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, 19 insertions, 6 deletions

diff --git a/arch/arm64/lib/copy_to_user.S b/arch/arm64/lib/copy_to_user.S
index a0aeeb9b7a28..a257b47e2dc4 100644
--- a/arch/arm64/lib/copy_to_user.S
+++ b/arch/arm64/lib/copy_to_user.S
@@ -15,7 +15,11 @@
  */
 
 #include <linux/linkage.h>
+
+#include <asm/alternative.h>
 #include <asm/assembler.h>
+#include <asm/cpufeature.h>
+#include <asm/sysreg.h>
 
 /*
  * Copy to user space from a kernel buffer (alignment handled by the hardware)
@@ -28,14 +32,21 @@
  *	x0 - bytes not copied
  */
 ENTRY(__copy_to_user)
-	add	x4, x0, x2			// upper user buffer boundary
-	subs	x2, x2, #8
+ALTERNATIVE("nop", __stringify(SET_PSTATE_PAN(0)), ARM64_HAS_PAN, \
+	    CONFIG_ARM64_PAN)
+	add	x5, x0, x2			// upper user buffer boundary
+	subs	x2, x2, #16
+	b.mi	1f
+0:
+	ldp	x3, x4, [x1], #16
+	subs	x2, x2, #16
+USER(9f, stp	x3, x4, [x0], #16)
+	b.pl	0b
+1:	adds	x2, x2, #8
 	b.mi	2f
-1:
 	ldr	x3, [x1], #8
-	subs	x2, x2, #8
+	sub	x2, x2, #8
 USER(9f, str	x3, [x0], #8	)
-	b.pl	1b
 2:	adds	x2, x2, #4
 	b.mi	3f
 	ldr	w3, [x1], #4
@@ -51,11 +62,13 @@ USER(9f, strh	w3, [x0], #2	)
 	ldrb	w3, [x1]
 USER(9f, strb	w3, [x0]	)
 5:	mov	x0, #0
+ALTERNATIVE("nop", __stringify(SET_PSTATE_PAN(1)), ARM64_HAS_PAN, \
+	    CONFIG_ARM64_PAN)
 	ret
 ENDPROC(__copy_to_user)
 
 	.section .fixup,"ax"
 	.align	2
-9:	sub	x0, x4, x0			// bytes not copied
+9:	sub	x0, x5, x0			// bytes not copied
 	ret
 	.previous