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:
  arch/tile: don't validate CROSS_COMPILE needlessly
  arch/tile: export only COMMAND_LINE_SIZE to userspace.
  arch/tile: rename ARCH_KMALLOC_MINALIGN to ARCH_DMA_MINALIGN
  arch/tile: Rename the hweight() implementations to __arch_hweight()
  arch/tile: extend syscall ABI to set r1 on return as well.
  arch/tile: Various cleanups.
  arch/tile: support backtracing on TILE-Gx
  arch/tile: Fix a couple of issues with the COMPAT code for TILE-Gx.
  arch/tile: Use separate, better minsec values for clocksource and sched_clock.
  arch/tile: correct a bug in freeing bootmem by VA for the optional second initrd.
  arch: tile: mm: pgtable.c: Removed duplicated #include
  arch: tile: kernel/proc.c Removed duplicated #include
  Add fanotify syscalls to <asm-generic/unistd.h>.
  arch/tile: support new kunmap_atomic() naming convention.
  tile: remove unused ISA_DMA_THRESHOLD define

Conflicts in arch/tile/configs/tile_defconfig (pick the mainline version
with the reduced defconfig).

4 files changed, 18 insertions, 47 deletions

diff --git a/arch/tile/lib/Makefile b/arch/tile/lib/Makefile
index 438af38bc9e..746dc81ed3c 100644
--- a/arch/tile/lib/Makefile
+++ b/arch/tile/lib/Makefile
@@ -7,7 +7,9 @@ lib-y = cacheflush.o checksum.o cpumask.o delay.o \
 	memcpy_$(BITS).o memchr_$(BITS).o memmove_$(BITS).o memset_$(BITS).o \
 	strchr_$(BITS).o strlen_$(BITS).o
 
-ifneq ($(CONFIG_TILEGX),y)
+ifeq ($(CONFIG_TILEGX),y)
+lib-y += memcpy_user_64.o
+else
 lib-y += atomic_32.o atomic_asm_32.o memcpy_tile64.o
 endif
 
diff --git a/arch/tile/lib/exports.c b/arch/tile/lib/exports.c
index 6bc7b52b4aa..ce5dbf56578 100644
--- a/arch/tile/lib/exports.c
+++ b/arch/tile/lib/exports.c
@@ -36,21 +36,29 @@ EXPORT_SYMBOL(clear_user_asm);
 EXPORT_SYMBOL(current_text_addr);
 EXPORT_SYMBOL(dump_stack);
 
-/* arch/tile/lib/__memcpy.S */
-/* NOTE: on TILE64, these symbols appear in arch/tile/lib/memcpy_tile64.c */
+/* arch/tile/lib/, various memcpy files */
 EXPORT_SYMBOL(memcpy);
 EXPORT_SYMBOL(__copy_to_user_inatomic);
 EXPORT_SYMBOL(__copy_from_user_inatomic);
 EXPORT_SYMBOL(__copy_from_user_zeroing);
+#ifdef __tilegx__
+EXPORT_SYMBOL(__copy_in_user_inatomic);
+#endif
 
 /* hypervisor glue */
 #include <hv/hypervisor.h>
 EXPORT_SYMBOL(hv_dev_open);
 EXPORT_SYMBOL(hv_dev_pread);
 EXPORT_SYMBOL(hv_dev_pwrite);
+EXPORT_SYMBOL(hv_dev_preada);
+EXPORT_SYMBOL(hv_dev_pwritea);
+EXPORT_SYMBOL(hv_dev_poll);
+EXPORT_SYMBOL(hv_dev_poll_cancel);
 EXPORT_SYMBOL(hv_dev_close);
+EXPORT_SYMBOL(hv_sysconf);
+EXPORT_SYMBOL(hv_confstr);
 
-/* -ltile-cc */
+/* libgcc.a */
 uint32_t __udivsi3(uint32_t dividend, uint32_t divisor);
 EXPORT_SYMBOL(__udivsi3);
 int32_t __divsi3(int32_t dividend, int32_t divisor);
@@ -70,8 +78,6 @@ EXPORT_SYMBOL(__moddi3);
 #ifndef __tilegx__
 uint64_t __ll_mul(uint64_t n0, uint64_t n1);
 EXPORT_SYMBOL(__ll_mul);
-#endif
-#ifndef __tilegx__
 int64_t __muldi3(int64_t, int64_t);
 EXPORT_SYMBOL(__muldi3);
 uint64_t __lshrdi3(uint64_t, unsigned int);
diff --git a/arch/tile/lib/memcpy_32.S b/arch/tile/lib/memcpy_32.S
index f92984bf60e..30c3b7ebb55 100644
--- a/arch/tile/lib/memcpy_32.S
+++ b/arch/tile/lib/memcpy_32.S
@@ -17,10 +17,6 @@
 
 #include <arch/chip.h>
 
-#if CHIP_HAS_WH64() || defined(MEMCPY_TEST_WH64)
-#define MEMCPY_USE_WH64
-#endif
-
 
 #include <linux/linkage.h>
 
@@ -160,7 +156,7 @@ EX:	{ sw r0, r3; addi r0, r0, 4; addi r2, r2, -4 }
 
 	{ addi r3, r1, 60; andi r9, r9, -64 }
 
-#ifdef MEMCPY_USE_WH64
+#if CHIP_HAS_WH64()
         /* No need to prefetch dst, we'll just do the wh64
          * right before we copy a line.
 	 */
@@ -173,7 +169,7 @@ EX:	{ lw r6, r3; addi r3, r3, 64 }
         /* Intentionally stall for a few cycles to leave L2 cache alone. */
         { bnzt zero, . }
 EX:	{ lw r7, r3; addi r3, r3, 64 }
-#ifndef MEMCPY_USE_WH64
+#if !CHIP_HAS_WH64()
         /* Prefetch the dest */
         /* Intentionally stall for a few cycles to leave L2 cache alone. */
         { bnzt zero, . }
@@ -288,15 +284,7 @@ EX:	{ lw r7, r3; addi r3, r3, 64 }
         /* Fill second L1D line. */
 EX:	{ lw r17, r17; addi r1, r1, 48; mvz r3, r13, r1 } /* r17 = WORD_4 */
 
-#ifdef MEMCPY_TEST_WH64
-        /* Issue a fake wh64 that clobbers the destination words
-         * with random garbage, for testing.
-         */
-	{ movei r19, 64; crc32_32 r10, r2, r9 }
-.Lwh64_test_loop:
-EX:	{ sw r9, r10; addi r9, r9, 4; addi r19, r19, -4 }
-        { bnzt r19, .Lwh64_test_loop; crc32_32 r10, r10, r19 }
-#elif CHIP_HAS_WH64()
+#if CHIP_HAS_WH64()
         /* Prepare destination line for writing. */
 EX:	{ wh64 r9; addi r9, r9, 64 }
 #else
@@ -340,7 +328,7 @@ EX:	{ lw r18, r1; addi r1, r1, 4 }                  /* r18 = WORD_8 */
 EX:	{ sw r0, r16; addi r0, r0, 4; add r16, r0, r2 } /* store(WORD_0) */
 EX:	{ sw r0, r13; addi r0, r0, 4; andi r16, r16, -64 } /* store(WORD_1) */
 EX:	{ sw r0, r14; addi r0, r0, 4; slt_u r16, r9, r16 } /* store(WORD_2) */
-#ifdef MEMCPY_USE_WH64
+#if CHIP_HAS_WH64()
 EX:	{ sw r0, r15; addi r0, r0, 4; addi r13, sp, -64 } /* store(WORD_3) */
 #else
         /* Back up the r9 to a cache line we are already storing to
diff --git a/arch/tile/lib/memset_32.c b/arch/tile/lib/memset_32.c
index bfde5d864df..d014c1fbcbc 100644
--- a/arch/tile/lib/memset_32.c
+++ b/arch/tile/lib/memset_32.c
@@ -141,7 +141,6 @@ void *memset(void *s, int c, size_t n)
 			 */
 			__insn_prefetch(&out32[ahead32]);
 
-#if 1
 #if CACHE_LINE_SIZE_IN_WORDS % 4 != 0
 #error "Unhandled CACHE_LINE_SIZE_IN_WORDS"
 #endif
@@ -157,30 +156,6 @@ void *memset(void *s, int c, size_t n)
 				*out32++ = v32;
 				*out32++ = v32;
 			}
-#else
-			/* Unfortunately, due to a code generator flaw this
-			 * allocates a separate register for each of these
-			 * stores, which requires a large number of spills,
-			 * which makes this procedure enormously bigger
-			 * (something like 70%)
-			 */
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			*out32++ = v32;
-			n32 -= 16;
-#endif
 
 			/* To save compiled code size, reuse this loop even
 			 * when we run out of prefetching to do by dropping