diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index 8d3d7a283eed..56961334bb7e 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -230,6 +230,9 @@ config NEED_RET_TO_USER
 config ARCH_MTD_XIP
 	bool
 
+config ARCH_WANT_KMAP_ATOMIC_FLUSH
+	bool
+
 config VECTORS_BASE
 	hex
 	default 0xffff0000 if MMU || CPU_HIGH_VECTOR
@@ -652,6 +655,7 @@ config ARCH_QCOM
 	select SPARSE_IRQ
 	select USE_OF
 	select PINCTRL
+	select ARCH_WANT_KMAP_ATOMIC_FLUSH
 	help
 	  Support for Qualcomm MSM/QSD based systems.  This runs on the
 	  apps processor of the MSM/QSD and depends on a shared memory
diff --git a/arch/arm/boot/dts/qcom/sdm630.dtsi b/arch/arm/boot/dts/qcom/sdm630.dtsi
index 38df4a63c5f5..ee8eaf48ad19 100644
--- a/arch/arm/boot/dts/qcom/sdm630.dtsi
+++ b/arch/arm/boot/dts/qcom/sdm630.dtsi
@@ -320,6 +320,17 @@
 			alignment = <0x0 0x400000>;
 			size = <0x0 0x5c00000>;
 		};
+
+		/* global autoconfigured region for contiguous allocations */
+		linux,cma {
+			compatible = "shared-dma-pool";
+			alloc-ranges = <0x0 0x00000000 0x0 0xffffffff>;
+			reusable;
+			alignment = <0x0 0x400000>;
+			size = <0x0 0x2000000>;
+			linux,cma-default;
+		};
+
 	};
 
 	bluetooth: bt_wcn3990 {
diff --git a/arch/arm/mm/highmem.c b/arch/arm/mm/highmem.c
index d02f8187b1cc..5d73327f8491 100644
--- a/arch/arm/mm/highmem.c
+++ b/arch/arm/mm/highmem.c
@@ -10,6 +10,7 @@
  * published by the Free Software Foundation.
  */
 
+#include <linux/cpu.h>
 #include <linux/module.h>
 #include <linux/highmem.h>
 #include <linux/interrupt.h>
@@ -147,3 +148,58 @@ void *kmap_atomic_pfn(unsigned long pfn)
 
 	return (void *)vaddr;
 }
+
+#ifdef CONFIG_ARCH_WANT_KMAP_ATOMIC_FLUSH
+static void kmap_remove_unused_cpu(int cpu)
+{
+	int start_idx, idx, type;
+
+	pagefault_disable();
+	type = kmap_atomic_idx();
+	start_idx = type + 1 + KM_TYPE_NR * cpu;
+
+	for (idx = start_idx; idx < KM_TYPE_NR + KM_TYPE_NR * cpu; idx++) {
+		unsigned long vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
+		pte_t ptep;
+
+		ptep = get_top_pte(vaddr);
+		if (ptep)
+			set_top_pte(vaddr, __pte(0));
+	}
+	pagefault_enable();
+}
+
+static void kmap_remove_unused(void *unused)
+{
+	kmap_remove_unused_cpu(smp_processor_id());
+}
+
+void kmap_atomic_flush_unused(void)
+{
+	on_each_cpu(kmap_remove_unused, NULL, 1);
+}
+
+static int hotplug_kmap_atomic_callback(struct notifier_block *nfb,
+					unsigned long action, void *hcpu)
+{
+	switch (action & (~CPU_TASKS_FROZEN)) {
+	case CPU_DYING:
+		kmap_remove_unused_cpu((int)hcpu);
+		break;
+	default:
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block hotplug_kmap_atomic_notifier = {
+	.notifier_call = hotplug_kmap_atomic_callback,
+};
+
+static int __init init_kmap_atomic(void)
+{
+	return register_hotcpu_notifier(&hotplug_kmap_atomic_notifier);
+}
+early_initcall(init_kmap_atomic);
+#endif
diff --git a/arch/arm/mm/init.c b/arch/arm/mm/init.c
index 107b5f1b864b..d3d718772381 100644
--- a/arch/arm/mm/init.c
+++ b/arch/arm/mm/init.c
@@ -625,6 +625,9 @@ struct section_perm {
 	pmdval_t mask;
 	pmdval_t prot;
 	pmdval_t clear;
+	pteval_t ptemask;
+	pteval_t pteprot;
+	pteval_t pteclear;
 };
 
 static struct section_perm nx_perms[] = {
@@ -634,6 +637,8 @@ static struct section_perm nx_perms[] = {
 		.end	= (unsigned long)_stext,
 		.mask	= ~PMD_SECT_XN,
 		.prot	= PMD_SECT_XN,
+		.ptemask = ~L_PTE_XN,
+		.pteprot = L_PTE_XN,
 	},
 	/* Make init RW (set NX). */
 	{
@@ -641,6 +646,8 @@ static struct section_perm nx_perms[] = {
 		.end	= (unsigned long)_sdata,
 		.mask	= ~PMD_SECT_XN,
 		.prot	= PMD_SECT_XN,
+		.ptemask = ~L_PTE_XN,
+		.pteprot = L_PTE_XN,
 	},
 #ifdef CONFIG_DEBUG_RODATA
 	/* Make rodata NX (set RO in ro_perms below). */
@@ -649,6 +656,8 @@ static struct section_perm nx_perms[] = {
 		.end    = (unsigned long)__init_begin,
 		.mask   = ~PMD_SECT_XN,
 		.prot   = PMD_SECT_XN,
+		.ptemask = ~L_PTE_XN,
+		.pteprot = L_PTE_XN,
 	},
 #endif
 };
@@ -667,6 +676,8 @@ static struct section_perm ro_perms[] = {
 		.prot   = PMD_SECT_APX | PMD_SECT_AP_WRITE,
 		.clear  = PMD_SECT_AP_WRITE,
 #endif
+		.ptemask = ~L_PTE_RDONLY,
+		.pteprot = L_PTE_RDONLY,
 	},
 };
 #endif
@@ -676,6 +687,35 @@ static struct section_perm ro_perms[] = {
  * copied into each mm). During startup, this is the init_mm. Is only
  * safe to be called with preemption disabled, as under stop_machine().
  */
+struct pte_data {
+	pteval_t mask;
+	pteval_t val;
+};
+
+static int __pte_update(pte_t *ptep, pgtable_t token, unsigned long addr,
+			void *d)
+{
+	struct pte_data *data = d;
+	pte_t pte = *ptep;
+
+	pte = __pte((pte_val(*ptep) & data->mask) | data->val);
+	set_pte_ext(ptep, pte, 0);
+
+	return 0;
+}
+
+static inline void pte_update(unsigned long addr, pteval_t mask,
+				  pteval_t prot, struct mm_struct *mm)
+{
+	struct pte_data data;
+
+	data.mask = mask;
+	data.val = prot;
+
+	apply_to_page_range(mm, addr, SECTION_SIZE, __pte_update, &data);
+	flush_tlb_kernel_range(addr, addr + SECTION_SIZE);
+}
+
 static inline void section_update(unsigned long addr, pmdval_t mask,
 				  pmdval_t prot, struct mm_struct *mm)
 {
@@ -724,11 +764,21 @@ void set_section_perms(struct section_perm *perms, int n, bool set,
 
 		for (addr = perms[i].start;
 		     addr < perms[i].end;
-		     addr += SECTION_SIZE)
-			section_update(addr, perms[i].mask,
-				set ? perms[i].prot : perms[i].clear, mm);
-	}
+		     addr += SECTION_SIZE) {
+			pmd_t *pmd;
 
+			pmd = pmd_offset(pud_offset(pgd_offset(mm, addr),
+						addr), addr);
+			if (pmd_bad(*pmd))
+				section_update(addr, perms[i].mask,
+					       set ? perms[i].prot : perms[i].clear,
+					       mm);
+			else
+				pte_update(addr, perms[i].ptemask,
+					       set ? perms[i].pteprot : perms[i].pteclear,
+					       mm);
+		}
+	}
 }
 
 static void update_sections_early(struct section_perm perms[], int n)