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

Pull second set of arm64 updates from Catalin Marinas:
 "A second pull request for this merging window, mainly with fixes and
  docs clarification:

   - Documentation clarification on CPU topology and booting
     requirements
   - Additional cache flushing during boot (needed in the presence of
     external caches or under virtualisation)
   - DMA range invalidation fix for non cache line aligned buffers
   - Build failure fix with !COMPAT
   - Kconfig update for STRICT_DEVMEM"

* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: Fix DMA range invalidation for cache line unaligned buffers
  arm64: Add missing Kconfig for CONFIG_STRICT_DEVMEM
  arm64: fix !CONFIG_COMPAT build failures
  Revert "arm64: virt: ensure visibility of __boot_cpu_mode"
  arm64: Relax the kernel cache requirements for boot
  arm64: Update the TCR_EL1 translation granule definitions for 16K pages
  ARM: topology: Make it clear that all CPUs need to be described

3 files changed, 34 insertions, 2 deletions

diff --git a/arch/arm64/kernel/head.S b/arch/arm64/kernel/head.S
index 1fe5d8d..0fd5650 100644
--- a/arch/arm64/kernel/head.S
+++ b/arch/arm64/kernel/head.S
@@ -26,6 +26,7 @@
 #include <asm/assembler.h>
 #include <asm/ptrace.h>
 #include <asm/asm-offsets.h>
+#include <asm/cache.h>
 #include <asm/cputype.h>
 #include <asm/memory.h>
 #include <asm/thread_info.h>
@@ -229,7 +230,11 @@ ENTRY(set_cpu_boot_mode_flag)
 	cmp	w20, #BOOT_CPU_MODE_EL2
 	b.ne	1f
 	add	x1, x1, #4
-1:	str	w20, [x1]			// This CPU has booted in EL1
+1:	dc	cvac, x1			// Clean potentially dirty cache line
+	dsb	sy
+	str	w20, [x1]			// This CPU has booted in EL1
+	dc	civac, x1			// Clean&invalidate potentially stale cache line
+	dsb	sy
 	ret
 ENDPROC(set_cpu_boot_mode_flag)
 
@@ -240,8 +245,9 @@ ENDPROC(set_cpu_boot_mode_flag)
  * This is not in .bss, because we set it sufficiently early that the boot-time
  * zeroing of .bss would clobber it.
  */
-	.pushsection	.data
+	.pushsection	.data..cacheline_aligned
 ENTRY(__boot_cpu_mode)
+	.align	L1_CACHE_SHIFT
 	.long	BOOT_CPU_MODE_EL2
 	.long	0
 	.popsection
@@ -408,6 +414,15 @@ ENDPROC(__calc_phys_offset)
  */
 __create_page_tables:
 	pgtbl	x25, x26, x24			// idmap_pg_dir and swapper_pg_dir addresses
+	mov	x27, lr
+
+	/*
+	 * Invalidate the idmap and swapper page tables to avoid potential
+	 * dirty cache lines being evicted.
+	 */
+	mov	x0, x25
+	add	x1, x26, #SWAPPER_DIR_SIZE
+	bl	__inval_cache_range
 
 	/*
 	 * Clear the idmap and swapper page tables.
@@ -467,6 +482,17 @@ __create_page_tables:
 	ldr	x5, =FIXADDR_TOP		// Fixed mapping virtual address
 	add	x0, x26, #2 * PAGE_SIZE		// section table address
 	create_pgd_entry x26, x0, x5, x6, x7
+
+	/*
+	 * Since the page tables have been populated with non-cacheable
+	 * accesses (MMU disabled), invalidate the idmap and swapper page
+	 * tables again to remove any speculatively loaded cache lines.
+	 */
+	mov	x0, x25
+	add	x1, x26, #SWAPPER_DIR_SIZE
+	bl	__inval_cache_range
+
+	mov	lr, x27
 	ret
 ENDPROC(__create_page_tables)
 	.ltorg
diff --git a/arch/arm64/kernel/perf_event.c b/arch/arm64/kernel/perf_event.c
index e868c72..baf5afb 100644
--- a/arch/arm64/kernel/perf_event.c
+++ b/arch/arm64/kernel/perf_event.c
@@ -1386,6 +1386,7 @@ user_backtrace(struct frame_tail __user *tail,
 	return buftail.fp;
 }
 
+#ifdef CONFIG_COMPAT
 /*
  * The registers we're interested in are at the end of the variable
  * length saved register structure. The fp points at the end of this
@@ -1430,6 +1431,7 @@ compat_user_backtrace(struct compat_frame_tail __user *tail,
 
 	return (struct compat_frame_tail __user *)compat_ptr(buftail.fp) - 1;
 }
+#endif /* CONFIG_COMPAT */
 
 void perf_callchain_user(struct perf_callchain_entry *entry,
 			 struct pt_regs *regs)
@@ -1451,6 +1453,7 @@ void perf_callchain_user(struct perf_callchain_entry *entry,
 		       tail && !((unsigned long)tail & 0xf))
 			tail = user_backtrace(tail, entry);
 	} else {
+#ifdef CONFIG_COMPAT
 		/* AARCH32 compat mode */
 		struct compat_frame_tail __user *tail;
 
@@ -1459,6 +1462,7 @@ void perf_callchain_user(struct perf_callchain_entry *entry,
 		while ((entry->nr < PERF_MAX_STACK_DEPTH) &&
 			tail && !((unsigned long)tail & 0x3))
 			tail = compat_user_backtrace(tail, entry);
+#endif
 	}
 }
 
diff --git a/arch/arm64/kernel/perf_regs.c b/arch/arm64/kernel/perf_regs.c
index f2d6f0a..422ebd6 100644
--- a/arch/arm64/kernel/perf_regs.c
+++ b/arch/arm64/kernel/perf_regs.c
@@ -2,6 +2,8 @@
 #include <linux/kernel.h>
 #include <linux/perf_event.h>
 #include <linux/bug.h>
+
+#include <asm/compat.h>
 #include <asm/perf_regs.h>
 #include <asm/ptrace.h>