Merge tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux

Pull arm64 updates from Will Deacon:

 - Support for new architectural features introduced in ARMv8.1:
   * Privileged Access Never (PAN) to catch user pointer dereferences in
     the kernel
   * Large System Extension (LSE) for building scalable atomics and locks
     (depends on locking/arch-atomic from tip, which is included here)
   * Hardware Dirty Bit Management (DBM) for updating clean PTEs
     automatically

 - Move our PSCI implementation out into drivers/firmware/, where it can
   be shared with arch/arm/. RMK has also pulled this component branch
   and has additional patches moving arch/arm/ over. MAINTAINERS is
   updated accordingly.

 - Better BUG implementation based on the BRK instruction for trapping

 - Leaf TLB invalidation for unmapping user pages

 - Support for PROBE_ONLY PCI configurations

 - Various cleanups and non-critical fixes, including:
   * Always flush FP/SIMD state over exec()
   * Restrict memblock additions based on range of linear mapping
   * Ensure *(LIST_POISON) generates a fatal fault
   * Context-tracking syscall return no longer corrupts return value when
     not forced on.
   * Alternatives patching synchronisation/stability improvements
   * Signed sub-word cmpxchg compare fix (tickled by HAVE_CMPXCHG_LOCAL)
   * Force SMP=y
   * Hide direct DCC access from userspace
   * Fix EFI stub memory allocation when DRAM starts at 0x0

* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (92 commits)
  arm64: flush FP/SIMD state correctly after execve()
  arm64: makefile: fix perf_callchain.o kconfig dependency
  arm64: set MAX_MEMBLOCK_ADDR according to linear region size
  of/fdt: make memblock maximum physical address arch configurable
  arm64: Fix source code file path in comments
  arm64: entry: always restore x0 from the stack on syscall return
  arm64: mdscr_el1: avoid exposing DCC to userspace
  arm64: kconfig: Move LIST_POISON to a safe value
  arm64: Add __exception_irq_entry definition for function graph
  arm64: mm: ensure patched kernel text is fetched from PoU
  arm64: alternatives: ensure secondary CPUs execute ISB after patching
  arm64: make ll/sc __cmpxchg_case_##name asm consistent
  arm64: dma-mapping: Simplify pgprot handling
  arm64: restore cpu suspend/resume functionality
  ARM64: PCI: do not enable resources on PROBE_ONLY systems
  arm64: cmpxchg: truncate sub-word signed types before comparison
  arm64: alternative: put secondary CPUs into polling loop during patch
  arm64/Documentation: clarify wording regarding memory below the Image
  arm64: lse: fix lse cmpxchg code indentation
  arm64: remove redundant object file list
  ...

1 files changed, 26 insertions, 19 deletions

diff --git a/arch/arm64/lib/bitops.S b/arch/arm64/lib/bitops.S
index 7dac371..43ac736 100644
--- a/arch/arm64/lib/bitops.S
+++ b/arch/arm64/lib/bitops.S
@@ -18,52 +18,59 @@
 
 #include <linux/linkage.h>
 #include <asm/assembler.h>
+#include <asm/lse.h>
 
 /*
  * x0: bits 5:0  bit offset
  *     bits 31:6 word offset
  * x1: address
  */
-	.macro	bitop, name, instr
+	.macro	bitop, name, llsc, lse
 ENTRY(	\name	)
 	and	w3, w0, #63		// Get bit offset
 	eor	w0, w0, w3		// Clear low bits
 	mov	x2, #1
 	add	x1, x1, x0, lsr #3	// Get word offset
+alt_lse "	prfm	pstl1strm, [x1]",	"nop"
 	lsl	x3, x2, x3		// Create mask
-1:	ldxr	x2, [x1]
-	\instr	x2, x2, x3
-	stxr	w0, x2, [x1]
-	cbnz	w0, 1b
+
+alt_lse	"1:	ldxr	x2, [x1]",		"\lse	x3, [x1]"
+alt_lse	"	\llsc	x2, x2, x3",		"nop"
+alt_lse	"	stxr	w0, x2, [x1]",		"nop"
+alt_lse	"	cbnz	w0, 1b",		"nop"
+
 	ret
 ENDPROC(\name	)
 	.endm
 
-	.macro	testop, name, instr
+	.macro	testop, name, llsc, lse
 ENTRY(	\name	)
 	and	w3, w0, #63		// Get bit offset
 	eor	w0, w0, w3		// Clear low bits
 	mov	x2, #1
 	add	x1, x1, x0, lsr #3	// Get word offset
+alt_lse "	prfm	pstl1strm, [x1]",	"nop"
 	lsl	x4, x2, x3		// Create mask
-1:	ldxr	x2, [x1]
-	lsr	x0, x2, x3		// Save old value of bit
-	\instr	x2, x2, x4		// toggle bit
-	stlxr	w5, x2, [x1]
-	cbnz	w5, 1b
-	dmb	ish
+
+alt_lse	"1:	ldxr	x2, [x1]",		"\lse	x4, x2, [x1]"
+	lsr	x0, x2, x3
+alt_lse	"	\llsc	x2, x2, x4",		"nop"
+alt_lse	"	stlxr	w5, x2, [x1]",		"nop"
+alt_lse	"	cbnz	w5, 1b",		"nop"
+alt_lse	"	dmb	ish",			"nop"
+
 	and	x0, x0, #1
-3:	ret
+	ret
 ENDPROC(\name	)
 	.endm
 
 /*
  * Atomic bit operations.
  */
-	bitop	change_bit, eor
-	bitop	clear_bit, bic
-	bitop	set_bit, orr
+	bitop	change_bit, eor, steor
+	bitop	clear_bit, bic, stclr
+	bitop	set_bit, orr, stset
 
-	testop	test_and_change_bit, eor
-	testop	test_and_clear_bit, bic
-	testop	test_and_set_bit, orr
+	testop	test_and_change_bit, eor, ldeoral
+	testop	test_and_clear_bit, bic, ldclral
+	testop	test_and_set_bit, orr, ldsetal