prerelease: 0.94: fix __mcctrl_os_read_write_cpu_register

Change-Id: Ibcfbe7796347cc9c2148cdea2519fe6c7ca9e97e

CMakeLists.txt

@@ -10,7 +10,7 @@ project(mckernel C ASM)
 set(MCKERNEL_VERSION "1.7.1")
 
 # See "Fedora Packaging Guidelines -- Versioning"
-set(MCKERNEL_RELEASE "0.91")
+set(MCKERNEL_RELEASE "0.94")
 
 set(CMAKE_MODULE_PATH ${CMAKE_SOURCE_DIR}/cmake/modules)
 # for rpmbuild
@@ -65,7 +65,7 @@ if(ENABLE_TOFU)
 endif()
 
 # when compiling on a compute-node
-execute_process(COMMAND bash -c "grep $(hostname) /etc/opt/FJSVfefs/config/fefs_node1.csv 2>/dev/null | cut -d, -f2"
+execute_process(COMMAND bash -c "grep $(hostname) /etc/opt/FJSVfefs/config/fefs_node1.csv 2>/dev/null | cut -d, -f2 | grep -o CN"
   OUTPUT_VARIABLE FUGAKU_NODE_TYPE OUTPUT_STRIP_TRAILING_WHITESPACE)
 if(FUGAKU_NODE_TYPE STREQUAL "CN")
 	option(ENABLE_FUGAKU_HACKS "Fugaku hacks" ON)

arch/arm64/kernel/cpu.c

@@ -1972,15 +1972,15 @@ int arch_cpu_read_write_register(
 	return ret;
 }
 
-int smp_call_func(cpu_set_t *__cpu_set, smp_func_t __func, void *__arg)
-{
-	/* TODO: skeleton for smp_call_func */
-	return -1;
-}
-
 void arch_flush_icache_all(void)
 {
 	asm("ic	ialluis");
 	dsb(ish);
 }
+
+int ihk_mc_get_smp_handler_irq(void)
+{
+	return LOCAL_SMP_FUNC_CALL_VECTOR;
+}
+
 /*** end of file ***/

arch/arm64/kernel/include/irq.h

@@ -17,6 +17,7 @@
 #define INTRID_STACK_TRACE	5
 #define INTRID_MULTI_INTR	6
 #define INTRID_MULTI_NMI	7
+#define LOCAL_SMP_FUNC_CALL_VECTOR   1 /* same as IKC */
 
 /* use PPI interrupt number */
 #define INTRID_PERF_OVF		23

arch/arm64/kernel/syscall.c

@@ -16,6 +16,7 @@
 #include <uio.h>
 #include <syscall.h>
 #include <rusage_private.h>
+#include <memory.h>
 #include <ihk/debug.h>
 
 void terminate_mcexec(int, int);
@@ -2250,8 +2251,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 		case 0:
 			memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
 			       sizeof(void *) * count);
-			memcpy(mpsr->nodes, mpsr->user_nodes,
-			       sizeof(int) * count);
+			if (mpsr->user_nodes) {
+				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);
@@ -2269,8 +2272,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 		case 0:
 			memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
 			       sizeof(void *) * count);
-			memcpy(mpsr->nodes, mpsr->user_nodes,
-			       sizeof(int) * count);
+			if (mpsr->user_nodes) {
+				memcpy(mpsr->nodes, mpsr->user_nodes,
+						sizeof(int) * count);
+			}
 			mpsr->nodes_ready = 1;
 			break;
 		case 1:
@@ -2292,8 +2297,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 			       sizeof(void *) * count);
 			break;
 		case 1:
-			memcpy(mpsr->nodes, mpsr->user_nodes,
-			       sizeof(int) * count);
+			if (mpsr->user_nodes) {
+				memcpy(mpsr->nodes, mpsr->user_nodes,
+						sizeof(int) * count);
+			}
 			mpsr->nodes_ready = 1;
 			break;
 		case 2:
@@ -2322,8 +2329,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 			       sizeof(void *) * (count / 2));
 			break;
 		case 2:
-			memcpy(mpsr->nodes, mpsr->user_nodes,
-			       sizeof(int) * count);
+			if (mpsr->user_nodes) {
+				memcpy(mpsr->nodes, mpsr->user_nodes,
+						sizeof(int) * count);
+			}
 			mpsr->nodes_ready = 1;
 			break;
 		case 3:
@@ -2349,13 +2358,15 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 	}
 
 	/* 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;
+	if (mpsr->user_nodes) {
+		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;
+			}
 		}
 	}
 
@@ -2387,7 +2398,7 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 
 		/* PTE valid? */
 		if (!mpsr->ptep[i] || !pte_is_present(mpsr->ptep[i])) {
-			mpsr->status[i] = -ENOENT;
+			mpsr->status[i] = -EFAULT;
 			mpsr->ptep[i] = NULL;
 			continue;
 		}
@@ -2451,6 +2462,26 @@ pte_out:
 	dkprintf("%s: phase %d done\n", __FUNCTION__, phase);
 	++phase;
 
+	/*
+	 * When nodes array is NULL, move_pages doesn't move any pages,
+	 * instead will return the node where each page
+	 * currently resides by status array.
+	 */
+	if (!mpsr->user_nodes) {
+		/* get nid in parallel */
+		for (i = i_s; i < i_e; i++) {
+			if (mpsr->status[i] < 0) {
+				continue;
+			}
+			mpsr->status[i] = phys_to_nid(
+					pte_get_phys(mpsr->ptep[i]));
+		}
+		mpsr->phase_ret = 0;
+		goto out;  // return node information
+	}
+
+	/* Processing of move pages */
+
 	if (cpu_index == 0) {
 		/* Allocate new pages on target NUMA nodes */
 		for (i = 0; i < count; i++) {
@@ -2463,8 +2494,11 @@ pte_out:
 
 			/* 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;
+				int nr_pages;
+
+				for (pgalign = 0, nr_pages = mpsr->nr_pages[i];
+				     nr_pages != 1; pgalign++, nr_pages >>= 1) {
+				}
 			}
 
 			dst = ihk_mc_alloc_aligned_pages_node(mpsr->nr_pages[i],

arch/x86_64/kernel/cpu.c

@@ -80,7 +80,11 @@ static void (*lapic_icr_write)(unsigned int h, unsigned int l);
 static void (*lapic_wait_icr_idle)(void);
 void (*x86_issue_ipi)(unsigned int apicid, unsigned int low);
 int running_on_kvm(void);
-static void smp_func_call_handler(void);
+void smp_func_call_handler(void);
+int ihk_mc_get_smp_handler_irq(void)
+{
+	return LOCAL_SMP_FUNC_CALL_VECTOR;
+}
 
 void init_processors_local(int max_id);
 void assign_processor_id(void);
@@ -2170,144 +2174,6 @@ int arch_cpu_read_write_register(
 	return 0;
 }
 
-/*
- * Generic remote CPU function invocation facility.
- */
-static void smp_func_call_handler(void)
-{
-	int irq_flags;
-	struct smp_func_call_request *req;
-	int reqs_left;
-
-reiterate:
-	req = NULL;
-	reqs_left = 0;
-
-	irq_flags = ihk_mc_spinlock_lock(
-			&cpu_local_var(smp_func_req_lock));
-
-	/* Take requests one-by-one */
-	if (!list_empty(&cpu_local_var(smp_func_req_list))) {
-		req = list_first_entry(&cpu_local_var(smp_func_req_list),
-			struct smp_func_call_request, list);
-		list_del(&req->list);
-
-		reqs_left = !list_empty(&cpu_local_var(smp_func_req_list));
-	}
-
-	ihk_mc_spinlock_unlock(&cpu_local_var(smp_func_req_lock),
-			irq_flags);
-
-	if (req) {
-		req->ret = req->sfcd->func(req->cpu_index,
-				req->sfcd->nr_cpus, req->sfcd->arg);
-		ihk_atomic_dec(&req->sfcd->cpus_left);
-	}
-
-	if (reqs_left)
-		goto reiterate;
-}
-
-int smp_call_func(cpu_set_t *__cpu_set, smp_func_t __func, void *__arg)
-{
-	int cpu, nr_cpus = 0;
-	int cpu_index = 0;
-	int this_cpu_index = 0;
-	struct smp_func_call_data sfcd;
-	struct smp_func_call_request *reqs;
-	int ret = 0;
-	int call_on_this_cpu = 0;
-	cpu_set_t cpu_set;
-
-	/* Sanity checks */
-	if (!__cpu_set || !__func) {
-		return -EINVAL;
-	}
-
-	/* Make sure it won't change in between */
-	cpu_set = *__cpu_set;
-
-	for_each_set_bit(cpu, (unsigned long *)&cpu_set,
-			sizeof(cpu_set) * BITS_PER_BYTE) {
-
-		if (cpu == ihk_mc_get_processor_id()) {
-			call_on_this_cpu = 1;
-		}
-		++nr_cpus;
-	}
-
-	if (!nr_cpus) {
-		return -EINVAL;
-	}
-
-	reqs = kmalloc(sizeof(*reqs) * nr_cpus, IHK_MC_AP_NOWAIT);
-	if (!reqs) {
-		ret = -ENOMEM;
-		goto free_out;
-	}
-
-	sfcd.nr_cpus = nr_cpus;
-	sfcd.func = __func;
-	sfcd.arg = __arg;
-	ihk_atomic_set(&sfcd.cpus_left,
-			call_on_this_cpu ? nr_cpus - 1 : nr_cpus);
-
-	/* Add requests and send IPIs */
-	cpu_index = 0;
-	for_each_set_bit(cpu, (unsigned long *)&cpu_set,
-			sizeof(cpu_set) * BITS_PER_BYTE) {
-		unsigned long irq_flags;
-
-		reqs[cpu_index].cpu_index = cpu_index;
-		reqs[cpu_index].ret = 0;
-
-		if (cpu == ihk_mc_get_processor_id()) {
-			this_cpu_index = cpu_index;
-			++cpu_index;
-			continue;
-		}
-
-		reqs[cpu_index].sfcd = &sfcd;
-
-		irq_flags =
-			ihk_mc_spinlock_lock(&get_cpu_local_var(cpu)->smp_func_req_lock);
-		list_add_tail(&reqs[cpu_index].list,
-				&get_cpu_local_var(cpu)->smp_func_req_list);
-		ihk_mc_spinlock_unlock(&get_cpu_local_var(cpu)->smp_func_req_lock,
-			irq_flags);
-
-		ihk_mc_interrupt_cpu(cpu, LOCAL_SMP_FUNC_CALL_VECTOR);
-
-		++cpu_index;
-	}
-
-	/* Is this CPU involved? */
-	if (call_on_this_cpu) {
-		reqs[this_cpu_index].ret =
-			__func(this_cpu_index, nr_cpus, __arg);
-	}
-
-	/* Wait for the rest of the CPUs */
-	while (ihk_atomic_read(&sfcd.cpus_left) > 0) {
-		cpu_pause();
-	}
-
-	/* Check return values, if error, report the first non-zero */
-	for (cpu_index = 0; cpu_index < nr_cpus; ++cpu_index) {
-		if (reqs[cpu_index].ret != 0) {
-			ret = reqs[cpu_index].ret;
-			goto free_out;
-		}
-	}
-
-	ret = 0;
-
-free_out:
-	kfree(reqs);
-
-	return ret;
-}
-
 extern int nmi_mode;
 extern long freeze_thaw(void *nmi_ctx);
 

arch/x86_64/kernel/syscall.c

@@ -32,6 +32,7 @@
 #include <limits.h>
 #include <syscall.h>
 #include <rusage_private.h>
+#include <memory.h>
 #include <ihk/debug.h>
 
 void terminate_mcexec(int, int);
@@ -2302,8 +2303,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 			case 0:
 				memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
 						sizeof(void *) * count);
-				memcpy(mpsr->nodes, mpsr->user_nodes,
-						sizeof(int) * count);
+				if (mpsr->user_nodes) {
+					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);
@@ -2321,8 +2324,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 			case 0:
 				memcpy(mpsr->virt_addr, mpsr->user_virt_addr,
 						sizeof(void *) * count);
-				memcpy(mpsr->nodes, mpsr->user_nodes,
-						sizeof(int) * count);
+				if (mpsr->user_nodes) {
+					memcpy(mpsr->nodes, mpsr->user_nodes,
+							sizeof(int) * count);
+				}
 				mpsr->nodes_ready = 1;
 				break;
 			case 1:
@@ -2344,8 +2349,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 						sizeof(void *) * count);
 				break;
 			case 1:
-				memcpy(mpsr->nodes, mpsr->user_nodes,
-						sizeof(int) * count);
+				if (mpsr->user_nodes) {
+					memcpy(mpsr->nodes, mpsr->user_nodes,
+							sizeof(int) * count);
+				}
 				mpsr->nodes_ready = 1;
 				break;
 			case 2:
@@ -2374,8 +2381,10 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 						sizeof(void *) * (count / 2));
 				break;
 			case 2:
-				memcpy(mpsr->nodes, mpsr->user_nodes,
-						sizeof(int) * count);
+				if (mpsr->user_nodes) {
+					memcpy(mpsr->nodes, mpsr->user_nodes,
+							sizeof(int) * count);
+				}
 				mpsr->nodes_ready = 1;
 				break;
 			case 3:
@@ -2401,13 +2410,15 @@ int move_pages_smp_handler(int cpu_index, int nr_cpus, void *arg)
 	}
 
 	/* 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;
+	if (mpsr->user_nodes) {
+		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;
+			}
 		}
 	}
 
@@ -2503,6 +2514,26 @@ pte_out:
 	dkprintf("%s: phase %d done\n", __FUNCTION__, phase);
 	++phase;
 
+	/*
+	 * When nodes array is NULL, move_pages doesn't move any pages,
+	 * instead will return the node where each page
+	 * currently resides by status array.
+	 */
+	if (!mpsr->user_nodes) {
+		/* get nid in parallel */
+		for (i = i_s; i < i_e; i++) {
+			if (mpsr->status[i] < 0) {
+				continue;
+			}
+			mpsr->status[i] = phys_to_nid(
+					pte_get_phys(mpsr->ptep[i]));
+		}
+		mpsr->phase_ret = 0;
+		goto out;  // return node information
+	}
+
+	/* Processing of move pages */
+
 	if (cpu_index == 0) {
 		/* Allocate new pages on target NUMA nodes */
 		for (i = 0; i < count; i++) {

docs/install.rst

@@ -129,11 +129,29 @@ Create the tarball and the spec file:
    make dist
    cp mckernel-<version>.tar.gz <rpmbuild>/SOURCES
 
+(optional) Edit the following line in ``scripts/mckernel.spec`` to change
+cmake options. For example:
+
+::
+
+   %cmake -DCMAKE_BUILD_TYPE=Release \
+	-DUNAME_R=%{kernel_version} \
+	-DKERNEL_DIR=%{kernel_dir} \
+	%{?cmake_libdir:-DCMAKE_INSTALL_LIBDIR=%{cmake_libdir}} \
+	%{?build_target:-DBUILD_TARGET=%{build_target}} \
+	%{?toolchain_file:-DCMAKE_TOOLCHAIN_FILE=%{toolchain_file}} \
+	-DENABLE_TOFU=ON -DENABLE_FUGAKU_HACKS=ON \
+	-DENABLE_KRM_WORKAROUND=OFF -DWITH_KRM=ON \
+	-DENABLE_FUGAKU_DEBUG=OFF \
+	.
+
 Create the rpm package:
 
 When not cross-compiling:
 """""""""""""""""""""""""
 
+Then build the rpm:
+
 ::
 
    rpmbuild -ba scripts/mckernel.spec

docs/users.rst

@@ -202,3 +202,21 @@ Limitations
 28. munlockall() is not supported and returns zero.
 
 29. scheduling behavior is not Linux compatible. For example, sometimes one of the two processes on the same CPU continues to run after yielding.
+
+30. (Fujitsu TCS-only) A job following the one in which __mcctrl_os_read_write_cpu_register() returns ``-ETIME`` fails because xos_hwb related CPU state isn't finalized. You can tell if the function returned ``-ETIME`` by checking if the following line appeared in the Linux kernel message:
+
+    ::
+
+       __mcctrl_os_read_write_cpu_register: ERROR sending IKC msg: -62
+
+    You can re-initialize xos_hwb related CPU state by the following command:
+
+    ::
+
+       sudo systemctl restart xos_hwb
+
+31. System calls can write the mcexec VMAs with PROT_WRITE flag not
+    set. This is because we never turn off PROT_WRITE of the mcexec
+    VMAs to circumvent the issue "set_host_vma(): do NOT read protect
+    Linux VMA".
+

docs/uti.rst

@@ -4,24 +4,41 @@ Advanced: Enable Utility Thread offloading Interface (UTI)
 UTI enables a runtime such as MPI runtime to spawn utility threads such
 as MPI asynchronous progress threads to Linux cores.
 
-Install capstone
-~~~~~~~~~~~~~~~~~~~~
+Install ``capstone`` and ``capstone-devel``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-When compute nodes don't have access to repositories
-""""""""""""""""""""""""""""""""""""""""""""""""""""
+When compute nodes don't have access to EPEL repository
+"""""""""""""""""""""""""""""""""""""""""""""""""""""""
 
-Install EPEL capstone-devel:
+Install EPEL ``capstone`` and ``capstone-devel``:
 
 ::
 
    sudo yum install https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm
-   sudo yum install capstone-devel
+   sudo yum install capstone capstone-devel
 
 
-When compute nodes don't have access to repositories
-""""""""""""""""""""""""""""""""""""""""""""""""""""
+When compute nodes don't have access to EPEL repository
+"""""""""""""""""""""""""""""""""""""""""""""""""""""""
 
-Ask the system administrator to install ``capstone-devel``. Note that it is in the EPEL repository.
+A. Ask the system administrator to install ``capstone`` and ``capstone-devel``. Note that it is in the EPEL repository.
+
+B. Download the rpm with the machine in which you are the administrator:
+
+::
+
+   sudo yum install https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm
+   sudo yum install yum-utils
+   yumdownloader capstone capstone-devel
+
+and then install it to your home directory of the login node:
+
+::
+
+   cd $HOME/$(uname -p)
+   rpm2cpio capstone-4.0.1-9.el8.aarch64.rpm | cpio -idv
+   rpm2cpio capstone-devel-4.0.1-9.el8.aarch64.rpm | cpio -idv
+   sed -i 's#/usr/#'"$HOME"'/'"$(uname -p)"'/usr/#' $HOME/$(uname -p)/usr/lib64/pkgconfig/capstone.pc
 
 
 Install syscall_intercept
@@ -31,7 +48,24 @@ Install syscall_intercept
 
    git clone https://github.com/RIKEN-SysSoft/syscall_intercept.git
    mkdir build && cd build
-   cmake <syscall_intercept>/arch/aarch64 -DCMAKE_INSTALL_PREFIX=<syscall-intercept-install> -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=gcc -DTREAT_WARNINGS_AS_ERRORS=OFF
+
+When ``capstone`` and ``capstone-devel`` are installed into the system directory:
+
+::
+
+   cmake ../syscall_intercept/arch/aarch64 -DCMAKE_INSTALL_PREFIX=${HOME}/$(uname -p)/usr -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=gcc -DTREAT_WARNINGS_AS_ERRORS=OFF
+
+When ``capstone`` and ``capstone-devel`` are installed into your home directory:
+
+::
+
+   CMAKE_PREFIX_PATH=${HOME}/$(uname -p)/usr cmake ../syscall_intercept/arch/aarch64 -DCMAKE_INSTALL_PREFIX=${HOME}/$(uname -p)/usr -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=gcc -DTREAT_WARNINGS_AS_ERRORS=OFF
+
+Install:
+
+::
+
+   make && make install && make test
 
 Install UTI for McKernel
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -48,16 +82,17 @@ Install:
 Install McKernel
 ~~~~~~~~~~~~~~~~~~~~
 
-Add ``-DENABLE_UTI=ON`` option to ``cmake``:
+``cmake`` with the additional options:
 
 ::
 
-   CMAKE_PREFIX_PATH=<syscall-intercept-install> cmake -DCMAKE_INSTALL_PREFIX=${HOME}/ihk+mckernel -DENABLE_UTI=ON $HOME/src/ihk+mckernel/mckernel
+   CMAKE_PREFIX_PATH=${HOME}/$(uname -p)/usr cmake -DCMAKE_INSTALL_PREFIX=${HOME}/ihk+mckernel -DENABLE_UTI=ON $HOME/src/ihk+mckernel/mckernel
+   make -j install
 
 Run programs
 ~~~~~~~~~~~~~~~~
 
-Add ``--enable-uti`` option to ``mcexec``:
+``mcexec`` with ``--enable-uti`` option:
 
 ::
 

executer/kernel/mcctrl/control.c

@@ -270,16 +270,17 @@ int mcexec_transfer_image(ihk_os_t os, struct remote_transfer *__user upt)
 		return -EFAULT;
 	}
 
+#ifdef CONFIG_MIC
 	if (pt.size > PAGE_SIZE) {
 		printk("mcexec_transfer_image(): ERROR: size exceeds PAGE_SIZE\n");
 		return -EFAULT;
 	}
-	
+
 	phys = ihk_device_map_memory(ihk_os_to_dev(os), pt.rphys, PAGE_SIZE);
-#ifdef CONFIG_MIC
 	rpm = ioremap_wc(phys, PAGE_SIZE);
 #else
-	rpm = ihk_device_map_virtual(ihk_os_to_dev(os), phys, PAGE_SIZE, NULL, 0);
+	phys = ihk_device_map_memory(ihk_os_to_dev(os), pt.rphys, pt.size);
+	rpm = ihk_device_map_virtual(ihk_os_to_dev(os), phys, pt.size, NULL, 0);
 #endif
 
 	if (!rpm) {
@@ -304,10 +305,11 @@ int mcexec_transfer_image(ihk_os_t os, struct remote_transfer *__user upt)
 
 #ifdef CONFIG_MIC
 	iounmap(rpm);
+	ihk_device_unmap_memory(ihk_os_to_dev(os), phys, PAGE_SIZE);
 #else
-	ihk_device_unmap_virtual(ihk_os_to_dev(os), rpm, PAGE_SIZE);
+	ihk_device_unmap_virtual(ihk_os_to_dev(os), rpm, pt.size);
+	ihk_device_unmap_memory(ihk_os_to_dev(os), phys, pt.size);
 #endif
-	ihk_device_unmap_memory(ihk_os_to_dev(os), phys, PAGE_SIZE);	
 
 	return ret;
 
@@ -3644,7 +3646,8 @@ int __mcctrl_os_read_write_cpu_register(ihk_os_t os, int cpu,
 	isp.op = op;
 	isp.pdesc = virt_to_phys(ldesc);
 
-	ret = mcctrl_ikc_send_wait(os, cpu, &isp, 0, NULL, &do_free, 1, ldesc);
+	/* 1 sec timeout for the case where McKernel can't respond */
+	ret = mcctrl_ikc_send_wait(os, cpu, &isp, -1000, NULL, &do_free, 1, ldesc);
 	if (ret != 0) {
 		printk("%s: ERROR sending IKC msg: %d\n", __FUNCTION__, ret);
 		goto out;

executer/kernel/mcctrl/ikc.c

@@ -536,9 +536,6 @@ int prepare_ikc_channels(ihk_os_t os)
 	usrdata->os = os;
 	ihk_host_os_set_usrdata(os, usrdata);
 
-	ihk_ikc_listen_port(os, &lp_ikc2linux);
-	ihk_ikc_listen_port(os, &lp_ikc2mckernel);
-
 	init_waitqueue_head(&usrdata->wq_procfs);
 	mutex_init(&usrdata->reserve_lock);
 	mutex_init(&usrdata->part_exec_lock);
@@ -555,6 +552,16 @@ int prepare_ikc_channels(ihk_os_t os)
 	INIT_LIST_HEAD(&usrdata->wakeup_descs_list);
 	spin_lock_init(&usrdata->wakeup_descs_lock);
 
+	/* ihk_ikc_listen_port should be performed after
+	 * usrdata->cpu_topology_list is initialized because the
+	 * function enables syscall_packet_handler which accesses
+	 * the list (the call path is sysfsm_packet_handler -->
+	 * sysfsm_work_main --> sysfsm_setup --> setup_sysfs_files
+	 * --> setup_cpus_sysfs_files).
+	 */
+	ihk_ikc_listen_port(os, &lp_ikc2linux);
+	ihk_ikc_listen_port(os, &lp_ikc2mckernel);
+
 	return 0;
 
 error:

executer/user/mcexec.c

@@ -1957,14 +1957,14 @@ opendev()
 			fprintf(stderr, "%s: warning: LD_PRELOAD line is too long\n", __FUNCTION__); \
 			return; \
 		} \
-		strncat(envbuf, elembuf, remainder); \
+		strncat(envbuf, elembuf, remainder - 1); \
 		remainder = PATH_MAX - (strlen(envbuf) + 1); \
 		nelem++; \
 	} while (0)
 
 static ssize_t find_libdir(char *libdir, size_t len)
 {
-	FILE *filep;
+	FILE *filep = NULL;
 	ssize_t rc;
 	size_t linelen = 0;
 	char *line = NULL;
@@ -2020,7 +2020,9 @@ static ssize_t find_libdir(char *libdir, size_t len)
 	}
 
 out:
-	pclose(filep);
+	if (filep) {
+		pclose(filep);
+	}
 	free(line);
 	return rc;
 }
@@ -4121,6 +4123,7 @@ int main_loop(struct thread_data_s *my_thread)
 #endif
 
 		case __NR_gettid:{
+			int rc = 0;
 			/*
 			 * Number of TIDs and the remote physical address where TIDs are
 			 * expected are passed in arg 4 and 5, respectively.
@@ -4132,6 +4135,7 @@ int main_loop(struct thread_data_s *my_thread)
 				int *tids = malloc(sizeof(int) * w.sr.args[4]);
 				if (!tids) {
 					fprintf(stderr, "__NR_gettid(): error allocating TIDs\n");
+					rc = -ENOMEM;
 					goto gettid_out;
 				}
 
@@ -4152,13 +4156,14 @@ int main_loop(struct thread_data_s *my_thread)
 				trans.direction = MCEXEC_UP_TRANSFER_TO_REMOTE;
 
 				if (ioctl(fd, MCEXEC_UP_TRANSFER, &trans) != 0) {
+					rc = -EFAULT;
 					fprintf(stderr, "__NR_gettid(): error transfering TIDs\n");
 				}
 
 				free(tids);
 			}
 gettid_out:
-			do_syscall_return(fd, cpu, 0, 0, 0, 0, 0);
+			do_syscall_return(fd, cpu, rc, 0, 0, 0, 0);
 			break;
 		}
 

kernel/ap.c

@@ -267,3 +267,154 @@ cpu_sysfs_setup(void)
 
 	return;
 } /* cpu_sysfs_setup() */
+
+/*
+ * Generic remote CPU function invocation facility.
+ */
+void smp_func_call_handler(void)
+{
+	unsigned long irq_flags;
+	struct smp_func_call_request *req;
+	int reqs_left;
+
+reiterate:
+	req = NULL;
+	reqs_left = 0;
+
+	irq_flags = ihk_mc_spinlock_lock(
+			&cpu_local_var(smp_func_req_lock));
+
+	/* Take requests one-by-one */
+	if (!list_empty(&cpu_local_var(smp_func_req_list))) {
+		req = list_first_entry(&cpu_local_var(smp_func_req_list),
+			struct smp_func_call_request, list);
+		list_del(&req->list);
+
+		reqs_left = !list_empty(&cpu_local_var(smp_func_req_list));
+	}
+
+	ihk_mc_spinlock_unlock(&cpu_local_var(smp_func_req_lock),
+			irq_flags);
+
+	if (req) {
+		req->ret = req->sfcd->func(req->cpu_index,
+				req->sfcd->nr_cpus, req->sfcd->arg);
+		ihk_atomic_dec(&req->sfcd->cpus_left);
+	}
+
+	if (reqs_left)
+		goto reiterate;
+}
+
+int smp_call_func(cpu_set_t *__cpu_set, smp_func_t __func, void *__arg)
+{
+	int cpu, nr_cpus = 0;
+	int cpu_index = 0;
+	int this_cpu_index = 0;
+	struct smp_func_call_data sfcd;
+	struct smp_func_call_request *reqs;
+	int ret = 0;
+	int call_on_this_cpu = 0;
+	cpu_set_t cpu_set;
+	int max_nr_cpus = 4;
+
+	/* Sanity checks */
+	if (!__cpu_set || !__func) {
+		return -EINVAL;
+	}
+
+	/* Make sure it won't change in between */
+	cpu_set = *__cpu_set;
+
+	for_each_set_bit(cpu, (unsigned long *)&cpu_set,
+			sizeof(cpu_set) * BITS_PER_BYTE) {
+
+		if (cpu == ihk_mc_get_processor_id()) {
+			call_on_this_cpu = 1;
+		}
+		++nr_cpus;
+
+		if (nr_cpus == max_nr_cpus)
+			break;
+	}
+
+	if (!nr_cpus) {
+		return -EINVAL;
+	}
+
+	reqs = kmalloc(sizeof(*reqs) * nr_cpus, IHK_MC_AP_NOWAIT);
+	if (!reqs) {
+		ret = -ENOMEM;
+		goto free_out;
+	}
+
+	kprintf("%s: interrupting %d CPUs for SMP call..\n", __func__, nr_cpus);
+	sfcd.nr_cpus = nr_cpus;
+	sfcd.func = __func;
+	sfcd.arg = __arg;
+	ihk_atomic_set(&sfcd.cpus_left,
+			call_on_this_cpu ? nr_cpus - 1 : nr_cpus);
+	smp_wmb();
+
+	/* Add requests and send IPIs */
+	cpu_index = 0;
+	for_each_set_bit(cpu, (unsigned long *)&cpu_set,
+			sizeof(cpu_set) * BITS_PER_BYTE) {
+		unsigned long irq_flags;
+
+		reqs[cpu_index].cpu_index = cpu_index;
+		reqs[cpu_index].ret = 0;
+
+		if (cpu == ihk_mc_get_processor_id()) {
+			this_cpu_index = cpu_index;
+			++cpu_index;
+			continue;
+		}
+
+		reqs[cpu_index].sfcd = &sfcd;
+
+		irq_flags =
+			ihk_mc_spinlock_lock(&get_cpu_local_var(cpu)->smp_func_req_lock);
+		list_add_tail(&reqs[cpu_index].list,
+				&get_cpu_local_var(cpu)->smp_func_req_list);
+		ihk_mc_spinlock_unlock(&get_cpu_local_var(cpu)->smp_func_req_lock,
+			irq_flags);
+
+		dkprintf("%s: interrupting IRQ: %d -> CPU: %d\n", __func__,
+			ihk_mc_get_smp_handler_irq(), cpu);
+		ihk_mc_interrupt_cpu(cpu, ihk_mc_get_smp_handler_irq());
+
+		++cpu_index;
+		if (cpu_index == max_nr_cpus)
+			break;
+	}
+
+	/* Is this CPU involved? */
+	if (call_on_this_cpu) {
+		reqs[this_cpu_index].ret =
+			__func(this_cpu_index, nr_cpus, __arg);
+	}
+
+	dkprintf("%s: waiting for remote CPUs..\n", __func__);
+	/* Wait for the rest of the CPUs */
+	while (smp_load_acquire(&sfcd.cpus_left.counter) > 0) {
+		cpu_pause();
+	}
+
+	/* Check return values, if error, report the first non-zero */
+	for (cpu_index = 0; cpu_index < nr_cpus; ++cpu_index) {
+		if (reqs[cpu_index].ret != 0) {
+			ret = reqs[cpu_index].ret;
+			goto free_out;
+		}
+	}
+
+	kprintf("%s: all CPUs finished SMP call successfully\n", __func__);
+	ret = 0;
+
+free_out:
+	kfree(reqs);
+
+	return ret;
+}
+

kernel/include/cls.h

@@ -106,9 +106,7 @@ struct cpu_local_var {
 	ihk_spinlock_t migq_lock;
 	struct list_head migq;
 	int in_interrupt;
-#ifdef ENABLE_FUGAKU_HACKS
 	int in_page_fault;
-#endif
 	int no_preempt;
 	int timer_enabled;
 	unsigned long nr_ctx_switches;

kernel/include/process.h

@@ -406,6 +406,7 @@ struct vm_range_numa_policy {
 	unsigned long start, end;
 	DECLARE_BITMAP(numa_mask, PROCESS_NUMA_MASK_BITS);
 	int numa_mem_policy;
+	int il_prev;
 };
 
 struct vm_regions {
@@ -797,6 +798,7 @@ struct process_vm {
 	long currss;
 	DECLARE_BITMAP(numa_mask, PROCESS_NUMA_MASK_BITS);
 	int numa_mem_policy;
+	int il_prev;
 	/* Protected by memory_range_lock */
 	struct rb_root vm_range_numa_policy_tree;
 	struct vm_range *range_cache[VM_RANGE_CACHE_SIZE];

kernel/mem.c

@@ -523,6 +523,18 @@ static void reserve_pages(struct ihk_page_allocator_desc *pa_allocator,
 	ihk_pagealloc_reserve(pa_allocator, start, end);
 }
 
+static int interleave_nodes(int off, unsigned long *numa_mask)
+{
+	int next;
+
+	next = find_next_bit(numa_mask, PROCESS_NUMA_MASK_BITS, off + 1);
+	if (next >= PROCESS_NUMA_MASK_BITS) {
+		next = find_first_bit(numa_mask, PROCESS_NUMA_MASK_BITS);
+	}
+
+	return next;
+}
+
 extern int cpu_local_var_initialized;
 static void *mckernel_allocate_aligned_pages_node(int npages, int p2align,
 		ihk_mc_ap_flag flag, int pref_node, int is_user, uintptr_t virt_addr)
@@ -538,7 +550,9 @@ static void *mckernel_allocate_aligned_pages_node(int npages, int p2align,
 	int numa_mem_policy = -1;
 	struct process_vm *vm;
 	struct vm_range *range = NULL;
-	int chk_shm = 0;
+	int chk_shm = 0, il_start, looping;
+	int *il_prev = NULL;
+	unsigned long *numa_mask = NULL;
 
 	if(npages <= 0)
 		return NULL;
@@ -549,31 +563,39 @@ static void *mckernel_allocate_aligned_pages_node(int npages, int p2align,
 			!cpu_local_var(current)->vm)
 		goto distance_based;
 
-	/* No explicitly requested NUMA or user policy? */
-	if ((pref_node == -1) && (!(flag & IHK_MC_AP_USER) ||
-				cpu_local_var(current)->vm->numa_mem_policy == MPOL_DEFAULT)) {
+	vm = cpu_local_var(current)->vm;
+	node = ihk_mc_get_numa_id();
 
-		if (virt_addr != -1) {
-			vm = cpu_local_var(current)->vm;
-			range_policy_iter = vm_range_policy_search(vm, virt_addr);
-			if (range_policy_iter) {
-				range = lookup_process_memory_range(vm, (uintptr_t)virt_addr, ((uintptr_t)virt_addr) + 1);
-				if (range) {
-					if( (range->memobj) && (range->memobj->flags == MF_SHM)) {
-						chk_shm = 1;
-					}
-				}
+	/* Get mempolicy user requested */
+	if (virt_addr != -1) {
+		range_policy_iter = vm_range_policy_search(vm, virt_addr);
+
+		if (range_policy_iter) {
+			range = lookup_process_memory_range(vm,
+					(uintptr_t)virt_addr,
+					((uintptr_t)virt_addr) + 1);
+			if ((range && (range->memobj->flags == MF_SHM))) {
+				chk_shm = 1;
 			}
+
+			/* Use range policy */
+			numa_mem_policy = range_policy_iter->numa_mem_policy;
+			numa_mask = range_policy_iter->numa_mask;
+			il_prev = &range_policy_iter->il_prev;
+		} else {
+			/* Use process policy */
+			numa_mem_policy = vm->numa_mem_policy;
+			numa_mask = vm->numa_mask;
+			il_prev = &vm->il_prev;
 		}
-
-
-		if ((!((range_policy_iter) && (range_policy_iter->numa_mem_policy != MPOL_DEFAULT))) && (chk_shm == 0))
-			goto distance_based;
 	}
 
-	node = ihk_mc_get_numa_id();
-	if (!memory_nodes[node].nodes_by_distance)
-		goto order_based;
+	/* No explicitly requested NUMA or user policy? */
+	if ((pref_node == -1) && !(flag & IHK_MC_AP_USER)) {
+		if ((numa_mem_policy == MPOL_DEFAULT) && (chk_shm == 0)) {
+			goto distance_based;
+		}
+	}
 
 	/* Explicit valid node? */
 	if (pref_node > -1 && pref_node < ihk_mc_get_nr_numa_nodes()) {
@@ -615,27 +637,6 @@ static void *mckernel_allocate_aligned_pages_node(int npages, int p2align,
 		}
 	}
 
-	if ((virt_addr != -1) && (chk_shm == 0)) {
-
-		vm = cpu_local_var(current)->vm;
-
-		if (!(range_policy_iter)) {
-			range_policy_iter = vm_range_policy_search(vm, virt_addr);
-		}
-
-		if (range_policy_iter) {
-			range = lookup_process_memory_range(vm, (uintptr_t)virt_addr, ((uintptr_t)virt_addr) + 1);
-			if ((range && (range->memobj->flags == MF_SHM))) {
-				chk_shm = 1;
-			} else {
-				numa_mem_policy = range_policy_iter->numa_mem_policy;
-			}
-		}
-	}
-
-	if (numa_mem_policy == -1)
-		numa_mem_policy = cpu_local_var(current)->vm->numa_mem_policy;
-
 	switch (numa_mem_policy) {
 		case MPOL_BIND:
 		case MPOL_PREFERRED:
@@ -644,9 +645,8 @@ static void *mckernel_allocate_aligned_pages_node(int npages, int p2align,
 			 * only the ones requested in user policy */
 			for (i = 0; i < ihk_mc_get_nr_numa_nodes(); ++i) {
 
-				/* Not part of user requested policy? */
 				if (!test_bit(memory_nodes[node].nodes_by_distance[i].id,
-						cpu_local_var(current)->proc->vm->numa_mask)) {
+						numa_mask)) {
 					continue;
 				}
 
@@ -687,7 +687,55 @@ static void *mckernel_allocate_aligned_pages_node(int npages, int p2align,
 			break;
 
 		case MPOL_INTERLEAVE:
-			/* TODO: */
+			/* Initialize interleave */
+			il_start = *il_prev;
+			looping = 0;
+
+retry_interleave:
+			/* Find next node */
+			numa_id = interleave_nodes(*il_prev, numa_mask);
+			*il_prev = numa_id;
+
+			if (il_start == *il_prev && looping) {
+				/* All interleave nodes are full */
+				pa = 0;
+				break;
+			}
+			looping = 1;
+#ifdef IHK_RBTREE_ALLOCATOR
+			{
+				if (rusage_check_oom(numa_id, npages, is_user)
+						== -ENOMEM) {
+					goto retry_interleave;
+				} else {
+					pa = ihk_numa_alloc_pages(
+							&memory_nodes[numa_id],
+							npages, p2align);
+				}
+#else
+			list_for_each_entry(pa_allocator,
+					&memory_nodes[numa_id].allocators,
+					list) {
+				if (rusage_check_oom(numa_id, npages, is_user)
+						== -ENOMEM) {
+					goto retry_interleave;
+				} else {
+					pa = ihk_pagealloc_alloc(pa_allocator,
+							npages, p2align);
+				}
+#endif
+
+				if (pa) {
+					rusage_page_add(numa_id, npages,
+							is_user);
+					dkprintf("%s: policy: CPU @ node %d allocated "
+							"%d pages from node %d\n",
+							__func__,
+							ihk_mc_get_numa_id(),
+							npages, node);
+
+				}
+			}
 			break;
 
 		default:
@@ -1395,7 +1443,6 @@ static void page_fault_handler(void *fault_addr, uint64_t reason, void *regs)
 			__FUNCTION__, fault_addr, reason, regs);
 
 	preempt_disable();
-#ifdef ENABLE_FUGAKU_HACKS
 	++cpu_local_var(in_page_fault);
 	if (cpu_local_var(in_page_fault) > 1) {
 		kprintf("%s: PF in PF??\n", __func__);
@@ -1408,7 +1455,6 @@ static void page_fault_handler(void *fault_addr, uint64_t reason, void *regs)
 			panic("PANIC");
 		}
 	}
-#endif
 
 	cpu_enable_interrupt();
 
@@ -1475,6 +1521,7 @@ out_linux:
 			__func__, thread ? thread->tid : -1, fault_addr,
 			reason, error);
 		unhandled_page_fault(thread, fault_addr, reason, regs);
+		--cpu_local_var(in_page_fault);
 		preempt_enable();
 
 #ifdef ENABLE_FUGAKU_DEBUG
@@ -1511,9 +1558,7 @@ out_linux:
 out_ok:
 #endif
 	error = 0;
-#ifdef ENABLE_FUGAKU_HACKS
 	--cpu_local_var(in_page_fault);
-#endif
 	preempt_enable();
 out:
 	dkprintf("%s: addr: %p, reason: %lx, regs: %p -> error: %d\n",
@@ -2885,3 +2930,44 @@ retry:
 
 	return ptep;
 }
+
+int phys_to_nid(unsigned long p)
+{
+   int i, numa_id = -1, _numa_id;
+   unsigned long _start, _end;
+
+   for (i = 0; i < ihk_mc_get_nr_memory_chunks(); i++) {
+	   ihk_mc_get_memory_chunk(i, &_start, &_end, &_numa_id);
+
+	   if (p >= _start && p < _end) {
+		   numa_id = _numa_id;
+		   goto out;
+	   }
+   }
+
+out:
+   return numa_id;
+}
+
+int lookup_node(struct process_vm *vm, void *addr)
+{
+	int node, err, reason = PF_POPULATE | PF_USER;
+	pte_t *ptep;
+
+	err = page_fault_process_vm(vm, (void *)addr, reason);
+	if (err) {
+		node = err;
+		goto out;
+	}
+
+	ptep = ihk_mc_pt_lookup_pte(vm->address_space->page_table,
+			(void *)addr, 0, NULL, NULL, NULL);
+	if (!ptep || !pte_is_present(ptep)) {
+		node = -ENOENT;
+		goto out;
+	}
+
+	node = phys_to_nid(pte_get_phys(ptep));
+out:
+	return node;
+}

kernel/syscall.c

@@ -9676,7 +9676,9 @@ SYSCALL_DECLARE(mbind)
 		return -EINVAL;
 	}
 
+#ifdef ENABLE_FUGAKU_HACKS
 	return 0;
+#endif
 
 	memset(numa_mask, 0, sizeof(numa_mask));
 
@@ -9921,6 +9923,10 @@ mbind_update_only:
 					sizeof(numa_mask));
 			}
 			range_policy->numa_mem_policy = mode;
+			if (mode == MPOL_INTERLEAVE) {
+				range_policy->il_prev =
+						PROCESS_NUMA_MASK_BITS - 1;
+			}
 
 			break;
 
@@ -10082,6 +10088,9 @@ SYSCALL_DECLARE(set_mempolicy)
 			}
 
 			vm->numa_mem_policy = mode;
+			if (mode == MPOL_INTERLEAVE) {
+				vm->il_prev = PROCESS_NUMA_MASK_BITS - 1;
+			}
 			error = 0;
 			break;
 
@@ -10144,6 +10153,20 @@ SYSCALL_DECLARE(get_mempolicy)
 		}
 	}
 
+	/* case of MPOL_F_NODE and MPOL_F_ADDR are specified */
+	if (flags & MPOL_F_NODE && flags & MPOL_F_ADDR) {
+		/* return the node ID which addr is allocated by mode */
+		int nid;
+
+		nid = lookup_node(vm, (void *)addr);
+		error = copy_to_user(mode, &nid, sizeof(int));
+		if (error) {
+			error = -EFAULT;
+			goto out;
+		}
+		goto out;
+	}
+
 	/* Special case of MPOL_F_MEMS_ALLOWED */
 	if (flags == MPOL_F_MEMS_ALLOWED) {
 		if (nodemask) {
@@ -10227,7 +10250,7 @@ SYSCALL_DECLARE(move_pages)
 	struct move_pages_smp_req mpsr;
 
 	struct process_vm *vm = cpu_local_var(current)->vm;
-	int ret = 0;
+	int i, ret = 0;
 
 	unsigned long t_s, t_e;
 
@@ -10237,18 +10260,20 @@ SYSCALL_DECLARE(move_pages)
 	if (pid) {
 		kprintf("%s: ERROR: only self (pid == 0)"
 				" is supported\n", __FUNCTION__);
-		return -EINVAL;
+		ret = -EINVAL;
+		goto out;
 	}
 
-	switch (flags) {
-	case MPOL_MF_MOVE_ALL:
+    /* Check flags */
+	if (flags & ~(MPOL_MF_MOVE|MPOL_MF_MOVE_ALL)) {
+		ret = -EINVAL;
+		goto out;
+	}
+	if (flags & MPOL_MF_MOVE_ALL) {
 		kprintf("%s: ERROR: MPOL_MF_MOVE_ALL"
 			" not supported\n", __func__);
-		return -EINVAL;
-	case MPOL_MF_MOVE:
-		break;
-	default:
-		return -EINVAL;
+		ret = -EINVAL;
+		goto out;
 	}
 
 	/* Allocate kernel arrays */
@@ -10296,7 +10321,7 @@ t_e = rdtsc(); kprintf("%s: init malloc: %lu \n", __FUNCTION__, t_e - t_s); t_s
 		goto dealloc_out;
 	}
 
-	if (verify_process_vm(cpu_local_var(current)->vm,
+	if (user_nodes && verify_process_vm(cpu_local_var(current)->vm,
 				user_nodes, sizeof(int) * count)) {
 		ret = -EFAULT;
 		goto dealloc_out;
@@ -10307,6 +10332,18 @@ t_e = rdtsc(); kprintf("%s: init malloc: %lu \n", __FUNCTION__, t_e - t_s); t_s
 		ret = -EFAULT;
 		goto dealloc_out;
 	}
+
+	/* Check node ID */
+	if (user_nodes) {
+		copy_from_user(nodes, user_nodes, sizeof(int) * count);
+		for (i = 0; i < count; i++) {
+			if (nodes[i] < 0 || nodes[i] >= ihk_mc_get_nr_numa_nodes()) {
+				ret = -ENODEV;
+				goto dealloc_out;
+			}
+		}
+	}
+
 t_e = rdtsc(); kprintf("%s: init verify: %lu \n", __FUNCTION__, t_e - t_s); t_s = t_e;
 
 #if 0
@@ -10399,6 +10436,7 @@ dealloc_out:
 	kfree(ptep);
 	kfree(dst_phys);
 
+out:
 	return ret;
 }
 

kernel/tofu/tof_utofu_main.c

@@ -1236,6 +1236,7 @@ static int tof_utofu_ioctl_alloc_stag(struct tof_utofu_device *dev, unsigned lon
 
 	readonly = (req.flags & 1) != 0;
 
+retry:
 	ihk_rwspinlock_read_lock_noirq(&vm->memory_range_lock);
 
 	/* Assume smallest page size at first */
@@ -1271,6 +1272,20 @@ static int tof_utofu_ioctl_alloc_stag(struct tof_utofu_device *dev, unsigned lon
 	}
 
 	if (!range) {
+		if (vm->region.stack_start <= start &&
+				vm->region.stack_end > end) {
+
+			ihk_rwspinlock_read_unlock_noirq(&vm->memory_range_lock);
+
+			if (page_fault_process_vm(vm, (void *)start,
+						PF_POPULATE | PF_WRITE | PF_USER) < 0) {
+				ret = -EINVAL;
+				goto out;
+			}
+
+			goto retry;
+		}
+
 		ret = -EINVAL;
 		goto unlock_out;
 	}
@@ -1358,6 +1373,7 @@ static int tof_utofu_ioctl_alloc_stag(struct tof_utofu_device *dev, unsigned lon
 unlock_out:
 	ihk_rwspinlock_read_unlock_noirq(&vm->memory_range_lock);
 
+out:
 	if(ret == 0){
 		if(copy_to_user((void *)arg, &req, sizeof(req)) != 0){
 			kprintf("%s: ret: %d\n", __func__, -EFAULT);

kernel/xpmem.c

@@ -2056,6 +2056,7 @@ static int xpmem_pin_page(
 
 	XPMEM_DEBUG("call: tgid=%d, vaddr=0x%lx", tg->tgid, vaddr);
 
+retry:
 	ihk_rwspinlock_read_lock_noirq(&src_vm->memory_range_lock);
 
 	range = lookup_process_memory_range(src_vm, vaddr, vaddr + 1);
@@ -2063,6 +2064,20 @@ static int xpmem_pin_page(
 	ihk_rwspinlock_read_unlock_noirq(&src_vm->memory_range_lock);
 
 	if (!range || range->start > vaddr) {
+		/*
+		 * Grow the stack if address falls into stack region
+		 * so that we can lookup range successfully.
+		 */
+		if (src_vm->region.stack_start <= vaddr &&
+				src_vm->region.stack_end > vaddr) {
+			if (page_fault_process_vm(src_vm, (void *)vaddr,
+						PF_POPULATE | PF_WRITE | PF_USER) < 0) {
+				return -ENOENT;
+			}
+
+			goto retry;
+		}
+
 		return -ENOENT;
 	}
 

lib/include/ihk/cpu.h

@@ -173,4 +173,7 @@ struct cpu_mapping;
 int arch_get_cpu_mapping(struct cpu_mapping **buf, int *nelemsp);
 int ihk_mc_ikc_arch_issue_host_ipi(int cpu, int vector);
 
+void smp_func_call_handler(void);
+int ihk_mc_get_smp_handler_irq(void);
+
 #endif

lib/include/memory.h

@@ -20,6 +20,8 @@ struct process_vm;
 
 unsigned long virt_to_phys(void *v);
 void *phys_to_virt(unsigned long p);
+int phys_to_nid(unsigned long p);
+int lookup_node(struct process_vm *vm, void *addr);
 int copy_from_user(void *dst, const void *src, size_t siz);
 int strlen_user(const char *s);
 int strcpy_from_user(char *dst, const char *src);

scripts/mckernel.spec.in

@@ -25,9 +25,17 @@ Source0: mckernel-%{version}.tar.gz
 
 Requires: systemd-libs numactl-libs libdwarf
 
-# kernel_module_package macro does not handle cross build...
+# don't use kernel_module_package so that one rpm including .ko and binaries are created
 %if "%{?_host_cpu}" == "x86_64" && "%{?_target_cpu}" == "aarch64"
 %define cross_compile 1
+%else
+BuildRequires: systemd-devel numactl-devel binutils-devel kernel-devel libdwarf-devel
+# Friendly reminder of the fact that kernel-rpm-macros is no longer included in kernel-devel
+%if 0%{?rhel} >= 8
+BuildRequires: redhat-rpm-config kernel-rpm-macros elfutils-libelf-devel
+%endif
+%endif
+
 %if 0%{?rhel} >= 8
 Requires: kernel >= %{krequires}
 %else
@@ -35,17 +43,6 @@ Requires: kernel = %{krequires}
 %endif
 Requires(post):   /usr/sbin/depmod
 Requires(postun): /usr/sbin/depmod
-%else
-BuildRequires: systemd-devel numactl-devel binutils-devel kernel-devel libdwarf-devel
-# Friendly reminder of the fact that kernel-rpm-macros is no longer included in kernel-devel
-%if 0%{?rhel} >= 8
-BuildRequires: redhat-rpm-config kernel-rpm-macros elfutils-libelf-devel kmod
-%endif
-%if %{defined kernel_module_package_buildreqs}
-BuildRequires: %kernel_module_package_buildreqs
-%kernel_module_package %{?kmod_flavors}
-%endif
-%endif
 
 %description
 Interface for Heterogeneous Kernels and McKernel.
@@ -78,6 +75,9 @@ This package contains headers and libraries required for build apps using IHK/Mc
 	%{?cmake_libdir:-DCMAKE_INSTALL_LIBDIR=%{cmake_libdir}} \
 	%{?build_target:-DBUILD_TARGET=%{build_target}} \
 	%{?toolchain_file:-DCMAKE_TOOLCHAIN_FILE=%{toolchain_file}} \
+	-DENABLE_TOFU=ON -DENABLE_FUGAKU_HACKS=ON \
+	-DENABLE_KRM_WORKAROUND=OFF -DWITH_KRM=ON \
+	-DENABLE_FUGAKU_DEBUG=OFF \
 	.
 %make_build
 
@@ -113,7 +113,6 @@ This package contains headers and libraries required for build apps using IHK/Mc
 %{_mandir}/man1/ihkosctl.1.gz
 %{_mandir}/man1/mcexec.1.gz
 
-%if 0%{?cross_compile}
 /lib/modules/%{kernel_version}/extra/mckernel/ihk.ko
 /lib/modules/%{kernel_version}/extra/mckernel/mcctrl.ko
 %ifarch x86_64
@@ -122,7 +121,6 @@ This package contains headers and libraries required for build apps using IHK/Mc
 %ifarch aarch64
 /lib/modules/%{kernel_version}/extra/mckernel/ihk-smp-arm64.ko
 %endif
-%endif
 
 %files	devel
 %{_includedir}/ihklib.h
@@ -135,8 +133,7 @@ This package contains headers and libraries required for build apps using IHK/Mc
 %{_includedir}/ihk/ihk_host_driver.h
 /lib/modules/%{kernel_version}/extra/mckernel/ihk/linux/core/Module.symvers
 
-%if 0%{?cross_compile}
-# scripts from /usr/lib/rpm/redhat/kmodtool (kernel_module_package) as well
+# taken from /usr/lib/rpm/redhat/kmodtool (kernel_module_package)
 %post
 if [ -e "/boot/System.map-%{kernel_version}" ]; then
     /usr/sbin/depmod -aeF "/boot/System.map-%{kernel_version}" "%{kernel_version}" > /dev/null || :
@@ -162,7 +159,6 @@ if [ -x "/sbin/weak-modules" ]; then
     printf '%s\n' "${modules[@]}" \
     | /sbin/weak-modules --remove-modules
 fi
-%endif
 
 %changelog
 * Tue Feb 12 2019 Dominique Martinet <dominique.martinet@cea.fr> - 1.6.0-0

test/issues/1523/C1523.sh

@@ -0,0 +1,30 @@
+#/bin/sh
+
+USELTP=1
+USEOSTEST=0
+
+MCREBOOT=0
+. ../../common.sh
+
+BOOTPARAM="${BOOTPARAM} -e anon_on_demand"
+mcreboot
+
+issue="1523"
+tid=01
+
+for tp in move_pages01 move_pages02 move_pages04 move_pages06 move_pages09 move_pages10
+do
+	tname=`printf "C${issue}T%02d" ${tid}`
+	echo "*** ${tname} start *******************************"
+	sudo $MCEXEC $LTPBIN/$tp 2>&1 | tee $tp.txt
+	ok=`grep PASS $tp.txt | wc -l`
+	ng=`grep FAIL $tp.txt | wc -l`
+	if [ $ng = 0 ]; then
+		echo "*** ${tname} PASSED ($ok)"
+	else
+		echo "*** ${tname} FAILED (ok=$ok ng=$ng)"
+	fi
+	let tid++
+	echo ""
+done
+

test/issues/1523/Makefile

@@ -0,0 +1,11 @@
+CFLAGS=-g
+LDFLAGS=
+
+TARGET=
+
+all: $(TARGET)
+
+test: all
+	./C1523.sh
+clean:
+	rm -f $(TARGET) *.o *.txt

test/issues/1523/README

@@ -0,0 +1,21 @@
+【Issue#1523 動作確認】
+□ テスト内容
+1. 以下のLTPがPASSすることを確認する
+  - move_pages01
+  - move_pages02
+  - move_pages04
+  - move_pages06
+  - move_pages09
+  - move_pages10
+
+□ 実行手順
+$ make test
+
+McKernelのインストール先や、OSTEST, LTPの配置場所は、
+$HOME/.mck_test_config を参照している
+.mck_test_config は、McKernelをビルドした際に生成されるmck_test_config.sample ファイルを
+$HOMEにコピーし、適宜編集する
+
+□ 実行結果
+x86_64_result.log aarch64_result.log 参照。
+すべての項目をPASSしていることを確認。

test/issues/1523/aarch64_result.log

@@ -0,0 +1,25 @@
+mcstop+release.sh ... done
+mcreboot.sh -c 37-43,49-55 -m 2G@2,2G@3 -r 37-43:36+49-55:48 -O -e anon_on_demand ... done
+*** C1523T01 start *******************************
+move_pages01    1  TPASS  :  pages are present in expected nodes
+*** C1523T01 PASSED (1)
+
+*** C1523T02 start *******************************
+move_pages02    1  TPASS  :  pages are present in expected nodes
+*** C1523T02 PASSED (1)
+
+*** C1523T03 start *******************************
+move_pages04    1  TPASS  :  status[1] has expected value
+*** C1523T03 PASSED (1)
+
+*** C1523T04 start *******************************
+move_pages06    1  TPASS  :  move_pages failed with ENODEV as expected
+*** C1523T04 PASSED (1)
+
+*** C1523T05 start *******************************
+move_pages09    1  TPASS  :  move_pages succeeded
+*** C1523T05 PASSED (1)
+
+*** C1523T06 start *******************************
+move_pages10    1  TPASS  :  move_pages failed with EINVAL as expected
+*** C1523T06 PASSED (1)

test/issues/1523/x86_64_result.log

@@ -0,0 +1,25 @@
+mcstop+release.sh ... done
+mcreboot.sh -c 1-7,9-15,17-23,25-31 -m 10G@0,10G@1 -r 1-7:0+9-15:8+17-23:16+25-31:24 -O -e anon_on_demand ... done
+*** C1523T01 start *******************************
+move_pages01    1  TPASS  :  pages are present in expected nodes
+*** C1523T01 PASSED (1)
+
+*** C1523T02 start *******************************
+move_pages02    1  TPASS  :  pages are present in expected nodes
+*** C1523T02 PASSED (1)
+
+*** C1523T03 start *******************************
+move_pages04    1  TPASS  :  status[1] has expected value
+*** C1523T03 PASSED (1)
+
+*** C1523T04 start *******************************
+move_pages06    1  TPASS  :  move_pages failed with ENODEV as expected
+*** C1523T04 PASSED (1)
+
+*** C1523T05 start *******************************
+move_pages09    1  TPASS  :  move_pages succeeded
+*** C1523T05 PASSED (1)
+
+*** C1523T06 start *******************************
+move_pages10    1  TPASS  :  move_pages failed with EINVAL as expected
+*** C1523T06 PASSED (1)

test/issues/959/C959.sh

@@ -0,0 +1,125 @@
+#/bin/sh
+
+USELTP=1
+USEOSTEST=1
+
+LTP_LIST="mbind01 get_mempolicy01"
+OSTEST_MBIND_LIST="1 3 5 9 12 14 15 16 20 24 26 28 30"
+
+BOOTPARAM="-c 1-7 -m 10G@0,10G@1 -O -e anon_on_demand"
+. ../../common.sh
+
+issue="959"
+tid=01
+arch=`uname -p`
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+sudo ${MCEXEC} ./check_mempol_il 1 30 6 3 3 3
+
+if [ $? -eq 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+sudo ${MCEXEC} ./check_mempol_il 2 30 6 3 3 3
+
+if [ $? -eq 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+sudo ${MCEXEC} ./check_mempol_il 1 30 6 2 0 6
+
+if [ $? -eq 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+sudo ${MCEXEC} ./check_mempol_il 2 30 6 2 0 6
+
+if [ $? -eq 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+BOOTPARAM="-c 1-7 -m 10G@0,2G@1 -O -e anon_on_demand"
+mcstop
+mcreboot
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+${IHKOSCTL} 0 clear_kmsg
+sudo ${MCEXEC} ./check_mempol_il 1 30 6 2 4 2
+ret=$?
+dbg_prints=`${IHKOSCTL} 0 kmsg | grep "TEST_959" | wc -l`
+
+if [ ${ret} -eq 0 -a ${dbg_prints} -gt 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+tname=`printf "C${issue}T%02d" ${tid}`
+echo "*** ${tname} start *******************************"
+sudo ${MCEXEC} ./check_mempol_il 2 30 6 2 4 2
+ret=$?
+dbg_prints=`${IHKOSCTL} 0 kmsg | grep "TEST_959" | wc -l`
+
+if [ ${ret} -eq 0 -a ${dbg_prints} -gt 0 ]; then
+	echo "*** ${tname} PASSED ******************************"
+else
+	echo "*** ${tname} FAILED ******************************"
+fi
+let tid++
+echo ""
+
+for tp in ${LTP_LIST}
+do
+	tname=`printf "C${issue}T%02d" ${tid}`
+	echo "*** ${tname} start *******************************"
+	sudo $MCEXEC $LTPBIN/$tp 2>&1 | tee $tp.txt
+	ok=`grep PASS $tp.txt | wc -l`
+	ng=`grep FAIL $tp.txt | wc -l`
+	if [ $ng = 0 ]; then
+		echo "*** ${tname} PASSED ($ok)"
+	else
+		echo "*** ${tname} FAILED (ok=$ok ng=$ng)"
+	fi
+	let tid++
+	echo ""
+done
+
+for tno in ${OSTEST_MBIND_LIST}
+do
+	tname=`printf "C${issue}T%02d" ${tid}`
+	echo "*** ${tname} start *******************************"
+	${MCEXEC} ${TESTMCK} -s mbind -n ${tno} -- -n 2 2>&1 | tee test_mck-mbind${tno}.txt
+	if [ $? = 0 ]; then
+		echo "*** ${tname} PASSED"
+	else
+		echo "*** ${tname} FAILED"
+	fi
+	let tid++
+	echo ""
+done

test/issues/959/Makefile

@@ -0,0 +1,14 @@
+include $(HOME)/.mck_test_config.mk
+
+CFLAGS=-g -O0 -Wall -I$(MCK_DIR)/include
+LDFLAGS=-L$(MCK_DIR)/lib64 -lihk -lnuma -Wl,-rpath=$(MCK_DIR)/lib64
+
+TARGET=check_mempol_il
+
+all: $(TARGET)
+
+test: all
+	./C959.sh
+clean:
+	rm -f $(TARGET) *.o *.txt
+

test/issues/959/README

@@ -0,0 +1,87 @@
+【Issue#959 動作確認】
+□ テスト内容
+本テストは2つのNUMAノード(node0, node1)を使用してMPOL_INTERLEAVEの動作を確認するテストである。
+2つ以上のNUMAノードを持つ環境で実行すること。
+
+1. INTERLEAVEするノードセットに十分なメモリ容量がある場合の動作確認
+C959T01: set_mempolicyによるmempolicy設定時の動作 (2ノード)
+	node0, node1 からそれぞれ10GBのメモリをMcKernelに割り当てた状態で
+	下記の処理を確認する
+	 (1) set_mempolicy() でプロセスのmempolicyを、node0, node1 でのINTERLEAVEに設定する
+	 (2) 6GBのメモリを確保し、書き込みを行う
+	 (3) McKernelの2つのNUMAノードから均等にメモリが使用されていることを確認する
+
+C959T02: mbindによるmempolicy設定時の動作 (2ノード)
+	node0, node1 からそれぞれ10GBのメモリをMcKernelに割り当てた状態で
+	下記の処理を確認する
+	 (1) set_mempolicy() でプロセスのmempolicyを、node0 でのBINDに設定する
+	 (2) 6GBのメモリを確保する
+	 (3) mbind() で(2)で確保した領域のmempolicyを、node0, node1 でのINTERLEAVEに設定する
+	 (4) McKernelの2つのNUMAノードから均等にメモリが使用されていることを確認する
+
+C959T03: set_mempolicyによるmempolicy設定時の動作 (1ノード)
+	node0, node1 からそれぞれ10GBのメモリをMcKernelに割り当てた状態で
+	下記の処理を確認する
+	 (1) set_mempolicy() でプロセスのmempolicyを、 node1 でのINTERLEAVEに設定する
+	 (2) 6GBのメモリを確保し、書き込みを行う
+	 (3) McKernelのnode1から 6GBが使用されていることを確認する
+
+C959T04: mbindによるmempolicy設定時の動作 (2ノード)
+	node0, node1 からそれぞれ10GBのメモリをMcKernelに割り当てた状態で
+	下記の処理を確認する
+	 (1) set_mempolicy() でプロセスのmempolicyを、node0 でのBINDに設定する
+	 (2) 6GBのメモリを確保する
+	 (3) mbind() で(2)で確保した領域のmempolicyを、node1 でのINTERLEAVEに設定する
+	 (4) McKernelのnode1から 6GBが使用されていることを確認する
+
+2. INTERLEAVEするノードセットにメモリ容量が不足している場合の動作確認
+C959T05: set_mempolicyによるmempolicy設定時の動作
+	node0 に10GB、 node1 に2GBのメモリをそれぞれMcKernelに割り当てた状態で
+	下記の処理を確認する
+	 (1) set_mempolicy() でプロセスのmempolicyを、node1 でのINTERLEAVEに設定する
+	 (2) 6GBのメモリを確保し、書き込みを行う
+	 (3) McKernelのnode0から4GB, node1から2GBがそれぞれ使用されていることを確認する
+
+C959T06: mbindによるmempolicy設定時の動作
+	node0 に10GB、 node1 に2GBのメモリをそれぞれMcKernelに割り当てた状態で
+	下記の処理を確認する
+	 (1) set_mempolicy() でプロセスのmempolicyを、node0 でのBINDに設定する
+	 (2) 6GBのメモリを確保する
+	 (3) mbind() で(2)で確保した領域のmempolicyを、node1 でのINTERLEAVEに設定する
+	 (4) McKernelのnode0から4GB, node1から2GBがそれぞれ使用されていることを確認する
+
+3. 以下のLTPを用いて既存のmbind機能に影響がないことを確認する
+	- mbind01
+ 	- get_mempolicy01
+
+4. 以下のOSTESTを用いて既存のmbind機能に影響がないことを確認する
+	- ostest-mbind.000
+	- ostest-mbind.001
+	- ostest-mbind.002
+	- ostest-mbind.003
+	- ostest-mbind.004
+	- ostest-mbind.005
+	- ostest-mbind.006
+	- ostest-mbind.007
+	- ostest-mbind.008
+	- ostest-mbind.009
+	- ostest-mbind.010
+	- ostest-mbind.011
+	- ostest-mbind.012
+
+□ 実行手順
+・下記の手順でテストを実行する
+$ cd <mckernel>
+$ patch -p0 < test/issues/959/test_print.patch
+(build mckernel)
+$ cd test/issues/959
+$ make test
+
+McKernelのインストール先や、OSTEST, LTPの配置場所は、
+$HOME/.mck_test_config を参照している
+.mck_test_config は、McKernelをビルドした際に生成されるmck_test_config.sample ファイルを
+$HOMEにコピーし、適宜編集する
+
+□ 実行結果
+x86_64result.log, aarch64_result.log 参照。
+すべての項目をPASSしていることを確認。

test/issues/959/aarch64_result.log

@@ -0,0 +1,314 @@
+mcstop+release.sh ... done
+mcreboot.sh -c 1-7 -m 10G@0,10G@1 -O -e anon_on_demand ... done
+*** C959T01 start *******************************
+INTERLEAVE BIT_MASK: 0x3
+set_mempolicy: INTERLEAVE  mask 0x3
+** Difference of numa_stat **
+[OK]  NUMA[0] 0xc0000000
+[OK]  NUMA[1] 0xc0000000
+*** C959T01 PASSED ******************************
+
+*** C959T02 start *******************************
+INTERLEAVE BIT_MASK: 0x3
+set_mempolicy: BIND  mask 0x1
+mbind        : INTERLEAVE  mask 0x3
+** Difference of numa_stat **
+[OK]  NUMA[0] 0xc0000000
+[OK]  NUMA[1] 0xc0000000
+*** C959T02 PASSED ******************************
+
+*** C959T03 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x0
+[OK]  NUMA[1] 0x180000000
+*** C959T03 PASSED ******************************
+
+*** C959T04 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: BIND  mask 0x1
+mbind        : INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x0
+[OK]  NUMA[1] 0x180000000
+*** C959T04 PASSED ******************************
+
+mcstop+release.sh ... done
+mcreboot.sh -c 1-7 -m 10G@0,2G@1 -O -e anon_on_demand ... done
+*** C959T05 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x100000000
+[OK]  NUMA[1] 0x80000000
+*** C959T05 PASSED ******************************
+
+*** C959T06 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: BIND  mask 0x1
+mbind        : INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x100000000
+[OK]  NUMA[1] 0x80000000
+*** C959T06 PASSED ******************************
+
+*** C959T07 start *******************************
+tst_test.c:1096: INFO: Timeout per run is 0h 05m 00s
+mbind01.c:181: INFO: case MPOL_DEFAULT
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_DEFAULT (target exists)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_BIND (no target)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_BIND
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_INTERLEAVE (no target)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_INTERLEAVE
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_PREFERRED (no target)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_PREFERRED
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case UNKNOWN_POLICY
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_DEFAULT (invalid flags)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_PREFERRED (invalid nodemask)
+mbind01.c:230: PASS: Test passed
+
+Summary:
+passed   11
+failed   0
+skipped  0
+warnings 0
+*** C959T07 PASSED (11)
+
+*** C959T08 start *******************************
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=-1 errno=14 (Bad address)
+RESULT: return value(ret)=-1 errno=14 (Bad address)
+EXPECT: return value(ret)=-1 errno=22 (Invalid argument)
+RESULT: return value(ret)=-1 errno=22 (Invalid argument)
+get_mempolicy01    0  TINFO  :  (case00) START
+get_mempolicy01    1  TPASS  :  (case00) END
+get_mempolicy01    0  TINFO  :  (case01) START
+get_mempolicy01    2  TPASS  :  (case01) END
+get_mempolicy01    0  TINFO  :  (case02) START
+get_mempolicy01    3  TPASS  :  (case02) END
+get_mempolicy01    0  TINFO  :  (case03) START
+get_mempolicy01    4  TPASS  :  (case03) END
+get_mempolicy01    0  TINFO  :  (case04) START
+get_mempolicy01    5  TPASS  :  (case04) END
+get_mempolicy01    0  TINFO  :  (case05) START
+get_mempolicy01    6  TPASS  :  (case05) END
+get_mempolicy01    0  TINFO  :  (case06) START
+get_mempolicy01    7  TPASS  :  (case06) END
+get_mempolicy01    0  TINFO  :  (case07) START
+get_mempolicy01    8  TPASS  :  (case07) END
+get_mempolicy01    0  TINFO  :  (case08) START
+get_mempolicy01    9  TPASS  :  (case08) END
+get_mempolicy01    0  TINFO  :  (case09) START
+get_mempolicy01   10  TPASS  :  (case09) END
+get_mempolicy01    0  TINFO  :  (case10) START
+get_mempolicy01   11  TPASS  :  (case10) END
+get_mempolicy01    0  TINFO  :  (case11) START
+get_mempolicy01   12  TPASS  :  (case11) END
+*** C959T08 PASSED (12)
+
+*** C959T09 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 1
+ARGS: -n 2
+RESULT: ok
+*** C959T09 PASSED
+
+*** C959T10 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 3
+ARGS: -n 2
+RESULT: ok
+*** C959T10 PASSED
+
+*** C959T11 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 5
+ARGS: -n 2
+RESULT: ok
+*** C959T11 PASSED
+
+*** C959T12 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 9
+ARGS: -n 2
+RESULT: ok
+*** C959T12 PASSED
+
+*** C959T13 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 12
+ARGS: -n 2
+RESULT: ok
+*** C959T13 PASSED
+
+*** C959T14 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 14
+ARGS: -n 2
+RESULT: ok
+*** C959T14 PASSED
+
+*** C959T15 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 15
+ARGS: -n 2
+region 0
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 1
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 2
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 3
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 4
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 5
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 6
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+RESULT: ok
+*** C959T15 PASSED
+
+*** C959T16 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 16
+ARGS: -n 2
+region 0
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 1
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 2
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 3
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 4
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 5
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 6
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+RESULT: ok
+*** C959T16 PASSED
+
+*** C959T17 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 20
+ARGS: -n 2
+region 0
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 1
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 2
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 3
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 4
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 5
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 6
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+RESULT: ok
+*** C959T17 PASSED
+
+*** C959T18 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 24
+ARGS: -n 2
+RESULT: ok
+*** C959T18 PASSED
+
+*** C959T19 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 26
+ARGS: -n 2
+nodemask = 0
+RESULT: ok
+*** C959T19 PASSED
+
+*** C959T20 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 28
+ARGS: -n 2
+RESULT: ok
+*** C959T20 PASSED
+
+*** C959T21 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 30
+ARGS: -n 2
+RESULT: ok
+*** C959T21 PASSED
+

test/issues/959/check_mempol_il.c

@@ -0,0 +1,139 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+#include <numaif.h>
+#include <sys/mman.h>
+#include <errno.h>
+#include <ihklib.h>
+#include <ihk/ihk_rusage.h>
+
+#define NUMA_NUM 2
+
+long long numa_stat_bfr[NUMA_NUM], numa_stat_aft[NUMA_NUM];
+long long exp_diff[NUMA_NUM];
+
+int
+get_current_numa_stat(long long *stat, int numa_cnt)
+{
+	int i, ret = 0;
+	struct ihk_os_rusage mck_rusage;
+
+	memset(&mck_rusage, 0, sizeof(mck_rusage));
+	ret = ihk_os_getrusage(0, &mck_rusage, sizeof(mck_rusage));
+	if (ret) {
+		perror("ihk_os_getrusage()");
+		goto out;
+	}
+
+	for (i = 0; i < numa_cnt; i++) {
+		if (mck_rusage.memory_numa_stat[i] != 0) {
+			stat[i] = mck_rusage.memory_numa_stat[i];
+		}
+	}
+out:
+	return ret;
+}
+
+int
+main(int argc, char **argv)
+{
+	void *p;
+	unsigned long mask, bind_mask = 1;
+	unsigned long ps;
+	int i, mode, pgshift, pgnum, exp_0, exp_1, ret = 0;
+
+	if (argc < 7) {
+		printf("error: too few arguments\n");
+		ret = -1;
+		goto out;
+	}
+
+	mode = atoi(argv[1]); /* 1: set_mempolicy, 2: mbind */
+	pgshift = atoi(argv[2]);
+	pgnum = atoi(argv[3]);
+	mask = atoi(argv[4]);
+	exp_0 = atoi(argv[5]);
+	exp_1 = atoi(argv[6]);
+
+	ps = 1UL << pgshift;
+	exp_diff[0] = exp_0 * ps;
+	exp_diff[1] = exp_1 * ps;
+
+	if (mode != 1 && mode != 2) {
+		printf("error: invalid mode\n");
+		ret = -1;
+		goto out;
+	}
+
+	printf("INTERLEAVE BIT_MASK: 0x%lx\n", mask);
+
+	get_current_numa_stat(numa_stat_bfr, NUMA_NUM);
+	switch (mode) {
+	case 1: /* set_mempolicy */
+		printf("set_mempolicy: INTERLEAVE  mask 0x%lx\n", mask);
+		if (set_mempolicy(MPOL_INTERLEAVE, &mask, NUMA_NUM)) {
+			perror("set_mempolicy");
+			ret = -1;
+			goto out;
+		}
+
+		p = mmap(NULL, ps * pgnum, PROT_READ|PROT_WRITE,
+				MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
+		if (p == ((void *)-1)) {
+			perror("mmap");
+			ret = -1;
+			goto out;
+		}
+		break;
+	case 2: /* mbind */
+		printf("set_mempolicy: BIND  mask 0x%lx\n", bind_mask);
+		if (set_mempolicy(MPOL_BIND, &bind_mask, NUMA_NUM)) {
+			perror("set_mempolicy");
+			ret = -1;
+			goto out;
+		}
+
+		p = mmap(NULL, ps * pgnum, PROT_READ|PROT_WRITE,
+				MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
+		if (p == ((void *)-1)) {
+			perror("mmap");
+			ret = -1;
+			goto out;
+		}
+
+		printf("mbind        : INTERLEAVE  mask 0x%lx\n", mask);
+		if (mbind(p, ps * pgnum, MPOL_INTERLEAVE, &mask,
+				NUMA_NUM, 0) == -1) {
+			perror("mbind");
+			ret = -1;
+			goto out;
+		}
+		break;
+	default:
+		printf("error: invalid mode\n");
+		ret = -1;
+		goto out;
+	}
+
+	memset(p, '0', ps * pgnum);
+
+	get_current_numa_stat(numa_stat_aft, NUMA_NUM);
+
+	printf("** Difference of numa_stat **\n");
+	for (i = 0; i < NUMA_NUM; i++) {
+		long long diff = numa_stat_aft[i] - numa_stat_bfr[i];
+
+		if (diff == exp_diff[i]) {
+			printf("[OK] ");
+		} else {
+			printf("[NG] ");
+			ret = -1;
+		}
+		printf(" NUMA[%d] 0x%llx\n", i, diff);
+	}
+
+	munmap(p, ps * pgnum);
+out:
+	return ret;
+}

test/issues/959/test_print.patch

@@ -0,0 +1,14 @@
+diff --git kernel/mem.c kernel/mem.c
+index e464eb2..7086c6a 100644
+--- kernel/mem.c
++++ kernel/mem.c
+@@ -749,6 +749,9 @@ retry_interleave:
+ #endif
+ 		dkprintf("%s: couldn't fulfill user policy for %d pages\n",
+ 			__FUNCTION__, npages);
++		if (numa_mem_policy == MPOL_INTERLEAVE) {
++			kprintf("TEST_959: reach HERE\n");
++		}
+ 	}
+ 
+ distance_based:

test/issues/959/x86_64_result.log

@@ -0,0 +1,313 @@
+mcstop+release.sh ... done
+mcreboot.sh -c 1-7 -m 10G@0,10G@1 -O -e anon_on_demand ... done
+*** C959T01 start *******************************
+INTERLEAVE BIT_MASK: 0x3
+set_mempolicy: INTERLEAVE  mask 0x3
+** Difference of numa_stat **
+[OK]  NUMA[0] 0xc0000000
+[OK]  NUMA[1] 0xc0000000
+*** C959T01 PASSED ******************************
+
+*** C959T02 start *******************************
+INTERLEAVE BIT_MASK: 0x3
+set_mempolicy: BIND  mask 0x1
+mbind        : INTERLEAVE  mask 0x3
+** Difference of numa_stat **
+[OK]  NUMA[0] 0xc0000000
+[OK]  NUMA[1] 0xc0000000
+*** C959T02 PASSED ******************************
+
+*** C959T03 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x0
+[OK]  NUMA[1] 0x180000000
+*** C959T03 PASSED ******************************
+
+*** C959T04 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: BIND  mask 0x1
+mbind        : INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x0
+[OK]  NUMA[1] 0x180000000
+*** C959T04 PASSED ******************************
+
+mcstop+release.sh ... done
+mcreboot.sh -c 1-7 -m 10G@0,2G@1 -O -e anon_on_demand ... done
+*** C959T05 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x100000000
+[OK]  NUMA[1] 0x80000000
+*** C959T05 PASSED ******************************
+
+*** C959T06 start *******************************
+INTERLEAVE BIT_MASK: 0x2
+set_mempolicy: BIND  mask 0x1
+mbind        : INTERLEAVE  mask 0x2
+** Difference of numa_stat **
+[OK]  NUMA[0] 0x100000000
+[OK]  NUMA[1] 0x80000000
+*** C959T06 PASSED ******************************
+
+*** C959T07 start *******************************
+tst_test.c:1096: INFO: Timeout per run is 0h 05m 00s
+mbind01.c:181: INFO: case MPOL_DEFAULT
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_DEFAULT (target exists)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_BIND (no target)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_BIND
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_INTERLEAVE (no target)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_INTERLEAVE
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_PREFERRED (no target)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_PREFERRED
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case UNKNOWN_POLICY
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_DEFAULT (invalid flags)
+mbind01.c:230: PASS: Test passed
+mbind01.c:181: INFO: case MPOL_PREFERRED (invalid nodemask)
+mbind01.c:230: PASS: Test passed
+
+Summary:
+passed   11
+failed   0
+skipped  0
+warnings 0
+*** C959T07 PASSED (11)
+
+*** C959T08 start *******************************
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=0 errno=0 (Success)
+RESULT: return value(ret)=0 errno=0 (Success)
+EXPECT: return value(ret)=-1 errno=14 (Bad address)
+RESULT: return value(ret)=-1 errno=14 (Bad address)
+EXPECT: return value(ret)=-1 errno=22 (Invalid argument)
+RESULT: return value(ret)=-1 errno=22 (Invalid argument)
+get_mempolicy01    0  TINFO  :  (case00) START
+get_mempolicy01    1  TPASS  :  (case00) END
+get_mempolicy01    0  TINFO  :  (case01) START
+get_mempolicy01    2  TPASS  :  (case01) END
+get_mempolicy01    0  TINFO  :  (case02) START
+get_mempolicy01    3  TPASS  :  (case02) END
+get_mempolicy01    0  TINFO  :  (case03) START
+get_mempolicy01    4  TPASS  :  (case03) END
+get_mempolicy01    0  TINFO  :  (case04) START
+get_mempolicy01    5  TPASS  :  (case04) END
+get_mempolicy01    0  TINFO  :  (case05) START
+get_mempolicy01    6  TPASS  :  (case05) END
+get_mempolicy01    0  TINFO  :  (case06) START
+get_mempolicy01    7  TPASS  :  (case06) END
+get_mempolicy01    0  TINFO  :  (case07) START
+get_mempolicy01    8  TPASS  :  (case07) END
+get_mempolicy01    0  TINFO  :  (case08) START
+get_mempolicy01    9  TPASS  :  (case08) END
+get_mempolicy01    0  TINFO  :  (case09) START
+get_mempolicy01   10  TPASS  :  (case09) END
+get_mempolicy01    0  TINFO  :  (case10) START
+get_mempolicy01   11  TPASS  :  (case10) END
+get_mempolicy01    0  TINFO  :  (case11) START
+get_mempolicy01   12  TPASS  :  (case11) END
+*** C959T08 PASSED (12)
+
+*** C959T09 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 1
+ARGS: -n 2
+RESULT: ok
+*** C959T09 PASSED
+
+*** C959T10 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 3
+ARGS: -n 2
+RESULT: ok
+*** C959T10 PASSED
+
+*** C959T11 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 5
+ARGS: -n 2
+RESULT: ok
+*** C959T11 PASSED
+
+*** C959T12 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 9
+ARGS: -n 2
+RESULT: ok
+*** C959T12 PASSED
+
+*** C959T13 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 12
+ARGS: -n 2
+RESULT: ok
+*** C959T13 PASSED
+
+*** C959T14 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 14
+ARGS: -n 2
+RESULT: ok
+*** C959T14 PASSED
+
+*** C959T15 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 15
+ARGS: -n 2
+region 0
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 1
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 2
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 3
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 4
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 5
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 6
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+RESULT: ok
+*** C959T15 PASSED
+
+*** C959T16 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 16
+ARGS: -n 2
+region 0
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 1
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 2
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 3
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 4
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 5
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 6
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+RESULT: ok
+*** C959T16 PASSED
+
+*** C959T17 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 20
+ARGS: -n 2
+region 0
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 1
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 2
+get      : mode = 3, node_mask = 3
+m_expect : mode = 3, node_mask = 3
+
+region 3
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 4
+get      : mode = 2, node_mask = 1
+m_expect : mode = 2, node_mask = 1
+
+region 5
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+region 6
+get      : mode = 0, node_mask = 0
+m_expect : mode = 0, node_mask = 0
+
+RESULT: ok
+*** C959T17 PASSED
+
+*** C959T18 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 24
+ARGS: -n 2
+RESULT: ok
+*** C959T18 PASSED
+
+*** C959T19 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 26
+ARGS: -n 2
+nodemask = 0
+RESULT: ok
+*** C959T19 PASSED
+
+*** C959T20 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 28
+ARGS: -n 2
+RESULT: ok
+*** C959T20 PASSED
+
+*** C959T21 start *******************************
+TEST_SUITE: mbind
+TEST_NUMBER: 30
+ARGS: -n 2
+RESULT: ok
+*** C959T21 PASSED