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

2 files changed, 34 insertions, 15 deletions

diff --git a/arch/arm64/mm/cache.S b/arch/arm64/mm/cache.S
index c46f48b33c14..fda756875fa6 100644
--- a/arch/arm64/mm/cache.S
+++ b/arch/arm64/mm/cache.S
@@ -168,6 +168,14 @@ ENTRY(__flush_dcache_area)
 ENDPROC(__flush_dcache_area)
 
 /*
+ *	__inval_cache_range(start, end)
+ *	- start   - start address of region
+ *	- end     - end address of region
+ */
+ENTRY(__inval_cache_range)
+	/* FALLTHROUGH */
+
+/*
  *	__dma_inv_range(start, end)
  *	- start   - virtual start address of region
  *	- end     - virtual end address of region
@@ -175,14 +183,22 @@ ENDPROC(__flush_dcache_area)
 __dma_inv_range:
 	dcache_line_size x2, x3
 	sub	x3, x2, #1
-	bic	x0, x0, x3
+	tst	x1, x3				// end cache line aligned?
 	bic	x1, x1, x3
-1:	dc	ivac, x0			// invalidate D / U line
-	add	x0, x0, x2
+	b.eq	1f
+	dc	civac, x1			// clean & invalidate D / U line
+1:	tst	x0, x3				// start cache line aligned?
+	bic	x0, x0, x3
+	b.eq	2f
+	dc	civac, x0			// clean & invalidate D / U line
+	b	3f
+2:	dc	ivac, x0			// invalidate D / U line
+3:	add	x0, x0, x2
 	cmp	x0, x1
-	b.lo	1b
+	b.lo	2b
 	dsb	sy
 	ret
+ENDPROC(__inval_cache_range)
 ENDPROC(__dma_inv_range)
 
 /*
diff --git a/arch/arm64/mm/proc.S b/arch/arm64/mm/proc.S
index e085ee6ef4e2..9042aff5e9e3 100644
--- a/arch/arm64/mm/proc.S
+++ b/arch/arm64/mm/proc.S
@@ -28,14 +28,21 @@
 
 #include "proc-macros.S"
 
-#ifndef CONFIG_SMP
-/* PTWs cacheable, inner/outer WBWA not shareable */
-#define TCR_FLAGS	TCR_IRGN_WBWA | TCR_ORGN_WBWA
+#ifdef CONFIG_ARM64_64K_PAGES
+#define TCR_TG_FLAGS	TCR_TG0_64K | TCR_TG1_64K
+#else
+#define TCR_TG_FLAGS	TCR_TG0_4K | TCR_TG1_4K
+#endif
+
+#ifdef CONFIG_SMP
+#define TCR_SMP_FLAGS	TCR_SHARED
 #else
-/* PTWs cacheable, inner/outer WBWA shareable */
-#define TCR_FLAGS	TCR_IRGN_WBWA | TCR_ORGN_WBWA | TCR_SHARED
+#define TCR_SMP_FLAGS	0
 #endif
 
+/* PTWs cacheable, inner/outer WBWA */
+#define TCR_CACHE_FLAGS	TCR_IRGN_WBWA | TCR_ORGN_WBWA
+
 #define MAIR(attr, mt)	((attr) << ((mt) * 8))
 
 /*
@@ -209,18 +216,14 @@ ENTRY(__cpu_setup)
 	 * Set/prepare TCR and TTBR. We use 512GB (39-bit) address range for
 	 * both user and kernel.
 	 */
-	ldr	x10, =TCR_TxSZ(VA_BITS) | TCR_FLAGS | \
-		      TCR_ASID16 | TCR_TBI0 | (1 << 31)
+	ldr	x10, =TCR_TxSZ(VA_BITS) | TCR_CACHE_FLAGS | TCR_SMP_FLAGS | \
+			TCR_TG_FLAGS | TCR_ASID16 | TCR_TBI0
 	/*
 	 * Read the PARange bits from ID_AA64MMFR0_EL1 and set the IPS bits in
 	 * TCR_EL1.
 	 */
 	mrs	x9, ID_AA64MMFR0_EL1
 	bfi	x10, x9, #32, #3
-#ifdef CONFIG_ARM64_64K_PAGES
-	orr	x10, x10, TCR_TG0_64K
-	orr	x10, x10, TCR_TG1_64K
-#endif
 	msr	tcr_el1, x10
 	ret					// return to head.S
 ENDPROC(__cpu_setup)