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

Pull arm64 festive updates from Will Deacon:
 "In the end, we ended up with quite a lot more than I expected:

   - Support for ARMv8.3 Pointer Authentication in userspace (CRIU and
     kernel-side support to come later)

   - Support for per-thread stack canaries, pending an update to GCC
     that is currently undergoing review

   - Support for kexec_file_load(), which permits secure boot of a kexec
     payload but also happens to improve the performance of kexec
     dramatically because we can avoid the sucky purgatory code from
     userspace. Kdump will come later (requires updates to libfdt).

   - Optimisation of our dynamic CPU feature framework, so that all
     detected features are enabled via a single stop_machine()
     invocation

   - KPTI whitelisting of Cortex-A CPUs unaffected by Meltdown, so that
     they can benefit from global TLB entries when KASLR is not in use

   - 52-bit virtual addressing for userspace (kernel remains 48-bit)

   - Patch in LSE atomics for per-cpu atomic operations

   - Custom preempt.h implementation to avoid unconditional calls to
     preempt_schedule() from preempt_enable()

   - Support for the new 'SB' Speculation Barrier instruction

   - Vectorised implementation of XOR checksumming and CRC32
     optimisations

   - Workaround for Cortex-A76 erratum #1165522

   - Improved compatibility with Clang/LLD

   - Support for TX2 system PMUS for profiling the L3 cache and DMC

   - Reflect read-only permissions in the linear map by default

   - Ensure MMIO reads are ordered with subsequent calls to Xdelay()

   - Initial support for memory hotplug

   - Tweak the threshold when we invalidate the TLB by-ASID, so that
     mremap() performance is improved for ranges spanning multiple PMDs.

   - Minor refactoring and cleanups"

* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (125 commits)
  arm64: kaslr: print PHYS_OFFSET in dump_kernel_offset()
  arm64: sysreg: Use _BITUL() when defining register bits
  arm64: cpufeature: Rework ptr auth hwcaps using multi_entry_cap_matches
  arm64: cpufeature: Reduce number of pointer auth CPU caps from 6 to 4
  arm64: docs: document pointer authentication
  arm64: ptr auth: Move per-thread keys from thread_info to thread_struct
  arm64: enable pointer authentication
  arm64: add prctl control for resetting ptrauth keys
  arm64: perf: strip PAC when unwinding userspace
  arm64: expose user PAC bit positions via ptrace
  arm64: add basic pointer authentication support
  arm64/cpufeature: detect pointer authentication
  arm64: Don't trap host pointer auth use to EL2
  arm64/kvm: hide ptrauth from guests
  arm64/kvm: consistently handle host HCR_EL2 flags
  arm64: add pointer authentication register bits
  arm64: add comments about EC exception levels
  arm64: perf: Treat EXCLUDE_EL* bit definitions as unsigned
  arm64: kpti: Whitelist Cortex-A CPUs that don't implement the CSV3 field
  arm64: enable per-task stack canaries
  ...

1 files changed, 11 insertions, 4 deletions

diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
index 5dfd23897dea..3a1870228946 100644
--- a/arch/arm64/include/asm/tlbflush.h
+++ b/arch/arm64/include/asm/tlbflush.h
@@ -21,6 +21,7 @@
 
 #ifndef __ASSEMBLY__
 
+#include <linux/mm_types.h>
 #include <linux/sched.h>
 #include <asm/cputype.h>
 #include <asm/mmu.h>
@@ -164,14 +165,20 @@ static inline void flush_tlb_mm(struct mm_struct *mm)
 	dsb(ish);
 }
 
-static inline void flush_tlb_page(struct vm_area_struct *vma,
-				  unsigned long uaddr)
+static inline void flush_tlb_page_nosync(struct vm_area_struct *vma,
+					 unsigned long uaddr)
 {
 	unsigned long addr = __TLBI_VADDR(uaddr, ASID(vma->vm_mm));
 
 	dsb(ishst);
 	__tlbi(vale1is, addr);
 	__tlbi_user(vale1is, addr);
+}
+
+static inline void flush_tlb_page(struct vm_area_struct *vma,
+				  unsigned long uaddr)
+{
+	flush_tlb_page_nosync(vma, uaddr);
 	dsb(ish);
 }
 
@@ -179,7 +186,7 @@ static inline void flush_tlb_page(struct vm_area_struct *vma,
  * This is meant to avoid soft lock-ups on large TLB flushing ranges and not
  * necessarily a performance improvement.
  */
-#define MAX_TLBI_OPS	1024UL
+#define MAX_TLBI_OPS	PTRS_PER_PTE
 
 static inline void __flush_tlb_range(struct vm_area_struct *vma,
 				     unsigned long start, unsigned long end,
@@ -188,7 +195,7 @@ static inline void __flush_tlb_range(struct vm_area_struct *vma,
 	unsigned long asid = ASID(vma->vm_mm);
 	unsigned long addr;
 
-	if ((end - start) > (MAX_TLBI_OPS * stride)) {
+	if ((end - start) >= (MAX_TLBI_OPS * stride)) {
 		flush_tlb_mm(vma->vm_mm);
 		return;
 	}