desc {
+	char lctx[4096]; /* TODO: Get the size from config.h */
+	char rctx[4096]; /* TODO: Get the size from config.h */
+	int mck_tid; /* TODO: Move this out for multiple migrated-to-Linux threads */
+	unsigned long key; /* struct task_struct* of mcexec thread, used to search struct host_thread */
+	int pid, tid; /* Used as the id of tracee when issuing MCEXEC_UP_TERMINATE_THREAD */
+	unsigned long uti_clv; /* copy of McKernel clv */
+
+	int fd; /* /dev/mcosX */
+	struct syscall_struct syscall_stack[UTI_SZ_SYSCALL_STACK]; /* stack of system call arguments and return values */
+	int syscall_stack_top; /* stack-pointer of syscall arguments list */
+	long syscalls[512], syscalls2[512]; /* Syscall profile counters */
+	int start_syscall_intercept; /* Used to sync between mcexec.c and syscall_intercept.c */
+};
+
+
+#endif
+

executer/kernel/mcctrl/Makefile.in

@@ -9,13 +9,22 @@ IHK_BASE=$(src)/../../../../ihk
 
 obj-m += mcctrl.o
 
-# POSTK_DEBUG_ARCH_DEP_1, arch depend "-mcmodel"
 # POSTK_DEBUG_ARCH_DEP_83, arch depend translate_rva_to_rpa() move
-ifeq ($(ARCH), arm64)
+ccflags-y := -I$(IHK_BASE)/linux/include \
-ccflags-y := -I$(IHK_BASE)/linux/include -I$(IHK_BASE)/linux/include/ihk/arch/$(ARCH) -I$(IHK_BASE)/ikc/include -I$(IHK_BASE)/ikc/include/ikc/arch/$(ARCH) -I$(IHK_BASE)/include -I$(IHK_BASE)/include/arch/$(ARCH) -I$(src)/../../include -I$(src)/arch/$(ARCH)/include -DMCEXEC_PATH=\"$(BINDIR)/mcexec\" -I@abs_builddir@
+             -I$(IHK_BASE)/linux/include/ihk/arch/$(ARCH) \
-else
+             -I$(IHK_BASE)/ikc/include \
-ccflags-y := -I$(IHK_BASE)/linux/include -I$(IHK_BASE)/linux/include/ihk/arch/$(ARCH) -I$(IHK_BASE)/ikc/include -I$(IHK_BASE)/ikc/include/ikc/arch/$(ARCH) -I$(IHK_BASE)/include -I$(IHK_BASE)/include/arch/$(ARCH) -I$(src)/../../../kernel/include -I$(src)/../../include -mcmodel=kernel -mno-red-zone -DMCEXEC_PATH=\"$(BINDIR)/mcexec\" -I@abs_builddir@ -I@abs_builddir@/../../../
+             -I$(IHK_BASE)/ikc/include/ikc/arch/$(ARCH) \
-endif
+             -I$(IHK_BASE)/include \
+             -I$(IHK_BASE)/include/arch/$(ARCH) \
+             -I$(src)/../../include \
+             -I$(src)/arch/$(ARCH)/include \
+             -I@abs_builddir@ \
+             -I@abs_builddir@/../../../ \
+             -I$(src)/../../../kernel/include \
+             -DMCEXEC_PATH=\"$(BINDIR)/mcexec\"
+
+# depending arch
+include @abs_builddir@/arch/$(ARCH)/Makefile
 
 mcctrl-y := driver.o control.o ikc.o syscall.o procfs.o binfmt_mcexec.o
 mcctrl-y += sysfs.o sysfs_files.o arch/$(ARCH)/archdeps.o