Merge git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile

* git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile: (26 commits)
  arch/tile: prefer "tilepro" as the name of the 32-bit architecture
  compat: include aio_abi.h for aio_context_t
  arch/tile: cleanups for tilegx compat mode
  arch/tile: allocate PCI IRQs later in boot
  arch/tile: support signal "exception-trace" hook
  arch/tile: use better definitions of xchg() and cmpxchg()
  include/linux/compat.h: coding-style fixes
  tile: add an RTC driver for the Tilera hypervisor
  arch/tile: finish enabling support for TILE-Gx 64-bit chip
  compat: fixes to allow working with tile arch
  arch/tile: update defconfig file to something more useful
  tile: do_hardwall_trap: do not play with task->sighand
  tile: replace mm->cpu_vm_mask with mm_cpumask()
  tile,mn10300: add device parameter to dma_cache_sync()
  audit: support the "standard" <asm-generic/unistd.h>
  arch/tile: clarify flush_buffer()/finv_buffer() function names
  arch/tile: kernel-related cleanups from removing static page size
  arch/tile: various header improvements for building drivers
  arch/tile: disable GX prefetcher during cache flush
  arch/tile: tolerate disabling CONFIG_BLK_DEV_INITRD
  ...

Makefile

@@ -220,6 +220,14 @@ ifeq ($(ARCH),sh64)
        SRCARCH := sh
 endif
 
+# Additional ARCH settings for tile
+ifeq ($(ARCH),tilepro)
+       SRCARCH := tile
+endif
+ifeq ($(ARCH),tilegx)
+       SRCARCH := tile
+endif
+
 # Where to locate arch specific headers
 hdr-arch  := $(SRCARCH)
 

arch/tile/Kconfig

@@ -339,6 +339,14 @@ config NO_IOPORT
 
 source "drivers/pci/Kconfig"
 
+config HOTPLUG
+	bool "Support for hot-pluggable devices"
+	---help---
+	  Say Y here if you want to plug devices into your computer while
+	  the system is running, and be able to use them quickly.  In many
+	  cases, the devices can likewise be unplugged at any time too.
+	  One well-known example of this is USB.
+
 source "drivers/pci/hotplug/Kconfig"
 
 endmenu

arch/tile/configs/tile_defconfig

-CONFIG_EXPERIMENTAL=y
-# CONFIG_SWAP is not set
-CONFIG_SYSVIPC=y
-CONFIG_BLK_DEV_INITRD=y
-CONFIG_INITRAMFS_SOURCE="usr/contents.txt"
-CONFIG_EXPERT=y
-# CONFIG_COMPAT_BRK is not set
-CONFIG_PROFILING=y
-CONFIG_MODULES=y
-CONFIG_MODULE_UNLOAD=y
-# CONFIG_BLK_DEV_BSG is not set
-# CONFIG_IOSCHED_DEADLINE is not set
-# CONFIG_IOSCHED_CFQ is not set
-CONFIG_NO_HZ=y
-CONFIG_HIGH_RES_TIMERS=y
-CONFIG_HZ_100=y
-CONFIG_NET=y
-CONFIG_PACKET=y
-CONFIG_UNIX=y
-CONFIG_INET=y
-CONFIG_IP_MULTICAST=y
-# CONFIG_INET_XFRM_MODE_TRANSPORT is not set
-# CONFIG_INET_XFRM_MODE_TUNNEL is not set
-# CONFIG_INET_LRO is not set
-# CONFIG_INET_DIAG is not set
-CONFIG_IPV6=y
-# CONFIG_WIRELESS is not set
-CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug"
-CONFIG_SCSI=y
-CONFIG_BLK_DEV_SD=y
-CONFIG_SCSI_CONSTANTS=y
-CONFIG_SCSI_LOGGING=y
-CONFIG_NETDEVICES=y
-CONFIG_TUN=y
-# CONFIG_NETDEV_10000 is not set
-# CONFIG_WLAN is not set
-# CONFIG_INPUT_MOUSEDEV is not set
-# CONFIG_INPUT_KEYBOARD is not set
-# CONFIG_INPUT_MOUSE is not set
-# CONFIG_SERIO is not set
-# CONFIG_VT is not set
-# CONFIG_LEGACY_PTYS is not set
-# CONFIG_HW_RANDOM is not set
-CONFIG_WATCHDOG=y
-CONFIG_WATCHDOG_NOWAYOUT=y
-# CONFIG_HID_SUPPORT is not set
-CONFIG_RTC_CLASS=y
-# CONFIG_RTC_INTF_SYSFS is not set
-# CONFIG_RTC_INTF_PROC is not set
-CONFIG_EXT2_FS=y
-CONFIG_EXT3_FS=y
-# CONFIG_EXT3_DEFAULTS_TO_ORDERED is not set
-CONFIG_FUSE_FS=y
-CONFIG_MSDOS_FS=y
-CONFIG_VFAT_FS=m
-CONFIG_TMPFS=y
-CONFIG_HUGETLBFS=y
-CONFIG_NFS_FS=m
-CONFIG_NFS_V3=y
-CONFIG_NLS_CODEPAGE_437=y
-CONFIG_NLS_ISO8859_1=y
-CONFIG_FRAME_WARN=2048
-CONFIG_MAGIC_SYSRQ=y
-CONFIG_DEBUG_KERNEL=y
-CONFIG_DETECT_HUNG_TASK=y
-CONFIG_DEBUG_SPINLOCK_SLEEP=y
-CONFIG_DEBUG_INFO=y
-CONFIG_DEBUG_VM=y
-# CONFIG_RCU_CPU_STALL_DETECTOR is not set
-CONFIG_DEBUG_STACKOVERFLOW=y
-CONFIG_DEBUG_EXTRA_FLAGS="-femit-struct-debug-baseonly"

arch/tile/include/arch/chip_tilegx.h

+/*
+ * Copyright 2011 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+/*
+ * @file
+ * Global header file.
+ * This header file specifies defines for TILE-Gx.
+ */
+
+#ifndef __ARCH_CHIP_H__
+#define __ARCH_CHIP_H__
+
+/** Specify chip version.
+ * When possible, prefer the CHIP_xxx symbols below for future-proofing.
+ * This is intended for cross-compiling; native compilation should
+ * use the predefined __tile_chip__ symbol.
+ */
+#define TILE_CHIP 10
+
+/** Specify chip revision.
+ * This provides for the case of a respin of a particular chip type;
+ * the normal value for this symbol is "0".
+ * This is intended for cross-compiling; native compilation should
+ * use the predefined __tile_chip_rev__ symbol.
+ */
+#define TILE_CHIP_REV 0
+
+/** The name of this architecture. */
+#define CHIP_ARCH_NAME "tilegx"
+
+/** The ELF e_machine type for binaries for this chip. */
+#define CHIP_ELF_TYPE() EM_TILEGX
+
+/** The alternate ELF e_machine type for binaries for this chip. */
+#define CHIP_COMPAT_ELF_TYPE() 0x2597
+
+/** What is the native word size of the machine? */
+#define CHIP_WORD_SIZE() 64
+
+/** How many bits of a virtual address are used. Extra bits must be
+ * the sign extension of the low bits.
+ */
+#define CHIP_VA_WIDTH() 42
+
+/** How many bits are in a physical address? */
+#define CHIP_PA_WIDTH() 40
+
+/** Size of the L2 cache, in bytes. */
+#define CHIP_L2_CACHE_SIZE() 262144
+
+/** Log size of an L2 cache line in bytes. */
+#define CHIP_L2_LOG_LINE_SIZE() 6
+
+/** Size of an L2 cache line, in bytes. */
+#define CHIP_L2_LINE_SIZE() (1 << CHIP_L2_LOG_LINE_SIZE())
+
+/** Associativity of the L2 cache. */
+#define CHIP_L2_ASSOC() 8
+
+/** Size of the L1 data cache, in bytes. */
+#define CHIP_L1D_CACHE_SIZE() 32768
+
+/** Log size of an L1 data cache line in bytes. */
+#define CHIP_L1D_LOG_LINE_SIZE() 6
+
+/** Size of an L1 data cache line, in bytes. */
+#define CHIP_L1D_LINE_SIZE() (1 << CHIP_L1D_LOG_LINE_SIZE())
+
+/** Associativity of the L1 data cache. */
+#define CHIP_L1D_ASSOC() 2
+
+/** Size of the L1 instruction cache, in bytes. */
+#define CHIP_L1I_CACHE_SIZE() 32768
+
+/** Log size of an L1 instruction cache line in bytes. */
+#define CHIP_L1I_LOG_LINE_SIZE() 6
+
+/** Size of an L1 instruction cache line, in bytes. */
+#define CHIP_L1I_LINE_SIZE() (1 << CHIP_L1I_LOG_LINE_SIZE())
+
+/** Associativity of the L1 instruction cache. */
+#define CHIP_L1I_ASSOC() 2
+
+/** Stride with which flush instructions must be issued. */
+#define CHIP_FLUSH_STRIDE() CHIP_L2_LINE_SIZE()
+
+/** Stride with which inv instructions must be issued. */
+#define CHIP_INV_STRIDE() CHIP_L2_LINE_SIZE()
+
+/** Stride with which finv instructions must be issued. */
+#define CHIP_FINV_STRIDE() CHIP_L2_LINE_SIZE()
+
+/** Can the local cache coherently cache data that is homed elsewhere? */
+#define CHIP_HAS_COHERENT_LOCAL_CACHE() 1
+
+/** How many simultaneous outstanding victims can the L2 cache have? */
+#define CHIP_MAX_OUTSTANDING_VICTIMS() 128
+
+/** Does the TLB support the NC and NOALLOC bits? */
+#define CHIP_HAS_NC_AND_NOALLOC_BITS() 1
+
+/** Does the chip support hash-for-home caching? */
+#define CHIP_HAS_CBOX_HOME_MAP() 1
+
+/** Number of entries in the chip's home map tables. */
+#define CHIP_CBOX_HOME_MAP_SIZE() 128
+
+/** Do uncacheable requests miss in the cache regardless of whether
+ * there is matching data? */
+#define CHIP_HAS_ENFORCED_UNCACHEABLE_REQUESTS() 1
+
+/** Does the mf instruction wait for victims? */
+#define CHIP_HAS_MF_WAITS_FOR_VICTIMS() 0
+
+/** Does the chip have an "inv" instruction that doesn't also flush? */
+#define CHIP_HAS_INV() 1
+
+/** Does the chip have a "wh64" instruction? */
+#define CHIP_HAS_WH64() 1
+
+/** Does this chip have a 'dword_align' instruction? */
+#define CHIP_HAS_DWORD_ALIGN() 0
+
+/** Number of performance counters. */
+#define CHIP_PERFORMANCE_COUNTERS() 4
+
+/** Does this chip have auxiliary performance counters? */
+#define CHIP_HAS_AUX_PERF_COUNTERS() 1
+
+/** Is the CBOX_MSR1 SPR supported? */
+#define CHIP_HAS_CBOX_MSR1() 0
+
+/** Is the TILE_RTF_HWM SPR supported? */
+#define CHIP_HAS_TILE_RTF_HWM() 1
+
+/** Is the TILE_WRITE_PENDING SPR supported? */
+#define CHIP_HAS_TILE_WRITE_PENDING() 0
+
+/** Is the PROC_STATUS SPR supported? */
+#define CHIP_HAS_PROC_STATUS_SPR() 1
+
+/** Is the DSTREAM_PF SPR supported? */
+#define CHIP_HAS_DSTREAM_PF() 1
+
+/** Log of the number of mshims we have. */
+#define CHIP_LOG_NUM_MSHIMS() 2
+
+/** Are the bases of the interrupt vector areas fixed? */
+#define CHIP_HAS_FIXED_INTVEC_BASE() 0
+
+/** Are the interrupt masks split up into 2 SPRs? */
+#define CHIP_HAS_SPLIT_INTR_MASK() 0
+
+/** Is the cycle count split up into 2 SPRs? */
+#define CHIP_HAS_SPLIT_CYCLE() 0
+
+/** Does the chip have a static network? */
+#define CHIP_HAS_SN() 0
+
+/** Does the chip have a static network processor? */
+#define CHIP_HAS_SN_PROC() 0
+
+/** Size of the L1 static network processor instruction cache, in bytes. */
+/* #define CHIP_L1SNI_CACHE_SIZE() -- does not apply to chip 10 */
+
+/** Does the chip have DMA support in each tile? */
+#define CHIP_HAS_TILE_DMA() 0
+
+/** Does the chip have the second revision of the directly accessible
+ *  dynamic networks?  This encapsulates a number of characteristics,
+ *  including the absence of the catch-all, the absence of inline message
+ *  tags, the absence of support for network context-switching, and so on.
+ */
+#define CHIP_HAS_REV1_XDN() 1
+
+/** Does the chip have cmpexch and similar (fetchadd, exch, etc.)? */
+#define CHIP_HAS_CMPEXCH() 1
+
+/** Does the chip have memory-mapped I/O support? */
+#define CHIP_HAS_MMIO() 1
+
+/** Does the chip have post-completion interrupts? */
+#define CHIP_HAS_POST_COMPLETION_INTERRUPTS() 1
+
+/** Does the chip have native single step support? */
+#define CHIP_HAS_SINGLE_STEP() 1
+
+#ifndef __OPEN_SOURCE__  /* features only relevant to hypervisor-level code */
+
+/** How many entries are present in the instruction TLB? */
+#define CHIP_ITLB_ENTRIES() 16
+
+/** How many entries are present in the data TLB? */
+#define CHIP_DTLB_ENTRIES() 32
+
+/** How many MAF entries does the XAUI shim have? */
+#define CHIP_XAUI_MAF_ENTRIES() 32
+
+/** Does the memory shim have a source-id table? */
+#define CHIP_HAS_MSHIM_SRCID_TABLE() 0
+
+/** Does the L1 instruction cache clear on reset? */
+#define CHIP_HAS_L1I_CLEAR_ON_RESET() 1
+
+/** Does the chip come out of reset with valid coordinates on all tiles?
+ * Note that if defined, this also implies that the upper left is 1,1.
+ */
+#define CHIP_HAS_VALID_TILE_COORD_RESET() 1
+
+/** Does the chip have unified packet formats? */
+#define CHIP_HAS_UNIFIED_PACKET_FORMATS() 1
+
+/** Does the chip support write reordering? */
+#define CHIP_HAS_WRITE_REORDERING() 1
+
+/** Does the chip support Y-X routing as well as X-Y? */
+#define CHIP_HAS_Y_X_ROUTING() 1
+
+/** Is INTCTRL_3 managed with the correct MPL? */
+#define CHIP_HAS_INTCTRL_3_STATUS_FIX() 1
+
+/** Is it possible to configure the chip to be big-endian? */
+#define CHIP_HAS_BIG_ENDIAN_CONFIG() 1
+
+/** Is the CACHE_RED_WAY_OVERRIDDEN SPR supported? */
+#define CHIP_HAS_CACHE_RED_WAY_OVERRIDDEN() 0
+
+/** Is the DIAG_TRACE_WAY SPR supported? */
+#define CHIP_HAS_DIAG_TRACE_WAY() 0
+
+/** Is the MEM_STRIPE_CONFIG SPR supported? */
+#define CHIP_HAS_MEM_STRIPE_CONFIG() 1
+
+/** Are the TLB_PERF SPRs supported? */
+#define CHIP_HAS_TLB_PERF() 1
+
+/** Is the VDN_SNOOP_SHIM_CTL SPR supported? */
+#define CHIP_HAS_VDN_SNOOP_SHIM_CTL() 0
+
+/** Does the chip support rev1 DMA packets? */
+#define CHIP_HAS_REV1_DMA_PACKETS() 1
+
+/** Does the chip have an IPI shim? */
+#define CHIP_HAS_IPI() 1
+
+#endif /* !__OPEN_SOURCE__ */
+#endif /* __ARCH_CHIP_H__ */

arch/tile/include/arch/icache.h

@@ -16,7 +16,7 @@
 /**
  * @file
  *
- * Support for invalidating bytes in the instruction
+ * Support for invalidating bytes in the instruction cache.
  */
 
 #ifndef __ARCH_ICACHE_H__
@@ -30,11 +30,10 @@
  *
  * @param addr The start of memory to be invalidated.
  * @param size The number of bytes to be invalidated.
- * @param page_size The system's page size, typically the PAGE_SIZE constant
- * in sys/page.h.  This value must be a power of two no larger
- * than the page containing the code to be invalidated. If the value
- * is smaller than the actual page size, this function will still
- * work, but may run slower than necessary.
+ * @param page_size The system's page size, e.g. getpagesize() in userspace.
+ * This value must be a power of two no larger than the page containing
+ * the code to be invalidated. If the value is smaller than the actual page
+ * size, this function will still work, but may run slower than necessary.
  */
 static __inline void
 invalidate_icache(const void* addr, unsigned long size,

arch/tile/include/arch/interrupts_64.h

+/*
+ * Copyright 2011 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#ifndef __ARCH_INTERRUPTS_H__
+#define __ARCH_INTERRUPTS_H__
+
+/** Mask for an interrupt. */
+#ifdef __ASSEMBLER__
+/* Note: must handle breaking interrupts into high and low words manually. */
+#define INT_MASK(intno) (1 << (intno))
+#else
+#define INT_MASK(intno) (1ULL << (intno))
+#endif
+
+
+/** Where a given interrupt executes */
+#define INTERRUPT_VECTOR(i, pl) (0xFC000000 + ((pl) << 24) + ((i) << 8))
+
+/** Where to store a vector for a given interrupt. */
+#define USER_INTERRUPT_VECTOR(i) INTERRUPT_VECTOR(i, 0)
+
+/** The base address of user-level interrupts. */
+#define USER_INTERRUPT_VECTOR_BASE INTERRUPT_VECTOR(0, 0)
+
+
+/** Additional synthetic interrupt. */
+#define INT_BREAKPOINT (63)
+
+#define INT_MEM_ERROR    0
+#define INT_SINGLE_STEP_3    1
+#define INT_SINGLE_STEP_2    2
+#define INT_SINGLE_STEP_1    3
+#define INT_SINGLE_STEP_0    4
+#define INT_IDN_COMPLETE    5
+#define INT_UDN_COMPLETE    6
+#define INT_ITLB_MISS    7
+#define INT_ILL    8
+#define INT_GPV    9
+#define INT_IDN_ACCESS   10
+#define INT_UDN_ACCESS   11
+#define INT_SWINT_3   12
+#define INT_SWINT_2   13
+#define INT_SWINT_1   14
+#define INT_SWINT_0   15
+#define INT_ILL_TRANS   16
+#define INT_UNALIGN_DATA   17
+#define INT_DTLB_MISS   18
+#define INT_DTLB_ACCESS   19
+#define INT_IDN_FIREWALL   20
+#define INT_UDN_FIREWALL   21
+#define INT_TILE_TIMER   22
+#define INT_AUX_TILE_TIMER   23
+#define INT_IDN_TIMER   24
+#define INT_UDN_TIMER   25
+#define INT_IDN_AVAIL   26
+#define INT_UDN_AVAIL   27
+#define INT_IPI_3   28
+#define INT_IPI_2   29
+#define INT_IPI_1   30
+#define INT_IPI_0   31
+#define INT_PERF_COUNT   32
+#define INT_AUX_PERF_COUNT   33
+#define INT_INTCTRL_3   34
+#define INT_INTCTRL_2   35
+#define INT_INTCTRL_1   36
+#define INT_INTCTRL_0   37
+#define INT_BOOT_ACCESS   38
+#define INT_WORLD_ACCESS   39
+#define INT_I_ASID   40
+#define INT_D_ASID   41
+#define INT_DOUBLE_FAULT   42
+
+#define NUM_INTERRUPTS 43
+
+#ifndef __ASSEMBLER__
+#define QUEUED_INTERRUPTS ( \
+    INT_MASK(INT_MEM_ERROR) | \
+    INT_MASK(INT_IDN_COMPLETE) | \
+    INT_MASK(INT_UDN_COMPLETE) | \
+    INT_MASK(INT_IDN_FIREWALL) | \
+    INT_MASK(INT_UDN_FIREWALL) | \
+    INT_MASK(INT_TILE_TIMER) | \
+    INT_MASK(INT_AUX_TILE_TIMER) | \
+    INT_MASK(INT_IDN_TIMER) | \
+    INT_MASK(INT_UDN_TIMER) | \
+    INT_MASK(INT_IDN_AVAIL) | \
+    INT_MASK(INT_UDN_AVAIL) | \
+    INT_MASK(INT_IPI_3) | \
+    INT_MASK(INT_IPI_2) | \
+    INT_MASK(INT_IPI_1) | \
+    INT_MASK(INT_IPI_0) | \
+    INT_MASK(INT_PERF_COUNT) | \
+    INT_MASK(INT_AUX_PERF_COUNT) | \
+    INT_MASK(INT_INTCTRL_3) | \
+    INT_MASK(INT_INTCTRL_2) | \
+    INT_MASK(INT_INTCTRL_1) | \
+    INT_MASK(INT_INTCTRL_0) | \
+    INT_MASK(INT_BOOT_ACCESS) | \
+    INT_MASK(INT_WORLD_ACCESS) | \
+    INT_MASK(INT_I_ASID) | \
+    INT_MASK(INT_D_ASID) | \
+    INT_MASK(INT_DOUBLE_FAULT) | \
+    0)
+#define NONQUEUED_INTERRUPTS ( \
+    INT_MASK(INT_SINGLE_STEP_3) | \
+    INT_MASK(INT_SINGLE_STEP_2) | \
+    INT_MASK(INT_SINGLE_STEP_1) | \
+    INT_MASK(INT_SINGLE_STEP_0) | \
+    INT_MASK(INT_ITLB_MISS) | \
+    INT_MASK(INT_ILL) | \
+    INT_MASK(INT_GPV) | \
+    INT_MASK(INT_IDN_ACCESS) | \
+    INT_MASK(INT_UDN_ACCESS) | \
+    INT_MASK(INT_SWINT_3) | \
+    INT_MASK(INT_SWINT_2) | \
+    INT_MASK(INT_SWINT_1) | \
+    INT_MASK(INT_SWINT_0) | \
+    INT_MASK(INT_ILL_TRANS) | \
+    INT_MASK(INT_UNALIGN_DATA) | \
+    INT_MASK(INT_DTLB_MISS) | \
+    INT_MASK(INT_DTLB_ACCESS) | \
+    0)
+#define CRITICAL_MASKED_INTERRUPTS ( \
+    INT_MASK(INT_MEM_ERROR) | \
+    INT_MASK(INT_SINGLE_STEP_3) | \
+    INT_MASK(INT_SINGLE_STEP_2) | \
+    INT_MASK(INT_SINGLE_STEP_1) | \
+    INT_MASK(INT_SINGLE_STEP_0) | \
+    INT_MASK(INT_IDN_COMPLETE) | \
+    INT_MASK(INT_UDN_COMPLETE) | \
+    INT_MASK(INT_IDN_FIREWALL) | \
+    INT_MASK(INT_UDN_FIREWALL) | \
+    INT_MASK(INT_TILE_TIMER) | \
+    INT_MASK(INT_AUX_TILE_TIMER) | \
+    INT_MASK(INT_IDN_TIMER) | \
+    INT_MASK(INT_UDN_TIMER) | \
+    INT_MASK(INT_IDN_AVAIL) | \
+    INT_MASK(INT_UDN_AVAIL) | \
+    INT_MASK(INT_IPI_3) | \
+    INT_MASK(INT_IPI_2) | \
+    INT_MASK(INT_IPI_1) | \
+    INT_MASK(INT_IPI_0) | \
+    INT_MASK(INT_PERF_COUNT) | \
+    INT_MASK(INT_AUX_PERF_COUNT) | \
+    INT_MASK(INT_INTCTRL_3) | \
+    INT_MASK(INT_INTCTRL_2) | \
+    INT_MASK(INT_INTCTRL_1) | \
+    INT_MASK(INT_INTCTRL_0) | \
+    0)
+#define CRITICAL_UNMASKED_INTERRUPTS ( \
+    INT_MASK(INT_ITLB_MISS) | \
+    INT_MASK(INT_ILL) | \
+    INT_MASK(INT_GPV) | \
+    INT_MASK(INT_IDN_ACCESS) | \
+    INT_MASK(INT_UDN_ACCESS) | \
+    INT_MASK(INT_SWINT_3) | \
+    INT_MASK(INT_SWINT_2) | \
+    INT_MASK(INT_SWINT_1) | \
+    INT_MASK(INT_SWINT_0) | \
+    INT_MASK(INT_ILL_TRANS) | \
+    INT_MASK(INT_UNALIGN_DATA) | \
+    INT_MASK(INT_DTLB_MISS) | \
+    INT_MASK(INT_DTLB_ACCESS) | \
+    INT_MASK(INT_BOOT_ACCESS) | \
+    INT_MASK(INT_WORLD_ACCESS) | \
+    INT_MASK(INT_I_ASID) | \
+    INT_MASK(INT_D_ASID) | \
+    INT_MASK(INT_DOUBLE_FAULT) | \
+    0)
+#define MASKABLE_INTERRUPTS ( \
+    INT_MASK(INT_MEM_ERROR) | \
+    INT_MASK(INT_SINGLE_STEP_3) | \
+    INT_MASK(INT_SINGLE_STEP_2) | \
+    INT_MASK(INT_SINGLE_STEP_1) | \
+    INT_MASK(INT_SINGLE_STEP_0) | \
+    INT_MASK(INT_IDN_COMPLETE) | \
+    INT_MASK(INT_UDN_COMPLETE) | \
+    INT_MASK(INT_IDN_FIREWALL) | \
+    INT_MASK(INT_UDN_FIREWALL) | \
+    INT_MASK(INT_TILE_TIMER) | \
+    INT_MASK(INT_AUX_TILE_TIMER) | \
+    INT_MASK(INT_IDN_TIMER) | \
+    INT_MASK(INT_UDN_TIMER) | \
+    INT_MASK(INT_IDN_AVAIL) | \
+    INT_MASK(INT_UDN_AVAIL) | \
+    INT_MASK(INT_IPI_3) | \
+    INT_MASK(INT_IPI_2) | \
+    INT_MASK(INT_IPI_1) | \
+    INT_MASK(INT_IPI_0) | \
+    INT_MASK(INT_PERF_COUNT) | \
+    INT_MASK(INT_AUX_PERF_COUNT) | \
+    INT_MASK(INT_INTCTRL_3) | \
+    INT_MASK(INT_INTCTRL_2) | \
+    INT_MASK(INT_INTCTRL_1) | \
+    INT_MASK(INT_INTCTRL_0) | \
+    0)
+#define UNMASKABLE_INTERRUPTS ( \
+    INT_MASK(INT_ITLB_MISS) | \
+    INT_MASK(INT_ILL) | \
+    INT_MASK(INT_GPV) | \
+    INT_MASK(INT_IDN_ACCESS) | \
+    INT_MASK(INT_UDN_ACCESS) | \
+    INT_MASK(INT_SWINT_3) | \
+    INT_MASK(INT_SWINT_2) | \
+    INT_MASK(INT_SWINT_1) | \
+    INT_MASK(INT_SWINT_0) | \
+    INT_MASK(INT_ILL_TRANS) | \
+    INT_MASK(INT_UNALIGN_DATA) | \
+    INT_MASK(INT_DTLB_MISS) | \
+    INT_MASK(INT_DTLB_ACCESS) | \
+    INT_MASK(INT_BOOT_ACCESS) | \
+    INT_MASK(INT_WORLD_ACCESS) | \
+    INT_MASK(INT_I_ASID) | \
+    INT_MASK(INT_D_ASID) | \
+    INT_MASK(INT_DOUBLE_FAULT) | \
+    0)
+#define SYNC_INTERRUPTS ( \
+    INT_MASK(INT_SINGLE_STEP_3) | \
+    INT_MASK(INT_SINGLE_STEP_2) | \
+    INT_MASK(INT_SINGLE_STEP_1) | \
+    INT_MASK(INT_SINGLE_STEP_0) | \
+    INT_MASK(INT_IDN_COMPLETE) | \
+    INT_MASK(INT_UDN_COMPLETE) | \
+    INT_MASK(INT_ITLB_MISS) | \
+    INT_MASK(INT_ILL) | \
+    INT_MASK(INT_GPV) | \
+    INT_MASK(INT_IDN_ACCESS) | \
+    INT_MASK(INT_UDN_ACCESS) | \
+    INT_MASK(INT_SWINT_3) | \
+    INT_MASK(INT_SWINT_2) | \
+    INT_MASK(INT_SWINT_1) | \
+    INT_MASK(INT_SWINT_0) | \
+    INT_MASK(INT_ILL_TRANS) | \
+    INT_MASK(INT_UNALIGN_DATA) | \
+    INT_MASK(INT_DTLB_MISS) | \
+    INT_MASK(INT_DTLB_ACCESS) | \
+    0)
+#define NON_SYNC_INTERRUPTS ( \
+    INT_MASK(INT_MEM_ERROR) | \
+    INT_MASK(INT_IDN_FIREWALL) | \
+    INT_MASK(INT_UDN_FIREWALL) | \
+    INT_MASK(INT_TILE_TIMER) | \
+    INT_MASK(INT_AUX_TILE_TIMER) | \
+    INT_MASK(INT_IDN_TIMER) | \
+    INT_MASK(INT_UDN_TIMER) | \
+    INT_MASK(INT_IDN_AVAIL) | \
+    INT_MASK(INT_UDN_AVAIL) | \
+    INT_MASK(INT_IPI_3) | \
+    INT_MASK(INT_IPI_2) | \
+    INT_MASK(INT_IPI_1) | \
+    INT_MASK(INT_IPI_0) | \
+    INT_MASK(INT_PERF_COUNT) | \
+    INT_MASK(INT_AUX_PERF_COUNT) | \
+    INT_MASK(INT_INTCTRL_3) | \
+    INT_MASK(INT_INTCTRL_2) | \
+    INT_MASK(INT_INTCTRL_1) | \
+    INT_MASK(INT_INTCTRL_0) | \
+    INT_MASK(INT_BOOT_ACCESS) | \
+    INT_MASK(INT_WORLD_ACCESS) | \
+    INT_MASK(INT_I_ASID) | \
+    INT_MASK(INT_D_ASID) | \
+    INT_MASK(INT_DOUBLE_FAULT) | \
+    0)
+#endif /* !__ASSEMBLER__ */
+#endif /* !__ARCH_INTERRUPTS_H__ */

arch/tile/include/arch/spr_def.h

@@ -12,6 +12,15 @@
  *   more details.
  */
 
+/* Include the proper base SPR definition file. */
+#ifdef __tilegx__
+#include <arch/spr_def_64.h>
+#else
+#include <arch/spr_def_32.h>
+#endif
+
+#ifdef __KERNEL__
+
 /*
  * In addition to including the proper base SPR definition file, depending
  * on machine architecture, this file defines several macros which allow
@@ -29,7 +38,6 @@
 #define _concat4(a, b, c, d)  __concat4(a, b, c, d)
 
 #ifdef __tilegx__
-#include <arch/spr_def_64.h>
 
 /* TILE-Gx dependent, protection-level dependent SPRs. */
 
@@ -65,7 +73,6 @@
 	_concat4(INT_SINGLE_STEP_, CONFIG_KERNEL_PL,,)
 
 #else
-#include <arch/spr_def_32.h>
 
 /* TILEPro dependent, protection-level dependent SPRs. */
 
@@ -102,3 +109,5 @@
 	_concat4(SPR_INTCTRL_, CONFIG_KERNEL_PL, _STATUS,)
 #define INT_INTCTRL_K \
 	_concat4(INT_INTCTRL_, CONFIG_KERNEL_PL,,)
+
+#endif /* __KERNEL__ */

arch/tile/include/arch/spr_def_64.h

+/*
+ * Copyright 2011 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#ifndef __DOXYGEN__
+
+#ifndef __ARCH_SPR_DEF_H__
+#define __ARCH_SPR_DEF_H__
+
+#define SPR_AUX_PERF_COUNT_0 0x2105
+#define SPR_AUX_PERF_COUNT_1 0x2106
+#define SPR_AUX_PERF_COUNT_CTL 0x2107
+#define SPR_AUX_PERF_COUNT_STS 0x2108
+#define SPR_CMPEXCH_VALUE 0x2780
+#define SPR_CYCLE 0x2781
+#define SPR_DONE 0x2705
+#define SPR_DSTREAM_PF 0x2706
+#define SPR_EVENT_BEGIN 0x2782
+#define SPR_EVENT_END 0x2783
+#define SPR_EX_CONTEXT_0_0 0x2580
+#define SPR_EX_CONTEXT_0_1 0x2581
+#define SPR_EX_CONTEXT_0_1__PL_SHIFT 0
+#define SPR_EX_CONTEXT_0_1__PL_RMASK 0x3
+#define SPR_EX_CONTEXT_0_1__PL_MASK  0x3
+#define SPR_EX_CONTEXT_0_1__ICS_SHIFT 2
+#define SPR_EX_CONTEXT_0_1__ICS_RMASK 0x1
+#define SPR_EX_CONTEXT_0_1__ICS_MASK  0x4
+#define SPR_EX_CONTEXT_1_0 0x2480
+#define SPR_EX_CONTEXT_1_1 0x2481
+#define SPR_EX_CONTEXT_1_1__PL_SHIFT 0
+#define SPR_EX_CONTEXT_1_1__PL_RMASK 0x3
+#define SPR_EX_CONTEXT_1_1__PL_MASK  0x3
+#define SPR_EX_CONTEXT_1_1__ICS_SHIFT 2
+#define SPR_EX_CONTEXT_1_1__ICS_RMASK 0x1
+#define SPR_EX_CONTEXT_1_1__ICS_MASK  0x4
+#define SPR_EX_CONTEXT_2_0 0x2380
+#define SPR_EX_CONTEXT_2_1 0x2381
+#define SPR_EX_CONTEXT_2_1__PL_SHIFT 0
+#define SPR_EX_CONTEXT_2_1__PL_RMASK 0x3
+#define SPR_EX_CONTEXT_2_1__PL_MASK  0x3
+#define SPR_EX_CONTEXT_2_1__ICS_SHIFT 2
+#define SPR_EX_CONTEXT_2_1__ICS_RMASK 0x1
+#define SPR_EX_CONTEXT_2_1__ICS_MASK  0x4
+#define SPR_FAIL 0x2707
+#define SPR_ILL_TRANS_REASON__I_STREAM_VA_RMASK 0x1
+#define SPR_INTCTRL_0_STATUS 0x2505
+#define SPR_INTCTRL_1_STATUS 0x2405
+#define SPR_INTCTRL_2_STATUS 0x2305
+#define SPR_INTERRUPT_CRITICAL_SECTION 0x2708
+#define SPR_INTERRUPT_MASK_0 0x2506
+#define SPR_INTERRUPT_MASK_1 0x2406
+#define SPR_INTERRUPT_MASK_2 0x2306
+#define SPR_INTERRUPT_MASK_RESET_0 0x2507
+#define SPR_INTERRUPT_MASK_RESET_1 0x2407
+#define SPR_INTERRUPT_MASK_RESET_2 0x2307
+#define SPR_INTERRUPT_MASK_SET_0 0x2508
+#define SPR_INTERRUPT_MASK_SET_1 0x2408
+#define SPR_INTERRUPT_MASK_SET_2 0x2308
+#define SPR_INTERRUPT_VECTOR_BASE_0 0x2509
+#define SPR_INTERRUPT_VECTOR_BASE_1 0x2409
+#define SPR_INTERRUPT_VECTOR_BASE_2 0x2309
+#define SPR_INTERRUPT_VECTOR_BASE_3 0x2209
+#define SPR_IPI_EVENT_0 0x1f05
+#define SPR_IPI_EVENT_1 0x1e05
+#define SPR_IPI_EVENT_2 0x1d05
+#define SPR_IPI_EVENT_RESET_0 0x1f06
+#define SPR_IPI_EVENT_RESET_1 0x1e06
+#define SPR_IPI_EVENT_RESET_2 0x1d06
+#define SPR_IPI_EVENT_SET_0 0x1f07
+#define SPR_IPI_EVENT_SET_1 0x1e07
+#define SPR_IPI_EVENT_SET_2 0x1d07
+#define SPR_IPI_MASK_0 0x1f08
+#define SPR_IPI_MASK_1 0x1e08
+#define SPR_IPI_MASK_2 0x1d08
+#define SPR_IPI_MASK_RESET_0 0x1f09
+#define SPR_IPI_MASK_RESET_1 0x1e09
+#define SPR_IPI_MASK_RESET_2 0x1d09
+#define SPR_IPI_MASK_SET_0 0x1f0a
+#define SPR_IPI_MASK_SET_1 0x1e0a
+#define SPR_IPI_MASK_SET_2 0x1d0a
+#define SPR_MPL_AUX_TILE_TIMER_SET_0 0x1700
+#define SPR_MPL_AUX_TILE_TIMER_SET_1 0x1701
+#define SPR_MPL_AUX_TILE_TIMER_SET_2 0x1702
+#define SPR_MPL_INTCTRL_0_SET_0 0x2500
+#define SPR_MPL_INTCTRL_0_SET_1 0x2501
+#define SPR_MPL_INTCTRL_0_SET_2 0x2502
+#define SPR_MPL_INTCTRL_1_SET_0 0x2400
+#define SPR_MPL_INTCTRL_1_SET_1 0x2401
+#define SPR_MPL_INTCTRL_1_SET_2 0x2402
+#define SPR_MPL_INTCTRL_2_SET_0 0x2300
+#define SPR_MPL_INTCTRL_2_SET_1 0x2301
+#define SPR_MPL_INTCTRL_2_SET_2 0x2302
+#define SPR_MPL_UDN_ACCESS_SET_0 0x0b00
+#define SPR_MPL_UDN_ACCESS_SET_1 0x0b01
+#define SPR_MPL_UDN_ACCESS_SET_2 0x0b02
+#define SPR_MPL_UDN_AVAIL_SET_0 0x1b00
+#define SPR_MPL_UDN_AVAIL_SET_1 0x1b01
+#define SPR_MPL_UDN_AVAIL_SET_2 0x1b02
+#define SPR_MPL_UDN_COMPLETE_SET_0 0x0600
+#define SPR_MPL_UDN_COMPLETE_SET_1 0x0601
+#define SPR_MPL_UDN_COMPLETE_SET_2 0x0602
+#define SPR_MPL_UDN_FIREWALL_SET_0 0x1500
+#define SPR_MPL_UDN_FIREWALL_SET_1 0x1501
+#define SPR_MPL_UDN_FIREWALL_SET_2 0x1502
+#define SPR_MPL_UDN_TIMER_SET_0 0x1900
+#define SPR_MPL_UDN_TIMER_SET_1 0x1901
+#define SPR_MPL_UDN_TIMER_SET_2 0x1902
+#define SPR_MPL_WORLD_ACCESS_SET_0 0x2700
+#define SPR_MPL_WORLD_ACCESS_SET_1 0x2701
+#define SPR_MPL_WORLD_ACCESS_SET_2 0x2702
+#define SPR_PASS 0x2709
+#define SPR_PERF_COUNT_0 0x2005
+#define SPR_PERF_COUNT_1 0x2006
+#define SPR_PERF_COUNT_CTL 0x2007
+#define SPR_PERF_COUNT_DN_CTL 0x2008
+#define SPR_PERF_COUNT_STS 0x2009
+#define SPR_PROC_STATUS 0x2784
+#define SPR_SIM_CONTROL 0x2785
+#define SPR_SINGLE_STEP_CONTROL_0 0x0405
+#define SPR_SINGLE_STEP_CONTROL_0__CANCELED_MASK  0x1
+#define SPR_SINGLE_STEP_CONTROL_0__INHIBIT_MASK  0x2
+#define SPR_SINGLE_STEP_CONTROL_1 0x0305
+#define SPR_SINGLE_STEP_CONTROL_1__CANCELED_MASK  0x1
+#define SPR_SINGLE_STEP_CONTROL_1__INHIBIT_MASK  0x2
+#define SPR_SINGLE_STEP_CONTROL_2 0x0205
+#define SPR_SINGLE_STEP_CONTROL_2__CANCELED_MASK  0x1
+#define SPR_SINGLE_STEP_CONTROL_2__INHIBIT_MASK  0x2
+#define SPR_SINGLE_STEP_EN_0_0 0x250a
+#define SPR_SINGLE_STEP_EN_0_1 0x240a
+#define SPR_SINGLE_STEP_EN_0_2 0x230a
+#define SPR_SINGLE_STEP_EN_1_0 0x250b
+#define SPR_SINGLE_STEP_EN_1_1 0x240b
+#define SPR_SINGLE_STEP_EN_1_2 0x230b
+#define SPR_SINGLE_STEP_EN_2_0 0x250c
+#define SPR_SINGLE_STEP_EN_2_1 0x240c
+#define SPR_SINGLE_STEP_EN_2_2 0x230c
+#define SPR_SYSTEM_SAVE_0_0 0x2582
+#define SPR_SYSTEM_SAVE_0_1 0x2583
+#define SPR_SYSTEM_SAVE_0_2 0x2584
+#define SPR_SYSTEM_SAVE_0_3 0x2585
+#define SPR_SYSTEM_SAVE_1_0 0x2482
+#define SPR_SYSTEM_SAVE_1_1 0x2483
+#define SPR_SYSTEM_SAVE_1_2 0x2484
+#define SPR_SYSTEM_SAVE_1_3 0x2485
+#define SPR_SYSTEM_SAVE_2_0 0x2382
+#define SPR_SYSTEM_SAVE_2_1 0x2383
+#define SPR_SYSTEM_SAVE_2_2 0x2384
+#define SPR_SYSTEM_SAVE_2_3 0x2385
+#define SPR_TILE_COORD 0x270b
+#define SPR_TILE_RTF_HWM 0x270c
+#define SPR_TILE_TIMER_CONTROL 0x1605
+#define SPR_UDN_AVAIL_EN 0x1b05
+#define SPR_UDN_DATA_AVAIL 0x0b80
+#define SPR_UDN_DEADLOCK_TIMEOUT 0x1906
+#define SPR_UDN_DEMUX_COUNT_0 0x0b05
+#define SPR_UDN_DEMUX_COUNT_1 0x0b06
+#define SPR_UDN_DEMUX_COUNT_2 0x0b07
+#define SPR_UDN_DEMUX_COUNT_3 0x0b08
+#define SPR_UDN_DIRECTION_PROTECT 0x1505
+
+#endif /* !defined(__ARCH_SPR_DEF_H__) */
+
+#endif /* !defined(__DOXYGEN__) */

arch/tile/include/asm/atomic.h

@@ -130,17 +130,52 @@ static inline int atomic_read(const atomic_t *v)
  */
 #define atomic_inc_not_zero(v)		atomic_add_unless((v), 1, 0)
 
-
-/*
- * We define xchg() and cmpxchg() in the included headers.
- * Note that we do not define __HAVE_ARCH_CMPXCHG, since that would imply
- * that cmpxchg() is an efficient operation, which is not particularly true.
- */
-
 /* Nonexistent functions intended to cause link errors. */
 extern unsigned long __xchg_called_with_bad_pointer(void);
 extern unsigned long __cmpxchg_called_with_bad_pointer(void);
 
+#define xchg(ptr, x)							\
+	({								\
+		typeof(*(ptr)) __x;					\
+		switch (sizeof(*(ptr))) {				\
+		case 4:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic_xchg( \
+				(atomic_t *)(ptr),			\
+				(u32)(typeof((x)-(x)))(x));		\
+			break;						\
+		case 8:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic64_xchg( \
+				(atomic64_t *)(ptr),			\
+				(u64)(typeof((x)-(x)))(x));		\
+			break;						\
+		default:						\
+			__xchg_called_with_bad_pointer();		\
+		}							\
+		__x;							\
+	})
+
+#define cmpxchg(ptr, o, n)						\
+	({								\
+		typeof(*(ptr)) __x;					\
+		switch (sizeof(*(ptr))) {				\
+		case 4:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic_cmpxchg( \
+				(atomic_t *)(ptr),			\
+				(u32)(typeof((o)-(o)))(o),		\
+				(u32)(typeof((n)-(n)))(n));		\
+			break;						\
+		case 8:							\
+			__x = (typeof(__x))(typeof(__x-__x))atomic64_cmpxchg( \
+				(atomic64_t *)(ptr),			\
+				(u64)(typeof((o)-(o)))(o),		\
+				(u64)(typeof((n)-(n)))(n));		\
+			break;						\
+		default:						\
+			__cmpxchg_called_with_bad_pointer();		\
+		}							\
+		__x;							\
+	})
+
 #define tas(ptr) (xchg((ptr), 1))
 
 #endif /* __ASSEMBLY__ */

arch/tile/include/asm/atomic_32.h

@@ -110,16 +110,6 @@ static inline void atomic_set(atomic_t *v, int n)
 	_atomic_xchg(v, n);
 }
 
-#define xchg(ptr, x) ((typeof(*(ptr))) \
-  ((sizeof(*(ptr)) == sizeof(atomic_t)) ? \
-   atomic_xchg((atomic_t *)(ptr), (long)(x)) : \
-   __xchg_called_with_bad_pointer()))
-
-#define cmpxchg(ptr, o, n) ((typeof(*(ptr))) \
-  ((sizeof(*(ptr)) == sizeof(atomic_t)) ? \
-   atomic_cmpxchg((atomic_t *)(ptr), (long)(o), (long)(n)) : \
-   __cmpxchg_called_with_bad_pointer()))
-
 /* A 64bit atomic type */
 
 typedef struct {

arch/tile/include/asm/atomic_64.h

+/*
+ * Copyright 2011 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ *
+ * Do not include directly; use <asm/atomic.h>.
+ */
+
+#ifndef _ASM_TILE_ATOMIC_64_H
+#define _ASM_TILE_ATOMIC_64_H
+
+#ifndef __ASSEMBLY__
+
+#include <arch/spr_def.h>
+
+/* First, the 32-bit atomic ops that are "real" on our 64-bit platform. */
+
+#define atomic_set(v, i) ((v)->counter = (i))
+
+/*
+ * The smp_mb() operations throughout are to support the fact that
+ * Linux requires memory barriers before and after the operation,
+ * on any routine which updates memory and returns a value.
+ */
+
+static inline int atomic_cmpxchg(atomic_t *v, int o, int n)
+{
+	int val;
+	__insn_mtspr(SPR_CMPEXCH_VALUE, o);
+	smp_mb();  /* barrier for proper semantics */
+	val = __insn_cmpexch4((void *)&v->counter, n);
+	smp_mb();  /* barrier for proper semantics */
+	return val;
+}
+
+static inline int atomic_xchg(atomic_t *v, int n)
+{
+	int val;
+	smp_mb();  /* barrier for proper semantics */
+	val = __insn_exch4((void *)&v->counter, n);
+	smp_mb();  /* barrier for proper semantics */
+	return val;
+}
+
+static inline void atomic_add(int i, atomic_t *v)
+{
+	__insn_fetchadd4((void *)&v->counter, i);
+}
+
+static inline int atomic_add_return(int i, atomic_t *v)
+{
+	int val;
+	smp_mb();  /* barrier for proper semantics */
+	val = __insn_fetchadd4((void *)&v->counter, i) + i;
+	barrier();  /* the "+ i" above will wait on memory */
+	return val;
+}
+
+static inline int atomic_add_unless(atomic_t *v, int a, int u)
+{
+	int guess, oldval = v->counter;
+	do {
+		if (oldval == u)
+			break;
+		guess = oldval;
+		oldval = atomic_cmpxchg(v, guess, guess + a);
+	} while (guess != oldval);
+	return oldval != u;
+}
+
+/* Now the true 64-bit operations. */
+
+#define ATOMIC64_INIT(i)	{ (i) }
+
+#define atomic64_read(v)		((v)->counter)
+#define atomic64_set(v, i) ((v)->counter = (i))
+
+static inline long atomic64_cmpxchg(atomic64_t *v, long o, long n)
+{
+	long val;
+	smp_mb();  /* barrier for proper semantics */
+	__insn_mtspr(SPR_CMPEXCH_VALUE, o);
+	val = __insn_cmpexch((void *)&v->counter, n);
+	smp_mb();  /* barrier for proper semantics */
+	return val;
+}
+
+static inline long atomic64_xchg(atomic64_t *v, long n)
+{
+	long val;
+	smp_mb();  /* barrier for proper semantics */
+	val = __insn_exch((void *)&v->counter, n);
+	smp_mb();  /* barrier for proper semantics */
+	return val;
+}
+
+static inline void atomic64_add(long i, atomic64_t *v)
+{
+	__insn_fetchadd((void *)&v->counter, i);
+}
+
+static inline long atomic64_add_return(long i, atomic64_t *v)
+{
+	int val;
+	smp_mb();  /* barrier for proper semantics */
+	val = __insn_fetchadd((void *)&v->counter, i) + i;
+	barrier();  /* the "+ i" above will wait on memory */
+	return val;
+}
+
+static inline long atomic64_add_unless(atomic64_t *v, long a, long u)
+{
+	long guess, oldval = v->counter;
+	do {
+		if (oldval == u)
+			break;
+		guess = oldval;
+		oldval = atomic64_cmpxchg(v, guess, guess + a);
+	} while (guess != oldval);
+	return oldval != u;
+}
+
+#define atomic64_sub_return(i, v)	atomic64_add_return(-(i), (v))
+#define atomic64_sub(i, v)		atomic64_add(-(i), (v))
+#define atomic64_inc_return(v)		atomic64_add_return(1, (v))
+#define atomic64_dec_return(v)		atomic64_sub_return(1, (v))
+#define atomic64_inc(v)			atomic64_add(1, (v))
+#define atomic64_dec(v)			atomic64_sub(1, (v))
+
+#define atomic64_inc_and_test(v)	(atomic64_inc_return(v) == 0)
+#define atomic64_dec_and_test(v)	(atomic64_dec_return(v) == 0)
+#define atomic64_sub_and_test(i, v)	(atomic64_sub_return((i), (v)) == 0)
+#define atomic64_add_negative(i, v)	(atomic64_add_return((i), (v)) < 0)
+
+#define atomic64_inc_not_zero(v)	atomic64_add_unless((v), 1, 0)
+
+/* Atomic dec and inc don't implement barrier, so provide them if needed. */
+#define smp_mb__before_atomic_dec()	smp_mb()
+#define smp_mb__after_atomic_dec()	smp_mb()
+#define smp_mb__before_atomic_inc()	smp_mb()
+#define smp_mb__after_atomic_inc()	smp_mb()
+
+/* Define this to indicate that cmpxchg is an efficient operation. */
+#define __HAVE_ARCH_CMPXCHG
+
+#endif /* !__ASSEMBLY__ */
+
+#endif /* _ASM_TILE_ATOMIC_64_H */

arch/tile/include/asm/backtrace.h

@@ -12,80 +12,41 @@
  *   more details.
  */
 
-#ifndef _TILE_BACKTRACE_H
-#define _TILE_BACKTRACE_H
-
-
+#ifndef _ASM_TILE_BACKTRACE_H
+#define _ASM_TILE_BACKTRACE_H
 
 #include <linux/types.h>
 
-#include <arch/chip.h>
-
-#if defined(__tile__)
-typedef unsigned long VirtualAddress;
-#elif CHIP_VA_WIDTH() > 32
-typedef unsigned long long VirtualAddress;
-#else
-typedef unsigned int VirtualAddress;
-#endif
-
-
-/** Reads 'size' bytes from 'address' and writes the data to 'result'.
+/* Reads 'size' bytes from 'address' and writes the data to 'result'.
  * Returns true if successful, else false (e.g. memory not readable).
  */
 typedef bool (*BacktraceMemoryReader)(void *result,
-				      VirtualAddress address,
+				      unsigned long address,
 				      unsigned int size,
 				      void *extra);
 
 typedef struct {
-	/** Current PC. */
-	VirtualAddress pc;
+	/* Current PC. */
+	unsigned long pc;
 
-	/** Current stack pointer value. */
-	VirtualAddress sp;
+	/* Current stack pointer value. */
+	unsigned long sp;
 
-	/** Current frame pointer value (i.e. caller's stack pointer) */
-	VirtualAddress fp;
+	/* Current frame pointer value (i.e. caller's stack pointer) */
+	unsigned long fp;
 
-	/** Internal use only: caller's PC for first frame. */
-	VirtualAddress initial_frame_caller_pc;
+	/* Internal use only: caller's PC for first frame. */
+	unsigned long initial_frame_caller_pc;
 
-	/** Internal use only: callback to read memory. */
+	/* Internal use only: callback to read memory. */
 	BacktraceMemoryReader read_memory_func;
 
-	/** Internal use only: arbitrary argument to read_memory_func. */
+	/* Internal use only: arbitrary argument to read_memory_func. */
 	void *read_memory_func_extra;
 
 } BacktraceIterator;
 
 
-/** Initializes a backtracer to start from the given location.
- *
- * If the frame pointer cannot be determined it is set to -1.
- *
- * @param state The state to be filled in.
- * @param read_memory_func A callback that reads memory. If NULL, a default
- *        value is provided.
- * @param read_memory_func_extra An arbitrary argument to read_memory_func.
- * @param pc The current PC.
- * @param lr The current value of the 'lr' register.
- * @param sp The current value of the 'sp' register.
- * @param r52 The current value of the 'r52' register.
- */
-extern void backtrace_init(BacktraceIterator *state,
-			   BacktraceMemoryReader read_memory_func,
-			   void *read_memory_func_extra,
-			   VirtualAddress pc, VirtualAddress lr,
-			   VirtualAddress sp, VirtualAddress r52);
-
-
-/** Advances the backtracing state to the calling frame, returning
- * true iff successful.
- */
-extern bool backtrace_next(BacktraceIterator *state);
-
-
 typedef enum {
 
 	/* We have no idea what the caller's pc is. */
@@ -138,7 +99,7 @@ enum {
 };
 
 
-/** Internal constants used to define 'info' operands. */
+/* Internal constants used to define 'info' operands. */
 enum {
 	/* 0 and 1 are reserved, as are all negative numbers. */
 
@@ -147,13 +108,10 @@ enum {
 	CALLER_SP_IN_R52_BASE = 4,
 
 	CALLER_SP_OFFSET_BASE = 8,
-
-	/* Marks the entry point of certain functions. */
-	ENTRY_POINT_INFO_OP = 16
 };
 
 
-/** Current backtracer state describing where it thinks the caller is. */
+/* Current backtracer state describing where it thinks the caller is. */
 typedef struct {
 	/*
 	 * Public fields
@@ -192,7 +150,13 @@ typedef struct {
 
 } CallerLocation;
 
+extern void backtrace_init(BacktraceIterator *state,
+                          BacktraceMemoryReader read_memory_func,
+                          void *read_memory_func_extra,
+                          unsigned long pc, unsigned long lr,
+                          unsigned long sp, unsigned long r52);
 
 
+extern bool backtrace_next(BacktraceIterator *state);
 
-#endif /* _TILE_BACKTRACE_H */
+#endif /* _ASM_TILE_BACKTRACE_H */

arch/tile/include/asm/bitops.h

@@ -122,6 +122,7 @@ static inline unsigned long __arch_hweight64(__u64 w)
 #include <asm-generic/bitops/lock.h>
 #include <asm-generic/bitops/find.h>
 #include <asm-generic/bitops/sched.h>
+#include <asm-generic/bitops/non-atomic.h>
 #include <asm-generic/bitops/le.h>
 
 #endif /* _ASM_TILE_BITOPS_H */

arch/tile/include/asm/bitops_32.h

@@ -126,7 +126,6 @@ static inline int test_and_change_bit(unsigned nr,
 #define smp_mb__before_clear_bit()	smp_mb()
 #define smp_mb__after_clear_bit()	do {} while (0)
 
-#include <asm-generic/bitops/non-atomic.h>
 #include <asm-generic/bitops/ext2-atomic.h>
 
 #endif /* _ASM_TILE_BITOPS_32_H */

arch/tile/include/asm/bitops_64.h

+/*
+ * Copyright 2011 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#ifndef _ASM_TILE_BITOPS_64_H
+#define _ASM_TILE_BITOPS_64_H
+
+#include <linux/compiler.h>
+#include <asm/atomic.h>
+#include <asm/system.h>
+
+/* See <asm/bitops.h> for API comments. */
+
+static inline void set_bit(unsigned nr, volatile unsigned long *addr)
+{
+	unsigned long mask = (1UL << (nr % BITS_PER_LONG));
+	__insn_fetchor((void *)(addr + nr / BITS_PER_LONG), mask);
+}
+
+static inline void clear_bit(unsigned nr, volatile unsigned long *addr)
+{
+	unsigned long mask = (1UL << (nr % BITS_PER_LONG));
+	__insn_fetchand((void *)(addr + nr / BITS_PER_LONG), ~mask);
+}
+
+#define smp_mb__before_clear_bit()	smp_mb()
+#define smp_mb__after_clear_bit()	smp_mb()
+
+
+static inline void change_bit(unsigned nr, volatile unsigned long *addr)
+{
+	unsigned long old, mask = (1UL << (nr % BITS_PER_LONG));
+	long guess, oldval;
+	addr += nr / BITS_PER_LONG;
+	old = *addr;
+	do {
+		guess = oldval;
+		oldval = atomic64_cmpxchg((atomic64_t *)addr,
+					  guess, guess ^ mask);
+	} while (guess != oldval);
+}
+
+
+/*
+ * The test_and_xxx_bit() routines require a memory fence before we
+ * start the operation, and after the operation completes.  We use
+ * smp_mb() before, and rely on the "!= 0" comparison, plus a compiler
+ * barrier(), to block until the atomic op is complete.
+ */
+
+static inline int test_and_set_bit(unsigned nr, volatile unsigned long *addr)
+{
+	int val;
+	unsigned long mask = (1UL << (nr % BITS_PER_LONG));
+	smp_mb();  /* barrier for proper semantics */
+	val = (__insn_fetchor((void *)(addr + nr / BITS_PER_LONG), mask)
+	       & mask) != 0;
+	barrier();
+	return val;
+}
+
+
+static inline int test_and_clear_bit(unsigned nr, volatile unsigned long *addr)
+{
+	int val;
+	unsigned long mask = (1UL << (nr % BITS_PER_LONG));
+	smp_mb();  /* barrier for proper semantics */
+	val = (__insn_fetchand((void *)(addr + nr / BITS_PER_LONG), ~mask)
+	       & mask) != 0;
+	barrier();
+	return val;
+}
+
+
+static inline int test_and_change_bit(unsigned nr,
+				      volatile unsigned long *addr)
+{
+	unsigned long mask = (1UL << (nr % BITS_PER_LONG));
+	long guess, oldval = *addr;
+	addr += nr / BITS_PER_LONG;
+	oldval = *addr;
+	do {
+		guess = oldval;
+		oldval = atomic64_cmpxchg((atomic64_t *)addr,
+					  guess, guess ^ mask);
+	} while (guess != oldval);
+	return (oldval & mask) != 0;
+}
+
+#define ext2_set_bit_atomic(lock, nr, addr)			\
+	test_and_set_bit((nr), (unsigned long *)(addr))
+#define ext2_clear_bit_atomic(lock, nr, addr)			\
+	test_and_clear_bit((nr), (unsigned long *)(addr))
+
+#endif /* _ASM_TILE_BITOPS_64_H */

arch/tile/include/asm/cacheflush.h

@@ -116,22 +116,28 @@ static inline void __finv_buffer(void *buffer, size_t size)
 }
 
 
-/* Invalidate a VA range, then memory fence. */
+/* Invalidate a VA range and wait for it to be complete. */
 static inline void inv_buffer(void *buffer, size_t size)
 {
 	__inv_buffer(buffer, size);
-	mb_incoherent();
+	mb();
 }
 
-/* Flush a VA range, then memory fence. */
-static inline void flush_buffer(void *buffer, size_t size)
+/*
+ * Flush a locally-homecached VA range and wait for the evicted
+ * cachelines to hit memory.
+ */
+static inline void flush_buffer_local(void *buffer, size_t size)
 {
 	__flush_buffer(buffer, size);
 	mb_incoherent();
 }
 
-/* Flush & invalidate a VA range, then memory fence. */
-static inline void finv_buffer(void *buffer, size_t size)
+/*
+ * Flush and invalidate a locally-homecached VA range and wait for the
+ * evicted cachelines to hit memory.
+ */
+static inline void finv_buffer_local(void *buffer, size_t size)
 {
 	__finv_buffer(buffer, size);
 	mb_incoherent();

arch/tile/include/asm/compat.h

@@ -215,8 +215,8 @@ struct compat_sigaction;
 struct compat_siginfo;
 struct compat_sigaltstack;
 long compat_sys_execve(const char __user *path,
-		       const compat_uptr_t __user *argv,
-		       const compat_uptr_t __user *envp, struct pt_regs *);
+		       compat_uptr_t __user *argv,
+		       compat_uptr_t __user *envp, struct pt_regs *);
 long compat_sys_rt_sigaction(int sig, struct compat_sigaction __user *act,
 			     struct compat_sigaction __user *oact,
 			     size_t sigsetsize);

arch/tile/include/asm/dma-mapping.h

@@ -65,7 +65,8 @@ extern void dma_sync_single_range_for_cpu(struct device *, dma_addr_t,
 extern void dma_sync_single_range_for_device(struct device *, dma_addr_t,
 					     unsigned long offset, size_t,
 					     enum dma_data_direction);
-extern void dma_cache_sync(void *vaddr, size_t, enum dma_data_direction);
+extern void dma_cache_sync(struct device *dev, void *vaddr, size_t,
+			   enum dma_data_direction);
 
 static inline int
 dma_mapping_error(struct device *dev, dma_addr_t dma_addr)

arch/tile/include/asm/fb.h

+#include <asm-generic/fb.h>

arch/tile/include/asm/io.h

@@ -52,6 +52,7 @@ extern void iounmap(volatile void __iomem *addr);
 #endif
 
 #define ioremap_nocache(physaddr, size)		ioremap(physaddr, size)
+#define ioremap_wc(physaddr, size)		ioremap(physaddr, size)
 #define ioremap_writethrough(physaddr, size)	ioremap(physaddr, size)
 #define ioremap_fullcache(physaddr, size)	ioremap(physaddr, size)
 
@@ -161,6 +162,15 @@ static inline void _tile_writeq(u64 val, unsigned long addr)
 #define iowrite32 writel
 #define iowrite64 writeq
 
+static inline void memset_io(void *dst, int val, size_t len)
+{
+	int x;
+	BUG_ON((unsigned long)dst & 0x3);
+	val = (val & 0xff) * 0x01010101;
+	for (x = 0; x < len; x += 4)
+		writel(val, dst + x);
+}
+
 static inline void memcpy_fromio(void *dst, const volatile void __iomem *src,
 				 size_t len)
 {
@@ -269,6 +279,11 @@ static inline void outsl(unsigned long addr, const void *buffer, int count)
 	ioport_panic();
 }
 
+#define ioread16be(addr)	be16_to_cpu(ioread16(addr))
+#define ioread32be(addr)	be32_to_cpu(ioread32(addr))
+#define iowrite16be(v, addr)	iowrite16(be16_to_cpu(v), (addr))
+#define iowrite32be(v, addr)	iowrite32(be32_to_cpu(v), (addr))
+
 #define ioread8_rep(p, dst, count) \
 	insb((unsigned long) (p), (dst), (count))
 #define ioread16_rep(p, dst, count) \
@@ -283,4 +298,7 @@ static inline void outsl(unsigned long addr, const void *buffer, int count)
 #define iowrite32_rep(p, src, count) \
 	outsl((unsigned long) (p), (src), (count))
 
+#define virt_to_bus     virt_to_phys
+#define bus_to_virt     phys_to_virt
+
 #endif /* _ASM_TILE_IO_H */

arch/tile/include/asm/irq.h

@@ -23,6 +23,8 @@
 /* IRQ numbers used for linux IPIs. */
 #define IRQ_RESCHEDULE 1
 
+#define irq_canonicalize(irq)   (irq)
+
 void ack_bad_irq(unsigned int irq);
 
 /*

arch/tile/include/asm/mmu_context.h

@@ -100,8 +100,8 @@ static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next,
 		__get_cpu_var(current_asid) = asid;
 
 		/* Clear cpu from the old mm, and set it in the new one. */
-		cpumask_clear_cpu(cpu, &prev->cpu_vm_mask);
-		cpumask_set_cpu(cpu, &next->cpu_vm_mask);
+		cpumask_clear_cpu(cpu, mm_cpumask(prev));
+		cpumask_set_cpu(cpu, mm_cpumask(next));
 
 		/* Re-load page tables */
 		install_page_table(next->pgd, asid);

arch/tile/include/asm/opcode-tile_32.h

@@ -1502,5 +1502,12 @@ extern int parse_insn_tile(tile_bundle_bits bits,
                            decoded[TILE_MAX_INSTRUCTIONS_PER_BUNDLE]);
 
 
+/* Given a set of bundle bits and a specific pipe, returns which
+ * instruction the bundle contains in that pipe.
+ */
+extern const struct tile_opcode *
+find_opcode(tile_bundle_bits bits, tile_pipeline pipe);
+
+
 
 #endif /* opcode_tile_h */

arch/tile/include/asm/page.h

@@ -16,7 +16,8 @@
 #define _ASM_TILE_PAGE_H
 
 #include <linux/const.h>
-#include <hv/pagesize.h>
+#include <hv/hypervisor.h>
+#include <arch/chip.h>
 
 /* PAGE_SHIFT and HPAGE_SHIFT determine the page sizes. */
 #define PAGE_SHIFT	HV_LOG2_PAGE_SIZE_SMALL
@@ -28,8 +29,6 @@
 #define PAGE_MASK	(~(PAGE_SIZE - 1))
 #define HPAGE_MASK	(~(HPAGE_SIZE - 1))
 
-#ifdef __KERNEL__
-
 /*
  * If the Kconfig doesn't specify, set a maximum zone order that
  * is enough so that we can create huge pages from small pages given
@@ -39,9 +38,6 @@
 #define CONFIG_FORCE_MAX_ZONEORDER (HPAGE_SHIFT - PAGE_SHIFT + 1)
 #endif
 
-#include <hv/hypervisor.h>
-#include <arch/chip.h>
-
 #ifndef __ASSEMBLY__
 
 #include <linux/types.h>
@@ -91,6 +87,10 @@ typedef struct page *pgtable_t;
 /* Must be a macro since it is used to create constants. */
 #define __pgprot(val) hv_pte(val)
 
+/* Rarely-used initializers, typically with a "zero" value. */
+#define __pte(x) hv_pte(x)
+#define __pgd(x) hv_pte(x)
+
 static inline u64 pgprot_val(pgprot_t pgprot)
 {
 	return hv_pte_val(pgprot);
@@ -110,6 +110,8 @@ static inline u64 pgd_val(pgd_t pgd)
 
 typedef HV_PTE pmd_t;
 
+#define __pmd(x) hv_pte(x)
+
 static inline u64 pmd_val(pmd_t pmd)
 {
 	return hv_pte_val(pmd);
@@ -318,7 +320,7 @@ static inline int pfn_valid(unsigned long pfn)
 
 /* Provide as macros since these require some other headers included. */
 #define page_to_pa(page) ((phys_addr_t)(page_to_pfn(page)) << PAGE_SHIFT)
-#define virt_to_page(kaddr) pfn_to_page(kaddr_to_pfn(kaddr))
+#define virt_to_page(kaddr) pfn_to_page(kaddr_to_pfn((void *)(kaddr)))
 #define page_to_virt(page) pfn_to_kaddr(page_to_pfn(page))
 
 struct mm_struct;
@@ -331,6 +333,4 @@ extern pte_t *virt_to_pte(struct mm_struct *mm, unsigned long addr);
 
 #include <asm-generic/memory_model.h>
 
-#endif /* __KERNEL__ */
-
 #endif /* _ASM_TILE_PAGE_H */

arch/tile/include/asm/parport.h

+#include <asm-generic/parport.h>

arch/tile/include/asm/pci.h

@@ -46,7 +46,8 @@ struct pci_controller {
  */
 #define PCI_DMA_BUS_IS_PHYS     1
 
-int __init tile_pci_init(void);
+int __devinit tile_pci_init(void);
+int __devinit pcibios_init(void);
 
 void __iomem *pci_iomap(struct pci_dev *dev, int bar, unsigned long max);
 static inline void pci_iounmap(struct pci_dev *dev, void __iomem *addr) {}

arch/tile/include/asm/pgtable_64.h

+/*
+ * Copyright 2011 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ *
+ */
+
+#ifndef _ASM_TILE_PGTABLE_64_H
+#define _ASM_TILE_PGTABLE_64_H
+
+/* The level-0 page table breaks the address space into 32-bit chunks. */
+#define PGDIR_SHIFT	HV_LOG2_L1_SPAN
+#define PGDIR_SIZE	HV_L1_SPAN
+#define PGDIR_MASK	(~(PGDIR_SIZE-1))
+#define PTRS_PER_PGD	HV_L0_ENTRIES
+#define SIZEOF_PGD	(PTRS_PER_PGD * sizeof(pgd_t))
+
+/*
+ * The level-1 index is defined by the huge page size.  A PMD is composed
+ * of PTRS_PER_PMD pgd_t's and is the middle level of the page table.
+ */
+#define PMD_SHIFT	HV_LOG2_PAGE_SIZE_LARGE
+#define PMD_SIZE	HV_PAGE_SIZE_LARGE
+#define PMD_MASK	(~(PMD_SIZE-1))
+#define PTRS_PER_PMD	(1 << (PGDIR_SHIFT - PMD_SHIFT))
+#define SIZEOF_PMD	(PTRS_PER_PMD * sizeof(pmd_t))
+
+/*
+ * The level-2 index is defined by the difference between the huge
+ * page size and the normal page size.  A PTE is composed of
+ * PTRS_PER_PTE pte_t's and is the bottom level of the page table.
+ * Note that the hypervisor docs use PTE for what we call pte_t, so
+ * this nomenclature is somewhat confusing.
+ */
+#define PTRS_PER_PTE (1 << (HV_LOG2_PAGE_SIZE_LARGE - HV_LOG2_PAGE_SIZE_SMALL))
+#define SIZEOF_PTE	(PTRS_PER_PTE * sizeof(pte_t))
+
+/*
+ * Align the vmalloc area to an L2 page table, and leave a guard page
+ * at the beginning and end.  The vmalloc code also puts in an internal
+ * guard page between each allocation.
+ */
+#define _VMALLOC_END	HUGE_VMAP_BASE
+#define VMALLOC_END	(_VMALLOC_END - PAGE_SIZE)
+#define VMALLOC_START	(_VMALLOC_START + PAGE_SIZE)
+
+#define HUGE_VMAP_END	(HUGE_VMAP_BASE + PGDIR_SIZE)
+
+#ifndef __ASSEMBLY__
+
+/* We have no pud since we are a three-level page table. */
+#include <asm-generic/pgtable-nopud.h>
+
+static inline int pud_none(pud_t pud)
+{
+	return pud_val(pud) == 0;
+}
+
+static inline int pud_present(pud_t pud)
+{
+	return pud_val(pud) & _PAGE_PRESENT;
+}
+
+#define pmd_ERROR(e) \
+	pr_err("%s:%d: bad pmd 0x%016llx.\n", __FILE__, __LINE__, pmd_val(e))
+
+static inline void pud_clear(pud_t *pudp)
+{
+	__pte_clear(&pudp->pgd);
+}
+
+static inline int pud_bad(pud_t pud)
+{
+	return ((pud_val(pud) & _PAGE_ALL) != _PAGE_TABLE);
+}
+
+/* Return the page-table frame number (ptfn) that a pud_t points at. */
+#define pud_ptfn(pud) hv_pte_get_ptfn((pud).pgd)
+
+/*
+ * A given kernel pud_t maps to a kernel pmd_t table at a specific
+ * virtual address.  Since kernel pmd_t tables can be aligned at
+ * sub-page granularity, this macro can return non-page-aligned
+ * pointers, despite its name.
+ */
+#define pud_page_vaddr(pud) \
+	(__va((phys_addr_t)pud_ptfn(pud) << HV_LOG2_PAGE_TABLE_ALIGN))
+
+/*
+ * A pud_t points to a pmd_t array.  Since we can have multiple per
+ * page, we don't have a one-to-one mapping of pud_t's to pages.
+ */
+#define pud_page(pud) pfn_to_page(HV_PTFN_TO_PFN(pud_ptfn(pud)))
+
+static inline unsigned long pud_index(unsigned long address)
+{
+	return (address >> PUD_SHIFT) & (PTRS_PER_PUD - 1);
+}
+
+#define pmd_offset(pud, address) \
+	((pmd_t *)pud_page_vaddr(*(pud)) + pmd_index(address))
+
+static inline void __set_pmd(pmd_t *pmdp, pmd_t pmdval)
+{
+	set_pte(pmdp, pmdval);
+}
+
+/* Create a pmd from a PTFN and pgprot. */
+static inline pmd_t ptfn_pmd(unsigned long ptfn, pgprot_t prot)
+{
+	return hv_pte_set_ptfn(prot, ptfn);
+}
+
+/* Return the page-table frame number (ptfn) that a pmd_t points at. */
+static inline unsigned long pmd_ptfn(pmd_t pmd)
+{
+	return hv_pte_get_ptfn(pmd);
+}
+
+static inline void pmd_clear(pmd_t *pmdp)
+{
+	__pte_clear(pmdp);
+}
+
+/* Normalize an address to having the correct high bits set. */
+#define pgd_addr_normalize pgd_addr_normalize
+static inline unsigned long pgd_addr_normalize(unsigned long addr)
+{
+	return ((long)addr << (CHIP_WORD_SIZE() - CHIP_VA_WIDTH())) >>
+		(CHIP_WORD_SIZE() - CHIP_VA_WIDTH());
+}
+
+/* We don't define any pgds for these addresses. */
+static inline int pgd_addr_invalid(unsigned long addr)
+{
+	return addr >= MEM_HV_START ||
+		(addr > MEM_LOW_END && addr < MEM_HIGH_START);
+}
+
+/*
+ * Use atomic instructions to provide atomicity against the hypervisor.
+ */
+#define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
+static inline int ptep_test_and_clear_young(struct vm_area_struct *vma,
+					    unsigned long addr, pte_t *ptep)
+{
+	return (__insn_fetchand(&ptep->val, ~HV_PTE_ACCESSED) >>
+		HV_PTE_INDEX_ACCESSED) & 0x1;
+}
+
+#define __HAVE_ARCH_PTEP_SET_WRPROTECT
+static inline void ptep_set_wrprotect(struct mm_struct *mm,
+				      unsigned long addr, pte_t *ptep)
+{
+	__insn_fetchand(&ptep->val, ~HV_PTE_WRITABLE);
+}
+
+#define __HAVE_ARCH_PTEP_GET_AND_CLEAR
+static inline pte_t ptep_get_and_clear(struct mm_struct *mm,
+				       unsigned long addr, pte_t *ptep)
+{
+	return hv_pte(__insn_exch(&ptep->val, 0UL));
+}
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* _ASM_TILE_PGTABLE_64_H */

arch/tile/include/asm/processor.h

@@ -215,6 +215,8 @@ static inline void release_thread(struct task_struct *dead_task)
 
 extern int kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
 
+extern int do_work_pending(struct pt_regs *regs, u32 flags);
+
 
 /*
  * Return saved (kernel) PC of a blocked thread.
@@ -255,10 +257,6 @@ static inline void cpu_relax(void)
 	barrier();
 }
 
-struct siginfo;
-extern void arch_coredump_signal(struct siginfo *, struct pt_regs *);
-#define arch_coredump_signal arch_coredump_signal
-
 /* Info on this processor (see fs/proc/cpuinfo.c) */
 struct seq_operations;
 extern const struct seq_operations cpuinfo_op;
@@ -269,9 +267,6 @@ extern char chip_model[64];
 /* Data on which physical memory controller corresponds to which NUMA node. */
 extern int node_controller[];
 
-/* Do we dump information to the console when a user application crashes? */
-extern int show_crashinfo;
-
 #if CHIP_HAS_CBOX_HOME_MAP()
 /* Does the heap allocator return hash-for-home pages by default? */
 extern int hash_default;

arch/tile/include/asm/serial.h

+#include <asm-generic/serial.h>

arch/tile/include/asm/signal.h

@@ -28,6 +28,10 @@ struct pt_regs;
 int restore_sigcontext(struct pt_regs *, struct sigcontext __user *);
 int setup_sigcontext(struct sigcontext __user *, struct pt_regs *);
 void do_signal(struct pt_regs *regs);
+void signal_fault(const char *type, struct pt_regs *,
+		  void __user *frame, int sig);
+void trace_unhandled_signal(const char *type, struct pt_regs *regs,
+			    unsigned long address, int signo);
 #endif
 
 #endif /* _ASM_TILE_SIGNAL_H */

arch/tile/include/asm/spinlock_64.h

+/*
+ * Copyright 2011 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   This program is distributed in the hope that it will be useful, but
+ *   WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ *
+ * 64-bit SMP ticket spinlocks, allowing only a single CPU anywhere
+ * (the type definitions are in asm/spinlock_types.h)
+ */
+
+#ifndef _ASM_TILE_SPINLOCK_64_H
+#define _ASM_TILE_SPINLOCK_64_H
+
+/* Shifts and masks for the various fields in "lock". */
+#define __ARCH_SPIN_CURRENT_SHIFT	17
+#define __ARCH_SPIN_NEXT_MASK		0x7fff
+#define __ARCH_SPIN_NEXT_OVERFLOW	0x8000
+
+/*
+ * Return the "current" portion of a ticket lock value,
+ * i.e. the number that currently owns the lock.
+ */
+static inline int arch_spin_current(u32 val)
+{
+	return val >> __ARCH_SPIN_CURRENT_SHIFT;
+}
+
+/*
+ * Return the "next" portion of a ticket lock value,
+ * i.e. the number that the next task to try to acquire the lock will get.
+ */
+static inline int arch_spin_next(u32 val)
+{
+	return val & __ARCH_SPIN_NEXT_MASK;
+}
+
+/* The lock is locked if a task would have to wait to get it. */
+static inline int arch_spin_is_locked(arch_spinlock_t *lock)
+{
+	u32 val = lock->lock;
+	return arch_spin_current(val) != arch_spin_next(val);
+}
+
+/* Bump the current ticket so the next task owns the lock. */
+static inline void arch_spin_unlock(arch_spinlock_t *lock)
+{
+	wmb();  /* guarantee anything modified under the lock is visible */
+	__insn_fetchadd4(&lock->lock, 1U << __ARCH_SPIN_CURRENT_SHIFT);
+}
+
+void arch_spin_unlock_wait(arch_spinlock_t *lock);
+
+void arch_spin_lock_slow(arch_spinlock_t *lock, u32 val);
+
+/* Grab the "next" ticket number and bump it atomically.
+ * If the current ticket is not ours, go to the slow path.
+ * We also take the slow path if the "next" value overflows.
+ */
+static inline void arch_spin_lock(arch_spinlock_t *lock)
+{
+	u32 val = __insn_fetchadd4(&lock->lock, 1);
+	u32 ticket = val & (__ARCH_SPIN_NEXT_MASK | __ARCH_SPIN_NEXT_OVERFLOW);
+	if (unlikely(arch_spin_current(val) != ticket))
+		arch_spin_lock_slow(lock, ticket);
+}
+
+/* Try to get the lock, and return whether we succeeded. */
+int arch_spin_trylock(arch_spinlock_t *lock);
+
+/* We cannot take an interrupt after getting a ticket, so don't enable them. */
+#define arch_spin_lock_flags(lock, flags) arch_spin_lock(lock)
+
+/*
+ * Read-write spinlocks, allowing multiple readers
+ * but only one writer.
+ *
+ * We use fetchadd() for readers, and fetchor() with the sign bit
+ * for writers.
+ */
+
+#define __WRITE_LOCK_BIT (1 << 31)
+
+static inline int arch_write_val_locked(int val)
+{
+	return val < 0;  /* Optimize "val & __WRITE_LOCK_BIT". */
+}
+
+/**
+ * read_can_lock - would read_trylock() succeed?
+ * @lock: the rwlock in question.
+ */
+static inline int arch_read_can_lock(arch_rwlock_t *rw)
+{
+	return !arch_write_val_locked(rw->lock);
+}
+
+/**
+ * write_can_lock - would write_trylock() succeed?
+ * @lock: the rwlock in question.
+ */
+static inline int arch_write_can_lock(arch_rwlock_t *rw)
+{
+	return rw->lock == 0;
+}
+
+extern void __read_lock_failed(arch_rwlock_t *rw);
+
+static inline void arch_read_lock(arch_rwlock_t *rw)
+{
+	u32 val = __insn_fetchaddgez4(&rw->lock, 1);
+	if (unlikely(arch_write_val_locked(val)))
+		__read_lock_failed(rw);
+}
+
+extern void __write_lock_failed(arch_rwlock_t *rw, u32 val);
+
+static inline void arch_write_lock(arch_rwlock_t *rw)
+{
+	u32 val = __insn_fetchor4(&rw->lock, __WRITE_LOCK_BIT);
+	if (unlikely(val != 0))
+		__write_lock_failed(rw, val);
+}
+
+static inline void arch_read_unlock(arch_rwlock_t *rw)
+{
+	__insn_mf();
+	__insn_fetchadd4(&rw->lock, -1);
+}
+
+static inline void arch_write_unlock(arch_rwlock_t *rw)
+{
+	__insn_mf();
+	rw->lock = 0;
+}
+
+static inline int arch_read_trylock(arch_rwlock_t *rw)
+{
+	return !arch_write_val_locked(__insn_fetchaddgez4(&rw->lock, 1));
+}
+
+static inline int arch_write_trylock(arch_rwlock_t *rw)
+{
+	u32 val = __insn_fetchor4(&rw->lock, __WRITE_LOCK_BIT);
+	if (likely(val == 0))
+		return 1;
+	if (!arch_write_val_locked(val))
+		__insn_fetchand4(&rw->lock, ~__WRITE_LOCK_BIT);
+	return 0;
+}
+
+#define arch_read_lock_flags(lock, flags) arch_read_lock(lock)
+#define arch_write_lock_flags(lock, flags) arch_write_lock(lock)
+
+#endif /* _ASM_TILE_SPINLOCK_64_H */

arch/tile/include/asm/stat.h

-#ifdef CONFIG_COMPAT
+#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 #define __ARCH_WANT_STAT64	/* Used for compat_sys_stat64() etc. */
 #endif
 #include <asm-generic/stat.h>

arch/tile/include/asm/swab.h

@@ -18,12 +18,6 @@
 /* Tile gcc is always >= 4.3.0, so we use __builtin_bswap. */
 #define __arch_swab32(x) __builtin_bswap32(x)
 #define __arch_swab64(x) __builtin_bswap64(x)
-
-/* Use the variant that is natural for the wordsize. */
-#ifdef CONFIG_64BIT
-#define __arch_swab16(x) (__builtin_bswap64(x) >> 48)
-#else
 #define __arch_swab16(x) (__builtin_bswap32(x) >> 16)
-#endif
 
 #endif /* _ASM_TILE_SWAB_H */

arch/tile/include/asm/thread_info.h

@@ -125,6 +125,7 @@ extern void cpu_idle_on_new_stack(struct thread_info *old_ti,
 #define TIF_SYSCALL_AUDIT	5	/* syscall auditing active */
 #define TIF_SECCOMP		6	/* secure computing */
 #define TIF_MEMDIE		7	/* OOM killer at work */
+#define TIF_NOTIFY_RESUME	8	/* callback before returning to user */
 
 #define _TIF_SIGPENDING		(1<<TIF_SIGPENDING)
 #define _TIF_NEED_RESCHED	(1<<TIF_NEED_RESCHED)
@@ -134,10 +135,12 @@ extern void cpu_idle_on_new_stack(struct thread_info *old_ti,
 #define _TIF_SYSCALL_AUDIT	(1<<TIF_SYSCALL_AUDIT)
 #define _TIF_SECCOMP		(1<<TIF_SECCOMP)
 #define _TIF_MEMDIE		(1<<TIF_MEMDIE)
+#define _TIF_NOTIFY_RESUME	(1<<TIF_NOTIFY_RESUME)
 
 /* Work to do on any return to user space. */
 #define _TIF_ALLWORK_MASK \
-  (_TIF_SIGPENDING|_TIF_NEED_RESCHED|_TIF_SINGLESTEP|_TIF_ASYNC_TLB)
+  (_TIF_SIGPENDING|_TIF_NEED_RESCHED|_TIF_SINGLESTEP|\
+   _TIF_ASYNC_TLB|_TIF_NOTIFY_RESUME)
 
 /*
  * Thread-synchronous status.

arch/tile/include/asm/traps.h

@@ -15,10 +15,14 @@
 #ifndef _ASM_TILE_TRAPS_H
 #define _ASM_TILE_TRAPS_H
 
+#include <arch/chip.h>
+
 /* mm/fault.c */
 void do_page_fault(struct pt_regs *, int fault_num,
 		   unsigned long address, unsigned long write);
+#if CHIP_HAS_TILE_DMA() || CHIP_HAS_SN_PROC()
 void do_async_page_fault(struct pt_regs *);
+#endif
 
 #ifndef __tilegx__
 /*

arch/tile/include/asm/unistd.h

@@ -15,7 +15,7 @@
 #if !defined(_ASM_TILE_UNISTD_H) || defined(__SYSCALL)
 #define _ASM_TILE_UNISTD_H
 
-#ifndef __LP64__
+#if !defined(__LP64__) || defined(__SYSCALL_COMPAT)
 /* Use the flavor of this syscall that matches the 32-bit API better. */
 #define __ARCH_WANT_SYNC_FILE_RANGE2
 #endif

arch/tile/include/hv/hypervisor.h

@@ -22,8 +22,6 @@
 
 #include <arch/chip.h>
 
-#include <hv/pagesize.h>
-
 /* Linux builds want unsigned long constants, but assembler wants numbers */
 #ifdef __ASSEMBLER__
 /** One, for assembler */
@@ -44,11 +42,21 @@
  */
 #define HV_L1_SPAN (__HV_SIZE_ONE << HV_LOG2_L1_SPAN)
 
+/** The log2 of the size of small pages, in bytes. This value should
+ * be verified at runtime by calling hv_sysconf(HV_SYSCONF_PAGE_SIZE_SMALL).
+ */
+#define HV_LOG2_PAGE_SIZE_SMALL 16
+
 /** The size of small pages, in bytes. This value should be verified
  * at runtime by calling hv_sysconf(HV_SYSCONF_PAGE_SIZE_SMALL).
  */
 #define HV_PAGE_SIZE_SMALL (__HV_SIZE_ONE << HV_LOG2_PAGE_SIZE_SMALL)
 
+/** The log2 of the size of large pages, in bytes. This value should be
+ * verified at runtime by calling hv_sysconf(HV_SYSCONF_PAGE_SIZE_LARGE).
+ */
+#define HV_LOG2_PAGE_SIZE_LARGE 24
+
 /** The size of large pages, in bytes. This value should be verified
  * at runtime by calling hv_sysconf(HV_SYSCONF_PAGE_SIZE_LARGE).
  */

arch/tile/kernel/traps.c

@@ -308,6 +308,7 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num,
 	info.si_addr = (void __user *)address;
 	if (signo == SIGILL)
 		info.si_trapno = fault_num;
+	trace_unhandled_signal("trap", regs, address, signo);
 	force_sig_info(signo, &info, current);
 }