Merge tag 'xtensa-20191201' of git://github.com/jcmvbkbc/linux-xtensa

Pull Xtensa updates from Max Filippov:

 - add support for execute in place (XIP) kernels

 - improvements in inline assembly: use named arguments and "m"
   constraints where possible

 - improve stack dumping

 - clean up system_call code and syscall tracing

 - various small fixes and cleanups

* tag 'xtensa-20191201' of git://github.com/jcmvbkbc/linux-xtensa: (30 commits)
  xtensa: clean up system_call/xtensa_rt_sigreturn interaction
  xtensa: fix system_call interaction with ptrace
  xtensa: rearrange syscall tracing
  xtensa: fix syscall_set_return_value
  xtensa: drop unneeded headers from coprocessor.S
  xtensa: entry: Remove unneeded need_resched() loop
  xtensa: use MEMBLOCK_ALLOC_ANYWHERE for KASAN shadow map
  xtensa: fix TLB sanity checker
  xtensa: get rid of __ARCH_USE_5LEVEL_HACK
  xtensa: mm: fix PMD folding implementation
  xtensa: make stack dump size configurable
  xtensa: improve stack dumping
  xtensa: use "m" constraint instead of "r" in futex.h assembly
  xtensa: use "m" constraint instead of "a" in cmpxchg.h assembly
  xtensa: use named assembly arguments in cmpxchg.h
  xtensa: use "m" constraint instead of "a" in atomic.h assembly
  xtensa: use named assembly arguments in atomic.h
  xtensa: use "m" constraint instead of "a" in bitops.h assembly
  xtensa: use named assembly arguments in bitops.h
  xtensa: use macros to generate *_bit and test_and_*_bit functions
  ...

37 files changed, 772 insertions, 663 deletions

diff --git a/Documentation/features/core/tracehook/arch-support.txt b/Documentation/features/core/tracehook/arch-support.txt
index d344b99aae1e..964667052eda 100644
--- a/Documentation/features/core/tracehook/arch-support.txt
+++ b/Documentation/features/core/tracehook/arch-support.txt
@@ -30,5 +30,5 @@
     |          um: | TODO |
     |   unicore32: | TODO |
     |         x86: |  ok  |
-    |      xtensa: | TODO |
+    |      xtensa: |  ok  |
     -----------------------
diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index c95a34702242..4a3fa295d8fe 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -21,8 +21,8 @@ config XTENSA
 	select GENERIC_PCI_IOMAP
 	select GENERIC_SCHED_CLOCK
 	select GENERIC_STRNCPY_FROM_USER if KASAN
-	select HAVE_ARCH_JUMP_LABEL
-	select HAVE_ARCH_KASAN if MMU
+	select HAVE_ARCH_JUMP_LABEL if !XIP_KERNEL
+	select HAVE_ARCH_KASAN if MMU && !XIP_KERNEL
 	select HAVE_ARCH_TRACEHOOK
 	select HAVE_DEBUG_KMEMLEAK
 	select HAVE_DMA_CONTIGUOUS
@@ -215,151 +215,6 @@ config HOTPLUG_CPU
 
 	  Say N if you want to disable CPU hotplug.
 
-config INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
-	bool "Initialize Xtensa MMU inside the Linux kernel code"
-	depends on !XTENSA_VARIANT_FSF && !XTENSA_VARIANT_DC232B
-	default y if XTENSA_VARIANT_DC233C || XTENSA_VARIANT_CUSTOM
-	help
-	  Earlier version initialized the MMU in the exception vector
-	  before jumping to _startup in head.S and had an advantage that
-	  it was possible to place a software breakpoint at 'reset' and
-	  then enter your normal kernel breakpoints once the MMU was mapped
-	  to the kernel mappings (0XC0000000).
-
-	  This unfortunately won't work for U-Boot and likely also wont
-	  work for using KEXEC to have a hot kernel ready for doing a
-	  KDUMP.
-
-	  So now the MMU is initialized in head.S but it's necessary to
-	  use hardware breakpoints (gdb 'hbreak' cmd) to break at _startup.
-	  xt-gdb can't place a Software Breakpoint in the  0XD region prior
-	  to mapping the MMU and after mapping even if the area of low memory
-	  was mapped gdb wouldn't remove the breakpoint on hitting it as the
-	  PC wouldn't match. Since Hardware Breakpoints are recommended for
-	  Linux configurations it seems reasonable to just assume they exist
-	  and leave this older mechanism for unfortunate souls that choose
-	  not to follow Tensilica's recommendation.
-
-	  Selecting this will cause U-Boot to set the KERNEL Load and Entry
-	  address at 0x00003000 instead of the mapped std of 0xD0003000.
-
-	  If in doubt, say Y.
-
-config MEMMAP_CACHEATTR
-	hex "Cache attributes for the memory address space"
-	depends on !MMU
-	default 0x22222222
-	help
-	  These cache attributes are set up for noMMU systems. Each hex digit
-	  specifies cache attributes for the corresponding 512MB memory
-	  region: bits 0..3 -- for addresses 0x00000000..0x1fffffff,
-	  bits 4..7 -- for addresses 0x20000000..0x3fffffff, and so on.
-
-	  Cache attribute values are specific for the MMU type.
-	  For region protection MMUs:
-	    1: WT cached,
-	    2: cache bypass,
-	    4: WB cached,
-	    f: illegal.
-	  For ful MMU:
-	    bit 0: executable,
-	    bit 1: writable,
-	    bits 2..3:
-	      0: cache bypass,
-	      1: WB cache,
-	      2: WT cache,
-	      3: special (c and e are illegal, f is reserved).
-	  For MPU:
-	    0: illegal,
-	    1: WB cache,
-	    2: WB, no-write-allocate cache,
-	    3: WT cache,
-	    4: cache bypass.
-
-config KSEG_PADDR
-	hex "Physical address of the KSEG mapping"
-	depends on INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX && MMU
-	default 0x00000000
-	help
-	  This is the physical address where KSEG is mapped. Please refer to
-	  the chosen KSEG layout help for the required address alignment.
-	  Unpacked kernel image (including vectors) must be located completely
-	  within KSEG.
-	  Physical memory below this address is not available to linux.
-
-	  If unsure, leave the default value here.
-
-config KERNEL_LOAD_ADDRESS
-	hex "Kernel load address"
-	default 0x60003000 if !MMU
-	default 0x00003000 if MMU && INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
-	default 0xd0003000 if MMU && !INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
-	help
-	  This is the address where the kernel is loaded.
-	  It is virtual address for MMUv2 configurations and physical address
-	  for all other configurations.
-
-	  If unsure, leave the default value here.
-
-config VECTORS_OFFSET
-	hex "Kernel vectors offset"
-	default 0x00003000
-	help
-	  This is the offset of the kernel image from the relocatable vectors
-	  base.
-
-	  If unsure, leave the default value here.
-
-choice
-	prompt "KSEG layout"
-	depends on MMU
-	default XTENSA_KSEG_MMU_V2
-
-config XTENSA_KSEG_MMU_V2
-	bool "MMUv2: 128MB cached + 128MB uncached"
-	help
-	  MMUv2 compatible kernel memory map: TLB way 5 maps 128MB starting
-	  at KSEG_PADDR to 0xd0000000 with cache and to 0xd8000000
-	  without cache.
-	  KSEG_PADDR must be aligned to 128MB.
-
-config XTENSA_KSEG_256M
-	bool "256MB cached + 256MB uncached"
-	depends on INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
-	help
-	  TLB way 6 maps 256MB starting at KSEG_PADDR to 0xb0000000
-	  with cache and to 0xc0000000 without cache.
-	  KSEG_PADDR must be aligned to 256MB.
-
-config XTENSA_KSEG_512M
-	bool "512MB cached + 512MB uncached"
-	depends on INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
-	help
-	  TLB way 6 maps 512MB starting at KSEG_PADDR to 0xa0000000
-	  with cache and to 0xc0000000 without cache.
-	  KSEG_PADDR must be aligned to 256MB.
-
-endchoice
-
-config HIGHMEM
-	bool "High Memory Support"
-	depends on MMU
-	help
-	  Linux can use the full amount of RAM in the system by
-	  default. However, the default MMUv2 setup only maps the
-	  lowermost 128 MB of memory linearly to the areas starting
-	  at 0xd0000000 (cached) and 0xd8000000 (uncached).
-	  When there are more than 128 MB memory in the system not
-	  all of it can be "permanently mapped" by the kernel.
-	  The physical memory that's not permanently mapped is called
-	  "high memory".
-
-	  If you are compiling a kernel which will never run on a
-	  machine with more than 128 MB total physical RAM, answer
-	  N here.
-
-	  If unsure, say Y.
-
 config FAST_SYSCALL_XTENSA
 	bool "Enable fast atomic syscalls"
 	default n
@@ -446,6 +301,9 @@ config XTENSA_CALIBRATE_CCOUNT
 config SERIAL_CONSOLE
 	def_bool n
 
+config PLATFORM_HAVE_XIP
+	def_bool n
+
 menu "Platform options"
 
 choice
@@ -472,6 +330,7 @@ config XTENSA_PLATFORM_XTFPGA
 	select PLATFORM_WANT_DEFAULT_MEM if !MMU
 	select SERIAL_CONSOLE
 	select XTENSA_CALIBRATE_CCOUNT
+	select PLATFORM_HAVE_XIP
 	help
 	  XTFPGA is the name of Tensilica board family (LX60, LX110, LX200, ML605).
 	  This hardware is capable of running a full Linux distribution.
@@ -563,34 +422,6 @@ config SIMDISK1_FILENAME
 	  Another simulated disk in a host file for a buildroot-independent
 	  storage.
 
-config FORCE_MAX_ZONEORDER
-	int "Maximum zone order"
-	default "11"
-	help
-	  The kernel memory allocator divides physically contiguous memory
-	  blocks into "zones", where each zone is a power of two number of
-	  pages.  This option selects the largest power of two that the kernel
-	  keeps in the memory allocator.  If you need to allocate very large
-	  blocks of physically contiguous memory, then you may need to
-	  increase this value.
-
-	  This config option is actually maximum order plus one. For example,
-	  a value of 11 means that the largest free memory block is 2^10 pages.
-
-config PLATFORM_WANT_DEFAULT_MEM
-	def_bool n
-
-config DEFAULT_MEM_START
-	hex
-	prompt "PAGE_OFFSET/PHYS_OFFSET" if !MMU && PLATFORM_WANT_DEFAULT_MEM
-	default 0x60000000 if PLATFORM_WANT_DEFAULT_MEM
-	default 0x00000000
-	help
-	  This is the base address used for both PAGE_OFFSET and PHYS_OFFSET
-	  in noMMU configurations.
-
-	  If unsure, leave the default value here.
-
 config XTFPGA_LCD
 	bool "Enable XTFPGA LCD driver"
 	depends on XTENSA_PLATFORM_XTFPGA
@@ -621,6 +452,221 @@ config XTFPGA_LCD_8BIT_ACCESS
 	  only be used with 8-bit interface. Please consult prototyping user
 	  guide for your board for the correct interface width.
 
+comment "Kernel memory layout"
+
+config INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
+	bool "Initialize Xtensa MMU inside the Linux kernel code"
+	depends on !XTENSA_VARIANT_FSF && !XTENSA_VARIANT_DC232B
+	default y if XTENSA_VARIANT_DC233C || XTENSA_VARIANT_CUSTOM
+	help
+	  Earlier version initialized the MMU in the exception vector
+	  before jumping to _startup in head.S and had an advantage that
+	  it was possible to place a software breakpoint at 'reset' and
+	  then enter your normal kernel breakpoints once the MMU was mapped
+	  to the kernel mappings (0XC0000000).
+
+	  This unfortunately won't work for U-Boot and likely also wont
+	  work for using KEXEC to have a hot kernel ready for doing a
+	  KDUMP.
+
+	  So now the MMU is initialized in head.S but it's necessary to
+	  use hardware breakpoints (gdb 'hbreak' cmd) to break at _startup.
+	  xt-gdb can't place a Software Breakpoint in the  0XD region prior
+	  to mapping the MMU and after mapping even if the area of low memory
+	  was mapped gdb wouldn't remove the breakpoint on hitting it as the
+	  PC wouldn't match. Since Hardware Breakpoints are recommended for
+	  Linux configurations it seems reasonable to just assume they exist
+	  and leave this older mechanism for unfortunate souls that choose
+	  not to follow Tensilica's recommendation.
+
+	  Selecting this will cause U-Boot to set the KERNEL Load and Entry
+	  address at 0x00003000 instead of the mapped std of 0xD0003000.
+
+	  If in doubt, say Y.
+
+config XIP_KERNEL
+	bool "Kernel Execute-In-Place from ROM"
+	depends on PLATFORM_HAVE_XIP
+	help
+	  Execute-In-Place allows the kernel to run from non-volatile storage
+	  directly addressable by the CPU, such as NOR flash. This saves RAM
+	  space since the text section of the kernel is not loaded from flash
+	  to RAM. Read-write sections, such as the data section and stack,
+	  are still copied to RAM. The XIP kernel is not compressed since
+	  it has to run directly from flash, so it will take more space to
+	  store it. The flash address used to link the kernel object files,
+	  and for storing it, is configuration dependent. Therefore, if you
+	  say Y here, you must know the proper physical address where to
+	  store the kernel image depending on your own flash memory usage.
+
+	  Also note that the make target becomes "make xipImage" rather than
+	  "make Image" or "make uImage". The final kernel binary to put in
+	  ROM memory will be arch/xtensa/boot/xipImage.
+
+	  If unsure, say N.
+
+config MEMMAP_CACHEATTR
+	hex "Cache attributes for the memory address space"
+	depends on !MMU
+	default 0x22222222
+	help
+	  These cache attributes are set up for noMMU systems. Each hex digit
+	  specifies cache attributes for the corresponding 512MB memory
+	  region: bits 0..3 -- for addresses 0x00000000..0x1fffffff,
+	  bits 4..7 -- for addresses 0x20000000..0x3fffffff, and so on.
+
+	  Cache attribute values are specific for the MMU type.
+	  For region protection MMUs:
+	    1: WT cached,
+	    2: cache bypass,
+	    4: WB cached,
+	    f: illegal.
+	  For ful MMU:
+	    bit 0: executable,
+	    bit 1: writable,
+	    bits 2..3:
+	      0: cache bypass,
+	      1: WB cache,
+	      2: WT cache,
+	      3: special (c and e are illegal, f is reserved).
+	  For MPU:
+	    0: illegal,
+	    1: WB cache,
+	    2: WB, no-write-allocate cache,
+	    3: WT cache,
+	    4: cache bypass.
+
+config KSEG_PADDR
+	hex "Physical address of the KSEG mapping"
+	depends on INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX && MMU
+	default 0x00000000
+	help
+	  This is the physical address where KSEG is mapped. Please refer to
+	  the chosen KSEG layout help for the required address alignment.
+	  Unpacked kernel image (including vectors) must be located completely
+	  within KSEG.
+	  Physical memory below this address is not available to linux.
+
+	  If unsure, leave the default value here.
+
+config KERNEL_VIRTUAL_ADDRESS
+	hex "Kernel virtual address"
+	depends on MMU && XIP_KERNEL
+	default 0xd0003000
+	help
+	  This is the virtual address where the XIP kernel is mapped.
+	  XIP kernel may be mapped into KSEG or KIO region, virtual address
+	  provided here must match kernel load address provided in
+	  KERNEL_LOAD_ADDRESS.
+
+config KERNEL_LOAD_ADDRESS
+	hex "Kernel load address"
+	default 0x60003000 if !MMU
+	default 0x00003000 if MMU && INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
+	default 0xd0003000 if MMU && !INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
+	help
+	  This is the address where the kernel is loaded.
+	  It is virtual address for MMUv2 configurations and physical address
+	  for all other configurations.
+
+	  If unsure, leave the default value here.
+
+config VECTORS_OFFSET
+	hex "Kernel vectors offset"
+	default 0x00003000
+	depends on !XIP_KERNEL
+	help
+	  This is the offset of the kernel image from the relocatable vectors
+	  base.
+
+	  If unsure, leave the default value here.
+
+config XIP_DATA_ADDR
+	hex "XIP kernel data virtual address"
+	depends on XIP_KERNEL
+	default 0x00000000
+	help
+	  This is the virtual address where XIP kernel data is copied.
+	  It must be within KSEG if MMU is used.
+
+config PLATFORM_WANT_DEFAULT_MEM
+	def_bool n
+
+config DEFAULT_MEM_START
+	hex
+	prompt "PAGE_OFFSET/PHYS_OFFSET" if !MMU && PLATFORM_WANT_DEFAULT_MEM
+	default 0x60000000 if PLATFORM_WANT_DEFAULT_MEM
+	default 0x00000000
+	help
+	  This is the base address used for both PAGE_OFFSET and PHYS_OFFSET
+	  in noMMU configurations.
+
+	  If unsure, leave the default value here.
+
+choice
+	prompt "KSEG layout"
+	depends on MMU
+	default XTENSA_KSEG_MMU_V2
+
+config XTENSA_KSEG_MMU_V2
+	bool "MMUv2: 128MB cached + 128MB uncached"
+	help
+	  MMUv2 compatible kernel memory map: TLB way 5 maps 128MB starting
+	  at KSEG_PADDR to 0xd0000000 with cache and to 0xd8000000
+	  without cache.
+	  KSEG_PADDR must be aligned to 128MB.
+
+config XTENSA_KSEG_256M
+	bool "256MB cached + 256MB uncached"
+	depends on INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
+	help
+	  TLB way 6 maps 256MB starting at KSEG_PADDR to 0xb0000000
+	  with cache and to 0xc0000000 without cache.
+	  KSEG_PADDR must be aligned to 256MB.
+
+config XTENSA_KSEG_512M
+	bool "512MB cached + 512MB uncached"
+	depends on INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
+	help
+	  TLB way 6 maps 512MB starting at KSEG_PADDR to 0xa0000000
+	  with cache and to 0xc0000000 without cache.
+	  KSEG_PADDR must be aligned to 256MB.
+
+endchoice
+
+config HIGHMEM
+	bool "High Memory Support"
+	depends on MMU
+	help
+	  Linux can use the full amount of RAM in the system by
+	  default. However, the default MMUv2 setup only maps the
+	  lowermost 128 MB of memory linearly to the areas starting
+	  at 0xd0000000 (cached) and 0xd8000000 (uncached).
+	  When there are more than 128 MB memory in the system not
+	  all of it can be "permanently mapped" by the kernel.
+	  The physical memory that's not permanently mapped is called
+	  "high memory".
+
+	  If you are compiling a kernel which will never run on a
+	  machine with more than 128 MB total physical RAM, answer
+	  N here.
+
+	  If unsure, say Y.
+
+config FORCE_MAX_ZONEORDER
+	int "Maximum zone order"
+	default "11"
+	help
+	  The kernel memory allocator divides physically contiguous memory
+	  blocks into "zones", where each zone is a power of two number of
+	  pages.  This option selects the largest power of two that the kernel
+	  keeps in the memory allocator.  If you need to allocate very large
+	  blocks of physically contiguous memory, then you may need to
+	  increase this value.
+
+	  This config option is actually maximum order plus one. For example,
+	  a value of 11 means that the largest free memory block is 2^10 pages.
+
 endmenu
 
 menu "Power management options"
diff --git a/arch/xtensa/Kconfig.debug b/arch/xtensa/Kconfig.debug
index 39de98e20018..83cc8d12fa0e 100644
--- a/arch/xtensa/Kconfig.debug
+++ b/arch/xtensa/Kconfig.debug
@@ -31,3 +31,10 @@ config S32C1I_SELFTEST
 	  It is easy to make wrong hardware configuration, this test should catch it early.
 
 	  Say 'N' on stable hardware.
+
+config PRINT_STACK_DEPTH
+	int "Stack depth to print" if DEBUG_KERNEL
+	default 64
+	help
+	  This option allows you to set the stack depth that the kernel
+	  prints in stack traces.
diff --git a/arch/xtensa/Makefile b/arch/xtensa/Makefile
index 1542018c9e57..67a7d151d1e7 100644
--- a/arch/xtensa/Makefile
+++ b/arch/xtensa/Makefile
@@ -87,7 +87,7 @@ drivers-$(CONFIG_OPROFILE)	+= arch/xtensa/oprofile/
 
 boot		:= arch/xtensa/boot
 
-all Image zImage uImage: vmlinux
+all Image zImage uImage xipImage: vmlinux
 	$(Q)$(MAKE) $(build)=$(boot) $@
 
 archheaders:
@@ -97,4 +97,5 @@ define archhelp
   @echo '* Image       - Kernel ELF image with reset vector'
   @echo '* zImage      - Compressed kernel image (arch/xtensa/boot/images/zImage.*)'
   @echo '* uImage      - U-Boot wrapped image'
+  @echo '  xipImage    - XIP image'
 endef
diff --git a/arch/xtensa/boot/Makefile b/arch/xtensa/boot/Makefile
index 294846117fc2..efb91bfda2b4 100644
--- a/arch/xtensa/boot/Makefile
+++ b/arch/xtensa/boot/Makefile
@@ -29,6 +29,7 @@ all: $(boot-y)
 Image: boot-elf
 zImage: boot-redboot
 uImage: $(obj)/uImage
+xipImage: $(obj)/xipImage
 
 boot-elf boot-redboot: $(addprefix $(obj)/,$(subdir-y))
 	$(Q)$(MAKE) $(build)=$(obj)/$@ $(MAKECMDGOALS)
@@ -50,3 +51,7 @@ UIMAGE_COMPRESSION = gzip
 $(obj)/uImage: vmlinux.bin.gz FORCE
 	$(call if_changed,uimage)
 	$(Q)$(kecho) '  Kernel: $@ is ready'
+
+$(obj)/xipImage: vmlinux FORCE
+	$(call if_changed,objcopy)
+	$(Q)$(kecho) '  Kernel: $@ is ready'
diff --git a/arch/xtensa/configs/xip_kc705_defconfig b/arch/xtensa/configs/xip_kc705_defconfig
new file mode 100644
index 000000000000..f9e85c082afc
--- /dev/null
+++ b/arch/xtensa/configs/xip_kc705_defconfig
@@ -0,0 +1,119 @@
+CONFIG_SYSVIPC=y
+CONFIG_POSIX_MQUEUE=y
+CONFIG_NO_HZ_IDLE=y
+CONFIG_HIGH_RES_TIMERS=y
+CONFIG_PREEMPT=y
+CONFIG_IRQ_TIME_ACCOUNTING=y
+CONFIG_BSD_PROCESS_ACCT=y
+CONFIG_MEMCG=y
+CONFIG_CGROUP_FREEZER=y
+CONFIG_CGROUP_DEVICE=y
+CONFIG_CGROUP_CPUACCT=y
+CONFIG_CGROUP_DEBUG=y
+CONFIG_NAMESPACES=y
+CONFIG_SCHED_AUTOGROUP=y
+CONFIG_RELAY=y
+CONFIG_BLK_DEV_INITRD=y
+CONFIG_EXPERT=y
+CONFIG_KALLSYMS_ALL=y
+CONFIG_PROFILING=y
+CONFIG_XTENSA_VARIANT_DC233C=y
+CONFIG_XTENSA_UNALIGNED_USER=y
+CONFIG_XIP_KERNEL=y
+CONFIG_XIP_DATA_ADDR=0xd0000000
+CONFIG_KERNEL_VIRTUAL_ADDRESS=0xe6000000
+CONFIG_KERNEL_LOAD_ADDRESS=0xf6000000
+CONFIG_XTENSA_KSEG_512M=y
+CONFIG_HIGHMEM=y
+CONFIG_XTENSA_PLATFORM_XTFPGA=y
+CONFIG_CMDLINE_BOOL=y
+CONFIG_CMDLINE="earlycon=uart8250,mmio32native,0xfd050020,115200n8 console=ttyS0,115200n8 ip=dhcp root=/dev/nfs rw debug memmap=0x38000000@0"
+CONFIG_USE_OF=y
+CONFIG_BUILTIN_DTB_SOURCE="kc705"
+# CONFIG_PARSE_BOOTPARAM is not set
+CONFIG_OPROFILE=y
+# CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS is not set
+# CONFIG_COMPACTION is not set
+CONFIG_NET=y
+CONFIG_PACKET=y
+CONFIG_UNIX=y
+CONFIG_INET=y
+CONFIG_IP_MULTICAST=y
+CONFIG_IP_PNP=y
+CONFIG_IP_PNP_DHCP=y
+CONFIG_IP_PNP_BOOTP=y
+CONFIG_IP_PNP_RARP=y
+# CONFIG_IPV6 is not set
+CONFIG_NETFILTER=y
+# CONFIG_WIRELESS is not set
+CONFIG_UEVENT_HELPER=y
+CONFIG_DEVTMPFS=y
+CONFIG_DEVTMPFS_MOUNT=y
+# CONFIG_STANDALONE is not set
+CONFIG_BLK_DEV_LOOP=y
+CONFIG_BLK_DEV_RAM=y
+CONFIG_SCSI=y
+CONFIG_BLK_DEV_SD=y
+CONFIG_NETDEVICES=y
+# CONFIG_NET_VENDOR_ARC is not set
+# CONFIG_NET_VENDOR_AURORA is not set
+# CONFIG_NET_VENDOR_BROADCOM is not set
+# CONFIG_NET_VENDOR_INTEL is not set
+# CONFIG_NET_VENDOR_MARVELL is not set
+# CONFIG_NET_VENDOR_MICREL is not set
+# CONFIG_NET_VENDOR_NATSEMI is not set
+# CONFIG_NET_VENDOR_SAMSUNG is not set
+# CONFIG_NET_VENDOR_SEEQ is not set
+# CONFIG_NET_VENDOR_SMSC is not set
+# CONFIG_NET_VENDOR_STMICRO is not set
+# CONFIG_NET_VENDOR_VIA is not set
+# CONFIG_NET_VENDOR_WIZNET is not set
+CONFIG_MARVELL_PHY=y
+# CONFIG_WLAN is not set
+# CONFIG_INPUT_KEYBOARD is not set
+# CONFIG_INPUT_MOUSE is not set
+# CONFIG_SERIO is not set
+CONFIG_DEVKMEM=y
+CONFIG_SERIAL_8250=y
+# CONFIG_SERIAL_8250_DEPRECATED_OPTIONS is not set
+CONFIG_SERIAL_8250_CONSOLE=y
+CONFIG_SERIAL_OF_PLATFORM=y
+# CONFIG_HWMON is not set
+CONFIG_WATCHDOG=y
+CONFIG_WATCHDOG_NOWAYOUT=y
+CONFIG_SOFT_WATCHDOG=y
+# CONFIG_VGA_CONSOLE is not set
+# CONFIG_USB_SUPPORT is not set
+# CONFIG_IOMMU_SUPPORT is not set
+CONFIG_EXT3_FS=y
+CONFIG_FANOTIFY=y
+CONFIG_VFAT_FS=y
+CONFIG_PROC_KCORE=y
+CONFIG_TMPFS=y
+CONFIG_TMPFS_POSIX_ACL=y
+CONFIG_NFS_FS=y
+CONFIG_NFS_V4=y
+CONFIG_NFS_SWAP=y
+CONFIG_ROOT_NFS=y
+CONFIG_SUNRPC_DEBUG=y
+CONFIG_NLS_CODEPAGE_437=y
+CONFIG_NLS_ISO8859_1=y
+CONFIG_CRYPTO_ECHAINIV=y
+CONFIG_CRYPTO_DEFLATE=y
+CONFIG_CRYPTO_LZO=y
+CONFIG_CRYPTO_ANSI_CPRNG=y
+CONFIG_PRINTK_TIME=y
+CONFIG_DYNAMIC_DEBUG=y
+CONFIG_DEBUG_INFO=y
+CONFIG_MAGIC_SYSRQ=y
+CONFIG_DETECT_HUNG_TASK=y
+# CONFIG_SCHED_DEBUG is not set
+CONFIG_SCHEDSTATS=y
+CONFIG_DEBUG_RT_MUTEXES=y
+CONFIG_DEBUG_SPINLOCK=y
+CONFIG_DEBUG_MUTEXES=y
+CONFIG_DEBUG_ATOMIC_SLEEP=y
+CONFIG_STACKTRACE=y
+CONFIG_RCU_TRACE=y
+# CONFIG_FTRACE is not set
+# CONFIG_S32C1I_SELFTEST is not set
diff --git a/arch/xtensa/include/asm/Kbuild b/arch/xtensa/include/asm/Kbuild
index ffa0cf7f66c3..3acc31e55e02 100644
--- a/arch/xtensa/include/asm/Kbuild
+++ b/arch/xtensa/include/asm/Kbuild
@@ -11,6 +11,7 @@ generic-y += exec.h
 generic-y += extable.h
 generic-y += fb.h
 generic-y += hardirq.h
+generic-y += hw_irq.h
 generic-y += irq_regs.h
 generic-y += irq_work.h
 generic-y += kdebug.h
@@ -30,6 +31,7 @@ generic-y += qspinlock.h
 generic-y += sections.h
 generic-y += topology.h
 generic-y += trace_clock.h
+generic-y += user.h
 generic-y += vga.h
 generic-y += word-at-a-time.h
 generic-y += xor.h
diff --git a/arch/xtensa/include/asm/atomic.h b/arch/xtensa/include/asm/atomic.h
index 7b00d26f472e..3e7c6134ed32 100644
--- a/arch/xtensa/include/asm/atomic.h
+++ b/arch/xtensa/include/asm/atomic.h
@@ -64,13 +64,13 @@ static inline void atomic_##op(int i, atomic_t *v)			\
 	int result;							\
 									\
 	__asm__ __volatile__(						\
-			"1:     l32ex   %1, %3\n"			\
-			"       " #op " %0, %1, %2\n"			\
-			"       s32ex   %0, %3\n"			\
-			"       getex   %0\n"				\
-			"       beqz    %0, 1b\n"			\
-			: "=&a" (result), "=&a" (tmp)			\
-			: "a" (i), "a" (v)				\
+			"1:     l32ex   %[tmp], %[addr]\n"		\
+			"       " #op " %[result], %[tmp], %[i]\n"	\
+			"       s32ex   %[result], %[addr]\n"		\
+			"       getex   %[result]\n"			\
+			"       beqz    %[result], 1b\n"		\
+			: [result] "=&a" (result), [tmp] "=&a" (tmp)	\
+			: [i] "a" (i), [addr] "a" (v)			\
 			: "memory"					\
 			);						\
 }									\
@@ -82,14 +82,14 @@ static inline int atomic_##op##_return(int i, atomic_t *v)		\
 	int result;							\
 									\
 	__asm__ __volatile__(						\
-			"1:     l32ex   %1, %3\n"			\
-			"       " #op " %0, %1, %2\n"			\
-			"       s32ex   %0, %3\n"			\
-			"       getex   %0\n"				\
-			"       beqz    %0, 1b\n"			\
-			"       " #op " %0, %1, %2\n"			\
-			: "=&a" (result), "=&a" (tmp)			\
-			: "a" (i), "a" (v)				\
+			"1:     l32ex   %[tmp], %[addr]\n"		\
+			"       " #op " %[result], %[tmp], %[i]\n"	\
+			"       s32ex   %[result], %[addr]\n"		\
+			"       getex   %[result]\n"			\
+			"       beqz    %[result], 1b\n"		\
+			"       " #op " %[result], %[tmp], %[i]\n"	\
+			: [result] "=&a" (result), [tmp] "=&a" (tmp)	\
+			: [i] "a" (i), [addr] "a" (v)			\
 			: "memory"					\
 			);						\
 									\
@@ -103,13 +103,13 @@ static inline int atomic_fetch_##op(int i, atomic_t *v)			\
 	int result;							\
 									\
 	__asm__ __volatile__(						\
-			"1:     l32ex   %1, %3\n"			\
-			"       " #op " %0, %1, %2\n"			\
-			"       s32ex   %0, %3\n"			\
-			"       getex   %0\n"				\
-			"       beqz    %0, 1b\n"			\
-			: "=&a" (result), "=&a" (tmp)			\
-			: "a" (i), "a" (v)				\
+			"1:     l32ex   %[tmp], %[addr]\n"		\
+			"       " #op " %[result], %[tmp], %[i]\n"	\
+			"       s32ex   %[result], %[addr]\n"		\
+			"       getex   %[result]\n"			\
+			"       beqz    %[result], 1b\n"		\
+			: [result] "=&a" (result), [tmp] "=&a" (tmp)	\
+			: [i] "a" (i), [addr] "a" (v)			\
 			: "memory"					\
 			);						\
 									\
@@ -124,13 +124,14 @@ static inline void atomic_##op(int i, atomic_t * v)			\
 	int result;							\
 									\
 	__asm__ __volatile__(						\
-			"1:     l32i    %1, %3, 0\n"			\
-			"       wsr     %1, scompare1\n"		\
-			"       " #op " %0, %1, %2\n"			\
-			"       s32c1i  %0, %3, 0\n"			\
-			"       bne     %0, %1, 1b\n"			\
-			: "=&a" (result), "=&a" (tmp)			\
-			: "a" (i), "a" (v)				\
+			"1:     l32i    %[tmp], %[mem]\n"		\
+			"       wsr     %[tmp], scompare1\n"		\
+			"       " #op " %[result], %[tmp], %[i]\n"	\
+			"       s32c1i  %[result], %[mem]\n"		\
+			"       bne     %[result], %[tmp], 1b\n"	\
+			: [result] "=&a" (result), [tmp] "=&a" (tmp),	\
+			  [mem] "+m" (*v)				\
+			: [i] "a" (i)					\
 			: "memory"					\
 			);						\
 }									\
@@ -142,14 +143,15 @@ static inline int atomic_##op##_return(int i, atomic_t * v)		\
 	int result;							\
 									\
 	__asm__ __volatile__(						\
-			"1:     l32i    %1, %3, 0\n"			\
-			"       wsr     %1, scompare1\n"		\
-			"       " #op " %0, %1, %2\n"			\
-			"       s32c1i  %0, %3, 0\n"			\
-			"       bne     %0, %1, 1b\n"			\
-			"       " #op " %0, %0, %2\n"			\
-			: "=&a" (result), "=&a" (tmp)			\
-			: "a" (i), "a" (v)				\
+			"1:     l32i    %[tmp], %[mem]\n"		\
+			"       wsr     %[tmp], scompare1\n"		\
+			"       " #op " %[result], %[tmp], %[i]\n"	\
+			"       s32c1i  %[result], %[mem]\n"		\
+			"       bne     %[result], %[tmp], 1b\n"	\
+			"       " #op " %[result], %[result], %[i]\n"	\
+			: [result] "=&a" (result), [tmp] "=&a" (tmp),	\
+			  [mem] "+m" (*v)				\
+			: [i] "a" (i)					\
 			: "memory"					\
 			);						\
 									\
@@ -163,13 +165,14 @@ static inline int atomic_fetch_##op(int i, atomic_t * v)		\
 	int result;							\
 									\
 	__asm__ __volatile__(						\
-			"1:     l32i    %1, %3, 0\n"			\
-			"       wsr     %1, scompare1\n"		\
-			"       " #op " %0, %1, %2\n"			\
-			"       s32c1i  %0, %3, 0\n"			\
-			"       bne     %0, %1, 1b\n"			\
-			: "=&a" (result), "=&a" (tmp)			\
-			: "a" (i), "a" (v)				\
+			"1:     l32i    %[tmp], %[mem]\n"		\
+			"       wsr     %[tmp], scompare1\n"		\
+			"       " #op " %[result], %[tmp], %[i]\n"	\
+			"       s32c1i  %[result], %[mem]\n"		\
+			"       bne     %[result], %[tmp], 1b\n"	\
+			: [result] "=&a" (result), [tmp] "=&a" (tmp),	\
+			  [mem] "+m" (*v)				\
+			: [i] "a" (i)					\
 			: "memory"					\
 			);						\
 									\
@@ -184,14 +187,14 @@ static inline void atomic_##op(int i, atomic_t * v)			\
 	unsigned int vval;						\
 									\
 	__asm__ __volatile__(						\
-			"       rsil    a15, "__stringify(TOPLEVEL)"\n"\
-			"       l32i    %0, %2, 0\n"			\
-			"       " #op " %0, %0, %1\n"			\
-			"       s32i    %0, %2, 0\n"			\
+			"       rsil    a15, "__stringify(TOPLEVEL)"\n"	\
+			"       l32i    %[result], %[mem]\n"		\
+			"       " #op " %[result], %[result], %[i]\n"	\
+			"       s32i    %[result], %[mem]\n"		\
 			"       wsr     a15, ps\n"			\
 			"       rsync\n"				\
-			: "=&a" (vval)					\
-			: "a" (i), "a" (v)				\
+			: [result] "=&a" (vval), [mem] "+m" (*v)	\
+			: [i] "a" (i)					\
 			: "a15", "memory"				\
 			);						\
 }									\
@@ -203,13 +206,13 @@ static inline int atomic_##op##_return(int i, atomic_t * v)		\
 									\
 	__asm__ __volatile__(						\
 			"       rsil    a15,"__stringify(TOPLEVEL)"\n"	\
-			"       l32i    %0, %2, 0\n"			\
-			"       " #op " %0, %0, %1\n"			\
-			"       s32i    %0, %2, 0\n"			\
+			"       l32i    %[result], %[mem]\n"		\
+			"       " #op " %[result], %[result], %[i]\n"	\
+			"       s32i    %[result], %[mem]\n"		\
 			"       wsr     a15, ps\n"			\
 			"       rsync\n"				\
-			: "=&a" (vval)					\
-			: "a" (i), "a" (v)				\
+			: [result] "=&a" (vval), [mem] "+m" (*v)	\
+			: [i] "a" (i)					\
 			: "a15", "memory"				\
 			);						\
 									\
@@ -223,13 +226,14 @@ static inline int atomic_fetch_##op(int i, atomic_t * v)		\
 									\
 	__asm__ __volatile__(						\
 			"       rsil    a15,"__stringify(TOPLEVEL)"\n"	\
-			"       l32i    %0, %3, 0\n"			\
-			"       " #op " %1, %0, %2\n"			\
-			"       s32i    %1, %3, 0\n"			\
+			"       l32i    %[result], %[mem]\n"		\
+			"       " #op " %[tmp], %[result], %[i]\n"	\
+			"       s32i    %[tmp], %[mem]\n"		\
 			"       wsr     a15, ps\n"			\
 			"       rsync\n"				\
-			: "=&a" (vval), "=&a" (tmp)			\
-			: "a" (i), "a" (v)				\
+			: [result] "=&a" (vval), [tmp] "=&a" (tmp),	\
+			  [mem] "+m" (*v)				\
+			: [i] "a" (i)					\
 			: "a15", "memory"				\
 			);						\
 									\
diff --git a/arch/xtensa/include/asm/bitops.h b/arch/xtensa/include/asm/bitops.h
index be8b2be5a98b..3f71d364ba90 100644
--- a/arch/xtensa/include/asm/bitops.h
+++ b/arch/xtensa/include/asm/bitops.h
@@ -98,247 +98,112 @@ static inline unsigned long __fls(unsigned long word)
 
 #if XCHAL_HAVE_EXCLUSIVE
 
-static inline void set_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32ex   %0, %2\n"
-			"       or      %0, %0, %1\n"
-			"       s32ex   %0, %2\n"
-			"       getex   %0\n"
-			"       beqz    %0, 1b\n"
-			: "=&a" (tmp)
-			: "a" (mask), "a" (p)
-			: "memory");
-}
-
-static inline void clear_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32ex   %0, %2\n"
-			"       and     %0, %0, %1\n"
-			"       s32ex   %0, %2\n"
-			"       getex   %0\n"
-			"       beqz    %0, 1b\n"
-			: "=&a" (tmp)
-			: "a" (~mask), "a" (p)
-			: "memory");
-}
-
-static inline void change_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32ex   %0, %2\n"
-			"       xor     %0, %0, %1\n"
-			"       s32ex   %0, %2\n"
-			"       getex   %0\n"
-			"       beqz    %0, 1b\n"
-			: "=&a" (tmp)
-			: "a" (mask), "a" (p)
-			: "memory");
-}
-
-static inline int
-test_and_set_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32ex   %1, %3\n"
-			"       or      %0, %1, %2\n"
-			"       s32ex   %0, %3\n"
-			"       getex   %0\n"
-			"       beqz    %0, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (mask), "a" (p)
-			: "memory");
-
-	return value & mask;
-}
-
-static inline int
-test_and_clear_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32ex   %1, %3\n"
-			"       and     %0, %1, %2\n"
-			"       s32ex   %0, %3\n"
-			"       getex   %0\n"
-			"       beqz    %0, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (~mask), "a" (p)
-			: "memory");
-
-	return value & mask;
-}
-
-static inline int
-test_and_change_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32ex   %1, %3\n"
-			"       xor     %0, %1, %2\n"
-			"       s32ex   %0, %3\n"
-			"       getex   %0\n"
-			"       beqz    %0, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (mask), "a" (p)
-			: "memory");
-
-	return value & mask;
+#define BIT_OP(op, insn, inv)						\
+static inline void op##_bit(unsigned int bit, volatile unsigned long *p)\
+{									\
+	unsigned long tmp;						\
+	unsigned long mask = 1UL << (bit & 31);				\
+									\
+	p += bit >> 5;							\
+									\
+	__asm__ __volatile__(						\
+			"1:     l32ex   %[tmp], %[addr]\n"		\
+			"      "insn"   %[tmp], %[tmp], %[mask]\n"	\
+			"       s32ex   %[tmp], %[addr]\n"		\
+			"       getex   %[tmp]\n"			\
+			"       beqz    %[tmp], 1b\n"			\
+			: [tmp] "=&a" (tmp)				\
+			: [mask] "a" (inv mask), [addr] "a" (p)		\
+			: "memory");					\
+}
+
+#define TEST_AND_BIT_OP(op, insn, inv)					\
+static inline int							\
+test_and_##op##_bit(unsigned int bit, volatile unsigned long *p)	\
+{									\
+	unsigned long tmp, value;					\
+	unsigned long mask = 1UL << (bit & 31);				\
+									\
+	p += bit >> 5;							\
+									\
+	__asm__ __volatile__(						\
+			"1:     l32ex   %[value], %[addr]\n"		\
+			"      "insn"   %[tmp], %[value], %[mask]\n"	\
+			"       s32ex   %[tmp], %[addr]\n"		\
+			"       getex   %[tmp]\n"			\
+			"       beqz    %[tmp], 1b\n"			\
+			: [tmp] "=&a" (tmp), [value] "=&a" (value)	\
+			: [mask] "a" (inv mask), [addr] "a" (p)		\
+			: "memory");					\
+									\
+	return value & mask;						\
 }
 
 #elif XCHAL_HAVE_S32C1I
 
-static inline void set_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %3, 0\n"
-			"       wsr     %1, scompare1\n"
-			"       or      %0, %1, %2\n"
-			"       s32c1i  %0, %3, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (mask), "a" (p)
-			: "memory");
-}
-
-static inline void clear_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %3, 0\n"
-			"       wsr     %1, scompare1\n"
-			"       and     %0, %1, %2\n"
-			"       s32c1i  %0, %3, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (~mask), "a" (p)
-			: "memory");
+#define BIT_OP(op, insn, inv)						\
+static inline void op##_bit(unsigned int bit, volatile unsigned long *p)\
+{									\
+	unsigned long tmp, value;					\
+	unsigned long mask = 1UL << (bit & 31);				\
+									\
+	p += bit >> 5;							\
+									\
+	__asm__ __volatile__(						\
+			"1:     l32i    %[value], %[mem]\n"		\
+			"       wsr     %[value], scompare1\n"		\
+			"      "insn"   %[tmp], %[value], %[mask]\n"	\
+			"       s32c1i  %[tmp], %[mem]\n"		\
+			"       bne     %[tmp], %[value], 1b\n"		\
+			: [tmp] "=&a" (tmp), [value] "=&a" (value),	\
+			  [mem] "+m" (*p)				\
+			: [mask] "a" (inv mask)				\
+			: "memory");					\
+}
+
+#define TEST_AND_BIT_OP(op, insn, inv)					\
+static inline int							\
+test_and_##op##_bit(unsigned int bit, volatile unsigned long *p)	\
+{									\
+	unsigned long tmp, value;					\
+	unsigned long mask = 1UL << (bit & 31);				\
+									\
+	p += bit >> 5;							\
+									\
+	__asm__ __volatile__(						\
+			"1:     l32i    %[value], %[mem]\n"		\
+			"       wsr     %[value], scompare1\n"		\
+			"      "insn"   %[tmp], %[value], %[mask]\n"	\
+			"       s32c1i  %[tmp], %[mem]\n"		\
+			"       bne     %[tmp], %[value], 1b\n"		\
+			: [tmp] "=&a" (tmp), [value] "=&a" (value),	\
+			  [mem] "+m" (*p)				\
+			: [mask] "a" (inv mask)				\
+			: "memory");					\
+									\
+	return tmp & mask;						\
 }
 
-static inline void change_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %3, 0\n"
-			"       wsr     %1, scompare1\n"
-			"       xor     %0, %1, %2\n"
-			"       s32c1i  %0, %3, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (mask), "a" (p)
-			: "memory");
-}
+#else
 
-static inline int
-test_and_set_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %3, 0\n"
-			"       wsr     %1, scompare1\n"
-			"       or      %0, %1, %2\n"
-			"       s32c1i  %0, %3, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (mask), "a" (p)
-			: "memory");
-
-	return tmp & mask;
-}
+#define BIT_OP(op, insn, inv)
+#define TEST_AND_BIT_OP(op, insn, inv)
 
-static inline int
-test_and_clear_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %3, 0\n"
-			"       wsr     %1, scompare1\n"
-			"       and     %0, %1, %2\n"
-			"       s32c1i  %0, %3, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (~mask), "a" (p)
-			: "memory");
-
-	return tmp & mask;
-}
+#include <asm-generic/bitops/atomic.h>
 
-static inline int
-test_and_change_bit(unsigned int bit, volatile unsigned long *p)
-{
-	unsigned long tmp, value;
-	unsigned long mask = 1UL << (bit & 31);
-
-	p += bit >> 5;
-
-	__asm__ __volatile__(
-			"1:     l32i    %1, %3, 0\n"
-			"       wsr     %1, scompare1\n"
-			"       xor     %0, %1, %2\n"
-			"       s32c1i  %0, %3, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (tmp), "=&a" (value)
-			: "a" (mask), "a" (p)
-			: "memory");
-
-	return tmp & mask;
-}
+#endif /* XCHAL_HAVE_S32C1I */
 
-#else
+#define BIT_OPS(op, insn, inv)		\
+	BIT_OP(op, insn, inv)		\
+	TEST_AND_BIT_OP(op, insn, inv)
 
-#include <asm-generic/bitops/atomic.h>
+BIT_OPS(set, "or", )
+BIT_OPS(clear, "and", ~)
+BIT_OPS(change, "xor", )
 
-#endif /* XCHAL_HAVE_S32C1I */
+#undef BIT_OPS
+#undef BIT_OP
+#undef TEST_AND_BIT_OP
 
 #include <asm-generic/bitops/find.h>
 #include <asm-generic/bitops/le.h>
diff --git a/arch/xtensa/include/asm/cache.h b/arch/xtensa/include/asm/cache.h
index b21fd133ff62..54e147ac26bf 100644
--- a/arch/xtensa/include/asm/cache.h
+++ b/arch/xtensa/include/asm/cache.h
@@ -31,4 +31,10 @@
 
 #define ARCH_DMA_MINALIGN	L1_CACHE_BYTES
 
+/*
+ * R/O after init is actually writable, it cannot go to .rodata
+ * according to vmlinux linker script.
+ */
+#define __ro_after_init __read_mostly
+
 #endif	/* _XTENSA_CACHE_H */
diff --git a/arch/xtensa/include/asm/cmpxchg.h b/arch/xtensa/include/asm/cmpxchg.h
index 7ccc5cbf441b..a175f8aec3fb 100644
--- a/arch/xtensa/include/asm/cmpxchg.h
+++ b/arch/xtensa/include/asm/cmpxchg.h
@@ -27,25 +27,25 @@ __cmpxchg_u32(volatile int *p, int old, int new)
 	unsigned long tmp, result;
 
 	__asm__ __volatile__(
-			"1:     l32ex   %0, %3\n"
-			"       bne     %0, %4, 2f\n"
-			"       mov     %1, %2\n"
-			"       s32ex   %1, %3\n"
-			"       getex   %1\n"
-			"       beqz    %1, 1b\n"
+			"1:     l32ex   %[result], %[addr]\n"
+			"       bne     %[result], %[cmp], 2f\n"
+			"       mov     %[tmp], %[new]\n"
+			"       s32ex   %[tmp], %[addr]\n"
+			"       getex   %[tmp]\n"
+			"       beqz    %[tmp], 1b\n"
 			"2:\n"
-			: "=&a" (result), "=&a" (tmp)
-			: "a" (new), "a" (p), "a" (old)
+			: [result] "=&a" (result), [tmp] "=&a" (tmp)
+			: [new] "a" (new), [addr] "a" (p), [cmp] "a" (old)
 			: "memory"
 			);
 
 	return result;
 #elif XCHAL_HAVE_S32C1I
 	__asm__ __volatile__(
-			"       wsr     %2, scompare1\n"
-			"       s32c1i  %0, %1, 0\n"
-			: "+a" (new)
-			: "a" (p), "a" (old)
+			"       wsr     %[cmp], scompare1\n"
+			"       s32c1i  %[new], %[mem]\n"
+			: [new] "+a" (new), [mem] "+m" (*p)
+			: [cmp] "a" (old)
 			: "memory"
 			);
 
@@ -53,14 +53,14 @@ __cmpxchg_u32(volatile int *p, int old, int new)
 #else
 	__asm__ __volatile__(
 			"       rsil    a15, "__stringify(TOPLEVEL)"\n"
-			"       l32i    %0, %1, 0\n"
-			"       bne     %0, %2, 1f\n"
-			"       s32i    %3, %1, 0\n"
+			"       l32i    %[old], %[mem]\n"
+			"       bne     %[old], %[cmp], 1f\n"
+			"       s32i    %[new], %[mem]\n"
 			"1:\n"
 			"       wsr     a15, ps\n"
 			"       rsync\n"
-			: "=&a" (old)
-			: "a" (p), "a" (old), "r" (new)
+			: [old] "=&a" (old), [mem] "+m" (*p)
+			: [cmp] "a" (old), [new] "r" (new)
 			: "a15", "memory");
 	return old;
 #endif
@@ -129,13 +129,13 @@ static inline unsigned long xchg_u32(volatile int * m, unsigned long val)
 	unsigned long tmp, result;
 
 	__asm__ __volatile__(
-			"1:     l32ex   %0, %3\n"
-			"       mov     %1, %2\n"
-			"       s32ex   %1, %3\n"
-			"       getex   %1\n"
-			"       beqz    %1, 1b\n"
-			: "=&a" (result), "=&a" (tmp)
-			: "a" (val), "a" (m)
+			"1:     l32ex   %[result], %[addr]\n"
+			"       mov     %[tmp], %[val]\n"
+			"       s32ex   %[tmp], %[addr]\n"
+			"       getex   %[tmp]\n"
+			"       beqz    %[tmp], 1b\n"
+			: [result] "=&a" (result), [tmp] "=&a" (tmp)
+			: [val] "a" (val), [addr] "a" (m)
 			: "memory"
 			);
 
@@ -143,13 +143,14 @@ static inline unsigned long xchg_u32(volatile int * m, unsigned long val)
 #elif XCHAL_HAVE_S32C1I
 	unsigned long tmp, result;
 	__asm__ __volatile__(
-			"1:     l32i    %1, %2, 0\n"
-			"       mov     %0, %3\n"
-			"       wsr     %1, scompare1\n"
-			"       s32c1i  %0, %2, 0\n"
-			"       bne     %0, %1, 1b\n"
-			: "=&a" (result), "=&a" (tmp)
-			: "a" (m), "a" (val)
+			"1:     l32i    %[tmp], %[mem]\n"
+			"       mov     %[result], %[val]\n"
+			"       wsr     %[tmp], scompare1\n"
+			"       s32c1i  %[result], %[mem]\n"
+			"       bne     %[result], %[tmp], 1b\n"
+			: [result] "=&a" (result), [tmp] "=&a" (tmp),
+			  [mem] "+m" (*m)
+			: [val] "a" (val)
 			: "memory"
 			);
 	return result;
@@ -157,12 +158,12 @@ static inline unsigned long xchg_u32(volatile int * m, unsigned long val)
 	unsigned long tmp;
 	__asm__ __volatile__(
 			"       rsil    a15, "__stringify(TOPLEVEL)"\n"
-			"       l32i    %0, %1, 0\n"
-			"       s32i    %2, %1, 0\n"
+			"       l32i    %[tmp], %[mem]\n"
+			"       s32i    %[val], %[mem]\n"
 			"       wsr     a15, ps\n"
 			"       rsync\n"
-			: "=&a" (tmp)
-			: "a" (m), "a" (val)
+			: [tmp] "=&a" (tmp), [mem] "+m" (*m)
+			: [val] "a" (val)
 			: "a15", "memory");
 	return tmp;
 #endif
diff --git a/arch/xtensa/include/asm/fixmap.h b/arch/xtensa/include/asm/fixmap.h
index 7e25c1b50ac0..cfb8696917e9 100644
--- a/arch/xtensa/include/asm/fixmap.h
+++ b/arch/xtensa/include/asm/fixmap.h
@@ -78,8 +78,10 @@ static inline unsigned long virt_to_fix(const unsigned long vaddr)
 
 #define kmap_get_fixmap_pte(vaddr) \
 	pte_offset_kernel( \
-		pmd_offset(pud_offset(pgd_offset_k(vaddr), (vaddr)), (vaddr)), \
-		(vaddr) \
-	)
+		pmd_offset(pud_offset(p4d_offset(pgd_offset_k(vaddr), \
+						 (vaddr)), \
+				      (vaddr)), \
+			   (vaddr)), \
+		(vaddr))
 
 #endif
diff --git a/arch/xtensa/include/asm/futex.h b/arch/xtensa/include/asm/futex.h
index 0c4457ca0a85..964611083224 100644
--- a/arch/xtensa/include/asm/futex.h
+++ b/arch/xtensa/include/asm/futex.h
@@ -43,10 +43,10 @@
 #elif XCHAL_HAVE_S32C1I
 #define __futex_atomic_op(insn, ret, old, uaddr, arg)	\
 	__asm__ __volatile(				\
-	"1:	l32i	%[oldval], %[addr], 0\n"	\
+	"1:	l32i	%[oldval], %[mem]\n"		\
 		insn "\n"				\
 	"	wsr	%[oldval], scompare1\n"		\
-	"2:	s32c1i	%[newval], %[addr], 0\n"	\
+	"2:	s32c1i	%[newval], %[mem]\n"		\
 	"	bne	%[newval], %[oldval], 1b\n"	\
 	"	movi	%[newval], 0\n"			\
 	"3:\n"						\
@@ -60,9 +60,9 @@
 	"	.section __ex_table,\"a\"\n"		\
 	"	.long 1b, 5b, 2b, 5b\n"			\
 	"	.previous\n"				\
-	: [oldval] "=&r" (old), [newval] "=&r" (ret)	\
-	: [addr] "r" (uaddr), [oparg] "r" (arg),	\
-	  [fault] "I" (-EFAULT)				\
+	: [oldval] "=&r" (old), [newval] "=&r" (ret),	\
+	  [mem] "+m" (*(uaddr))				\
+	: [oparg] "r" (arg), [fault] "I" (-EFAULT)	\
 	: "memory")
 #endif
 
diff --git a/arch/xtensa/include/asm/hw_irq.h b/arch/xtensa/include/asm/hw_irq.h
deleted file mode 100644
index 3ddbea759b2b..000000000000
--- a/arch/xtensa/include/asm/hw_irq.h
+++ /dev/null
@@ -1,14 +0,0 @@
-/*
- * include/asm-xtensa/hw_irq.h
- *
- * This file is subject to the terms and conditions of the GNU General
- * Public License.  See the file "COPYING" in the main directory of
- * this archive for more details.
- *
- * Copyright (C) 2002 - 2005 Tensilica Inc.
- */
-
-#ifndef _XTENSA_HW_IRQ_H
-#define _XTENSA_HW_IRQ_H
-
-#endif
diff --git a/arch/xtensa/include/asm/initialize_mmu.h b/arch/xtensa/include/asm/initialize_mmu.h
index 3b054d2bede0..e3e1d9a1ef69 100644
--- a/arch/xtensa/include/asm/initialize_mmu.h
+++ b/arch/xtensa/include/asm/initialize_mmu.h
@@ -23,6 +23,7 @@
 #ifndef _XTENSA_INITIALIZE_MMU_H
 #define _XTENSA_INITIALIZE_MMU_H
 
+#include <linux/init.h>
 #include <asm/pgtable.h>
 #include <asm/vectors.h>
 
@@ -183,7 +184,7 @@
 #endif
 
 #if XCHAL_HAVE_MPU
-	.data
+	__REFCONST
 	.align	4
 .Lattribute_table:
 	.long 0x000000, 0x1fff00, 0x1ddf00, 0x1eef00
diff --git a/arch/xtensa/include/asm/kmem_layout.h b/arch/xtensa/include/asm/kmem_layout.h
index 9c12babc016c..7cbf68ca7106 100644
--- a/arch/xtensa/include/asm/kmem_layout.h
+++ b/arch/xtensa/include/asm/kmem_layout.h
@@ -11,6 +11,7 @@
 #ifndef _XTENSA_KMEM_LAYOUT_H
 #define _XTENSA_KMEM_LAYOUT_H
 
+#include <asm/core.h>
 #include <asm/types.h>
 
 #ifdef CONFIG_MMU
@@ -65,6 +66,34 @@
 
 #endif
 
+/* KIO definition */
+
+#if XCHAL_HAVE_PTP_MMU
+#define XCHAL_KIO_CACHED_VADDR		0xe0000000
+#define XCHAL_KIO_BYPASS_VADDR		0xf0000000
+#define XCHAL_KIO_DEFAULT_PADDR		0xf0000000
+#else
+#define XCHAL_KIO_BYPASS_VADDR		XCHAL_KIO_PADDR
+#define XCHAL_KIO_DEFAULT_PADDR		0x90000000
+#endif
+#define XCHAL_KIO_SIZE			0x10000000
+
+#if (!XCHAL_HAVE_PTP_MMU || XCHAL_HAVE_SPANNING_WAY) && defined(CONFIG_OF)
+#define XCHAL_KIO_PADDR			xtensa_get_kio_paddr()
+#ifndef __ASSEMBLY__
+extern unsigned long xtensa_kio_paddr;
+
+static inline unsigned long xtensa_get_kio_paddr(void)
+{
+	return xtensa_kio_paddr;
+}
+#endif
+#else
+#define XCHAL_KIO_PADDR			XCHAL_KIO_DEFAULT_PADDR
+#endif
+
+/* KERNEL_STACK definition */
+
 #ifndef CONFIG_KASAN
 #define KERNEL_STACK_SHIFT	13
 #else
diff --git a/arch/xtensa/include/asm/page.h b/arch/xtensa/include/asm/page.h
index 09c56cba442e..f4771c29c7e9 100644
--- a/arch/xtensa/include/asm/page.h
+++ b/arch/xtensa/include/asm/page.h
@@ -169,7 +169,18 @@ static inline unsigned long ___pa(unsigned long va)
 	if (off >= XCHAL_KSEG_SIZE)
 		off -= XCHAL_KSEG_SIZE;
 
+#ifndef CONFIG_XIP_KERNEL
 	return off + PHYS_OFFSET;
+#else
+	if (off < XCHAL_KSEG_SIZE)
+		return off + PHYS_OFFSET;
+
+	off -= XCHAL_KSEG_SIZE;
+	if (off >= XCHAL_KIO_SIZE)
+		off -= XCHAL_KIO_SIZE;
+
+	return off + XCHAL_KIO_PADDR;
+#endif
 }
 #define __pa(x)	___pa((unsigned long)(x))
 #else
diff --git a/arch/xtensa/include/asm/pgtable.h b/arch/xtensa/include/asm/pgtable.h
index 3f7fe5a8c286..27ac17c9da09 100644
--- a/arch/xtensa/include/asm/pgtable.h
+++ b/arch/xtensa/include/asm/pgtable.h
@@ -8,7 +8,6 @@
 #ifndef _XTENSA_PGTABLE_H
 #define _XTENSA_PGTABLE_H
 
-#define __ARCH_USE_5LEVEL_HACK
 #include <asm/page.h>
 #include <asm/kmem_layout.h>
 #include <asm-generic/pgtable-nopmd.h>
@@ -371,9 +370,6 @@ ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 
 #define pgd_index(address)	((address) >> PGDIR_SHIFT)
 
-/* Find an entry in the second-level page table.. */
-#define pmd_offset(dir,address) ((pmd_t*)(dir))
-
 /* Find an entry in the third-level page table.. */
 #define pte_index(address)	(((address) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
 #define pte_offset_kernel(dir,addr) 					\
diff --git a/arch/xtensa/include/asm/processor.h b/arch/xtensa/include/asm/processor.h
index 7495520d7a3e..6fa903daf2a2 100644
--- a/arch/xtensa/include/asm/processor.h
+++ b/arch/xtensa/include/asm/processor.h
@@ -195,6 +195,7 @@ struct thread_struct {
 /* Clearing a0 terminates the backtrace. */
 #define start_thread(regs, new_pc, new_sp) \
 	do { \
+		unsigned long syscall = (regs)->syscall; \
 		memset((regs), 0, sizeof(*(regs))); \
 		(regs)->pc = (new_pc); \
 		(regs)->ps = USER_PS_VALUE; \
@@ -204,7 +205,7 @@ struct thread_struct {
 		(regs)->depc = 0; \
 		(regs)->windowbase = 0; \
 		(regs)->windowstart = 1; \
-		(regs)->syscall = NO_SYSCALL; \
+		(regs)->syscall = syscall; \
 	} while (0)
 
 /* Forward declaration */
diff --git a/arch/xtensa/include/asm/syscall.h b/arch/xtensa/include/asm/syscall.h
index 359ab40e935a..f9a671cbf933 100644
--- a/arch/xtensa/include/asm/syscall.h
+++ b/arch/xtensa/include/asm/syscall.h
@@ -51,7 +51,7 @@ static inline void syscall_set_return_value(struct task_struct *task,
 					    struct pt_regs *regs,
 					    int error, long val)
 {
-	regs->areg[0] = (long) error ? error : val;
+	regs->areg[2] = (long) error ? error : val;
 }
 
 #define SYSCALL_MAX_ARGS 6
@@ -79,7 +79,7 @@ static inline void syscall_set_arguments(struct task_struct *task,
 		regs->areg[reg[i]] = args[i];
 }
 
-asmlinkage long xtensa_rt_sigreturn(struct pt_regs*);
+asmlinkage long xtensa_rt_sigreturn(void);
 asmlinkage long xtensa_shmat(int, char __user *, int);
 asmlinkage long xtensa_fadvise64_64(int, int,
 				    unsigned long long, unsigned long long);
diff --git a/arch/xtensa/include/asm/uaccess.h b/arch/xtensa/include/asm/uaccess.h
index 3f80386f1883..47b7702aaa40 100644
--- a/arch/xtensa/include/asm/uaccess.h
+++ b/arch/xtensa/include/asm/uaccess.h
@@ -132,13 +132,13 @@ do {									\
 #define __check_align_1  ""
 
 #define __check_align_2				\
-	"   _bbci.l %[addr], 0, 1f	\n"	\
+	"   _bbci.l %[mem] * 0, 1f	\n"	\
 	"   movi    %[err], %[efault]	\n"	\
 	"   _j      2f			\n"
 
 #define __check_align_4				\
-	"   _bbsi.l %[addr], 0, 0f	\n"	\
-	"   _bbci.l %[addr], 1, 1f	\n"	\
+	"   _bbsi.l %[mem] * 0, 0f	\n"	\
+	"   _bbci.l %[mem] * 0 + 1, 1f	\n"	\
 	"0: movi    %[err], %[efault]	\n"	\
 	"   _j      2f			\n"
 
@@ -154,7 +154,7 @@ do {									\
 #define __put_user_asm(x_, addr_, err_, align, insn, cb)\
 __asm__ __volatile__(					\
 	__check_align_##align				\
-	"1: "insn"  %[x], %[addr], 0	\n"		\
+	"1: "insn"  %[x], %[mem]	\n"		\
 	"2:				\n"		\
 	"   .section  .fixup,\"ax\"	\n"		\
 	"   .align 4			\n"		\
@@ -167,8 +167,8 @@ __asm__ __volatile__(					\
 	"   .section  __ex_table,\"a\"	\n"		\
 	"   .long	1b, 5b		\n"		\
 	"   .previous"					\
-	:[err] "+r"(err_), [tmp] "=r"(cb)		\
-	:[x] "r"(x_), [addr] "r"(addr_), [efault] "i"(-EFAULT))
+	:[err] "+r"(err_), [tmp] "=r"(cb), [mem] "=m"(*(addr_))		\
+	:[x] "r"(x_), [efault] "i"(-EFAULT))
 
 #define __get_user_nocheck(x, ptr, size)			\
 ({								\
@@ -222,7 +222,7 @@ do {							\
 	u32 __x = 0;					\
 	__asm__ __volatile__(				\
 		__check_align_##align			\
-		"1: "insn"  %[x], %[addr], 0	\n"	\
+		"1: "insn"  %[x], %[mem]	\n"	\
 		"2:				\n"	\
 		"   .section  .fixup,\"ax\"	\n"	\
 		"   .align 4			\n"	\
@@ -236,7 +236,7 @@ do {							\
 		"   .long	1b, 5b		\n"	\
 		"   .previous"				\
 		:[err] "+r"(err_), [tmp] "=r"(cb), [x] "+r"(__x) \
-		:[addr] "r"(addr_), [efault] "i"(-EFAULT)); \
+		:[mem] "m"(*(addr_)), [efault] "i"(-EFAULT)); \
 	(x_) = (__force __typeof__(*(addr_)))__x;	\
 } while (0)
 
diff --git a/arch/xtensa/include/asm/user.h b/arch/xtensa/include/asm/user.h
deleted file mode 100644
index 2c3ed23354a8..000000000000
--- a/arch/xtensa/include/asm/user.h
+++ /dev/null
@@ -1,20 +0,0 @@
-/*
- * include/asm-xtensa/user.h
- *
- * Xtensa Processor version.
- *
- * This file is subject to the terms and conditions of the GNU General Public
- * License.  See the file "COPYING" in the main directory of this archive
- * for more details.
- *
- * Copyright (C) 2001 - 2005 Tensilica Inc.
- */
-
-#ifndef _XTENSA_USER_H
-#define _XTENSA_USER_H
-
-/* This file usually defines a 'struct user' structure. However, it it only
- * used for a.out file, which are not supported on Xtensa.
- */
-
-#endif	/* _XTENSA_USER_H */
diff --git a/arch/xtensa/include/asm/vectors.h b/arch/xtensa/include/asm/vectors.h
index 79fe3007919e..fd99b25037a7 100644
--- a/arch/xtensa/include/asm/vectors.h
+++ b/arch/xtensa/include/asm/vectors.h
@@ -21,50 +21,18 @@
 #include <asm/core.h>
 #include <asm/kmem_layout.h>
 
-#if XCHAL_HAVE_PTP_MMU
-#define XCHAL_KIO_CACHED_VADDR		0xe0000000
-#define XCHAL_KIO_BYPASS_VADDR		0xf0000000
-#define XCHAL_KIO_DEFAULT_PADDR		0xf0000000
+#if defined(CONFIG_MMU) && XCHAL_HAVE_PTP_MMU && XCHAL_HAVE_SPANNING_WAY
+#ifdef CONFIG_KERNEL_VIRTUAL_ADDRESS
+#define KERNELOFFSET			CONFIG_KERNEL_VIRTUAL_ADDRESS
 #else
-#define XCHAL_KIO_BYPASS_VADDR		XCHAL_KIO_PADDR
-#define XCHAL_KIO_DEFAULT_PADDR		0x90000000
-#endif
-#define XCHAL_KIO_SIZE			0x10000000
-
-#if (!XCHAL_HAVE_PTP_MMU || XCHAL_HAVE_SPANNING_WAY) && defined(CONFIG_OF)
-#define XCHAL_KIO_PADDR			xtensa_get_kio_paddr()
-#ifndef __ASSEMBLY__
-extern unsigned long xtensa_kio_paddr;
-
-static inline unsigned long xtensa_get_kio_paddr(void)
-{
-	return xtensa_kio_paddr;
-}
-#endif
-#else
-#define XCHAL_KIO_PADDR			XCHAL_KIO_DEFAULT_PADDR
-#endif
-
-#if defined(CONFIG_MMU)
-
-#if XCHAL_HAVE_PTP_MMU && XCHAL_HAVE_SPANNING_WAY
-/* Image Virtual Start Address */
-#define KERNELOFFSET			(XCHAL_KSEG_CACHED_VADDR + \
-					 CONFIG_KERNEL_LOAD_ADDRESS - \
+#define KERNELOFFSET			(CONFIG_KERNEL_LOAD_ADDRESS + \
+					 XCHAL_KSEG_CACHED_VADDR - \
 					 XCHAL_KSEG_PADDR)
+#endif
 #else
 #define KERNELOFFSET			CONFIG_KERNEL_LOAD_ADDRESS
 #endif
 
-#else /* !defined(CONFIG_MMU) */
-  /* MMU Not being used - Virtual == Physical */
-
-/* Location of the start of the kernel text, _start */
-#define KERNELOFFSET			CONFIG_KERNEL_LOAD_ADDRESS
-
-
-#endif /* CONFIG_MMU */
-
 #define RESET_VECTOR1_VADDR		(XCHAL_RESET_VECTOR1_VADDR)
 #ifdef CONFIG_VECTORS_OFFSET
 #define VECBASE_VADDR			(KERNELOFFSET - CONFIG_VECTORS_OFFSET)
diff --git a/arch/xtensa/kernel/coprocessor.S b/arch/xtensa/kernel/coprocessor.S
index 80828b95a51f..bb8e499b9900 100644
--- a/arch/xtensa/kernel/coprocessor.S
+++ b/arch/xtensa/kernel/coprocessor.S
@@ -15,17 +15,9 @@
 #include <linux/linkage.h>
 #include <asm/asm-offsets.h>
 #include <asm/asmmacro.h>
-#include <asm/processor.h>
 #include <asm/coprocessor.h>
-#include <asm/thread_info.h>
-#include <asm/asm-uaccess.h>
-#include <asm/unistd.h>
-#include <asm/ptrace.h>
 #include <asm/current.h>
-#include <asm/pgtable.h>
-#include <asm/page.h>
-#include <asm/signal.h>
-#include <asm/tlbflush.h>
+#include <asm/regs.h>
 
 #if XTENSA_HAVE_COPROCESSORS
 
diff --git a/arch/xtensa/kernel/entry.S b/arch/xtensa/kernel/entry.S
index 9e3676879168..be897803834a 100644
--- a/arch/xtensa/kernel/entry.S
+++ b/arch/xtensa/kernel/entry.S
@@ -529,7 +529,7 @@ common_exception_return:
 	l32i	a4, a2, TI_PRE_COUNT
 	bnez	a4, 4f
 	call4	preempt_schedule_irq
-	j	1b
+	j	4f
 #endif
 
 #if XTENSA_FAKE_NMI
@@ -1876,8 +1876,7 @@ ENDPROC(fast_store_prohibited)
 
 ENTRY(system_call)
 
-	/* reserve 4 bytes on stack for function parameter */
-	abi_entry(4)
+	abi_entry_default
 
 	/* regs->syscall = regs->areg[2] */
 
@@ -1892,11 +1891,10 @@ ENTRY(system_call)
 
 	mov	a6, a2
 	call4	do_syscall_trace_enter
+	beqz	a6, .Lsyscall_exit
 	l32i	a7, a2, PT_SYSCALL
 
 1:
-	s32i	a7, a1, 4
-
 	/* syscall = sys_call_table[syscall_nr] */
 
 	movi	a4, sys_call_table
@@ -1906,8 +1904,6 @@ ENTRY(system_call)
 
 	addx4	a4, a7, a4
 	l32i	a4, a4, 0
-	movi	a5, sys_ni_syscall;
-	beq	a4, a5, 1f
 
 	/* Load args: arg0 - arg5 are passed via regs. */
 
@@ -1918,25 +1914,19 @@ ENTRY(system_call)
 	l32i	a10, a2, PT_AREG8
 	l32i	a11, a2, PT_AREG9
 
-	/* Pass one additional argument to the syscall: pt_regs (on stack) */
-	s32i	a2, a1, 0
-
 	callx4	a4
 
 1:	/* regs->areg[2] = return_value */
 
 	s32i	a6, a2, PT_AREG2
 	bnez	a3, 1f
-	abi_ret(4)
+.Lsyscall_exit:
+	abi_ret_default
 
 1:
-	l32i	a4, a1, 4
-	l32i	a3, a2, PT_SYSCALL
-	s32i	a4, a2, PT_SYSCALL
 	mov	a6, a2
 	call4	do_syscall_trace_leave
-	s32i	a3, a2, PT_SYSCALL
-	abi_ret(4)
+	abi_ret_default
 
 ENDPROC(system_call)
 
diff --git a/arch/xtensa/kernel/head.S b/arch/xtensa/kernel/head.S
index 4ae998b5a348..e0c1fac0910f 100644
--- a/arch/xtensa/kernel/head.S
+++ b/arch/xtensa/kernel/head.S
@@ -260,6 +260,13 @@ ENTRY(_startup)
 	___invalidate_icache_all a2 a3
 	isync
 
+#ifdef CONFIG_XIP_KERNEL
+	/* Setup bootstrap CPU stack in XIP kernel */
+
+	movi	a1, start_info
+	l32i	a1, a1, 0
+#endif
+
 	movi	a6, 0
 	xsr	a6, excsave1
 
@@ -355,10 +362,10 @@ ENDPROC(cpu_restart)
  * DATA section
  */
 
-        .section ".data.init.refok"
-        .align  4
+	__REFDATA
+	.align  4
 ENTRY(start_info)
-        .long   init_thread_union + KERNEL_STACK_SIZE
+	.long	init_thread_union + KERNEL_STACK_SIZE
 
 /*
  * BSS section
diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c
index db278a9e80c7..9e1c49134c07 100644
--- a/arch/xtensa/kernel/process.c
+++ b/arch/xtensa/kernel/process.c
@@ -264,6 +264,8 @@ int copy_thread(unsigned long clone_flags, unsigned long usp_thread_fn,
 			       &regs->areg[XCHAL_NUM_AREGS - len/4], len);
 		}
 
+		childregs->syscall = regs->syscall;
+
 		/* The thread pointer is passed in the '4th argument' (= a5) */
 		if (clone_flags & CLONE_SETTLS)
 			childregs->threadptr = childregs->areg[5];
diff --git a/arch/xtensa/kernel/ptrace.c b/arch/xtensa/kernel/ptrace.c
index b964f0b2d886..145742d70a9f 100644
--- a/arch/xtensa/kernel/ptrace.c
+++ b/arch/xtensa/kernel/ptrace.c
@@ -542,14 +542,28 @@ long arch_ptrace(struct task_struct *child, long request,
 	return ret;
 }
 
-void do_syscall_trace_enter(struct pt_regs *regs)
+void do_syscall_trace_leave(struct pt_regs *regs);
+int do_syscall_trace_enter(struct pt_regs *regs)
 {
+	if (regs->syscall == NO_SYSCALL)
+		regs->areg[2] = -ENOSYS;
+
 	if (test_thread_flag(TIF_SYSCALL_TRACE) &&
-	    tracehook_report_syscall_entry(regs))
+	    tracehook_report_syscall_entry(regs)) {
+		regs->areg[2] = -ENOSYS;
 		regs->syscall = NO_SYSCALL;
+		return 0;
+	}
+
+	if (regs->syscall == NO_SYSCALL) {
+		do_syscall_trace_leave(regs);
+		return 0;
+	}
 
 	if (test_thread_flag(TIF_SYSCALL_TRACEPOINT))
 		trace_sys_enter(regs, syscall_get_nr(current, regs));
+
+	return 1;
 }
 
 void do_syscall_trace_leave(struct pt_regs *regs)
diff --git a/arch/xtensa/kernel/setup.c b/arch/xtensa/kernel/setup.c
index e0e1e1892b86..0f93b67c7a5a 100644
--- a/arch/xtensa/kernel/setup.c
+++ b/arch/xtensa/kernel/setup.c
@@ -308,6 +308,10 @@ extern char _Level6InterruptVector_text_end;
 extern char _SecondaryResetVector_text_start;
 extern char _SecondaryResetVector_text_end;
 #endif
+#ifdef CONFIG_XIP_KERNEL
+extern char _xip_start[];
+extern char _xip_end[];
+#endif
 
 static inline int __init_memblock mem_reserve(unsigned long start,
 					      unsigned long end)
@@ -339,6 +343,9 @@ void __init setup_arch(char **cmdline_p)
 #endif
 
 	mem_reserve(__pa(_stext), __pa(_end));
+#ifdef CONFIG_XIP_KERNEL
+	mem_reserve(__pa(_xip_start), __pa(_xip_end));
+#endif
 
 #ifdef CONFIG_VECTORS_OFFSET
 	mem_reserve(__pa(&_WindowVectors_text_start),
diff --git a/arch/xtensa/kernel/signal.c b/arch/xtensa/kernel/signal.c
index dae83cddd6ca..76cee341507b 100644
--- a/arch/xtensa/kernel/signal.c
+++ b/arch/xtensa/kernel/signal.c
@@ -236,9 +236,9 @@ restore_sigcontext(struct pt_regs *regs, struct rt_sigframe __user *frame)
  * Do a signal return; undo the signal stack.
  */
 
-asmlinkage long xtensa_rt_sigreturn(long a0, long a1, long a2, long a3,
-				    long a4, long a5, struct pt_regs *regs)
+asmlinkage long xtensa_rt_sigreturn(void)
 {
+	struct pt_regs *regs = current_pt_regs();
 	struct rt_sigframe __user *frame;
 	sigset_t set;
 	int ret;
diff --git a/arch/xtensa/kernel/traps.c b/arch/xtensa/kernel/traps.c
index 4a6c495ce9b6..87bd68dd7687 100644
--- a/arch/xtensa/kernel/traps.c
+++ b/arch/xtensa/kernel/traps.c
@@ -491,32 +491,27 @@ void show_trace(struct task_struct *task, unsigned long *sp)
 
 	pr_info("Call Trace:\n");
 	walk_stackframe(sp, show_trace_cb, NULL);
-#ifndef CONFIG_KALLSYMS
-	pr_cont("\n");
-#endif
 }
 
-static int kstack_depth_to_print = 24;
+#define STACK_DUMP_ENTRY_SIZE 4
+#define STACK_DUMP_LINE_SIZE 32
+static size_t kstack_depth_to_print = CONFIG_PRINT_STACK_DEPTH;
 
 void show_stack(struct task_struct *task, unsigned long *sp)
 {
-	int i = 0;
-	unsigned long *stack;
+	size_t len;
 
 	if (!sp)
 		sp = stack_pointer(task);
-	stack = sp;
 
-	pr_info("Stack:\n");
+	len = min((-(size_t)sp) & (THREAD_SIZE - STACK_DUMP_ENTRY_SIZE),
+		  kstack_depth_to_print * STACK_DUMP_ENTRY_SIZE);
 
-	for (i = 0; i < kstack_depth_to_print; i++) {
-		if (kstack_end(sp))
-			break;
-		pr_cont(" %08lx", *sp++);
-		if (i % 8 == 7)
-			pr_cont("\n");
-	}
-	show_trace(task, stack);
+	pr_info("Stack:\n");
+	print_hex_dump(KERN_INFO, " ", DUMP_PREFIX_NONE,
+		       STACK_DUMP_LINE_SIZE, STACK_DUMP_ENTRY_SIZE,
+		       sp, len, false);
+	show_trace(task, sp);
 }
 
 DEFINE_SPINLOCK(die_lock);
diff --git a/arch/xtensa/kernel/vmlinux.lds.S b/arch/xtensa/kernel/vmlinux.lds.S
index 0043d5858f14..409c05cac15e 100644
--- a/arch/xtensa/kernel/vmlinux.lds.S
+++ b/arch/xtensa/kernel/vmlinux.lds.S
@@ -119,7 +119,7 @@ SECTIONS
     SCHED_TEXT
     CPUIDLE_TEXT
     LOCK_TEXT
-
+    *(.fixup)
   }
   _etext = .;
   PROVIDE (etext = .);
@@ -128,12 +128,11 @@ SECTIONS
 
   RO_DATA(4096)
 
-  /*  Relocation table */
-
-  .fixup   : { *(.fixup) }
-
   /* Data section */
 
+#ifdef CONFIG_XIP_KERNEL
+  INIT_TEXT_SECTION(PAGE_SIZE)
+#else
   _sdata = .;
   RW_DATA(XCHAL_ICACHE_LINESIZE, PAGE_SIZE, THREAD_SIZE)
   _edata = .;
@@ -147,6 +146,11 @@ SECTIONS
   .init.data :
   {
     INIT_DATA
+  }
+#endif
+
+  .init.rodata :
+  {
     . = ALIGN(0x4);
     __tagtable_begin = .;
     *(.taglist)
@@ -187,12 +191,16 @@ SECTIONS
     RELOCATE_ENTRY(_DebugInterruptVector_text,
 		   .DebugInterruptVector.text);
 #endif
+#ifdef CONFIG_XIP_KERNEL
+    RELOCATE_ENTRY(_xip_data, .data);
+    RELOCATE_ENTRY(_xip_init_data, .init.data);
+#else
 #if defined(CONFIG_SMP)
     RELOCATE_ENTRY(_SecondaryResetVector_text,
 		   .SecondaryResetVector.text);
 #endif
+#endif
 
-  
     __boot_reloc_table_end = ABSOLUTE(.) ;
 
     INIT_SETUP(XCHAL_ICACHE_LINESIZE)
@@ -278,7 +286,7 @@ SECTIONS
   . = (LOADADDR( .DoubleExceptionVector.text ) + SIZEOF( .DoubleExceptionVector.text ) + 3) & ~ 3;
 
 #endif
-#if defined(CONFIG_SMP)
+#if !defined(CONFIG_XIP_KERNEL) && defined(CONFIG_SMP)
 
   SECTION_VECTOR (_SecondaryResetVector_text,
 		  .SecondaryResetVector.text,
@@ -291,12 +299,48 @@ SECTIONS
 
   . = ALIGN(PAGE_SIZE);
 
+#ifndef CONFIG_XIP_KERNEL
   __init_end = .;
 
   BSS_SECTION(0, 8192, 0)
+#endif
 
   _end = .;
 
+#ifdef CONFIG_XIP_KERNEL
+  . = CONFIG_XIP_DATA_ADDR;
+
+  _xip_start = .;
+
+#undef LOAD_OFFSET
+#define LOAD_OFFSET \
+  (CONFIG_XIP_DATA_ADDR - (LOADADDR(.dummy) + SIZEOF(.dummy) + 3) & ~ 3)
+
+  _xip_data_start = .;
+  _sdata = .;
+  RW_DATA(XCHAL_ICACHE_LINESIZE, PAGE_SIZE, THREAD_SIZE)
+  _edata = .;
+  _xip_data_end = .;
+
+  /* Initialization data: */
+
+  STRUCT_ALIGN();
+
+  _xip_init_data_start = .;
+  __init_begin = .;
+  .init.data :
+  {
+    INIT_DATA
+  }
+  _xip_init_data_end = .;
+  __init_end = .;
+  BSS_SECTION(0, 8192, 0)
+
+  _xip_end = .;
+
+#undef LOAD_OFFSET
+#endif
+
   DWARF_DEBUG
 
   .xt.prop 0 : { KEEP(*(.xt.prop .xt.prop.* .gnu.linkonce.prop.*)) }
diff --git a/arch/xtensa/mm/fault.c b/arch/xtensa/mm/fault.c
index f81b1478da61..bee30a77cd70 100644
--- a/arch/xtensa/mm/fault.c
+++ b/arch/xtensa/mm/fault.c
@@ -197,6 +197,8 @@ vmalloc_fault:
 		struct mm_struct *act_mm = current->active_mm;
 		int index = pgd_index(address);
 		pgd_t *pgd, *pgd_k;
+		p4d_t *p4d, *p4d_k;
+		pud_t *pud, *pud_k;
 		pmd_t *pmd, *pmd_k;
 		pte_t *pte_k;
 
@@ -211,8 +213,18 @@ vmalloc_fault:
 
 		pgd_val(*pgd) = pgd_val(*pgd_k);
 
-		pmd = pmd_offset(pgd, address);
-		pmd_k = pmd_offset(pgd_k, address);
+		p4d = p4d_offset(pgd, address);
+		p4d_k = p4d_offset(pgd_k, address);
+		if (!p4d_present(*p4d) || !p4d_present(*p4d_k))
+			goto bad_page_fault;
+
+		pud = pud_offset(p4d, address);
+		pud_k = pud_offset(p4d_k, address);
+		if (!pud_present(*pud) || !pud_present(*pud_k))
+			goto bad_page_fault;
+
+		pmd = pmd_offset(pud, address);
+		pmd_k = pmd_offset(pud_k, address);
 		if (!pmd_present(*pmd) || !pmd_present(*pmd_k))
 			goto bad_page_fault;
 
diff --git a/arch/xtensa/mm/init.c b/arch/xtensa/mm/init.c
index d898ed67d890..19c625e6d81f 100644
--- a/arch/xtensa/mm/init.c
+++ b/arch/xtensa/mm/init.c
@@ -193,8 +193,8 @@ void __init mem_init(void)
 		((max_low_pfn - min_low_pfn) * PAGE_SIZE) >> 20,
 		(unsigned long)_text, (unsigned long)_etext,
 		(unsigned long)(_etext - _text) >> 10,
-		(unsigned long)__start_rodata, (unsigned long)_sdata,
-		(unsigned long)(_sdata - __start_rodata) >> 10,
+		(unsigned long)__start_rodata, (unsigned long)__end_rodata,
+		(unsigned long)(__end_rodata - __start_rodata) >> 10,
 		(unsigned long)_sdata, (unsigned long)_edata,
 		(unsigned long)(_edata - _sdata) >> 10,
 		(unsigned long)__init_begin, (unsigned long)__init_end,
diff --git a/arch/xtensa/mm/kasan_init.c b/arch/xtensa/mm/kasan_init.c
index af7152560bc3..e3baa21ff24c 100644
--- a/arch/xtensa/mm/kasan_init.c
+++ b/arch/xtensa/mm/kasan_init.c
@@ -20,7 +20,9 @@ void __init kasan_early_init(void)
 {
 	unsigned long vaddr = KASAN_SHADOW_START;
 	pgd_t *pgd = pgd_offset_k(vaddr);
-	pmd_t *pmd = pmd_offset(pgd, vaddr);
+	p4d_t *p4d = p4d_offset(pgd, vaddr);
+	pud_t *pud = pud_offset(p4d, vaddr);
+	pmd_t *pmd = pmd_offset(pud, vaddr);
 	int i;
 
 	for (i = 0; i < PTRS_PER_PTE; ++i)
@@ -42,7 +44,9 @@ static void __init populate(void *start, void *end)
 	unsigned long i, j;
 	unsigned long vaddr = (unsigned long)start;
 	pgd_t *pgd = pgd_offset_k(vaddr);
-	pmd_t *pmd = pmd_offset(pgd, vaddr);
+	p4d_t *p4d = p4d_offset(pgd, vaddr);
+	pud_t *pud = pud_offset(p4d, vaddr);
+	pmd_t *pmd = pmd_offset(pud, vaddr);
 	pte_t *pte = memblock_alloc(n_pages * sizeof(pte_t), PAGE_SIZE);
 
 	if (!pte)
@@ -56,7 +60,9 @@ static void __init populate(void *start, void *end)
 
 		for (k = 0; k < PTRS_PER_PTE; ++k, ++j) {
 			phys_addr_t phys =
-				memblock_phys_alloc(PAGE_SIZE, PAGE_SIZE);
+				memblock_phys_alloc_range(PAGE_SIZE, PAGE_SIZE,
+							  0,
+							  MEMBLOCK_ALLOC_ANYWHERE);
 
 			if (!phys)
 				panic("Failed to allocate page table page\n");
diff --git a/arch/xtensa/mm/mmu.c b/arch/xtensa/mm/mmu.c
index 03678c4afc39..37e478a27877 100644
--- a/arch/xtensa/mm/mmu.c
+++ b/arch/xtensa/mm/mmu.c
@@ -22,7 +22,9 @@
 static void * __init init_pmd(unsigned long vaddr, unsigned long n_pages)
 {
 	pgd_t *pgd = pgd_offset_k(vaddr);
-	pmd_t *pmd = pmd_offset(pgd, vaddr);
+	p4d_t *p4d = p4d_offset(pgd, vaddr);
+	pud_t *pud = pud_offset(p4d, vaddr);
+	pmd_t *pmd = pmd_offset(pud, vaddr);
 	pte_t *pte;
 	unsigned long i;
 
diff --git a/arch/xtensa/mm/tlb.c b/arch/xtensa/mm/tlb.c
index 59153d0aa890..f436cf2efd8b 100644
--- a/arch/xtensa/mm/tlb.c
+++ b/arch/xtensa/mm/tlb.c
@@ -169,6 +169,8 @@ static unsigned get_pte_for_vaddr(unsigned vaddr)
 	struct task_struct *task = get_current();
 	struct mm_struct *mm = task->mm;
 	pgd_t *pgd;
+	p4d_t *p4d;
+	pud_t *pud;
 	pmd_t *pmd;
 	pte_t *pte;
 
@@ -177,7 +179,13 @@ static unsigned get_pte_for_vaddr(unsigned vaddr)
 	pgd = pgd_offset(mm, vaddr);
 	if (pgd_none_or_clear_bad(pgd))
 		return 0;
-	pmd = pmd_offset(pgd, vaddr);
+	p4d = p4d_offset(pgd, vaddr);
+	if (p4d_none_or_clear_bad(p4d))
+		return 0;
+	pud = pud_offset(p4d, vaddr);
+	if (pud_none_or_clear_bad(pud))
+		return 0;
+	pmd = pmd_offset(pud, vaddr);
 	if (pmd_none_or_clear_bad(pmd))
 		return 0;
 	pte = pte_offset_map(pmd, vaddr);
@@ -216,6 +224,8 @@ static int check_tlb_entry(unsigned w, unsigned e, bool dtlb)
 	unsigned tlbidx = w | (e << PAGE_SHIFT);
 	unsigned r0 = dtlb ?
 		read_dtlb_virtual(tlbidx) : read_itlb_virtual(tlbidx);
+	unsigned r1 = dtlb ?
+		read_dtlb_translation(tlbidx) : read_itlb_translation(tlbidx);
 	unsigned vpn = (r0 & PAGE_MASK) | (e << PAGE_SHIFT);
 	unsigned pte = get_pte_for_vaddr(vpn);
 	unsigned mm_asid = (get_rasid_register() >> 8) & ASID_MASK;
@@ -231,8 +241,6 @@ static int check_tlb_entry(unsigned w, unsigned e, bool dtlb)
 	}
 
 	if (tlb_asid == mm_asid) {
-		unsigned r1 = dtlb ? read_dtlb_translation(tlbidx) :
-			read_itlb_translation(tlbidx);
 		if ((pte ^ r1) & PAGE_MASK) {
 			pr_err("%cTLB: way: %u, entry: %u, mapping: %08x->%08x, PTE: %08x\n",
 					dtlb ? 'D' : 'I', w, e, r0, r1, pte);