Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Pull s390 updates from Martin Schwidefsky:
 "There is only one new feature in this pull for the 4.4 merge window,
  most of it is small enhancements, cleanup and bug fixes:

   - Add the s390 backend for the software dirty bit tracking.  This
     adds two new pgtable functions pte_clear_soft_dirty and
     pmd_clear_soft_dirty which is why there is a hit to
     arch/x86/include/asm/pgtable.h in this pull request.

   - A series of cleanup patches for the AP bus, this includes the
     removal of the support for two outdated crypto cards (PCICC and
     PCICA).

   - The irq handling / signaling on buffer full in the runtime
     instrumentation code is dropped.

   - Some micro optimizations: remove unnecessary memory barriers for a
     couple of functions: [smb_]rmb, [smb_]wmb, atomics, bitops, and for
     spin_unlock.  Use the builtin bswap if available and make
     test_and_set_bit_lock more cache friendly.

   - Statistics and a tracepoint for the diagnose calls to the
     hypervisor.

   - The CPU measurement facility support to sample KVM guests is
     improved.

   - The vector instructions are now always enabled for user space
     processes if the hardware has the vector facility.  This simplifies
     the FPU handling code.  The fpu-internal.h header is split into fpu
     internals, api and types just like x86.

   - Cleanup and improvements for the common I/O layer.

   - Rework udelay to solve a problem with kprobe.  udelay has busy loop
     semantics but still uses an idle processor state for the wait"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (66 commits)
  s390: remove runtime instrumentation interrupts
  s390/cio: de-duplicate subchannel validation
  s390/css: unneeded initialization in for_each_subchannel
  s390/Kconfig: use builtin bswap
  s390/dasd: fix disconnected device with valid path mask
  s390/dasd: fix invalid PAV assignment after suspend/resume
  s390/dasd: fix double free in dasd_eckd_read_conf
  s390/kernel: fix ptrace peek/poke for floating point registers
  s390/cio: move ccw_device_stlck functions
  s390/cio: move ccw_device_call_handler
  s390/topology: reduce per_cpu() invocations
  s390/nmi: reduce size of percpu variable
  s390/nmi: fix terminology
  s390/nmi: remove casts
  s390/nmi: remove pointless error strings
  s390: don't store registers on disabled wait anymore
  s390: get rid of __set_psw_mask()
  s390/fpu: split fpu-internal.h into fpu internals, api, and type headers
  s390/dasd: fix list_del corruption after lcu changes
  s390/spinlock: remove unneeded serializations at unlock
  ...

1 files changed, 131 insertions, 89 deletions

diff --git a/drivers/s390/cio/cmf.c b/drivers/s390/cio/cmf.c
index 23054f8..b2afad5 100644
--- a/drivers/s390/cio/cmf.c
+++ b/drivers/s390/cio/cmf.c
@@ -113,7 +113,6 @@ module_param(format, bint, 0444);
  * @readall:	read a measurement block in a common format
  * @reset:	clear the data in the associated measurement block and
  *		reset its time stamp
- * @align:	align an allocated block so that the hardware can use it
  */
 struct cmb_operations {
 	int  (*alloc)  (struct ccw_device *);
@@ -122,7 +121,6 @@ struct cmb_operations {
 	u64  (*read)   (struct ccw_device *, int);
 	int  (*readall)(struct ccw_device *, struct cmbdata *);
 	void (*reset)  (struct ccw_device *);
-	void *(*align) (void *);
 /* private: */
 	struct attribute_group *attr_group;
 };
@@ -186,9 +184,8 @@ static inline void cmf_activate(void *area, unsigned int onoff)
 static int set_schib(struct ccw_device *cdev, u32 mme, int mbfc,
 		     unsigned long address)
 {
-	struct subchannel *sch;
-
-	sch = to_subchannel(cdev->dev.parent);
+	struct subchannel *sch = to_subchannel(cdev->dev.parent);
+	int ret;
 
 	sch->config.mme = mme;
 	sch->config.mbfc = mbfc;
@@ -198,7 +195,15 @@ static int set_schib(struct ccw_device *cdev, u32 mme, int mbfc,
 	else
 		sch->config.mbi = address;
 
-	return cio_commit_config(sch);
+	ret = cio_commit_config(sch);
+	if (!mme && ret == -ENODEV) {
+		/*
+		 * The task was to disable measurement block updates but
+		 * the subchannel is already gone. Report success.
+		 */
+		ret = 0;
+	}
+	return ret;
 }
 
 struct set_schib_struct {
@@ -314,7 +319,7 @@ static int cmf_copy_block(struct ccw_device *cdev)
 			return -EBUSY;
 	}
 	cmb_data = cdev->private->cmb;
-	hw_block = cmbops->align(cmb_data->hw_block);
+	hw_block = cmb_data->hw_block;
 	if (!memcmp(cmb_data->last_block, hw_block, cmb_data->size))
 		/* No need to copy. */
 		return 0;
@@ -425,7 +430,7 @@ static void cmf_generic_reset(struct ccw_device *cdev)
 		 * Need to reset hw block as well to make the hardware start
 		 * from 0 again.
 		 */
-		memset(cmbops->align(cmb_data->hw_block), 0, cmb_data->size);
+		memset(cmb_data->hw_block, 0, cmb_data->size);
 		cmb_data->last_update = 0;
 	}
 	cdev->private->cmb_start_time = get_tod_clock();
@@ -606,12 +611,6 @@ static void free_cmb(struct ccw_device *cdev)
 	spin_lock_irq(cdev->ccwlock);
 
 	priv = cdev->private;
-
-	if (list_empty(&priv->cmb_list)) {
-		/* already freed */
-		goto out;
-	}
-
 	cmb_data = priv->cmb;
 	priv->cmb = NULL;
 	if (cmb_data)
@@ -626,7 +625,6 @@ static void free_cmb(struct ccw_device *cdev)
 		free_pages((unsigned long)cmb_area.mem, get_order(size));
 		cmb_area.mem = NULL;
 	}
-out:
 	spin_unlock_irq(cdev->ccwlock);
 	spin_unlock(&cmb_area.lock);
 }
@@ -755,11 +753,6 @@ static void reset_cmb(struct ccw_device *cdev)
 	cmf_generic_reset(cdev);
 }
 
-static void * align_cmb(void *area)
-{
-	return area;
-}
-
 static struct attribute_group cmf_attr_group;
 
 static struct cmb_operations cmbops_basic = {
@@ -769,7 +762,6 @@ static struct cmb_operations cmbops_basic = {
 	.read	= read_cmb,
 	.readall    = readall_cmb,
 	.reset	    = reset_cmb,
-	.align	    = align_cmb,
 	.attr_group = &cmf_attr_group,
 };
 
@@ -804,64 +796,57 @@ struct cmbe {
 	u32 device_busy_time;
 	u32 initial_command_response_time;
 	u32 reserved[7];
-};
+} __packed __aligned(64);
 
-/*
- * kmalloc only guarantees 8 byte alignment, but we need cmbe
- * pointers to be naturally aligned. Make sure to allocate
- * enough space for two cmbes.
- */
-static inline struct cmbe *cmbe_align(struct cmbe *c)
-{
-	unsigned long addr;
-	addr = ((unsigned long)c + sizeof (struct cmbe) - sizeof(long)) &
-				 ~(sizeof (struct cmbe) - sizeof(long));
-	return (struct cmbe*)addr;
-}
+static struct kmem_cache *cmbe_cache;
 
 static int alloc_cmbe(struct ccw_device *cdev)
 {
-	struct cmbe *cmbe;
 	struct cmb_data *cmb_data;
-	int ret;
+	struct cmbe *cmbe;
+	int ret = -ENOMEM;
 
-	cmbe = kzalloc (sizeof (*cmbe) * 2, GFP_KERNEL);
+	cmbe = kmem_cache_zalloc(cmbe_cache, GFP_KERNEL);
 	if (!cmbe)
-		return -ENOMEM;
-	cmb_data = kzalloc(sizeof(struct cmb_data), GFP_KERNEL);
-	if (!cmb_data) {
-		ret = -ENOMEM;
+		return ret;
+
+	cmb_data = kzalloc(sizeof(*cmb_data), GFP_KERNEL);
+	if (!cmb_data)
 		goto out_free;
-	}
+
 	cmb_data->last_block = kzalloc(sizeof(struct cmbe), GFP_KERNEL);
-	if (!cmb_data->last_block) {
-		ret = -ENOMEM;
+	if (!cmb_data->last_block)
 		goto out_free;
-	}
-	cmb_data->size = sizeof(struct cmbe);
-	spin_lock_irq(cdev->ccwlock);
-	if (cdev->private->cmb) {
-		spin_unlock_irq(cdev->ccwlock);
-		ret = -EBUSY;
-		goto out_free;
-	}
+
+	cmb_data->size = sizeof(*cmbe);
 	cmb_data->hw_block = cmbe;
+
+	spin_lock(&cmb_area.lock);
+	spin_lock_irq(cdev->ccwlock);
+	if (cdev->private->cmb)
+		goto out_unlock;
+
 	cdev->private->cmb = cmb_data;
-	spin_unlock_irq(cdev->ccwlock);
 
 	/* activate global measurement if this is the first channel */
-	spin_lock(&cmb_area.lock);
 	if (list_empty(&cmb_area.list))
 		cmf_activate(NULL, 1);
 	list_add_tail(&cdev->private->cmb_list, &cmb_area.list);
-	spin_unlock(&cmb_area.lock);
 
+	spin_unlock_irq(cdev->ccwlock);
+	spin_unlock(&cmb_area.lock);
 	return 0;
+
+out_unlock:
+	spin_unlock_irq(cdev->ccwlock);
+	spin_unlock(&cmb_area.lock);
+	ret = -EBUSY;
 out_free:
 	if (cmb_data)
 		kfree(cmb_data->last_block);
 	kfree(cmb_data);
-	kfree(cmbe);
+	kmem_cache_free(cmbe_cache, cmbe);
+
 	return ret;
 }
 
@@ -869,19 +854,21 @@ static void free_cmbe(struct ccw_device *cdev)
 {
 	struct cmb_data *cmb_data;
 
+	spin_lock(&cmb_area.lock);
 	spin_lock_irq(cdev->ccwlock);
 	cmb_data = cdev->private->cmb;
 	cdev->private->cmb = NULL;
-	if (cmb_data)
+	if (cmb_data) {
 		kfree(cmb_data->last_block);
+		kmem_cache_free(cmbe_cache, cmb_data->hw_block);
+	}
 	kfree(cmb_data);
-	spin_unlock_irq(cdev->ccwlock);
 
 	/* deactivate global measurement if this is the last channel */
-	spin_lock(&cmb_area.lock);
 	list_del_init(&cdev->private->cmb_list);
 	if (list_empty(&cmb_area.list))
 		cmf_activate(NULL, 0);
+	spin_unlock_irq(cdev->ccwlock);
 	spin_unlock(&cmb_area.lock);
 }
 
@@ -897,7 +884,7 @@ static int set_cmbe(struct ccw_device *cdev, u32 mme)
 		return -EINVAL;
 	}
 	cmb_data = cdev->private->cmb;
-	mba = mme ? (unsigned long) cmbe_align(cmb_data->hw_block) : 0;
+	mba = mme ? (unsigned long) cmb_data->hw_block : 0;
 	spin_unlock_irqrestore(cdev->ccwlock, flags);
 
 	return set_schib_wait(cdev, mme, 1, mba);
@@ -1022,11 +1009,6 @@ static void reset_cmbe(struct ccw_device *cdev)
 	cmf_generic_reset(cdev);
 }
 
-static void * align_cmbe(void *area)
-{
-	return cmbe_align(area);
-}
-
 static struct attribute_group cmf_attr_group_ext;
 
 static struct cmb_operations cmbops_extended = {
@@ -1036,7 +1018,6 @@ static struct cmb_operations cmbops_extended = {
 	.read	    = read_cmbe,
 	.readall    = readall_cmbe,
 	.reset	    = reset_cmbe,
-	.align	    = align_cmbe,
 	.attr_group = &cmf_attr_group_ext,
 };
 
@@ -1171,23 +1152,28 @@ static ssize_t cmb_enable_show(struct device *dev,
 			       struct device_attribute *attr,
 			       char *buf)
 {
-	return sprintf(buf, "%d\n", to_ccwdev(dev)->private->cmb ? 1 : 0);
+	struct ccw_device *cdev = to_ccwdev(dev);
+	int enabled;
+
+	spin_lock_irq(cdev->ccwlock);
+	enabled = !!cdev->private->cmb;
+	spin_unlock_irq(cdev->ccwlock);
+
+	return sprintf(buf, "%d\n", enabled);
 }
 
 static ssize_t cmb_enable_store(struct device *dev,
 				struct device_attribute *attr, const char *buf,
 				size_t c)
 {
-	struct ccw_device *cdev;
-	int ret;
+	struct ccw_device *cdev = to_ccwdev(dev);
 	unsigned long val;
+	int ret;
 
 	ret = kstrtoul(buf, 16, &val);
 	if (ret)
 		return ret;
 
-	cdev = to_ccwdev(dev);
-
 	switch (val) {
 	case 0:
 		ret = disable_cmf(cdev);
@@ -1195,12 +1181,13 @@ static ssize_t cmb_enable_store(struct device *dev,
 	case 1:
 		ret = enable_cmf(cdev);
 		break;
+	default:
+		ret = -EINVAL;
 	}
 
-	return c;
+	return ret ? ret : c;
 }
-
-DEVICE_ATTR(cmb_enable, 0644, cmb_enable_show, cmb_enable_store);
+DEVICE_ATTR_RW(cmb_enable);
 
 int ccw_set_cmf(struct ccw_device *cdev, int enable)
 {
@@ -1220,41 +1207,71 @@ int enable_cmf(struct ccw_device *cdev)
 {
 	int ret;
 
+	device_lock(&cdev->dev);
+	get_device(&cdev->dev);
 	ret = cmbops->alloc(cdev);
-	cmbops->reset(cdev);
 	if (ret)
-		return ret;
+		goto out;
+	cmbops->reset(cdev);
+	ret = sysfs_create_group(&cdev->dev.kobj, cmbops->attr_group);
+	if (ret) {
+		cmbops->free(cdev);
+		goto out;
+	}
 	ret = cmbops->set(cdev, 2);
 	if (ret) {
+		sysfs_remove_group(&cdev->dev.kobj, cmbops->attr_group);
 		cmbops->free(cdev);
-		return ret;
 	}
-	ret = sysfs_create_group(&cdev->dev.kobj, cmbops->attr_group);
-	if (!ret)
-		return 0;
-	cmbops->set(cdev, 0);  //FIXME: this can fail
-	cmbops->free(cdev);
+out:
+	if (ret)
+		put_device(&cdev->dev);
+
+	device_unlock(&cdev->dev);
 	return ret;
 }
 
 /**
- * disable_cmf() - switch off the channel measurement for a specific device
+ * __disable_cmf() - switch off the channel measurement for a specific device
  *  @cdev:	The ccw device to be disabled
  *
  *  Returns %0 for success or a negative error value.
  *
  *  Context:
- *    non-atomic
+ *    non-atomic, device_lock() held.
  */
-int disable_cmf(struct ccw_device *cdev)
+int __disable_cmf(struct ccw_device *cdev)
 {
 	int ret;
 
 	ret = cmbops->set(cdev, 0);
 	if (ret)
 		return ret;
-	cmbops->free(cdev);
+
 	sysfs_remove_group(&cdev->dev.kobj, cmbops->attr_group);
+	cmbops->free(cdev);
+	put_device(&cdev->dev);
+
+	return ret;
+}
+
+/**
+ * disable_cmf() - switch off the channel measurement for a specific device
+ *  @cdev:	The ccw device to be disabled
+ *
+ *  Returns %0 for success or a negative error value.
+ *
+ *  Context:
+ *    non-atomic
+ */
+int disable_cmf(struct ccw_device *cdev)
+{
+	int ret;
+
+	device_lock(&cdev->dev);
+	ret = __disable_cmf(cdev);
+	device_unlock(&cdev->dev);
+
 	return ret;
 }
 
@@ -1295,10 +1312,32 @@ int cmf_reenable(struct ccw_device *cdev)
 	return cmbops->set(cdev, 2);
 }
 
+/**
+ * cmf_reactivate() - reactivate measurement block updates
+ *
+ * Use this during resume from hibernate.
+ */
+void cmf_reactivate(void)
+{
+	spin_lock(&cmb_area.lock);
+	if (!list_empty(&cmb_area.list))
+		cmf_activate(cmb_area.mem, 1);
+	spin_unlock(&cmb_area.lock);
+}
+
+static int __init init_cmbe(void)
+{
+	cmbe_cache = kmem_cache_create("cmbe_cache", sizeof(struct cmbe),
+				       __alignof__(struct cmbe), 0, NULL);
+
+	return cmbe_cache ? 0 : -ENOMEM;
+}
+
 static int __init init_cmf(void)
 {
 	char *format_string;
-	char *detect_string = "parameter";
+	char *detect_string;
+	int ret;
 
 	/*
 	 * If the user did not give a parameter, see if we are running on a
@@ -1324,15 +1363,18 @@ static int __init init_cmf(void)
 	case CMF_EXTENDED:
 		format_string = "extended";
 		cmbops = &cmbops_extended;
+
+		ret = init_cmbe();
+		if (ret)
+			return ret;
 		break;
 	default:
-		return 1;
+		return -EINVAL;
 	}
 	pr_info("Channel measurement facility initialized using format "
 		"%s (mode %s)\n", format_string, detect_string);
 	return 0;
 }
-
 module_init(init_cmf);