Merge tag 'char-misc-4.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc

Pull char/misc driver updates from Greg KH:
 "Here is the big set of char/misc patches for 4.20-rc1.

  Loads of things here, we have new code in all of these driver
  subsystems:
   - fpga
   - stm
   - extcon
   - nvmem
   - eeprom
   - hyper-v
   - gsmi
   - coresight
   - thunderbolt
   - vmw_balloon
   - goldfish
   - soundwire
  along with lots of fixes and minor changes to other small drivers.

  All of these have been in linux-next for a while with no reported
  issues"

* tag 'char-misc-4.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc: (245 commits)
  Documentation/security-bugs: Clarify treatment of embargoed information
  lib: Fix ia64 bootloader linkage
  MAINTAINERS: Clarify UIO vs UIOVEC maintainer
  docs/uio: fix a grammar nitpick
  docs: fpga: document programming fpgas using regions
  fpga: add devm_fpga_region_create
  fpga: bridge: add devm_fpga_bridge_create
  fpga: mgr: add devm_fpga_mgr_create
  hv_balloon: Replace spin_is_locked() with lockdep
  sgi-xp: Replace spin_is_locked() with lockdep
  eeprom: New ee1004 driver for DDR4 memory
  eeprom: at25: remove unneeded 'at25_remove'
  w1: IAD Register is yet readable trough iad sys file. Fix snprintf (%u for unsigned, count for max size).
  misc: mic: scif: remove set but not used variables 'src_dma_addr, dst_dma_addr'
  misc: mic: fix a DMA pool free failure
  platform: goldfish: pipe: Add a blank line to separate varibles and code
  platform: goldfish: pipe: Remove redundant casting
  platform: goldfish: pipe: Call misc_deregister if init fails
  platform: goldfish: pipe: Move the file-scope goldfish_pipe_dev variable into the driver state
  platform: goldfish: pipe: Move the file-scope goldfish_pipe_miscdev variable into the driver state
  ...

32 files changed, 1411 insertions, 821 deletions

diff --git a/drivers/misc/ad525x_dpot-i2c.c b/drivers/misc/ad525x_dpot-i2c.c
index 4f832002d116..1827c69959fb 100644
--- a/drivers/misc/ad525x_dpot-i2c.c
+++ b/drivers/misc/ad525x_dpot-i2c.c
@@ -114,6 +114,6 @@ static struct i2c_driver ad_dpot_i2c_driver = {
 
 module_i2c_driver(ad_dpot_i2c_driver);
 
-MODULE_AUTHOR("Michael Hennerich <hennerich@blackfin.uclinux.org>");
+MODULE_AUTHOR("Michael Hennerich <michael.hennerich@analog.com>");
 MODULE_DESCRIPTION("digital potentiometer I2C bus driver");
 MODULE_LICENSE("GPL");
diff --git a/drivers/misc/ad525x_dpot-spi.c b/drivers/misc/ad525x_dpot-spi.c
index 39a7f517ee7e..0383ec153725 100644
--- a/drivers/misc/ad525x_dpot-spi.c
+++ b/drivers/misc/ad525x_dpot-spi.c
@@ -140,7 +140,7 @@ static struct spi_driver ad_dpot_spi_driver = {
 
 module_spi_driver(ad_dpot_spi_driver);
 
-MODULE_AUTHOR("Michael Hennerich <hennerich@blackfin.uclinux.org>");
+MODULE_AUTHOR("Michael Hennerich <michael.hennerich@analog.com>");
 MODULE_DESCRIPTION("digital potentiometer SPI bus driver");
 MODULE_LICENSE("GPL");
 MODULE_ALIAS("spi:ad_dpot");
diff --git a/drivers/misc/ad525x_dpot.c b/drivers/misc/ad525x_dpot.c
index bc591b7168db..a0afadefcc49 100644
--- a/drivers/misc/ad525x_dpot.c
+++ b/drivers/misc/ad525x_dpot.c
@@ -1,7 +1,7 @@
 /*
  * ad525x_dpot: Driver for the Analog Devices digital potentiometers
  * Copyright (c) 2009-2010 Analog Devices, Inc.
- * Author: Michael Hennerich <hennerich@blackfin.uclinux.org>
+ * Author: Michael Hennerich <michael.hennerich@analog.com>
  *
  * DEVID		#Wipers		#Positions	Resistor Options (kOhm)
  * AD5258		1		64		1, 10, 50, 100
@@ -64,7 +64,7 @@
  * Author: Chris Verges <chrisv@cyberswitching.com>
  *
  * derived from ad5252.c
- * Copyright (c) 2006-2011 Michael Hennerich <hennerich@blackfin.uclinux.org>
+ * Copyright (c) 2006-2011 Michael Hennerich <michael.hennerich@analog.com>
  *
  * Licensed under the GPL-2 or later.
  */
@@ -760,6 +760,6 @@ EXPORT_SYMBOL(ad_dpot_remove);
 
 
 MODULE_AUTHOR("Chris Verges <chrisv@cyberswitching.com>, "
-	      "Michael Hennerich <hennerich@blackfin.uclinux.org>");
+	      "Michael Hennerich <michael.hennerich@analog.com>");
 MODULE_DESCRIPTION("Digital potentiometer driver");
 MODULE_LICENSE("GPL");
diff --git a/drivers/misc/apds990x.c b/drivers/misc/apds990x.c
index ed9412d750b7..24876c615c3c 100644
--- a/drivers/misc/apds990x.c
+++ b/drivers/misc/apds990x.c
@@ -188,7 +188,6 @@ struct apds990x_chip {
 #define APDS_LUX_DEFAULT_RATE		200
 
 static const u8 again[]	= {1, 8, 16, 120}; /* ALS gain steps */
-static const u8 ir_currents[]	= {100, 50, 25, 12}; /* IRled currents in mA */
 
 /* Following two tables must match i.e 10Hz rate means 1 as persistence value */
 static const u16 arates_hz[] = {10, 5, 2, 1};
diff --git a/drivers/misc/bh1770glc.c b/drivers/misc/bh1770glc.c
index 9c62bf064f77..17e81ce9925b 100644
--- a/drivers/misc/bh1770glc.c
+++ b/drivers/misc/bh1770glc.c
@@ -180,9 +180,6 @@ static const char reg_vleds[] = "Vleds";
 static const s16 prox_rates_hz[] = {100, 50, 33, 25, 14, 10, 5, 2};
 static const s16 prox_rates_ms[] = {10, 20, 30, 40, 70, 100, 200, 500};
 
-/* Supported IR-led currents in mA */
-static const u8 prox_curr_ma[] = {5, 10, 20, 50, 100, 150, 200};
-
 /*
  * Supported stand alone rates in ms from chip data sheet
  * {100, 200, 500, 1000, 2000};
diff --git a/drivers/misc/cxl/flash.c b/drivers/misc/cxl/flash.c
index 43917898fb9a..4d6836f19489 100644
--- a/drivers/misc/cxl/flash.c
+++ b/drivers/misc/cxl/flash.c
@@ -92,8 +92,8 @@ static int update_property(struct device_node *dn, const char *name,
 
 	val = (u32 *)new_prop->value;
 	rc = cxl_update_properties(dn, new_prop);
-	pr_devel("%s: update property (%s, length: %i, value: %#x)\n",
-		  dn->name, name, vd, be32_to_cpu(*val));
+	pr_devel("%pOFn: update property (%s, length: %i, value: %#x)\n",
+		  dn, name, vd, be32_to_cpu(*val));
 
 	if (rc) {
 		kfree(new_prop->name);
diff --git a/drivers/misc/cxl/guest.c b/drivers/misc/cxl/guest.c
index 3bc0c15d4d85..5d28d9e454f5 100644
--- a/drivers/misc/cxl/guest.c
+++ b/drivers/misc/cxl/guest.c
@@ -1018,8 +1018,6 @@ err1:
 
 void cxl_guest_remove_afu(struct cxl_afu *afu)
 {
-	pr_devel("in %s - AFU(%d)\n", __func__, afu->slice);
-
 	if (!afu)
 		return;
 
diff --git a/drivers/misc/echo/echo.c b/drivers/misc/echo/echo.c
index 8a5adc0d2e88..3ebe5d75ad6a 100644
--- a/drivers/misc/echo/echo.c
+++ b/drivers/misc/echo/echo.c
@@ -381,7 +381,7 @@ int16_t oslec_update(struct oslec_state *ec, int16_t tx, int16_t rx)
 	 */
 	ec->factor = 0;
 	ec->shift = 0;
-	if ((ec->nonupdate_dwell == 0)) {
+	if (!ec->nonupdate_dwell) {
 		int p, logp, shift;
 
 		/* Determine:
diff --git a/drivers/misc/eeprom/Kconfig b/drivers/misc/eeprom/Kconfig
index 68a1ac929917..fe7a1d27a017 100644
--- a/drivers/misc/eeprom/Kconfig
+++ b/drivers/misc/eeprom/Kconfig
@@ -111,4 +111,15 @@ config EEPROM_IDT_89HPESX
 	  This driver can also be built as a module. If so, the module
 	  will be called idt_89hpesx.
 
+config EEPROM_EE1004
+	tristate "SPD EEPROMs on DDR4 memory modules"
+	depends on I2C && SYSFS
+	help
+	  Enable this driver to get read support to SPD EEPROMs following
+	  the JEDEC EE1004 standard. These are typically found on DDR4
+	  SDRAM memory modules.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called ee1004.
+
 endmenu
diff --git a/drivers/misc/eeprom/Makefile b/drivers/misc/eeprom/Makefile
index 2aab60ef3e3e..a9b4b6579b75 100644
--- a/drivers/misc/eeprom/Makefile
+++ b/drivers/misc/eeprom/Makefile
@@ -7,3 +7,4 @@ obj-$(CONFIG_EEPROM_93CX6)	+= eeprom_93cx6.o
 obj-$(CONFIG_EEPROM_93XX46)	+= eeprom_93xx46.o
 obj-$(CONFIG_EEPROM_DIGSY_MTC_CFG) += digsy_mtc_eeprom.o
 obj-$(CONFIG_EEPROM_IDT_89HPESX) += idt_89hpesx.o
+obj-$(CONFIG_EEPROM_EE1004)	+= ee1004.o
diff --git a/drivers/misc/eeprom/at25.c b/drivers/misc/eeprom/at25.c
index 840afb398f9e..99de6939cd5a 100644
--- a/drivers/misc/eeprom/at25.c
+++ b/drivers/misc/eeprom/at25.c
@@ -366,7 +366,7 @@ static int at25_probe(struct spi_device *spi)
 	at25->nvmem_config.word_size = 1;
 	at25->nvmem_config.size = chip.byte_len;
 
-	at25->nvmem = nvmem_register(&at25->nvmem_config);
+	at25->nvmem = devm_nvmem_register(&spi->dev, &at25->nvmem_config);
 	if (IS_ERR(at25->nvmem))
 		return PTR_ERR(at25->nvmem);
 
@@ -379,16 +379,6 @@ static int at25_probe(struct spi_device *spi)
 	return 0;
 }
 
-static int at25_remove(struct spi_device *spi)
-{
-	struct at25_data	*at25;
-
-	at25 = spi_get_drvdata(spi);
-	nvmem_unregister(at25->nvmem);
-
-	return 0;
-}
-
 /*-------------------------------------------------------------------------*/
 
 static const struct of_device_id at25_of_match[] = {
@@ -403,7 +393,6 @@ static struct spi_driver at25_driver = {
 		.of_match_table = at25_of_match,
 	},
 	.probe		= at25_probe,
-	.remove		= at25_remove,
 };
 
 module_spi_driver(at25_driver);
diff --git a/drivers/misc/eeprom/ee1004.c b/drivers/misc/eeprom/ee1004.c
new file mode 100644
index 000000000000..276c1690ea1b
--- /dev/null
+++ b/drivers/misc/eeprom/ee1004.c
@@ -0,0 +1,281 @@
+/*
+ * ee1004 - driver for DDR4 SPD EEPROMs
+ *
+ * Copyright (C) 2017 Jean Delvare
+ *
+ * Based on the at24 driver:
+ * Copyright (C) 2005-2007 David Brownell
+ * Copyright (C) 2008 Wolfram Sang, Pengutronix
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/i2c.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/mod_devicetable.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+
+/*
+ * DDR4 memory modules use special EEPROMs following the Jedec EE1004
+ * specification. These are 512-byte EEPROMs using a single I2C address
+ * in the 0x50-0x57 range for data. One of two 256-byte page is selected
+ * by writing a command to I2C address 0x36 or 0x37 on the same I2C bus.
+ *
+ * Therefore we need to request these 2 additional addresses, and serialize
+ * access to all such EEPROMs with a single mutex.
+ *
+ * We assume it is safe to read up to 32 bytes at once from these EEPROMs.
+ * We use SMBus access even if I2C is available, these EEPROMs are small
+ * enough, and reading from them infrequent enough, that we favor simplicity
+ * over performance.
+ */
+
+#define EE1004_ADDR_SET_PAGE		0x36
+#define EE1004_EEPROM_SIZE		512
+#define EE1004_PAGE_SIZE		256
+#define EE1004_PAGE_SHIFT		8
+
+/*
+ * Mutex protects ee1004_set_page and ee1004_dev_count, and must be held
+ * from page selection to end of read.
+ */
+static DEFINE_MUTEX(ee1004_bus_lock);
+static struct i2c_client *ee1004_set_page[2];
+static unsigned int ee1004_dev_count;
+static int ee1004_current_page;
+
+static const struct i2c_device_id ee1004_ids[] = {
+	{ "ee1004", 0 },
+	{ }
+};
+MODULE_DEVICE_TABLE(i2c, ee1004_ids);
+
+/*-------------------------------------------------------------------------*/
+
+static ssize_t ee1004_eeprom_read(struct i2c_client *client, char *buf,
+				  unsigned int offset, size_t count)
+{
+	int status;
+
+	if (count > I2C_SMBUS_BLOCK_MAX)
+		count = I2C_SMBUS_BLOCK_MAX;
+	/* Can't cross page boundaries */
+	if (unlikely(offset + count > EE1004_PAGE_SIZE))
+		count = EE1004_PAGE_SIZE - offset;
+
+	status = i2c_smbus_read_i2c_block_data_or_emulated(client, offset,
+							   count, buf);
+	dev_dbg(&client->dev, "read %zu@%d --> %d\n", count, offset, status);
+
+	return status;
+}
+
+static ssize_t ee1004_read(struct file *filp, struct kobject *kobj,
+			   struct bin_attribute *bin_attr,
+			   char *buf, loff_t off, size_t count)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct i2c_client *client = to_i2c_client(dev);
+	size_t requested = count;
+	int page;
+
+	if (unlikely(!count))
+		return count;
+
+	page = off >> EE1004_PAGE_SHIFT;
+	if (unlikely(page > 1))
+		return 0;
+	off &= (1 << EE1004_PAGE_SHIFT) - 1;
+
+	/*
+	 * Read data from chip, protecting against concurrent access to
+	 * other EE1004 SPD EEPROMs on the same adapter.
+	 */
+	mutex_lock(&ee1004_bus_lock);
+
+	while (count) {
+		int status;
+
+		/* Select page */
+		if (page != ee1004_current_page) {
+			/* Data is ignored */
+			status = i2c_smbus_write_byte(ee1004_set_page[page],
+						      0x00);
+			if (status < 0) {
+				dev_err(dev, "Failed to select page %d (%d)\n",
+					page, status);
+				mutex_unlock(&ee1004_bus_lock);
+				return status;
+			}
+			dev_dbg(dev, "Selected page %d\n", page);
+			ee1004_current_page = page;
+		}
+
+		status = ee1004_eeprom_read(client, buf, off, count);
+		if (status < 0) {
+			mutex_unlock(&ee1004_bus_lock);
+			return status;
+		}
+		buf += status;
+		off += status;
+		count -= status;
+
+		if (off == EE1004_PAGE_SIZE) {
+			page++;
+			off = 0;
+		}
+	}
+
+	mutex_unlock(&ee1004_bus_lock);
+
+	return requested;
+}
+
+static const struct bin_attribute eeprom_attr = {
+	.attr = {
+		.name = "eeprom",
+		.mode = 0444,
+	},
+	.size = EE1004_EEPROM_SIZE,
+	.read = ee1004_read,
+};
+
+static int ee1004_probe(struct i2c_client *client,
+			const struct i2c_device_id *id)
+{
+	int err, cnr = 0;
+	const char *slow = NULL;
+
+	/* Make sure we can operate on this adapter */
+	if (!i2c_check_functionality(client->adapter,
+				     I2C_FUNC_SMBUS_READ_BYTE |
+				     I2C_FUNC_SMBUS_READ_I2C_BLOCK)) {
+		if (i2c_check_functionality(client->adapter,
+				     I2C_FUNC_SMBUS_READ_BYTE |
+				     I2C_FUNC_SMBUS_READ_WORD_DATA))
+			slow = "word";
+		else if (i2c_check_functionality(client->adapter,
+				     I2C_FUNC_SMBUS_READ_BYTE |
+				     I2C_FUNC_SMBUS_READ_BYTE_DATA))
+			slow = "byte";
+		else
+			return -EPFNOSUPPORT;
+	}
+
+	/* Use 2 dummy devices for page select command */
+	mutex_lock(&ee1004_bus_lock);
+	if (++ee1004_dev_count == 1) {
+		for (cnr = 0; cnr < 2; cnr++) {
+			ee1004_set_page[cnr] = i2c_new_dummy(client->adapter,
+						EE1004_ADDR_SET_PAGE + cnr);
+			if (!ee1004_set_page[cnr]) {
+				dev_err(&client->dev,
+					"address 0x%02x unavailable\n",
+					EE1004_ADDR_SET_PAGE + cnr);
+				err = -EADDRINUSE;
+				goto err_clients;
+			}
+		}
+	} else if (i2c_adapter_id(client->adapter) !=
+		   i2c_adapter_id(ee1004_set_page[0]->adapter)) {
+		dev_err(&client->dev,
+			"Driver only supports devices on a single I2C bus\n");
+		err = -EOPNOTSUPP;
+		goto err_clients;
+	}
+
+	/* Remember current page to avoid unneeded page select */
+	err = i2c_smbus_read_byte(ee1004_set_page[0]);
+	if (err == -ENXIO) {
+		/* Nack means page 1 is selected */
+		ee1004_current_page = 1;
+	} else if (err < 0) {
+		/* Anything else is a real error, bail out */
+		goto err_clients;
+	} else {
+		/* Ack means page 0 is selected, returned value meaningless */
+		ee1004_current_page = 0;
+	}
+	dev_dbg(&client->dev, "Currently selected page: %d\n",
+		ee1004_current_page);
+	mutex_unlock(&ee1004_bus_lock);
+
+	/* Create the sysfs eeprom file */
+	err = sysfs_create_bin_file(&client->dev.kobj, &eeprom_attr);
+	if (err)
+		goto err_clients_lock;
+
+	dev_info(&client->dev,
+		 "%u byte EE1004-compliant SPD EEPROM, read-only\n",
+		 EE1004_EEPROM_SIZE);
+	if (slow)
+		dev_notice(&client->dev,
+			   "Falling back to %s reads, performance will suffer\n",
+			   slow);
+
+	return 0;
+
+ err_clients_lock:
+	mutex_lock(&ee1004_bus_lock);
+ err_clients:
+	if (--ee1004_dev_count == 0) {
+		for (cnr--; cnr >= 0; cnr--) {
+			i2c_unregister_device(ee1004_set_page[cnr]);
+			ee1004_set_page[cnr] = NULL;
+		}
+	}
+	mutex_unlock(&ee1004_bus_lock);
+
+	return err;
+}
+
+static int ee1004_remove(struct i2c_client *client)
+{
+	int i;
+
+	sysfs_remove_bin_file(&client->dev.kobj, &eeprom_attr);
+
+	/* Remove page select clients if this is the last device */
+	mutex_lock(&ee1004_bus_lock);
+	if (--ee1004_dev_count == 0) {
+		for (i = 0; i < 2; i++) {
+			i2c_unregister_device(ee1004_set_page[i]);
+			ee1004_set_page[i] = NULL;
+		}
+	}
+	mutex_unlock(&ee1004_bus_lock);
+
+	return 0;
+}
+
+/*-------------------------------------------------------------------------*/
+
+static struct i2c_driver ee1004_driver = {
+	.driver = {
+		.name = "ee1004",
+	},
+	.probe = ee1004_probe,
+	.remove = ee1004_remove,
+	.id_table = ee1004_ids,
+};
+
+static int __init ee1004_init(void)
+{
+	return i2c_add_driver(&ee1004_driver);
+}
+module_init(ee1004_init);
+
+static void __exit ee1004_exit(void)
+{
+	i2c_del_driver(&ee1004_driver);
+}
+module_exit(ee1004_exit);
+
+MODULE_DESCRIPTION("Driver for EE1004-compliant DDR4 SPD EEPROMs");
+MODULE_AUTHOR("Jean Delvare");
+MODULE_LICENSE("GPL");
diff --git a/drivers/misc/eeprom/eeprom_93xx46.c b/drivers/misc/eeprom/eeprom_93xx46.c
index 38766968bfa2..c6dd9ad9bf7b 100644
--- a/drivers/misc/eeprom/eeprom_93xx46.c
+++ b/drivers/misc/eeprom/eeprom_93xx46.c
@@ -439,7 +439,7 @@ static int eeprom_93xx46_probe(struct spi_device *spi)
 		return -ENODEV;
 	}
 
-	edev = kzalloc(sizeof(*edev), GFP_KERNEL);
+	edev = devm_kzalloc(&spi->dev, sizeof(*edev), GFP_KERNEL);
 	if (!edev)
 		return -ENOMEM;
 
@@ -449,8 +449,7 @@ static int eeprom_93xx46_probe(struct spi_device *spi)
 		edev->addrlen = 6;
 	else {
 		dev_err(&spi->dev, "unspecified address type\n");
-		err = -EINVAL;
-		goto fail;
+		return -EINVAL;
 	}
 
 	mutex_init(&edev->lock);
@@ -473,11 +472,9 @@ static int eeprom_93xx46_probe(struct spi_device *spi)
 	edev->nvmem_config.word_size = 1;
 	edev->nvmem_config.size = edev->size;
 
-	edev->nvmem = nvmem_register(&edev->nvmem_config);
-	if (IS_ERR(edev->nvmem)) {
-		err = PTR_ERR(edev->nvmem);
-		goto fail;
-	}
+	edev->nvmem = devm_nvmem_register(&spi->dev, &edev->nvmem_config);
+	if (IS_ERR(edev->nvmem))
+		return PTR_ERR(edev->nvmem);
 
 	dev_info(&spi->dev, "%d-bit eeprom %s\n",
 		(pd->flags & EE_ADDR8) ? 8 : 16,
@@ -490,21 +487,15 @@ static int eeprom_93xx46_probe(struct spi_device *spi)
 
 	spi_set_drvdata(spi, edev);
 	return 0;
-fail:
-	kfree(edev);
-	return err;
 }
 
 static int eeprom_93xx46_remove(struct spi_device *spi)
 {
 	struct eeprom_93xx46_dev *edev = spi_get_drvdata(spi);
 
-	nvmem_unregister(edev->nvmem);
-
 	if (!(edev->pdata->flags & EE_READONLY))
 		device_remove_file(&spi->dev, &dev_attr_erase);
 
-	kfree(edev);
 	return 0;
 }
 
diff --git a/drivers/misc/genwqe/card_base.c b/drivers/misc/genwqe/card_base.c
index c7cd3675bcd1..d137d0fab9bf 100644
--- a/drivers/misc/genwqe/card_base.c
+++ b/drivers/misc/genwqe/card_base.c
@@ -24,7 +24,6 @@
  * controlled from here.
  */
 
-#include <linux/module.h>
 #include <linux/types.h>
 #include <linux/pci.h>
 #include <linux/err.h>
diff --git a/drivers/misc/genwqe/card_ddcb.c b/drivers/misc/genwqe/card_ddcb.c
index 656449cb4476..9a65bd9d6152 100644
--- a/drivers/misc/genwqe/card_ddcb.c
+++ b/drivers/misc/genwqe/card_ddcb.c
@@ -27,7 +27,6 @@
  */
 
 #include <linux/types.h>
-#include <linux/module.h>
 #include <linux/sched.h>
 #include <linux/wait.h>
 #include <linux/pci.h>
diff --git a/drivers/misc/genwqe/card_utils.c b/drivers/misc/genwqe/card_utils.c
index 8679e0bd8ec2..3fcb9a2fe1c9 100644
--- a/drivers/misc/genwqe/card_utils.c
+++ b/drivers/misc/genwqe/card_utils.c
@@ -23,14 +23,12 @@
  */
 
 #include <linux/kernel.h>
-#include <linux/dma-mapping.h>
 #include <linux/sched.h>
 #include <linux/vmalloc.h>
 #include <linux/page-flags.h>
 #include <linux/scatterlist.h>
 #include <linux/hugetlb.h>
 #include <linux/iommu.h>
-#include <linux/delay.h>
 #include <linux/pci.h>
 #include <linux/dma-mapping.h>
 #include <linux/ctype.h>
@@ -298,7 +296,7 @@ static int genwqe_sgl_size(int num_pages)
 int genwqe_alloc_sync_sgl(struct genwqe_dev *cd, struct genwqe_sgl *sgl,
 			  void __user *user_addr, size_t user_size, int write)
 {
-	int rc;
+	int ret = -ENOMEM;
 	struct pci_dev *pci_dev = cd->pci_dev;
 
 	sgl->fpage_offs = offset_in_page((unsigned long)user_addr);
@@ -318,7 +316,7 @@ int genwqe_alloc_sync_sgl(struct genwqe_dev *cd, struct genwqe_sgl *sgl,
 	if (get_order(sgl->sgl_size) > MAX_ORDER) {
 		dev_err(&pci_dev->dev,
 			"[%s] err: too much memory requested!\n", __func__);
-		return -ENOMEM;
+		return ret;
 	}
 
 	sgl->sgl = __genwqe_alloc_consistent(cd, sgl->sgl_size,
@@ -326,7 +324,7 @@ int genwqe_alloc_sync_sgl(struct genwqe_dev *cd, struct genwqe_sgl *sgl,
 	if (sgl->sgl == NULL) {
 		dev_err(&pci_dev->dev,
 			"[%s] err: no memory available!\n", __func__);
-		return -ENOMEM;
+		return ret;
 	}
 
 	/* Only use buffering on incomplete pages */
@@ -339,7 +337,7 @@ int genwqe_alloc_sync_sgl(struct genwqe_dev *cd, struct genwqe_sgl *sgl,
 		/* Sync with user memory */
 		if (copy_from_user(sgl->fpage + sgl->fpage_offs,
 				   user_addr, sgl->fpage_size)) {
-			rc = -EFAULT;
+			ret = -EFAULT;
 			goto err_out;
 		}
 	}
@@ -352,7 +350,7 @@ int genwqe_alloc_sync_sgl(struct genwqe_dev *cd, struct genwqe_sgl *sgl,
 		/* Sync with user memory */
 		if (copy_from_user(sgl->lpage, user_addr + user_size -
 				   sgl->lpage_size, sgl->lpage_size)) {
-			rc = -EFAULT;
+			ret = -EFAULT;
 			goto err_out2;
 		}
 	}
@@ -374,7 +372,8 @@ int genwqe_alloc_sync_sgl(struct genwqe_dev *cd, struct genwqe_sgl *sgl,
 	sgl->sgl = NULL;
 	sgl->sgl_dma_addr = 0;
 	sgl->sgl_size = 0;
-	return -ENOMEM;
+
+	return ret;
 }
 
 int genwqe_setup_sgl(struct genwqe_dev *cd, struct genwqe_sgl *sgl,
diff --git a/drivers/misc/kgdbts.c b/drivers/misc/kgdbts.c
index 6193270e7b3d..de20bdaa148d 100644
--- a/drivers/misc/kgdbts.c
+++ b/drivers/misc/kgdbts.c
@@ -985,6 +985,12 @@ static void kgdbts_run_tests(void)
 	int nmi_sleep = 0;
 	int i;
 
+	verbose = 0;
+	if (strstr(config, "V1"))
+		verbose = 1;
+	if (strstr(config, "V2"))
+		verbose = 2;
+
 	ptr = strchr(config, 'F');
 	if (ptr)
 		fork_test = simple_strtol(ptr + 1, NULL, 10);
@@ -1068,13 +1074,6 @@ static int kgdbts_option_setup(char *opt)
 		return -ENOSPC;
 	}
 	strcpy(config, opt);
-
-	verbose = 0;
-	if (strstr(config, "V1"))
-		verbose = 1;
-	if (strstr(config, "V2"))
-		verbose = 2;
-
 	return 0;
 }
 
@@ -1086,9 +1085,6 @@ static int configure_kgdbts(void)
 
 	if (!strlen(config) || isspace(config[0]))
 		goto noconfig;
-	err = kgdbts_option_setup(config);
-	if (err)
-		goto noconfig;
 
 	final_ack = 0;
 	run_plant_and_detach_test(1);
diff --git a/drivers/misc/lkdtm/usercopy.c b/drivers/misc/lkdtm/usercopy.c
index 389475b25bb7..d5a0e7f1813b 100644
--- a/drivers/misc/lkdtm/usercopy.c
+++ b/drivers/misc/lkdtm/usercopy.c
@@ -18,7 +18,7 @@
  * hardened usercopy checks by added "unconst" to all the const copies,
  * and making sure "cache_size" isn't optimized into a const.
  */
-static volatile size_t unconst = 0;
+static volatile size_t unconst;
 static volatile size_t cache_size = 1024;
 static struct kmem_cache *whitelist_cache;
 
diff --git a/drivers/misc/mei/bus-fixup.c b/drivers/misc/mei/bus-fixup.c
index a6f41f96f2a1..80215c312f0e 100644
--- a/drivers/misc/mei/bus-fixup.c
+++ b/drivers/misc/mei/bus-fixup.c
@@ -17,7 +17,6 @@
 #include <linux/kernel.h>
 #include <linux/sched.h>
 #include <linux/module.h>
-#include <linux/moduleparam.h>
 #include <linux/device.h>
 #include <linux/slab.h>
 #include <linux/uuid.h>
diff --git a/drivers/misc/mei/main.c b/drivers/misc/mei/main.c
index 4d77a6ae183a..87281b3695e6 100644
--- a/drivers/misc/mei/main.c
+++ b/drivers/misc/mei/main.c
@@ -599,10 +599,10 @@ static __poll_t mei_poll(struct file *file, poll_table *wait)
 			mei_cl_read_start(cl, mei_cl_mtu(cl), file);
 	}
 
-	if (req_events & (POLLOUT | POLLWRNORM)) {
+	if (req_events & (EPOLLOUT | EPOLLWRNORM)) {
 		poll_wait(file, &cl->tx_wait, wait);
 		if (cl->tx_cb_queued < dev->tx_queue_limit)
-			mask |= POLLOUT | POLLWRNORM;
+			mask |= EPOLLOUT | EPOLLWRNORM;
 	}
 
 out:
diff --git a/drivers/misc/mic/scif/scif_dma.c b/drivers/misc/mic/scif/scif_dma.c
index 6369aeaa7056..18b8ed57c4ac 100644
--- a/drivers/misc/mic/scif/scif_dma.c
+++ b/drivers/misc/mic/scif/scif_dma.c
@@ -1035,8 +1035,6 @@ scif_rma_list_dma_copy_unaligned(struct scif_copy_work *work,
 			}
 			dma_async_issue_pending(chan);
 		}
-		if (ret < 0)
-			goto err;
 		offset += loop_len;
 		temp += loop_len;
 		temp_phys += loop_len;
@@ -1553,9 +1551,8 @@ static int scif_rma_list_dma_copy_wrapper(struct scif_endpt *epd,
 	int src_cache_off, dst_cache_off;
 	s64 src_offset = work->src_offset, dst_offset = work->dst_offset;
 	u8 *temp = NULL;
-	bool src_local = true, dst_local = false;
+	bool src_local = true;
 	struct scif_dma_comp_cb *comp_cb;
-	dma_addr_t src_dma_addr, dst_dma_addr;
 	int err;
 
 	if (is_dma_copy_aligned(chan->device, 1, 1, 1))
@@ -1569,12 +1566,8 @@ static int scif_rma_list_dma_copy_wrapper(struct scif_endpt *epd,
 
 	if (work->loopback)
 		return scif_rma_list_cpu_copy(work);
-	src_dma_addr = __scif_off_to_dma_addr(work->src_window, src_offset);
-	dst_dma_addr = __scif_off_to_dma_addr(work->dst_window, dst_offset);
 	src_local = work->src_window->type == SCIF_WINDOW_SELF;
-	dst_local = work->dst_window->type == SCIF_WINDOW_SELF;
 
-	dst_local = dst_local;
 	/* Allocate dma_completion cb */
 	comp_cb = kzalloc(sizeof(*comp_cb), GFP_KERNEL);
 	if (!comp_cb)
diff --git a/drivers/misc/mic/scif/scif_fence.c b/drivers/misc/mic/scif/scif_fence.c
index cac3bcc308a7..7bb929f05d85 100644
--- a/drivers/misc/mic/scif/scif_fence.c
+++ b/drivers/misc/mic/scif/scif_fence.c
@@ -272,7 +272,7 @@ static int _scif_prog_signal(scif_epd_t epd, dma_addr_t dst, u64 val)
 dma_fail:
 	if (!x100)
 		dma_pool_free(ep->remote_dev->signal_pool, status,
-			      status->src_dma_addr);
+			      src - offsetof(struct scif_status, val));
 alloc_fail:
 	return err;
 }
diff --git a/drivers/misc/sgi-gru/grukservices.c b/drivers/misc/sgi-gru/grukservices.c
index 030769018461..4b23d586fc3f 100644
--- a/drivers/misc/sgi-gru/grukservices.c
+++ b/drivers/misc/sgi-gru/grukservices.c
@@ -634,7 +634,7 @@ static int send_noop_message(void *cb, struct gru_message_queue_desc *mqd,
 			break;
 		case CBSS_PAGE_OVERFLOW:
 			STAT(mesq_noop_page_overflow);
-			/* fallthru */
+			/* fall through */
 		default:
 			BUG();
 		}
@@ -792,7 +792,7 @@ static int send_message_failure(void *cb, struct gru_message_queue_desc *mqd,
 		break;
 	case CBSS_PAGE_OVERFLOW:
 		STAT(mesq_page_overflow);
-		/* fallthru */
+		/* fall through */
 	default:
 		BUG();
 	}
diff --git a/drivers/misc/sgi-xp/xpc_channel.c b/drivers/misc/sgi-xp/xpc_channel.c
index 05a890ce2ab8..8e6607fc8a67 100644
--- a/drivers/misc/sgi-xp/xpc_channel.c
+++ b/drivers/misc/sgi-xp/xpc_channel.c
@@ -28,7 +28,7 @@ xpc_process_connect(struct xpc_channel *ch, unsigned long *irq_flags)
 {
 	enum xp_retval ret;
 
-	DBUG_ON(!spin_is_locked(&ch->lock));
+	lockdep_assert_held(&ch->lock);
 
 	if (!(ch->flags & XPC_C_OPENREQUEST) ||
 	    !(ch->flags & XPC_C_ROPENREQUEST)) {
@@ -82,7 +82,7 @@ xpc_process_disconnect(struct xpc_channel *ch, unsigned long *irq_flags)
 	struct xpc_partition *part = &xpc_partitions[ch->partid];
 	u32 channel_was_connected = (ch->flags & XPC_C_WASCONNECTED);
 
-	DBUG_ON(!spin_is_locked(&ch->lock));
+	lockdep_assert_held(&ch->lock);
 
 	if (!(ch->flags & XPC_C_DISCONNECTING))
 		return;
@@ -755,7 +755,7 @@ xpc_disconnect_channel(const int line, struct xpc_channel *ch,
 {
 	u32 channel_was_connected = (ch->flags & XPC_C_CONNECTED);
 
-	DBUG_ON(!spin_is_locked(&ch->lock));
+	lockdep_assert_held(&ch->lock);
 
 	if (ch->flags & (XPC_C_DISCONNECTING | XPC_C_DISCONNECTED))
 		return;
diff --git a/drivers/misc/sgi-xp/xpc_partition.c b/drivers/misc/sgi-xp/xpc_partition.c
index 0c3ef6f1df54..3eba1c420cc0 100644
--- a/drivers/misc/sgi-xp/xpc_partition.c
+++ b/drivers/misc/sgi-xp/xpc_partition.c
@@ -98,8 +98,7 @@ xpc_get_rsvd_page_pa(int nasid)
 			len = L1_CACHE_ALIGN(len);
 
 		if (len > buf_len) {
-			if (buf_base != NULL)
-				kfree(buf_base);
+			kfree(buf_base);
 			buf_len = L1_CACHE_ALIGN(len);
 			buf = xpc_kmalloc_cacheline_aligned(buf_len, GFP_KERNEL,
 							    &buf_base);
diff --git a/drivers/misc/sgi-xp/xpc_sn2.c b/drivers/misc/sgi-xp/xpc_sn2.c
index 5a12d2a54049..0ae69b9390ce 100644
--- a/drivers/misc/sgi-xp/xpc_sn2.c
+++ b/drivers/misc/sgi-xp/xpc_sn2.c
@@ -1671,7 +1671,7 @@ xpc_teardown_msg_structures_sn2(struct xpc_channel *ch)
 {
 	struct xpc_channel_sn2 *ch_sn2 = &ch->sn.sn2;
 
-	DBUG_ON(!spin_is_locked(&ch->lock));
+	lockdep_assert_held(&ch->lock);
 
 	ch_sn2->remote_msgqueue_pa = 0;
 
diff --git a/drivers/misc/sgi-xp/xpc_uv.c b/drivers/misc/sgi-xp/xpc_uv.c
index 340b44d9e8cf..0441abe87880 100644
--- a/drivers/misc/sgi-xp/xpc_uv.c
+++ b/drivers/misc/sgi-xp/xpc_uv.c
@@ -1183,7 +1183,7 @@ xpc_teardown_msg_structures_uv(struct xpc_channel *ch)
 {
 	struct xpc_channel_uv *ch_uv = &ch->sn.uv;
 
-	DBUG_ON(!spin_is_locked(&ch->lock));
+	lockdep_assert_held(&ch->lock);
 
 	kfree(ch_uv->cached_notify_gru_mq_desc);
 	ch_uv->cached_notify_gru_mq_desc = NULL;
diff --git a/drivers/misc/sram.c b/drivers/misc/sram.c
index 74b183baf044..80d8cbe8c01a 100644
--- a/drivers/misc/sram.c
+++ b/drivers/misc/sram.c
@@ -323,10 +323,8 @@ static int sram_reserve_regions(struct sram_dev *sram, struct resource *res)
 		cur_start = block->start + block->size;
 	}
 
- err_chunks:
-	if (child)
-		of_node_put(child);
-
+err_chunks:
+	of_node_put(child);
 	kfree(rblocks);
 
 	return ret;
diff --git a/drivers/misc/vmw_balloon.c b/drivers/misc/vmw_balloon.c
index 2543ef1ece17..9b0b3fa4f836 100644
--- a/drivers/misc/vmw_balloon.c
+++ b/drivers/misc/vmw_balloon.c
@@ -25,6 +25,9 @@
 #include <linux/workqueue.h>
 #include <linux/debugfs.h>
 #include <linux/seq_file.h>
+#include <linux/rwsem.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
 #include <linux/vmw_vmci_defs.h>
 #include <linux/vmw_vmci_api.h>
 #include <asm/hypervisor.h>
@@ -37,20 +40,20 @@ MODULE_ALIAS("vmware_vmmemctl");
 MODULE_LICENSE("GPL");
 
 /*
- * Use __GFP_HIGHMEM to allow pages from HIGHMEM zone. We don't
- * allow wait (__GFP_RECLAIM) for NOSLEEP page allocations. Use
- * __GFP_NOWARN, to suppress page allocation failure warnings.
+ * Use __GFP_HIGHMEM to allow pages from HIGHMEM zone. We don't allow wait
+ * (__GFP_RECLAIM) for huge page allocations. Use __GFP_NOWARN, to suppress page
+ * allocation failure warnings. Disallow access to emergency low-memory pools.
  */
-#define VMW_PAGE_ALLOC_NOSLEEP		(__GFP_HIGHMEM|__GFP_NOWARN)
+#define VMW_HUGE_PAGE_ALLOC_FLAGS	(__GFP_HIGHMEM|__GFP_NOWARN|	\
+					 __GFP_NOMEMALLOC)
 
 /*
- * Use GFP_HIGHUSER when executing in a separate kernel thread
- * context and allocation can sleep.  This is less stressful to
- * the guest memory system, since it allows the thread to block
- * while memory is reclaimed, and won't take pages from emergency
- * low-memory pools.
+ * Use __GFP_HIGHMEM to allow pages from HIGHMEM zone. We allow lightweight
+ * reclamation (__GFP_NORETRY). Use __GFP_NOWARN, to suppress page allocation
+ * failure warnings. Disallow access to emergency low-memory pools.
  */
-#define VMW_PAGE_ALLOC_CANSLEEP		(GFP_HIGHUSER)
+#define VMW_PAGE_ALLOC_FLAGS		(__GFP_HIGHMEM|__GFP_NOWARN|	\
+					 __GFP_NOMEMALLOC|__GFP_NORETRY)
 
 /* Maximum number of refused pages we accumulate during inflation cycle */
 #define VMW_BALLOON_MAX_REFUSED		16
@@ -77,225 +80,420 @@ enum vmwballoon_capabilities {
 					| VMW_BALLOON_BATCHED_2M_CMDS \
 					| VMW_BALLOON_SIGNALLED_WAKEUP_CMD)
 
-#define VMW_BALLOON_2M_SHIFT		(9)
-#define VMW_BALLOON_NUM_PAGE_SIZES	(2)
+#define VMW_BALLOON_2M_ORDER		(PMD_SHIFT - PAGE_SHIFT)
 
-/*
- * Backdoor commands availability:
- *
- * START, GET_TARGET and GUEST_ID are always available,
- *
- * VMW_BALLOON_BASIC_CMDS:
- *	LOCK and UNLOCK commands,
- * VMW_BALLOON_BATCHED_CMDS:
- *	BATCHED_LOCK and BATCHED_UNLOCK commands.
- * VMW BALLOON_BATCHED_2M_CMDS:
- *	BATCHED_2M_LOCK and BATCHED_2M_UNLOCK commands,
- * VMW VMW_BALLOON_SIGNALLED_WAKEUP_CMD:
- *	VMW_BALLOON_CMD_VMCI_DOORBELL_SET command.
- */
-#define VMW_BALLOON_CMD_START			0
-#define VMW_BALLOON_CMD_GET_TARGET		1
-#define VMW_BALLOON_CMD_LOCK			2
-#define VMW_BALLOON_CMD_UNLOCK			3
-#define VMW_BALLOON_CMD_GUEST_ID		4
-#define VMW_BALLOON_CMD_BATCHED_LOCK		6
-#define VMW_BALLOON_CMD_BATCHED_UNLOCK		7
-#define VMW_BALLOON_CMD_BATCHED_2M_LOCK		8
-#define VMW_BALLOON_CMD_BATCHED_2M_UNLOCK	9
-#define VMW_BALLOON_CMD_VMCI_DOORBELL_SET	10
-
-
-/* error codes */
-#define VMW_BALLOON_SUCCESS		        0
-#define VMW_BALLOON_FAILURE		        -1
-#define VMW_BALLOON_ERROR_CMD_INVALID	        1
-#define VMW_BALLOON_ERROR_PPN_INVALID	        2
-#define VMW_BALLOON_ERROR_PPN_LOCKED	        3
-#define VMW_BALLOON_ERROR_PPN_UNLOCKED	        4
-#define VMW_BALLOON_ERROR_PPN_PINNED	        5
-#define VMW_BALLOON_ERROR_PPN_NOTNEEDED	        6
-#define VMW_BALLOON_ERROR_RESET		        7
-#define VMW_BALLOON_ERROR_BUSY		        8
+enum vmballoon_page_size_type {
+	VMW_BALLOON_4K_PAGE,
+	VMW_BALLOON_2M_PAGE,
+	VMW_BALLOON_LAST_SIZE = VMW_BALLOON_2M_PAGE
+};
 
-#define VMW_BALLOON_SUCCESS_WITH_CAPABILITIES	(0x03000000)
+#define VMW_BALLOON_NUM_PAGE_SIZES	(VMW_BALLOON_LAST_SIZE + 1)
 
-/* Batch page description */
+static const char * const vmballoon_page_size_names[] = {
+	[VMW_BALLOON_4K_PAGE]			= "4k",
+	[VMW_BALLOON_2M_PAGE]			= "2M"
+};
 
-/*
- * Layout of a page in the batch page:
+enum vmballoon_op {
+	VMW_BALLOON_INFLATE,
+	VMW_BALLOON_DEFLATE
+};
+
+enum vmballoon_op_stat_type {
+	VMW_BALLOON_OP_STAT,
+	VMW_BALLOON_OP_FAIL_STAT
+};
+
+#define VMW_BALLOON_OP_STAT_TYPES	(VMW_BALLOON_OP_FAIL_STAT + 1)
+
+/**
+ * enum vmballoon_cmd_type - backdoor commands.
  *
- * +-------------+----------+--------+
- * |             |          |        |
- * | Page number | Reserved | Status |
- * |             |          |        |
- * +-------------+----------+--------+
- * 64  PAGE_SHIFT          6         0
+ * Availability of the commands is as followed:
  *
- * The reserved field should be set to 0.
+ * %VMW_BALLOON_CMD_START, %VMW_BALLOON_CMD_GET_TARGET and
+ * %VMW_BALLOON_CMD_GUEST_ID are always available.
+ *
+ * If the host reports %VMW_BALLOON_BASIC_CMDS are supported then
+ * %VMW_BALLOON_CMD_LOCK and %VMW_BALLOON_CMD_UNLOCK commands are available.
+ *
+ * If the host reports %VMW_BALLOON_BATCHED_CMDS are supported then
+ * %VMW_BALLOON_CMD_BATCHED_LOCK and VMW_BALLOON_CMD_BATCHED_UNLOCK commands
+ * are available.
+ *
+ * If the host reports %VMW_BALLOON_BATCHED_2M_CMDS are supported then
+ * %VMW_BALLOON_CMD_BATCHED_2M_LOCK and %VMW_BALLOON_CMD_BATCHED_2M_UNLOCK
+ * are supported.
+ *
+ * If the host reports  VMW_BALLOON_SIGNALLED_WAKEUP_CMD is supported then
+ * VMW_BALLOON_CMD_VMCI_DOORBELL_SET command is supported.
+ *
+ * @VMW_BALLOON_CMD_START: Communicating supported version with the hypervisor.
+ * @VMW_BALLOON_CMD_GET_TARGET: Gets the balloon target size.
+ * @VMW_BALLOON_CMD_LOCK: Informs the hypervisor about a ballooned page.
+ * @VMW_BALLOON_CMD_UNLOCK: Informs the hypervisor about a page that is about
+ *			    to be deflated from the balloon.
+ * @VMW_BALLOON_CMD_GUEST_ID: Informs the hypervisor about the type of OS that
+ *			      runs in the VM.
+ * @VMW_BALLOON_CMD_BATCHED_LOCK: Inform the hypervisor about a batch of
+ *				  ballooned pages (up to 512).
+ * @VMW_BALLOON_CMD_BATCHED_UNLOCK: Inform the hypervisor about a batch of
+ *				  pages that are about to be deflated from the
+ *				  balloon (up to 512).
+ * @VMW_BALLOON_CMD_BATCHED_2M_LOCK: Similar to @VMW_BALLOON_CMD_BATCHED_LOCK
+ *				     for 2MB pages.
+ * @VMW_BALLOON_CMD_BATCHED_2M_UNLOCK: Similar to
+ *				       @VMW_BALLOON_CMD_BATCHED_UNLOCK for 2MB
+ *				       pages.
+ * @VMW_BALLOON_CMD_VMCI_DOORBELL_SET: A command to set doorbell notification
+ *				       that would be invoked when the balloon
+ *				       size changes.
+ * @VMW_BALLOON_CMD_LAST: Value of the last command.
  */
-#define VMW_BALLOON_BATCH_MAX_PAGES	(PAGE_SIZE / sizeof(u64))
-#define VMW_BALLOON_BATCH_STATUS_MASK	((1UL << 5) - 1)
-#define VMW_BALLOON_BATCH_PAGE_MASK	(~((1UL << PAGE_SHIFT) - 1))
-
-struct vmballoon_batch_page {
-	u64 pages[VMW_BALLOON_BATCH_MAX_PAGES];
+enum vmballoon_cmd_type {
+	VMW_BALLOON_CMD_START,
+	VMW_BALLOON_CMD_GET_TARGET,
+	VMW_BALLOON_CMD_LOCK,
+	VMW_BALLOON_CMD_UNLOCK,
+	VMW_BALLOON_CMD_GUEST_ID,
+	/* No command 5 */
+	VMW_BALLOON_CMD_BATCHED_LOCK = 6,
+	VMW_BALLOON_CMD_BATCHED_UNLOCK,
+	VMW_BALLOON_CMD_BATCHED_2M_LOCK,
+	VMW_BALLOON_CMD_BATCHED_2M_UNLOCK,
+	VMW_BALLOON_CMD_VMCI_DOORBELL_SET,
+	VMW_BALLOON_CMD_LAST = VMW_BALLOON_CMD_VMCI_DOORBELL_SET,
 };
 
-static u64 vmballoon_batch_get_pa(struct vmballoon_batch_page *batch, int idx)
-{
-	return batch->pages[idx] & VMW_BALLOON_BATCH_PAGE_MASK;
-}
+#define VMW_BALLOON_CMD_NUM	(VMW_BALLOON_CMD_LAST + 1)
+
+enum vmballoon_error_codes {
+	VMW_BALLOON_SUCCESS,
+	VMW_BALLOON_ERROR_CMD_INVALID,
+	VMW_BALLOON_ERROR_PPN_INVALID,
+	VMW_BALLOON_ERROR_PPN_LOCKED,
+	VMW_BALLOON_ERROR_PPN_UNLOCKED,
+	VMW_BALLOON_ERROR_PPN_PINNED,
+	VMW_BALLOON_ERROR_PPN_NOTNEEDED,
+	VMW_BALLOON_ERROR_RESET,
+	VMW_BALLOON_ERROR_BUSY
+};
 
-static int vmballoon_batch_get_status(struct vmballoon_batch_page *batch,
-				int idx)
-{
-	return (int)(batch->pages[idx] & VMW_BALLOON_BATCH_STATUS_MASK);
-}
+#define VMW_BALLOON_SUCCESS_WITH_CAPABILITIES	(0x03000000)
 
-static void vmballoon_batch_set_pa(struct vmballoon_batch_page *batch, int idx,
-				u64 pa)
-{
-	batch->pages[idx] = pa;
-}
+#define VMW_BALLOON_CMD_WITH_TARGET_MASK			\
+	((1UL << VMW_BALLOON_CMD_GET_TARGET)		|	\
+	 (1UL << VMW_BALLOON_CMD_LOCK)			|	\
+	 (1UL << VMW_BALLOON_CMD_UNLOCK)		|	\
+	 (1UL << VMW_BALLOON_CMD_BATCHED_LOCK)		|	\
+	 (1UL << VMW_BALLOON_CMD_BATCHED_UNLOCK)	|	\
+	 (1UL << VMW_BALLOON_CMD_BATCHED_2M_LOCK)	|	\
+	 (1UL << VMW_BALLOON_CMD_BATCHED_2M_UNLOCK))
+
+static const char * const vmballoon_cmd_names[] = {
+	[VMW_BALLOON_CMD_START]			= "start",
+	[VMW_BALLOON_CMD_GET_TARGET]		= "target",
+	[VMW_BALLOON_CMD_LOCK]			= "lock",
+	[VMW_BALLOON_CMD_UNLOCK]		= "unlock",
+	[VMW_BALLOON_CMD_GUEST_ID]		= "guestType",
+	[VMW_BALLOON_CMD_BATCHED_LOCK]		= "batchLock",
+	[VMW_BALLOON_CMD_BATCHED_UNLOCK]	= "batchUnlock",
+	[VMW_BALLOON_CMD_BATCHED_2M_LOCK]	= "2m-lock",
+	[VMW_BALLOON_CMD_BATCHED_2M_UNLOCK]	= "2m-unlock",
+	[VMW_BALLOON_CMD_VMCI_DOORBELL_SET]	= "doorbellSet"
+};
 
+enum vmballoon_stat_page {
+	VMW_BALLOON_PAGE_STAT_ALLOC,
+	VMW_BALLOON_PAGE_STAT_ALLOC_FAIL,
+	VMW_BALLOON_PAGE_STAT_REFUSED_ALLOC,
+	VMW_BALLOON_PAGE_STAT_REFUSED_FREE,
+	VMW_BALLOON_PAGE_STAT_FREE,
+	VMW_BALLOON_PAGE_STAT_LAST = VMW_BALLOON_PAGE_STAT_FREE
+};
 
-#define VMWARE_BALLOON_CMD(cmd, arg1, arg2, result)		\
-({								\
-	unsigned long __status, __dummy1, __dummy2, __dummy3;	\
-	__asm__ __volatile__ ("inl %%dx" :			\
-		"=a"(__status),					\
-		"=c"(__dummy1),					\
-		"=d"(__dummy2),					\
-		"=b"(result),					\
-		"=S" (__dummy3) :				\
-		"0"(VMW_BALLOON_HV_MAGIC),			\
-		"1"(VMW_BALLOON_CMD_##cmd),			\
-		"2"(VMW_BALLOON_HV_PORT),			\
-		"3"(arg1),					\
-		"4" (arg2) :					\
-		"memory");					\
-	if (VMW_BALLOON_CMD_##cmd == VMW_BALLOON_CMD_START)	\
-		result = __dummy1;				\
-	result &= -1UL;						\
-	__status & -1UL;					\
-})
+#define VMW_BALLOON_PAGE_STAT_NUM	(VMW_BALLOON_PAGE_STAT_LAST + 1)
 
-#ifdef CONFIG_DEBUG_FS
-struct vmballoon_stats {
-	unsigned int timer;
-	unsigned int doorbell;
-
-	/* allocation statistics */
-	unsigned int alloc[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int alloc_fail[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int sleep_alloc;
-	unsigned int sleep_alloc_fail;
-	unsigned int refused_alloc[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int refused_free[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int free[VMW_BALLOON_NUM_PAGE_SIZES];
-
-	/* monitor operations */
-	unsigned int lock[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int lock_fail[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int unlock[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int unlock_fail[VMW_BALLOON_NUM_PAGE_SIZES];
-	unsigned int target;
-	unsigned int target_fail;
-	unsigned int start;
-	unsigned int start_fail;
-	unsigned int guest_type;
-	unsigned int guest_type_fail;
-	unsigned int doorbell_set;
-	unsigned int doorbell_unset;
+enum vmballoon_stat_general {
+	VMW_BALLOON_STAT_TIMER,
+	VMW_BALLOON_STAT_DOORBELL,
+	VMW_BALLOON_STAT_RESET,
+	VMW_BALLOON_STAT_LAST = VMW_BALLOON_STAT_RESET
 };
 
-#define STATS_INC(stat) (stat)++
-#else
-#define STATS_INC(stat)
-#endif
+#define VMW_BALLOON_STAT_NUM		(VMW_BALLOON_STAT_LAST + 1)
 
-struct vmballoon;
 
-struct vmballoon_ops {
-	void (*add_page)(struct vmballoon *b, int idx, struct page *p);
-	int (*lock)(struct vmballoon *b, unsigned int num_pages,
-			bool is_2m_pages, unsigned int *target);
-	int (*unlock)(struct vmballoon *b, unsigned int num_pages,
-			bool is_2m_pages, unsigned int *target);
+static DEFINE_STATIC_KEY_TRUE(vmw_balloon_batching);
+static DEFINE_STATIC_KEY_FALSE(balloon_stat_enabled);
+
+struct vmballoon_ctl {
+	struct list_head pages;
+	struct list_head refused_pages;
+	unsigned int n_refused_pages;
+	unsigned int n_pages;
+	enum vmballoon_page_size_type page_size;
+	enum vmballoon_op op;
 };
 
 struct vmballoon_page_size {
 	/* list of reserved physical pages */
 	struct list_head pages;
-
-	/* transient list of non-balloonable pages */
-	struct list_head refused_pages;
-	unsigned int n_refused_pages;
 };
 
+/**
+ * struct vmballoon_batch_entry - a batch entry for lock or unlock.
+ *
+ * @status: the status of the operation, which is written by the hypervisor.
+ * @reserved: reserved for future use. Must be set to zero.
+ * @pfn: the physical frame number of the page to be locked or unlocked.
+ */
+struct vmballoon_batch_entry {
+	u64 status : 5;
+	u64 reserved : PAGE_SHIFT - 5;
+	u64 pfn : 52;
+} __packed;
+
 struct vmballoon {
 	struct vmballoon_page_size page_sizes[VMW_BALLOON_NUM_PAGE_SIZES];
 
-	/* supported page sizes. 1 == 4k pages only, 2 == 4k and 2m pages */
-	unsigned supported_page_sizes;
+	/**
+	 * @max_page_size: maximum supported page size for ballooning.
+	 *
+	 * Protected by @conf_sem
+	 */
+	enum vmballoon_page_size_type max_page_size;
+
+	/**
+	 * @size: balloon actual size in basic page size (frames).
+	 *
+	 * While we currently do not support size which is bigger than 32-bit,
+	 * in preparation for future support, use 64-bits.
+	 */
+	atomic64_t size;
 
-	/* balloon size in pages */
-	unsigned int size;
-	unsigned int target;
+	/**
+	 * @target: balloon target size in basic page size (frames).
+	 *
+	 * We do not protect the target under the assumption that setting the
+	 * value is always done through a single write. If this assumption ever
+	 * breaks, we would have to use X_ONCE for accesses, and suffer the less
+	 * optimized code. Although we may read stale target value if multiple
+	 * accesses happen at once, the performance impact should be minor.
+	 */
+	unsigned long target;
 
-	/* reset flag */
+	/**
+	 * @reset_required: reset flag
+	 *
+	 * Setting this flag may introduce races, but the code is expected to
+	 * handle them gracefully. In the worst case, another operation will
+	 * fail as reset did not take place. Clearing the flag is done while
+	 * holding @conf_sem for write.
+	 */
 	bool reset_required;
 
+	/**
+	 * @capabilities: hypervisor balloon capabilities.
+	 *
+	 * Protected by @conf_sem.
+	 */
 	unsigned long capabilities;
 
-	struct vmballoon_batch_page *batch_page;
+	/**
+	 * @batch_page: pointer to communication batch page.
+	 *
+	 * When batching is used, batch_page points to a page, which holds up to
+	 * %VMW_BALLOON_BATCH_MAX_PAGES entries for locking or unlocking.
+	 */
+	struct vmballoon_batch_entry *batch_page;
+
+	/**
+	 * @batch_max_pages: maximum pages that can be locked/unlocked.
+	 *
+	 * Indicates the number of pages that the hypervisor can lock or unlock
+	 * at once, according to whether batching is enabled. If batching is
+	 * disabled, only a single page can be locked/unlock on each operation.
+	 *
+	 * Protected by @conf_sem.
+	 */
 	unsigned int batch_max_pages;
-	struct page *page;
 
-	const struct vmballoon_ops *ops;
+	/**
+	 * @page: page to be locked/unlocked by the hypervisor
+	 *
+	 * @page is only used when batching is disabled and a single page is
+	 * reclaimed on each iteration.
+	 *
+	 * Protected by @comm_lock.
+	 */
+	struct page *page;
 
-#ifdef CONFIG_DEBUG_FS
 	/* statistics */
-	struct vmballoon_stats stats;
+	struct vmballoon_stats *stats;
 
+#ifdef CONFIG_DEBUG_FS
 	/* debugfs file exporting statistics */
 	struct dentry *dbg_entry;
 #endif
 
-	struct sysinfo sysinfo;
-
 	struct delayed_work dwork;
 
+	/**
+	 * @vmci_doorbell.
+	 *
+	 * Protected by @conf_sem.
+	 */
 	struct vmci_handle vmci_doorbell;
+
+	/**
+	 * @conf_sem: semaphore to protect the configuration and the statistics.
+	 */
+	struct rw_semaphore conf_sem;
+
+	/**
+	 * @comm_lock: lock to protect the communication with the host.
+	 *
+	 * Lock ordering: @conf_sem -> @comm_lock .
+	 */
+	spinlock_t comm_lock;
 };
 
 static struct vmballoon balloon;
 
+struct vmballoon_stats {
+	/* timer / doorbell operations */
+	atomic64_t general_stat[VMW_BALLOON_STAT_NUM];
+
+	/* allocation statistics for huge and small pages */
+	atomic64_t
+	       page_stat[VMW_BALLOON_PAGE_STAT_NUM][VMW_BALLOON_NUM_PAGE_SIZES];
+
+	/* Monitor operations: total operations, and failures */
+	atomic64_t ops[VMW_BALLOON_CMD_NUM][VMW_BALLOON_OP_STAT_TYPES];
+};
+
+static inline bool is_vmballoon_stats_on(void)
+{
+	return IS_ENABLED(CONFIG_DEBUG_FS) &&
+		static_branch_unlikely(&balloon_stat_enabled);
+}
+
+static inline void vmballoon_stats_op_inc(struct vmballoon *b, unsigned int op,
+					  enum vmballoon_op_stat_type type)
+{
+	if (is_vmballoon_stats_on())
+		atomic64_inc(&b->stats->ops[op][type]);
+}
+
+static inline void vmballoon_stats_gen_inc(struct vmballoon *b,
+					   enum vmballoon_stat_general stat)
+{
+	if (is_vmballoon_stats_on())
+		atomic64_inc(&b->stats->general_stat[stat]);
+}
+
+static inline void vmballoon_stats_gen_add(struct vmballoon *b,
+					   enum vmballoon_stat_general stat,
+					   unsigned int val)
+{
+	if (is_vmballoon_stats_on())
+		atomic64_add(val, &b->stats->general_stat[stat]);
+}
+
+static inline void vmballoon_stats_page_inc(struct vmballoon *b,
+					    enum vmballoon_stat_page stat,
+					    enum vmballoon_page_size_type size)
+{
+	if (is_vmballoon_stats_on())
+		atomic64_inc(&b->stats->page_stat[stat][size]);
+}
+
+static inline void vmballoon_stats_page_add(struct vmballoon *b,
+					    enum vmballoon_stat_page stat,
+					    enum vmballoon_page_size_type size,
+					    unsigned int val)
+{
+	if (is_vmballoon_stats_on())
+		atomic64_add(val, &b->stats->page_stat[stat][size]);
+}
+
+static inline unsigned long
+__vmballoon_cmd(struct vmballoon *b, unsigned long cmd, unsigned long arg1,
+		unsigned long arg2, unsigned long *result)
+{
+	unsigned long status, dummy1, dummy2, dummy3, local_result;
+
+	vmballoon_stats_op_inc(b, cmd, VMW_BALLOON_OP_STAT);
+
+	asm volatile ("inl %%dx" :
+		"=a"(status),
+		"=c"(dummy1),
+		"=d"(dummy2),
+		"=b"(local_result),
+		"=S"(dummy3) :
+		"0"(VMW_BALLOON_HV_MAGIC),
+		"1"(cmd),
+		"2"(VMW_BALLOON_HV_PORT),
+		"3"(arg1),
+		"4"(arg2) :
+		"memory");
+
+	/* update the result if needed */
+	if (result)
+		*result = (cmd == VMW_BALLOON_CMD_START) ? dummy1 :
+							   local_result;
+
+	/* update target when applicable */
+	if (status == VMW_BALLOON_SUCCESS &&
+	    ((1ul << cmd) & VMW_BALLOON_CMD_WITH_TARGET_MASK))
+		WRITE_ONCE(b->target, local_result);
+
+	if (status != VMW_BALLOON_SUCCESS &&
+	    status != VMW_BALLOON_SUCCESS_WITH_CAPABILITIES) {
+		vmballoon_stats_op_inc(b, cmd, VMW_BALLOON_OP_FAIL_STAT);
+		pr_debug("%s: %s [0x%lx,0x%lx) failed, returned %ld\n",
+			 __func__, vmballoon_cmd_names[cmd], arg1, arg2,
+			 status);
+	}
+
+	/* mark reset required accordingly */
+	if (status == VMW_BALLOON_ERROR_RESET)
+		b->reset_required = true;
+
+	return status;
+}
+
+static __always_inline unsigned long
+vmballoon_cmd(struct vmballoon *b, unsigned long cmd, unsigned long arg1,
+	      unsigned long arg2)
+{
+	unsigned long dummy;
+
+	return __vmballoon_cmd(b, cmd, arg1, arg2, &dummy);
+}
+
 /*
  * Send "start" command to the host, communicating supported version
  * of the protocol.
  */
-static bool vmballoon_send_start(struct vmballoon *b, unsigned long req_caps)
+static int vmballoon_send_start(struct vmballoon *b, unsigned long req_caps)
 {
-	unsigned long status, capabilities, dummy = 0;
-	bool success;
-
-	STATS_INC(b->stats.start);
+	unsigned long status, capabilities;
 
-	status = VMWARE_BALLOON_CMD(START, req_caps, dummy, capabilities);
+	status = __vmballoon_cmd(b, VMW_BALLOON_CMD_START, req_caps, 0,
+				 &capabilities);
 
 	switch (status) {
 	case VMW_BALLOON_SUCCESS_WITH_CAPABILITIES:
 		b->capabilities = capabilities;
-		success = true;
 		break;
 	case VMW_BALLOON_SUCCESS:
 		b->capabilities = VMW_BALLOON_BASIC_CMDS;
-		success = true;
 		break;
 	default:
-		success = false;
+		return -EIO;
 	}
 
 	/*
@@ -303,626 +501,693 @@ static bool vmballoon_send_start(struct vmballoon *b, unsigned long req_caps)
 	 * reason disabled, do not use 2MB pages, since otherwise the legacy
 	 * mechanism is used with 2MB pages, causing a failure.
 	 */
+	b->max_page_size = VMW_BALLOON_4K_PAGE;
 	if ((b->capabilities & VMW_BALLOON_BATCHED_2M_CMDS) &&
 	    (b->capabilities & VMW_BALLOON_BATCHED_CMDS))
-		b->supported_page_sizes = 2;
-	else
-		b->supported_page_sizes = 1;
-
-	if (!success) {
-		pr_debug("%s - failed, hv returns %ld\n", __func__, status);
-		STATS_INC(b->stats.start_fail);
-	}
-	return success;
-}
+		b->max_page_size = VMW_BALLOON_2M_PAGE;
 
-static bool vmballoon_check_status(struct vmballoon *b, unsigned long status)
-{
-	switch (status) {
-	case VMW_BALLOON_SUCCESS:
-		return true;
 
-	case VMW_BALLOON_ERROR_RESET:
-		b->reset_required = true;
-		/* fall through */
-
-	default:
-		return false;
-	}
+	return 0;
 }
 
-/*
+/**
+ * vmballoon_send_guest_id - communicate guest type to the host.
+ *
+ * @b: pointer to the balloon.
+ *
  * Communicate guest type to the host so that it can adjust ballooning
  * algorithm to the one most appropriate for the guest. This command
  * is normally issued after sending "start" command and is part of
  * standard reset sequence.
+ *
+ * Return: zero on success or appropriate error code.
  */
-static bool vmballoon_send_guest_id(struct vmballoon *b)
+static int vmballoon_send_guest_id(struct vmballoon *b)
 {
-	unsigned long status, dummy = 0;
-
-	status = VMWARE_BALLOON_CMD(GUEST_ID, VMW_BALLOON_GUEST_ID, dummy,
-				dummy);
-
-	STATS_INC(b->stats.guest_type);
+	unsigned long status;
 
-	if (vmballoon_check_status(b, status))
-		return true;
+	status = vmballoon_cmd(b, VMW_BALLOON_CMD_GUEST_ID,
+			       VMW_BALLOON_GUEST_ID, 0);
 
-	pr_debug("%s - failed, hv returns %ld\n", __func__, status);
-	STATS_INC(b->stats.guest_type_fail);
-	return false;
+	return status == VMW_BALLOON_SUCCESS ? 0 : -EIO;
 }
 
-static u16 vmballoon_page_size(bool is_2m_page)
+/**
+ * vmballoon_page_order() - return the order of the page
+ * @page_size: the size of the page.
+ *
+ * Return: the allocation order.
+ */
+static inline
+unsigned int vmballoon_page_order(enum vmballoon_page_size_type page_size)
 {
-	if (is_2m_page)
-		return 1 << VMW_BALLOON_2M_SHIFT;
+	return page_size == VMW_BALLOON_2M_PAGE ? VMW_BALLOON_2M_ORDER : 0;
+}
 
-	return 1;
+/**
+ * vmballoon_page_in_frames() - returns the number of frames in a page.
+ * @page_size: the size of the page.
+ *
+ * Return: the number of 4k frames.
+ */
+static inline unsigned int
+vmballoon_page_in_frames(enum vmballoon_page_size_type page_size)
+{
+	return 1 << vmballoon_page_order(page_size);
 }
 
-/*
- * Retrieve desired balloon size from the host.
+/**
+ * vmballoon_send_get_target() - Retrieve desired balloon size from the host.
+ *
+ * @b: pointer to the balloon.
+ *
+ * Return: zero on success, EINVAL if limit does not fit in 32-bit, as required
+ * by the host-guest protocol and EIO if an error occurred in communicating with
+ * the host.
  */
-static bool vmballoon_send_get_target(struct vmballoon *b, u32 *new_target)
+static int vmballoon_send_get_target(struct vmballoon *b)
 {
 	unsigned long status;
-	unsigned long target;
 	unsigned long limit;
-	unsigned long dummy = 0;
-	u32 limit32;
 
-	/*
-	 * si_meminfo() is cheap. Moreover, we want to provide dynamic
-	 * max balloon size later. So let us call si_meminfo() every
-	 * iteration.
-	 */
-	si_meminfo(&b->sysinfo);
-	limit = b->sysinfo.totalram;
+	limit = totalram_pages;
 
 	/* Ensure limit fits in 32-bits */
-	limit32 = (u32)limit;
-	if (limit != limit32)
-		return false;
-
-	/* update stats */
-	STATS_INC(b->stats.target);
+	if (limit != (u32)limit)
+		return -EINVAL;
 
-	status = VMWARE_BALLOON_CMD(GET_TARGET, limit, dummy, target);
-	if (vmballoon_check_status(b, status)) {
-		*new_target = target;
-		return true;
-	}
+	status = vmballoon_cmd(b, VMW_BALLOON_CMD_GET_TARGET, limit, 0);
 
-	pr_debug("%s - failed, hv returns %ld\n", __func__, status);
-	STATS_INC(b->stats.target_fail);
-	return false;
+	return status == VMW_BALLOON_SUCCESS ? 0 : -EIO;
 }
 
-/*
- * Notify the host about allocated page so that host can use it without
- * fear that guest will need it. Host may reject some pages, we need to
- * check the return value and maybe submit a different page.
+/**
+ * vmballoon_alloc_page_list - allocates a list of pages.
+ *
+ * @b: pointer to the balloon.
+ * @ctl: pointer for the %struct vmballoon_ctl, which defines the operation.
+ * @req_n_pages: the number of requested pages.
+ *
+ * Tries to allocate @req_n_pages. Add them to the list of balloon pages in
+ * @ctl.pages and updates @ctl.n_pages to reflect the number of pages.
+ *
+ * Return: zero on success or error code otherwise.
  */
-static int vmballoon_send_lock_page(struct vmballoon *b, unsigned long pfn,
-				unsigned int *hv_status, unsigned int *target)
+static int vmballoon_alloc_page_list(struct vmballoon *b,
+				     struct vmballoon_ctl *ctl,
+				     unsigned int req_n_pages)
 {
-	unsigned long status, dummy = 0;
-	u32 pfn32;
-
-	pfn32 = (u32)pfn;
-	if (pfn32 != pfn)
-		return -EINVAL;
-
-	STATS_INC(b->stats.lock[false]);
+	struct page *page;
+	unsigned int i;
 
-	*hv_status = status = VMWARE_BALLOON_CMD(LOCK, pfn, dummy, *target);
-	if (vmballoon_check_status(b, status))
-		return 0;
+	for (i = 0; i < req_n_pages; i++) {
+		if (ctl->page_size == VMW_BALLOON_2M_PAGE)
+			page = alloc_pages(VMW_HUGE_PAGE_ALLOC_FLAGS,
+					   VMW_BALLOON_2M_ORDER);
+		else
+			page = alloc_page(VMW_PAGE_ALLOC_FLAGS);
 
-	pr_debug("%s - ppn %lx, hv returns %ld\n", __func__, pfn, status);
-	STATS_INC(b->stats.lock_fail[false]);
-	return -EIO;
-}
+		/* Update statistics */
+		vmballoon_stats_page_inc(b, VMW_BALLOON_PAGE_STAT_ALLOC,
+					 ctl->page_size);
 
-static int vmballoon_send_batched_lock(struct vmballoon *b,
-		unsigned int num_pages, bool is_2m_pages, unsigned int *target)
-{
-	unsigned long status;
-	unsigned long pfn = PHYS_PFN(virt_to_phys(b->batch_page));
+		if (page) {
+			/* Success. Add the page to the list and continue. */
+			list_add(&page->lru, &ctl->pages);
+			continue;
+		}
 
-	STATS_INC(b->stats.lock[is_2m_pages]);
+		/* Allocation failed. Update statistics and stop. */
+		vmballoon_stats_page_inc(b, VMW_BALLOON_PAGE_STAT_ALLOC_FAIL,
+					 ctl->page_size);
+		break;
+	}
 
-	if (is_2m_pages)
-		status = VMWARE_BALLOON_CMD(BATCHED_2M_LOCK, pfn, num_pages,
-				*target);
-	else
-		status = VMWARE_BALLOON_CMD(BATCHED_LOCK, pfn, num_pages,
-				*target);
+	ctl->n_pages = i;
 
-	if (vmballoon_check_status(b, status))
-		return 0;
-
-	pr_debug("%s - batch ppn %lx, hv returns %ld\n", __func__, pfn, status);
-	STATS_INC(b->stats.lock_fail[is_2m_pages]);
-	return 1;
+	return req_n_pages == ctl->n_pages ? 0 : -ENOMEM;
 }
 
-/*
- * Notify the host that guest intends to release given page back into
- * the pool of available (to the guest) pages.
+/**
+ * vmballoon_handle_one_result - Handle lock/unlock result for a single page.
+ *
+ * @b: pointer for %struct vmballoon.
+ * @page: pointer for the page whose result should be handled.
+ * @page_size: size of the page.
+ * @status: status of the operation as provided by the hypervisor.
  */
-static bool vmballoon_send_unlock_page(struct vmballoon *b, unsigned long pfn,
-							unsigned int *target)
+static int vmballoon_handle_one_result(struct vmballoon *b, struct page *page,
+				       enum vmballoon_page_size_type page_size,
+				       unsigned long status)
 {
-	unsigned long status, dummy = 0;
-	u32 pfn32;
-
-	pfn32 = (u32)pfn;
-	if (pfn32 != pfn)
-		return false;
+	/* On success do nothing. The page is already on the balloon list. */
+	if (likely(status == VMW_BALLOON_SUCCESS))
+		return 0;
 
-	STATS_INC(b->stats.unlock[false]);
+	pr_debug("%s: failed comm pfn %lx status %lu page_size %s\n", __func__,
+		 page_to_pfn(page), status,
+		 vmballoon_page_size_names[page_size]);
 
-	status = VMWARE_BALLOON_CMD(UNLOCK, pfn, dummy, *target);
-	if (vmballoon_check_status(b, status))
-		return true;
+	/* Error occurred */
+	vmballoon_stats_page_inc(b, VMW_BALLOON_PAGE_STAT_REFUSED_ALLOC,
+				 page_size);
 
-	pr_debug("%s - ppn %lx, hv returns %ld\n", __func__, pfn, status);
-	STATS_INC(b->stats.unlock_fail[false]);
-	return false;
+	return -EIO;
 }
 
-static bool vmballoon_send_batched_unlock(struct vmballoon *b,
-		unsigned int num_pages, bool is_2m_pages, unsigned int *target)
+/**
+ * vmballoon_status_page - returns the status of (un)lock operation
+ *
+ * @b: pointer to the balloon.
+ * @idx: index for the page for which the operation is performed.
+ * @p: pointer to where the page struct is returned.
+ *
+ * Following a lock or unlock operation, returns the status of the operation for
+ * an individual page. Provides the page that the operation was performed on on
+ * the @page argument.
+ *
+ * Returns: The status of a lock or unlock operation for an individual page.
+ */
+static unsigned long vmballoon_status_page(struct vmballoon *b, int idx,
+					   struct page **p)
 {
-	unsigned long status;
-	unsigned long pfn = PHYS_PFN(virt_to_phys(b->batch_page));
-
-	STATS_INC(b->stats.unlock[is_2m_pages]);
-
-	if (is_2m_pages)
-		status = VMWARE_BALLOON_CMD(BATCHED_2M_UNLOCK, pfn, num_pages,
-				*target);
-	else
-		status = VMWARE_BALLOON_CMD(BATCHED_UNLOCK, pfn, num_pages,
-				*target);
+	if (static_branch_likely(&vmw_balloon_batching)) {
+		/* batching mode */
+		*p = pfn_to_page(b->batch_page[idx].pfn);
+		return b->batch_page[idx].status;
+	}
 
-	if (vmballoon_check_status(b, status))
-		return true;
+	/* non-batching mode */
+	*p = b->page;
 
-	pr_debug("%s - batch ppn %lx, hv returns %ld\n", __func__, pfn, status);
-	STATS_INC(b->stats.unlock_fail[is_2m_pages]);
-	return false;
+	/*
+	 * If a failure occurs, the indication will be provided in the status
+	 * of the entire operation, which is considered before the individual
+	 * page status. So for non-batching mode, the indication is always of
+	 * success.
+	 */
+	return VMW_BALLOON_SUCCESS;
 }
 
-static struct page *vmballoon_alloc_page(gfp_t flags, bool is_2m_page)
+/**
+ * vmballoon_lock_op - notifies the host about inflated/deflated pages.
+ * @b: pointer to the balloon.
+ * @num_pages: number of inflated/deflated pages.
+ * @page_size: size of the page.
+ * @op: the type of operation (lock or unlock).
+ *
+ * Notify the host about page(s) that were ballooned (or removed from the
+ * balloon) so that host can use it without fear that guest will need it (or
+ * stop using them since the VM does). Host may reject some pages, we need to
+ * check the return value and maybe submit a different page. The pages that are
+ * inflated/deflated are pointed by @b->page.
+ *
+ * Return: result as provided by the hypervisor.
+ */
+static unsigned long vmballoon_lock_op(struct vmballoon *b,
+				       unsigned int num_pages,
+				       enum vmballoon_page_size_type page_size,
+				       enum vmballoon_op op)
 {
-	if (is_2m_page)
-		return alloc_pages(flags, VMW_BALLOON_2M_SHIFT);
+	unsigned long cmd, pfn;
 
-	return alloc_page(flags);
-}
+	lockdep_assert_held(&b->comm_lock);
 
-static void vmballoon_free_page(struct page *page, bool is_2m_page)
-{
-	if (is_2m_page)
-		__free_pages(page, VMW_BALLOON_2M_SHIFT);
-	else
-		__free_page(page);
+	if (static_branch_likely(&vmw_balloon_batching)) {
+		if (op == VMW_BALLOON_INFLATE)
+			cmd = page_size == VMW_BALLOON_2M_PAGE ?
+				VMW_BALLOON_CMD_BATCHED_2M_LOCK :
+				VMW_BALLOON_CMD_BATCHED_LOCK;
+		else
+			cmd = page_size == VMW_BALLOON_2M_PAGE ?
+				VMW_BALLOON_CMD_BATCHED_2M_UNLOCK :
+				VMW_BALLOON_CMD_BATCHED_UNLOCK;
+
+		pfn = PHYS_PFN(virt_to_phys(b->batch_page));
+	} else {
+		cmd = op == VMW_BALLOON_INFLATE ? VMW_BALLOON_CMD_LOCK :
+						  VMW_BALLOON_CMD_UNLOCK;
+		pfn = page_to_pfn(b->page);
+
+		/* In non-batching mode, PFNs must fit in 32-bit */
+		if (unlikely(pfn != (u32)pfn))
+			return VMW_BALLOON_ERROR_PPN_INVALID;
+	}
+
+	return vmballoon_cmd(b, cmd, pfn, num_pages);
 }
 
-/*
- * Quickly release all pages allocated for the balloon. This function is
- * called when host decides to "reset" balloon for one reason or another.
- * Unlike normal "deflate" we do not (shall not) notify host of the pages
- * being released.
+/**
+ * vmballoon_add_page - adds a page towards lock/unlock operation.
+ *
+ * @b: pointer to the balloon.
+ * @idx: index of the page to be ballooned in this batch.
+ * @p: pointer to the page that is about to be ballooned.
+ *
+ * Adds the page to be ballooned. Must be called while holding @comm_lock.
  */
-static void vmballoon_pop(struct vmballoon *b)
+static void vmballoon_add_page(struct vmballoon *b, unsigned int idx,
+			       struct page *p)
 {
-	struct page *page, *next;
-	unsigned is_2m_pages;
-
-	for (is_2m_pages = 0; is_2m_pages < VMW_BALLOON_NUM_PAGE_SIZES;
-			is_2m_pages++) {
-		struct vmballoon_page_size *page_size =
-				&b->page_sizes[is_2m_pages];
-		u16 size_per_page = vmballoon_page_size(is_2m_pages);
-
-		list_for_each_entry_safe(page, next, &page_size->pages, lru) {
-			list_del(&page->lru);
-			vmballoon_free_page(page, is_2m_pages);
-			STATS_INC(b->stats.free[is_2m_pages]);
-			b->size -= size_per_page;
-			cond_resched();
-		}
-	}
+	lockdep_assert_held(&b->comm_lock);
 
-	/* Clearing the batch_page unconditionally has no adverse effect */
-	free_page((unsigned long)b->batch_page);
-	b->batch_page = NULL;
+	if (static_branch_likely(&vmw_balloon_batching))
+		b->batch_page[idx] = (struct vmballoon_batch_entry)
+					{ .pfn = page_to_pfn(p) };
+	else
+		b->page = p;
 }
 
-/*
- * Notify the host of a ballooned page. If host rejects the page put it on the
- * refuse list, those refused page are then released at the end of the
- * inflation cycle.
+/**
+ * vmballoon_lock - lock or unlock a batch of pages.
+ *
+ * @b: pointer to the balloon.
+ * @ctl: pointer for the %struct vmballoon_ctl, which defines the operation.
+ *
+ * Notifies the host of about ballooned pages (after inflation or deflation,
+ * according to @ctl). If the host rejects the page put it on the
+ * @ctl refuse list. These refused page are then released when moving to the
+ * next size of pages.
+ *
+ * Note that we neither free any @page here nor put them back on the ballooned
+ * pages list. Instead we queue it for later processing. We do that for several
+ * reasons. First, we do not want to free the page under the lock. Second, it
+ * allows us to unify the handling of lock and unlock. In the inflate case, the
+ * caller will check if there are too many refused pages and release them.
+ * Although it is not identical to the past behavior, it should not affect
+ * performance.
  */
-static int vmballoon_lock_page(struct vmballoon *b, unsigned int num_pages,
-				bool is_2m_pages, unsigned int *target)
+static int vmballoon_lock(struct vmballoon *b, struct vmballoon_ctl *ctl)
 {
-	int locked, hv_status;
-	struct page *page = b->page;
-	struct vmballoon_page_size *page_size = &b->page_sizes[false];
-
-	/* is_2m_pages can never happen as 2m pages support implies batching */
-
-	locked = vmballoon_send_lock_page(b, page_to_pfn(page), &hv_status,
-								target);
-	if (locked) {
-		STATS_INC(b->stats.refused_alloc[false]);
-
-		if (locked == -EIO &&
-		    (hv_status == VMW_BALLOON_ERROR_RESET ||
-		     hv_status == VMW_BALLOON_ERROR_PPN_NOTNEEDED)) {
-			vmballoon_free_page(page, false);
-			return -EIO;
-		}
-
-		/*
-		 * Place page on the list of non-balloonable pages
-		 * and retry allocation, unless we already accumulated
-		 * too many of them, in which case take a breather.
-		 */
-		if (page_size->n_refused_pages < VMW_BALLOON_MAX_REFUSED) {
-			page_size->n_refused_pages++;
-			list_add(&page->lru, &page_size->refused_pages);
-		} else {
-			vmballoon_free_page(page, false);
-		}
-		return locked;
-	}
-
-	/* track allocated page */
-	list_add(&page->lru, &page_size->pages);
+	unsigned long batch_status;
+	struct page *page;
+	unsigned int i, num_pages;
 
-	/* update balloon size */
-	b->size++;
+	num_pages = ctl->n_pages;
+	if (num_pages == 0)
+		return 0;
 
-	return 0;
-}
+	/* communication with the host is done under the communication lock */
+	spin_lock(&b->comm_lock);
 
-static int vmballoon_lock_batched_page(struct vmballoon *b,
-		unsigned int num_pages, bool is_2m_pages, unsigned int *target)
-{
-	int locked, i;
-	u16 size_per_page = vmballoon_page_size(is_2m_pages);
+	i = 0;
+	list_for_each_entry(page, &ctl->pages, lru)
+		vmballoon_add_page(b, i++, page);
 
-	locked = vmballoon_send_batched_lock(b, num_pages, is_2m_pages,
-			target);
-	if (locked > 0) {
-		for (i = 0; i < num_pages; i++) {
-			u64 pa = vmballoon_batch_get_pa(b->batch_page, i);
-			struct page *p = pfn_to_page(pa >> PAGE_SHIFT);
+	batch_status = vmballoon_lock_op(b, ctl->n_pages, ctl->page_size,
+					 ctl->op);
 
-			vmballoon_free_page(p, is_2m_pages);
-		}
+	/*
+	 * Iterate over the pages in the provided list. Since we are changing
+	 * @ctl->n_pages we are saving the original value in @num_pages and
+	 * use this value to bound the loop.
+	 */
+	for (i = 0; i < num_pages; i++) {
+		unsigned long status;
 
-		return -EIO;
-	}
+		status = vmballoon_status_page(b, i, &page);
 
-	for (i = 0; i < num_pages; i++) {
-		u64 pa = vmballoon_batch_get_pa(b->batch_page, i);
-		struct page *p = pfn_to_page(pa >> PAGE_SHIFT);
-		struct vmballoon_page_size *page_size =
-				&b->page_sizes[is_2m_pages];
+		/*
+		 * Failure of the whole batch overrides a single operation
+		 * results.
+		 */
+		if (batch_status != VMW_BALLOON_SUCCESS)
+			status = batch_status;
 
-		locked = vmballoon_batch_get_status(b->batch_page, i);
+		/* Continue if no error happened */
+		if (!vmballoon_handle_one_result(b, page, ctl->page_size,
+						 status))
+			continue;
 
-		switch (locked) {
-		case VMW_BALLOON_SUCCESS:
-			list_add(&p->lru, &page_size->pages);
-			b->size += size_per_page;
-			break;
-		case VMW_BALLOON_ERROR_PPN_PINNED:
-		case VMW_BALLOON_ERROR_PPN_INVALID:
-			if (page_size->n_refused_pages
-					< VMW_BALLOON_MAX_REFUSED) {
-				list_add(&p->lru, &page_size->refused_pages);
-				page_size->n_refused_pages++;
-				break;
-			}
-			/* Fallthrough */
-		case VMW_BALLOON_ERROR_RESET:
-		case VMW_BALLOON_ERROR_PPN_NOTNEEDED:
-			vmballoon_free_page(p, is_2m_pages);
-			break;
-		default:
-			/* This should never happen */
-			WARN_ON_ONCE(true);
-		}
+		/*
+		 * Error happened. Move the pages to the refused list and update
+		 * the pages number.
+		 */
+		list_move(&page->lru, &ctl->refused_pages);
+		ctl->n_pages--;
+		ctl->n_refused_pages++;
 	}
 
-	return 0;
+	spin_unlock(&b->comm_lock);
+
+	return batch_status == VMW_BALLOON_SUCCESS ? 0 : -EIO;
 }
 
-/*
- * Release the page allocated for the balloon. Note that we first notify
- * the host so it can make sure the page will be available for the guest
- * to use, if needed.
+/**
+ * vmballoon_release_page_list() - Releases a page list
+ *
+ * @page_list: list of pages to release.
+ * @n_pages: pointer to the number of pages.
+ * @page_size: whether the pages in the list are 2MB (or else 4KB).
+ *
+ * Releases the list of pages and zeros the number of pages.
  */
-static int vmballoon_unlock_page(struct vmballoon *b, unsigned int num_pages,
-		bool is_2m_pages, unsigned int *target)
+static void vmballoon_release_page_list(struct list_head *page_list,
+				       int *n_pages,
+				       enum vmballoon_page_size_type page_size)
 {
-	struct page *page = b->page;
-	struct vmballoon_page_size *page_size = &b->page_sizes[false];
-
-	/* is_2m_pages can never happen as 2m pages support implies batching */
+	struct page *page, *tmp;
 
-	if (!vmballoon_send_unlock_page(b, page_to_pfn(page), target)) {
-		list_add(&page->lru, &page_size->pages);
-		return -EIO;
+	list_for_each_entry_safe(page, tmp, page_list, lru) {
+		list_del(&page->lru);
+		__free_pages(page, vmballoon_page_order(page_size));
 	}
 
-	/* deallocate page */
-	vmballoon_free_page(page, false);
-	STATS_INC(b->stats.free[false]);
+	*n_pages = 0;
+}
 
-	/* update balloon size */
-	b->size--;
 
-	return 0;
+/*
+ * Release pages that were allocated while attempting to inflate the
+ * balloon but were refused by the host for one reason or another.
+ */
+static void vmballoon_release_refused_pages(struct vmballoon *b,
+					    struct vmballoon_ctl *ctl)
+{
+	vmballoon_stats_page_inc(b, VMW_BALLOON_PAGE_STAT_REFUSED_FREE,
+				 ctl->page_size);
+
+	vmballoon_release_page_list(&ctl->refused_pages, &ctl->n_refused_pages,
+				    ctl->page_size);
 }
 
-static int vmballoon_unlock_batched_page(struct vmballoon *b,
-				unsigned int num_pages, bool is_2m_pages,
-				unsigned int *target)
+/**
+ * vmballoon_change - retrieve the required balloon change
+ *
+ * @b: pointer for the balloon.
+ *
+ * Return: the required change for the balloon size. A positive number
+ * indicates inflation, a negative number indicates a deflation.
+ */
+static int64_t vmballoon_change(struct vmballoon *b)
 {
-	int locked, i, ret = 0;
-	bool hv_success;
-	u16 size_per_page = vmballoon_page_size(is_2m_pages);
+	int64_t size, target;
 
-	hv_success = vmballoon_send_batched_unlock(b, num_pages, is_2m_pages,
-			target);
-	if (!hv_success)
-		ret = -EIO;
+	size = atomic64_read(&b->size);
+	target = READ_ONCE(b->target);
 
-	for (i = 0; i < num_pages; i++) {
-		u64 pa = vmballoon_batch_get_pa(b->batch_page, i);
-		struct page *p = pfn_to_page(pa >> PAGE_SHIFT);
-		struct vmballoon_page_size *page_size =
-				&b->page_sizes[is_2m_pages];
+	/*
+	 * We must cast first because of int sizes
+	 * Otherwise we might get huge positives instead of negatives
+	 */
 
-		locked = vmballoon_batch_get_status(b->batch_page, i);
-		if (!hv_success || locked != VMW_BALLOON_SUCCESS) {
-			/*
-			 * That page wasn't successfully unlocked by the
-			 * hypervisor, re-add it to the list of pages owned by
-			 * the balloon driver.
-			 */
-			list_add(&p->lru, &page_size->pages);
-		} else {
-			/* deallocate page */
-			vmballoon_free_page(p, is_2m_pages);
-			STATS_INC(b->stats.free[is_2m_pages]);
-
-			/* update balloon size */
-			b->size -= size_per_page;
-		}
-	}
+	if (b->reset_required)
+		return 0;
+
+	/* consider a 2MB slack on deflate, unless the balloon is emptied */
+	if (target < size && target != 0 &&
+	    size - target < vmballoon_page_in_frames(VMW_BALLOON_2M_PAGE))
+		return 0;
 
-	return ret;
+	return target - size;
 }
 
-/*
- * Release pages that were allocated while attempting to inflate the
- * balloon but were refused by the host for one reason or another.
+/**
+ * vmballoon_enqueue_page_list() - Enqueues list of pages after inflation.
+ *
+ * @b: pointer to balloon.
+ * @pages: list of pages to enqueue.
+ * @n_pages: pointer to number of pages in list. The value is zeroed.
+ * @page_size: whether the pages are 2MB or 4KB pages.
+ *
+ * Enqueues the provides list of pages in the ballooned page list, clears the
+ * list and zeroes the number of pages that was provided.
  */
-static void vmballoon_release_refused_pages(struct vmballoon *b,
-		bool is_2m_pages)
+static void vmballoon_enqueue_page_list(struct vmballoon *b,
+					struct list_head *pages,
+					unsigned int *n_pages,
+					enum vmballoon_page_size_type page_size)
 {
-	struct page *page, *next;
-	struct vmballoon_page_size *page_size =
-			&b->page_sizes[is_2m_pages];
+	struct vmballoon_page_size *page_size_info = &b->page_sizes[page_size];
 
-	list_for_each_entry_safe(page, next, &page_size->refused_pages, lru) {
-		list_del(&page->lru);
-		vmballoon_free_page(page, is_2m_pages);
-		STATS_INC(b->stats.refused_free[is_2m_pages]);
-	}
-
-	page_size->n_refused_pages = 0;
+	list_splice_init(pages, &page_size_info->pages);
+	*n_pages = 0;
 }
 
-static void vmballoon_add_page(struct vmballoon *b, int idx, struct page *p)
+/**
+ * vmballoon_dequeue_page_list() - Dequeues page lists for deflation.
+ *
+ * @b: pointer to balloon.
+ * @pages: list of pages to enqueue.
+ * @n_pages: pointer to number of pages in list. The value is zeroed.
+ * @page_size: whether the pages are 2MB or 4KB pages.
+ * @n_req_pages: the number of requested pages.
+ *
+ * Dequeues the number of requested pages from the balloon for deflation. The
+ * number of dequeued pages may be lower, if not enough pages in the requested
+ * size are available.
+ */
+static void vmballoon_dequeue_page_list(struct vmballoon *b,
+					struct list_head *pages,
+					unsigned int *n_pages,
+					enum vmballoon_page_size_type page_size,
+					unsigned int n_req_pages)
 {
-	b->page = p;
-}
+	struct vmballoon_page_size *page_size_info = &b->page_sizes[page_size];
+	struct page *page, *tmp;
+	unsigned int i = 0;
 
-static void vmballoon_add_batched_page(struct vmballoon *b, int idx,
-				struct page *p)
-{
-	vmballoon_batch_set_pa(b->batch_page, idx,
-			(u64)page_to_pfn(p) << PAGE_SHIFT);
+	list_for_each_entry_safe(page, tmp, &page_size_info->pages, lru) {
+		list_move(&page->lru, pages);
+		if (++i == n_req_pages)
+			break;
+	}
+	*n_pages = i;
 }
 
-/*
- * Inflate the balloon towards its target size. Note that we try to limit
- * the rate of allocation to make sure we are not choking the rest of the
- * system.
+/**
+ * vmballoon_inflate() - Inflate the balloon towards its target size.
+ *
+ * @b: pointer to the balloon.
  */
 static void vmballoon_inflate(struct vmballoon *b)
 {
-	unsigned int num_pages = 0;
-	int error = 0;
-	gfp_t flags = VMW_PAGE_ALLOC_NOSLEEP;
-	bool is_2m_pages;
+	int64_t to_inflate_frames;
+	struct vmballoon_ctl ctl = {
+		.pages = LIST_HEAD_INIT(ctl.pages),
+		.refused_pages = LIST_HEAD_INIT(ctl.refused_pages),
+		.page_size = b->max_page_size,
+		.op = VMW_BALLOON_INFLATE
+	};
 
-	pr_debug("%s - size: %d, target %d\n", __func__, b->size, b->target);
+	while ((to_inflate_frames = vmballoon_change(b)) > 0) {
+		unsigned int to_inflate_pages, page_in_frames;
+		int alloc_error, lock_error = 0;
 
-	/*
-	 * First try NOSLEEP page allocations to inflate balloon.
-	 *
-	 * If we do not throttle nosleep allocations, we can drain all
-	 * free pages in the guest quickly (if the balloon target is high).
-	 * As a side-effect, draining free pages helps to inform (force)
-	 * the guest to start swapping if balloon target is not met yet,
-	 * which is a desired behavior. However, balloon driver can consume
-	 * all available CPU cycles if too many pages are allocated in a
-	 * second. Therefore, we throttle nosleep allocations even when
-	 * the guest is not under memory pressure. OTOH, if we have already
-	 * predicted that the guest is under memory pressure, then we
-	 * slowdown page allocations considerably.
-	 */
+		VM_BUG_ON(!list_empty(&ctl.pages));
+		VM_BUG_ON(ctl.n_pages != 0);
 
-	/*
-	 * Start with no sleep allocation rate which may be higher
-	 * than sleeping allocation rate.
-	 */
-	is_2m_pages = b->supported_page_sizes == VMW_BALLOON_NUM_PAGE_SIZES;
+		page_in_frames = vmballoon_page_in_frames(ctl.page_size);
 
-	pr_debug("%s - goal: %d",  __func__, b->target - b->size);
+		to_inflate_pages = min_t(unsigned long, b->batch_max_pages,
+					 DIV_ROUND_UP_ULL(to_inflate_frames,
+							  page_in_frames));
 
-	while (!b->reset_required &&
-		b->size + num_pages * vmballoon_page_size(is_2m_pages)
-		< b->target) {
-		struct page *page;
+		/* Start by allocating */
+		alloc_error = vmballoon_alloc_page_list(b, &ctl,
+							to_inflate_pages);
 
-		if (flags == VMW_PAGE_ALLOC_NOSLEEP)
-			STATS_INC(b->stats.alloc[is_2m_pages]);
-		else
-			STATS_INC(b->stats.sleep_alloc);
-
-		page = vmballoon_alloc_page(flags, is_2m_pages);
-		if (!page) {
-			STATS_INC(b->stats.alloc_fail[is_2m_pages]);
-
-			if (is_2m_pages) {
-				b->ops->lock(b, num_pages, true, &b->target);
-
-				/*
-				 * ignore errors from locking as we now switch
-				 * to 4k pages and we might get different
-				 * errors.
-				 */
-
-				num_pages = 0;
-				is_2m_pages = false;
-				continue;
-			}
-
-			if (flags == VMW_PAGE_ALLOC_CANSLEEP) {
-				/*
-				 * CANSLEEP page allocation failed, so guest
-				 * is under severe memory pressure. We just log
-				 * the event, but do not stop the inflation
-				 * due to its negative impact on performance.
-				 */
-				STATS_INC(b->stats.sleep_alloc_fail);
+		/* Actually lock the pages by telling the hypervisor */
+		lock_error = vmballoon_lock(b, &ctl);
+
+		/*
+		 * If an error indicates that something serious went wrong,
+		 * stop the inflation.
+		 */
+		if (lock_error)
+			break;
+
+		/* Update the balloon size */
+		atomic64_add(ctl.n_pages * page_in_frames, &b->size);
+
+		vmballoon_enqueue_page_list(b, &ctl.pages, &ctl.n_pages,
+					    ctl.page_size);
+
+		/*
+		 * If allocation failed or the number of refused pages exceeds
+		 * the maximum allowed, move to the next page size.
+		 */
+		if (alloc_error ||
+		    ctl.n_refused_pages >= VMW_BALLOON_MAX_REFUSED) {
+			if (ctl.page_size == VMW_BALLOON_4K_PAGE)
 				break;
-			}
 
 			/*
-			 * NOSLEEP page allocation failed, so the guest is
-			 * under memory pressure. Slowing down page alloctions
-			 * seems to be reasonable, but doing so might actually
-			 * cause the hypervisor to throttle us down, resulting
-			 * in degraded performance. We will count on the
-			 * scheduler and standard memory management mechanisms
-			 * for now.
+			 * Ignore errors from locking as we now switch to 4k
+			 * pages and we might get different errors.
 			 */
-			flags = VMW_PAGE_ALLOC_CANSLEEP;
-			continue;
-		}
-
-		b->ops->add_page(b, num_pages++, page);
-		if (num_pages == b->batch_max_pages) {
-			error = b->ops->lock(b, num_pages, is_2m_pages,
-					&b->target);
-			num_pages = 0;
-			if (error)
-				break;
+			vmballoon_release_refused_pages(b, &ctl);
+			ctl.page_size--;
 		}
 
 		cond_resched();
 	}
 
-	if (num_pages > 0)
-		b->ops->lock(b, num_pages, is_2m_pages, &b->target);
-
-	vmballoon_release_refused_pages(b, true);
-	vmballoon_release_refused_pages(b, false);
+	/*
+	 * Release pages that were allocated while attempting to inflate the
+	 * balloon but were refused by the host for one reason or another,
+	 * and update the statistics.
+	 */
+	if (ctl.n_refused_pages != 0)
+		vmballoon_release_refused_pages(b, &ctl);
 }
 
-/*
+/**
+ * vmballoon_deflate() - Decrease the size of the balloon.
+ *
+ * @b: pointer to the balloon
+ * @n_frames: the number of frames to deflate. If zero, automatically
+ * calculated according to the target size.
+ * @coordinated: whether to coordinate with the host
+ *
  * Decrease the size of the balloon allowing guest to use more memory.
+ *
+ * Return: The number of deflated frames (i.e., basic page size units)
  */
-static void vmballoon_deflate(struct vmballoon *b)
+static unsigned long vmballoon_deflate(struct vmballoon *b, uint64_t n_frames,
+				       bool coordinated)
 {
-	unsigned is_2m_pages;
-
-	pr_debug("%s - size: %d, target %d\n", __func__, b->size, b->target);
+	unsigned long deflated_frames = 0;
+	unsigned long tried_frames = 0;
+	struct vmballoon_ctl ctl = {
+		.pages = LIST_HEAD_INIT(ctl.pages),
+		.refused_pages = LIST_HEAD_INIT(ctl.refused_pages),
+		.page_size = VMW_BALLOON_4K_PAGE,
+		.op = VMW_BALLOON_DEFLATE
+	};
 
 	/* free pages to reach target */
-	for (is_2m_pages = 0; is_2m_pages < b->supported_page_sizes;
-			is_2m_pages++) {
-		struct page *page, *next;
-		unsigned int num_pages = 0;
-		struct vmballoon_page_size *page_size =
-				&b->page_sizes[is_2m_pages];
-
-		list_for_each_entry_safe(page, next, &page_size->pages, lru) {
-			if (b->reset_required ||
-				(b->target > 0 &&
-					b->size - num_pages
-					* vmballoon_page_size(is_2m_pages)
-				< b->target + vmballoon_page_size(true)))
-				break;
+	while (true) {
+		unsigned int to_deflate_pages, n_unlocked_frames;
+		unsigned int page_in_frames;
+		int64_t to_deflate_frames;
+		bool deflated_all;
+
+		page_in_frames = vmballoon_page_in_frames(ctl.page_size);
+
+		VM_BUG_ON(!list_empty(&ctl.pages));
+		VM_BUG_ON(ctl.n_pages);
+		VM_BUG_ON(!list_empty(&ctl.refused_pages));
+		VM_BUG_ON(ctl.n_refused_pages);
+
+		/*
+		 * If we were requested a specific number of frames, we try to
+		 * deflate this number of frames. Otherwise, deflation is
+		 * performed according to the target and balloon size.
+		 */
+		to_deflate_frames = n_frames ? n_frames - tried_frames :
+					       -vmballoon_change(b);
+
+		/* break if no work to do */
+		if (to_deflate_frames <= 0)
+			break;
+
+		/*
+		 * Calculate the number of frames based on current page size,
+		 * but limit the deflated frames to a single chunk
+		 */
+		to_deflate_pages = min_t(unsigned long, b->batch_max_pages,
+					 DIV_ROUND_UP_ULL(to_deflate_frames,
+							  page_in_frames));
+
+		/* First take the pages from the balloon pages. */
+		vmballoon_dequeue_page_list(b, &ctl.pages, &ctl.n_pages,
+					    ctl.page_size, to_deflate_pages);
+
+		/*
+		 * Before pages are moving to the refused list, count their
+		 * frames as frames that we tried to deflate.
+		 */
+		tried_frames += ctl.n_pages * page_in_frames;
+
+		/*
+		 * Unlock the pages by communicating with the hypervisor if the
+		 * communication is coordinated (i.e., not pop). We ignore the
+		 * return code. Instead we check if all the pages we manage to
+		 * unlock all the pages. If we failed, we will move to the next
+		 * page size, and would eventually try again later.
+		 */
+		if (coordinated)
+			vmballoon_lock(b, &ctl);
+
+		/*
+		 * Check if we deflated enough. We will move to the next page
+		 * size if we did not manage to do so. This calculation takes
+		 * place now, as once the pages are released, the number of
+		 * pages is zeroed.
+		 */
+		deflated_all = (ctl.n_pages == to_deflate_pages);
+
+		/* Update local and global counters */
+		n_unlocked_frames = ctl.n_pages * page_in_frames;
+		atomic64_sub(n_unlocked_frames, &b->size);
+		deflated_frames += n_unlocked_frames;
 
-			list_del(&page->lru);
-			b->ops->add_page(b, num_pages++, page);
+		vmballoon_stats_page_add(b, VMW_BALLOON_PAGE_STAT_FREE,
+					 ctl.page_size, ctl.n_pages);
 
-			if (num_pages == b->batch_max_pages) {
-				int error;
+		/* free the ballooned pages */
+		vmballoon_release_page_list(&ctl.pages, &ctl.n_pages,
+					    ctl.page_size);
 
-				error = b->ops->unlock(b, num_pages,
-						is_2m_pages, &b->target);
-				num_pages = 0;
-				if (error)
-					return;
-			}
+		/* Return the refused pages to the ballooned list. */
+		vmballoon_enqueue_page_list(b, &ctl.refused_pages,
+					    &ctl.n_refused_pages,
+					    ctl.page_size);
 
-			cond_resched();
+		/* If we failed to unlock all the pages, move to next size. */
+		if (!deflated_all) {
+			if (ctl.page_size == b->max_page_size)
+				break;
+			ctl.page_size++;
 		}
 
-		if (num_pages > 0)
-			b->ops->unlock(b, num_pages, is_2m_pages, &b->target);
+		cond_resched();
 	}
-}
 
-static const struct vmballoon_ops vmballoon_basic_ops = {
-	.add_page = vmballoon_add_page,
-	.lock = vmballoon_lock_page,
-	.unlock = vmballoon_unlock_page
-};
+	return deflated_frames;
+}
 
-static const struct vmballoon_ops vmballoon_batched_ops = {
-	.add_page = vmballoon_add_batched_page,
-	.lock = vmballoon_lock_batched_page,
-	.unlock = vmballoon_unlock_batched_page
-};
+/**
+ * vmballoon_deinit_batching - disables batching mode.
+ *
+ * @b: pointer to &struct vmballoon.
+ *
+ * Disables batching, by deallocating the page for communication with the
+ * hypervisor and disabling the static key to indicate that batching is off.
+ */
+static void vmballoon_deinit_batching(struct vmballoon *b)
+{
+	free_page((unsigned long)b->batch_page);
+	b->batch_page = NULL;
+	static_branch_disable(&vmw_balloon_batching);
+	b->batch_max_pages = 1;
+}
 
-static bool vmballoon_init_batching(struct vmballoon *b)
+/**
+ * vmballoon_init_batching - enable batching mode.
+ *
+ * @b: pointer to &struct vmballoon.
+ *
+ * Enables batching, by allocating a page for communication with the hypervisor
+ * and enabling the static_key to use batching.
+ *
+ * Return: zero on success or an appropriate error-code.
+ */
+static int vmballoon_init_batching(struct vmballoon *b)
 {
 	struct page *page;
 
 	page = alloc_page(GFP_KERNEL | __GFP_ZERO);
 	if (!page)
-		return false;
+		return -ENOMEM;
 
 	b->batch_page = page_address(page);
-	return true;
+	b->batch_max_pages = PAGE_SIZE / sizeof(struct vmballoon_batch_entry);
+
+	static_branch_enable(&vmw_balloon_batching);
+
+	return 0;
 }
 
 /*
@@ -932,7 +1197,7 @@ static void vmballoon_doorbell(void *client_data)
 {
 	struct vmballoon *b = client_data;
 
-	STATS_INC(b->stats.doorbell);
+	vmballoon_stats_gen_inc(b, VMW_BALLOON_STAT_DOORBELL);
 
 	mod_delayed_work(system_freezable_wq, &b->dwork, 0);
 }
@@ -942,11 +1207,8 @@ static void vmballoon_doorbell(void *client_data)
  */
 static void vmballoon_vmci_cleanup(struct vmballoon *b)
 {
-	int error;
-
-	VMWARE_BALLOON_CMD(VMCI_DOORBELL_SET, VMCI_INVALID_ID,
-			VMCI_INVALID_ID, error);
-	STATS_INC(b->stats.doorbell_unset);
+	vmballoon_cmd(b, VMW_BALLOON_CMD_VMCI_DOORBELL_SET,
+		      VMCI_INVALID_ID, VMCI_INVALID_ID);
 
 	if (!vmci_handle_is_invalid(b->vmci_doorbell)) {
 		vmci_doorbell_destroy(b->vmci_doorbell);
@@ -954,12 +1216,19 @@ static void vmballoon_vmci_cleanup(struct vmballoon *b)
 	}
 }
 
-/*
- * Initialize vmci doorbell, to get notified as soon as balloon changes
+/**
+ * vmballoon_vmci_init - Initialize vmci doorbell.
+ *
+ * @b: pointer to the balloon.
+ *
+ * Return: zero on success or when wakeup command not supported. Error-code
+ * otherwise.
+ *
+ * Initialize vmci doorbell, to get notified as soon as balloon changes.
  */
 static int vmballoon_vmci_init(struct vmballoon *b)
 {
-	unsigned long error, dummy;
+	unsigned long error;
 
 	if ((b->capabilities & VMW_BALLOON_SIGNALLED_WAKEUP_CMD) == 0)
 		return 0;
@@ -971,10 +1240,9 @@ static int vmballoon_vmci_init(struct vmballoon *b)
 	if (error != VMCI_SUCCESS)
 		goto fail;
 
-	error = VMWARE_BALLOON_CMD(VMCI_DOORBELL_SET, b->vmci_doorbell.context,
-				   b->vmci_doorbell.resource, dummy);
-
-	STATS_INC(b->stats.doorbell_set);
+	error =	__vmballoon_cmd(b, VMW_BALLOON_CMD_VMCI_DOORBELL_SET,
+				b->vmci_doorbell.context,
+				b->vmci_doorbell.resource, NULL);
 
 	if (error != VMW_BALLOON_SUCCESS)
 		goto fail;
@@ -985,6 +1253,23 @@ fail:
 	return -EIO;
 }
 
+/**
+ * vmballoon_pop - Quickly release all pages allocate for the balloon.
+ *
+ * @b: pointer to the balloon.
+ *
+ * This function is called when host decides to "reset" balloon for one reason
+ * or another. Unlike normal "deflate" we do not (shall not) notify host of the
+ * pages being released.
+ */
+static void vmballoon_pop(struct vmballoon *b)
+{
+	unsigned long size;
+
+	while ((size = atomic64_read(&b->size)))
+		vmballoon_deflate(b, size, false);
+}
+
 /*
  * Perform standard reset sequence by popping the balloon (in case it
  * is not  empty) and then restarting protocol. This operation normally
@@ -994,18 +1279,18 @@ static void vmballoon_reset(struct vmballoon *b)
 {
 	int error;
 
+	down_write(&b->conf_sem);
+
 	vmballoon_vmci_cleanup(b);
 
 	/* free all pages, skipping monitor unlock */
 	vmballoon_pop(b);
 
-	if (!vmballoon_send_start(b, VMW_BALLOON_CAPABILITIES))
+	if (vmballoon_send_start(b, VMW_BALLOON_CAPABILITIES))
 		return;
 
 	if ((b->capabilities & VMW_BALLOON_BATCHED_CMDS) != 0) {
-		b->ops = &vmballoon_batched_ops;
-		b->batch_max_pages = VMW_BALLOON_BATCH_MAX_PAGES;
-		if (!vmballoon_init_batching(b)) {
+		if (vmballoon_init_batching(b)) {
 			/*
 			 * We failed to initialize batching, inform the monitor
 			 * about it by sending a null capability.
@@ -1016,52 +1301,70 @@ static void vmballoon_reset(struct vmballoon *b)
 			return;
 		}
 	} else if ((b->capabilities & VMW_BALLOON_BASIC_CMDS) != 0) {
-		b->ops = &vmballoon_basic_ops;
-		b->batch_max_pages = 1;
+		vmballoon_deinit_batching(b);
 	}
 
+	vmballoon_stats_gen_inc(b, VMW_BALLOON_STAT_RESET);
 	b->reset_required = false;
 
 	error = vmballoon_vmci_init(b);
 	if (error)
 		pr_err("failed to initialize vmci doorbell\n");
 
-	if (!vmballoon_send_guest_id(b))
+	if (vmballoon_send_guest_id(b))
 		pr_err("failed to send guest ID to the host\n");
+
+	up_write(&b->conf_sem);
 }
 
-/*
- * Balloon work function: reset protocol, if needed, get the new size and
- * adjust balloon as needed. Repeat in 1 sec.
+/**
+ * vmballoon_work - periodic balloon worker for reset, inflation and deflation.
+ *
+ * @work: pointer to the &work_struct which is provided by the workqueue.
+ *
+ * Resets the protocol if needed, gets the new size and adjusts balloon as
+ * needed. Repeat in 1 sec.
  */
 static void vmballoon_work(struct work_struct *work)
 {
 	struct delayed_work *dwork = to_delayed_work(work);
 	struct vmballoon *b = container_of(dwork, struct vmballoon, dwork);
-	unsigned int target;
-
-	STATS_INC(b->stats.timer);
+	int64_t change = 0;
 
 	if (b->reset_required)
 		vmballoon_reset(b);
 
-	if (!b->reset_required && vmballoon_send_get_target(b, &target)) {
-		/* update target, adjust size */
-		b->target = target;
+	down_read(&b->conf_sem);
+
+	/*
+	 * Update the stats while holding the semaphore to ensure that
+	 * @stats_enabled is consistent with whether the stats are actually
+	 * enabled
+	 */
+	vmballoon_stats_gen_inc(b, VMW_BALLOON_STAT_TIMER);
+
+	if (!vmballoon_send_get_target(b))
+		change = vmballoon_change(b);
+
+	if (change != 0) {
+		pr_debug("%s - size: %llu, target %lu\n", __func__,
+			 atomic64_read(&b->size), READ_ONCE(b->target));
 
-		if (b->size < target)
+		if (change > 0)
 			vmballoon_inflate(b);
-		else if (target == 0 ||
-				b->size > target + vmballoon_page_size(true))
-			vmballoon_deflate(b);
+		else  /* (change < 0) */
+			vmballoon_deflate(b, 0, true);
 	}
 
+	up_read(&b->conf_sem);
+
 	/*
 	 * We are using a freezable workqueue so that balloon operations are
 	 * stopped while the system transitions to/from sleep/hibernation.
 	 */
 	queue_delayed_work(system_freezable_wq,
 			   dwork, round_jiffies_relative(HZ));
+
 }
 
 /*
@@ -1069,64 +1372,100 @@ static void vmballoon_work(struct work_struct *work)
  */
 #ifdef CONFIG_DEBUG_FS
 
+static const char * const vmballoon_stat_page_names[] = {
+	[VMW_BALLOON_PAGE_STAT_ALLOC]		= "alloc",
+	[VMW_BALLOON_PAGE_STAT_ALLOC_FAIL]	= "allocFail",
+	[VMW_BALLOON_PAGE_STAT_REFUSED_ALLOC]	= "errAlloc",
+	[VMW_BALLOON_PAGE_STAT_REFUSED_FREE]	= "errFree",
+	[VMW_BALLOON_PAGE_STAT_FREE]		= "free"
+};
+
+static const char * const vmballoon_stat_names[] = {
+	[VMW_BALLOON_STAT_TIMER]		= "timer",
+	[VMW_BALLOON_STAT_DOORBELL]		= "doorbell",
+	[VMW_BALLOON_STAT_RESET]		= "reset",
+};
+
+static int vmballoon_enable_stats(struct vmballoon *b)
+{
+	int r = 0;
+
+	down_write(&b->conf_sem);
+
+	/* did we somehow race with another reader which enabled stats? */
+	if (b->stats)
+		goto out;
+
+	b->stats = kzalloc(sizeof(*b->stats), GFP_KERNEL);
+
+	if (!b->stats) {
+		/* allocation failed */
+		r = -ENOMEM;
+		goto out;
+	}
+	static_key_enable(&balloon_stat_enabled.key);
+out:
+	up_write(&b->conf_sem);
+	return r;
+}
+
+/**
+ * vmballoon_debug_show - shows statistics of balloon operations.
+ * @f: pointer to the &struct seq_file.
+ * @offset: ignored.
+ *
+ * Provides the statistics that can be accessed in vmmemctl in the debugfs.
+ * To avoid the overhead - mainly that of memory - of collecting the statistics,
+ * we only collect statistics after the first time the counters are read.
+ *
+ * Return: zero on success or an error code.
+ */
 static int vmballoon_debug_show(struct seq_file *f, void *offset)
 {
 	struct vmballoon *b = f->private;
-	struct vmballoon_stats *stats = &b->stats;
+	int i, j;
+
+	/* enables stats if they are disabled */
+	if (!b->stats) {
+		int r = vmballoon_enable_stats(b);
+
+		if (r)
+			return r;
+	}
 
 	/* format capabilities info */
-	seq_printf(f,
-		   "balloon capabilities:   %#4x\n"
-		   "used capabilities:      %#4lx\n"
-		   "is resetting:           %c\n",
-		   VMW_BALLOON_CAPABILITIES, b->capabilities,
-		   b->reset_required ? 'y' : 'n');
+	seq_printf(f, "%-22s: %#16x\n", "balloon capabilities",
+		   VMW_BALLOON_CAPABILITIES);
+	seq_printf(f, "%-22s: %#16lx\n", "used capabilities", b->capabilities);
+	seq_printf(f, "%-22s: %16s\n", "is resetting",
+		   b->reset_required ? "y" : "n");
 
 	/* format size info */
-	seq_printf(f,
-		   "target:             %8d pages\n"
-		   "current:            %8d pages\n",
-		   b->target, b->size);
-
-	seq_printf(f,
-		   "\n"
-		   "timer:              %8u\n"
-		   "doorbell:           %8u\n"
-		   "start:              %8u (%4u failed)\n"
-		   "guestType:          %8u (%4u failed)\n"
-		   "2m-lock:            %8u (%4u failed)\n"
-		   "lock:               %8u (%4u failed)\n"
-		   "2m-unlock:          %8u (%4u failed)\n"
-		   "unlock:             %8u (%4u failed)\n"
-		   "target:             %8u (%4u failed)\n"
-		   "prim2mAlloc:        %8u (%4u failed)\n"
-		   "primNoSleepAlloc:   %8u (%4u failed)\n"
-		   "primCanSleepAlloc:  %8u (%4u failed)\n"
-		   "prim2mFree:         %8u\n"
-		   "primFree:           %8u\n"
-		   "err2mAlloc:         %8u\n"
-		   "errAlloc:           %8u\n"
-		   "err2mFree:          %8u\n"
-		   "errFree:            %8u\n"
-		   "doorbellSet:        %8u\n"
-		   "doorbellUnset:      %8u\n",
-		   stats->timer,
-		   stats->doorbell,
-		   stats->start, stats->start_fail,
-		   stats->guest_type, stats->guest_type_fail,
-		   stats->lock[true],  stats->lock_fail[true],
-		   stats->lock[false],  stats->lock_fail[false],
-		   stats->unlock[true], stats->unlock_fail[true],
-		   stats->unlock[false], stats->unlock_fail[false],
-		   stats->target, stats->target_fail,
-		   stats->alloc[true], stats->alloc_fail[true],
-		   stats->alloc[false], stats->alloc_fail[false],
-		   stats->sleep_alloc, stats->sleep_alloc_fail,
-		   stats->free[true],
-		   stats->free[false],
-		   stats->refused_alloc[true], stats->refused_alloc[false],
-		   stats->refused_free[true], stats->refused_free[false],
-		   stats->doorbell_set, stats->doorbell_unset);
+	seq_printf(f, "%-22s: %16lu\n", "target", READ_ONCE(b->target));
+	seq_printf(f, "%-22s: %16llu\n", "current", atomic64_read(&b->size));
+
+	for (i = 0; i < VMW_BALLOON_CMD_NUM; i++) {
+		if (vmballoon_cmd_names[i] == NULL)
+			continue;
+
+		seq_printf(f, "%-22s: %16llu (%llu failed)\n",
+			   vmballoon_cmd_names[i],
+			   atomic64_read(&b->stats->ops[i][VMW_BALLOON_OP_STAT]),
+			   atomic64_read(&b->stats->ops[i][VMW_BALLOON_OP_FAIL_STAT]));
+	}
+
+	for (i = 0; i < VMW_BALLOON_STAT_NUM; i++)
+		seq_printf(f, "%-22s: %16llu\n",
+			   vmballoon_stat_names[i],
+			   atomic64_read(&b->stats->general_stat[i]));
+
+	for (i = 0; i < VMW_BALLOON_PAGE_STAT_NUM; i++) {
+		for (j = 0; j < VMW_BALLOON_NUM_PAGE_SIZES; j++)
+			seq_printf(f, "%-18s(%s): %16llu\n",
+				   vmballoon_stat_page_names[i],
+				   vmballoon_page_size_names[j],
+				   atomic64_read(&b->stats->page_stat[i][j]));
+	}
 
 	return 0;
 }
@@ -1161,7 +1500,10 @@ static int __init vmballoon_debugfs_init(struct vmballoon *b)
 
 static void __exit vmballoon_debugfs_exit(struct vmballoon *b)
 {
+	static_key_disable(&balloon_stat_enabled.key);
 	debugfs_remove(b->dbg_entry);
+	kfree(b->stats);
+	b->stats = NULL;
 }
 
 #else
@@ -1179,8 +1521,9 @@ static inline void vmballoon_debugfs_exit(struct vmballoon *b)
 
 static int __init vmballoon_init(void)
 {
+	enum vmballoon_page_size_type page_size;
 	int error;
-	unsigned is_2m_pages;
+
 	/*
 	 * Check if we are running on VMware's hypervisor and bail out
 	 * if we are not.
@@ -1188,11 +1531,10 @@ static int __init vmballoon_init(void)
 	if (x86_hyper_type != X86_HYPER_VMWARE)
 		return -ENODEV;
 
-	for (is_2m_pages = 0; is_2m_pages < VMW_BALLOON_NUM_PAGE_SIZES;
-			is_2m_pages++) {
-		INIT_LIST_HEAD(&balloon.page_sizes[is_2m_pages].pages);
-		INIT_LIST_HEAD(&balloon.page_sizes[is_2m_pages].refused_pages);
-	}
+	for (page_size = VMW_BALLOON_4K_PAGE;
+	     page_size <= VMW_BALLOON_LAST_SIZE; page_size++)
+		INIT_LIST_HEAD(&balloon.page_sizes[page_size].pages);
+
 
 	INIT_DELAYED_WORK(&balloon.dwork, vmballoon_work);
 
@@ -1200,6 +1542,8 @@ static int __init vmballoon_init(void)
 	if (error)
 		return error;
 
+	spin_lock_init(&balloon.comm_lock);
+	init_rwsem(&balloon.conf_sem);
 	balloon.vmci_doorbell = VMCI_INVALID_HANDLE;
 	balloon.batch_page = NULL;
 	balloon.page = NULL;
diff --git a/drivers/misc/vmw_vmci/vmci_driver.c b/drivers/misc/vmw_vmci/vmci_driver.c
index d7eaf1eb11e7..003bfba40758 100644
--- a/drivers/misc/vmw_vmci/vmci_driver.c
+++ b/drivers/misc/vmw_vmci/vmci_driver.c
@@ -113,5 +113,5 @@ module_exit(vmci_drv_exit);
 
 MODULE_AUTHOR("VMware, Inc.");
 MODULE_DESCRIPTION("VMware Virtual Machine Communication Interface.");
-MODULE_VERSION("1.1.5.0-k");
+MODULE_VERSION("1.1.6.0-k");
 MODULE_LICENSE("GPL v2");
diff --git a/drivers/misc/vmw_vmci/vmci_host.c b/drivers/misc/vmw_vmci/vmci_host.c
index 83e0c95d20a4..edfffc9699ba 100644
--- a/drivers/misc/vmw_vmci/vmci_host.c
+++ b/drivers/misc/vmw_vmci/vmci_host.c
@@ -15,7 +15,6 @@
 
 #include <linux/vmw_vmci_defs.h>
 #include <linux/vmw_vmci_api.h>
-#include <linux/moduleparam.h>
 #include <linux/miscdevice.h>
 #include <linux/interrupt.h>
 #include <linux/highmem.h>
@@ -448,15 +447,12 @@ static int vmci_host_do_alloc_queuepair(struct vmci_host_dev *vmci_host_dev,
 	struct vmci_handle handle;
 	int vmci_status;
 	int __user *retptr;
-	u32 cid;
 
 	if (vmci_host_dev->ct_type != VMCIOBJ_CONTEXT) {
 		vmci_ioctl_err("only valid for contexts\n");
 		return -EINVAL;
 	}
 
-	cid = vmci_ctx_get_id(vmci_host_dev->context);
-
 	if (vmci_host_dev->user_version < VMCI_VERSION_NOVMVM) {
 		struct vmci_qp_alloc_info_vmvm alloc_info;
 		struct vmci_qp_alloc_info_vmvm __user *info = uptr;
diff --git a/drivers/misc/vmw_vmci/vmci_resource.c b/drivers/misc/vmw_vmci/vmci_resource.c
index 1ab6e8737a5f..da1ee2e1ba99 100644
--- a/drivers/misc/vmw_vmci/vmci_resource.c
+++ b/drivers/misc/vmw_vmci/vmci_resource.c
@@ -57,7 +57,8 @@ static struct vmci_resource *vmci_resource_lookup(struct vmci_handle handle,
 
 		if (r->type == type &&
 		    rid == handle.resource &&
-		    (cid == handle.context || cid == VMCI_INVALID_ID)) {
+		    (cid == handle.context || cid == VMCI_INVALID_ID ||
+		     handle.context == VMCI_INVALID_ID)) {
 			resource = r;
 			break;
 		}