summaryrefslogtreecommitdiffstats
path: root/arch/hexagon/include/asm/pgalloc.h
diff options
context:
space:
mode:
Diffstat (limited to 'arch/hexagon/include/asm/pgalloc.h')
-rw-r--r--arch/hexagon/include/asm/pgalloc.h96
1 files changed, 96 insertions, 0 deletions
diff --git a/arch/hexagon/include/asm/pgalloc.h b/arch/hexagon/include/asm/pgalloc.h
new file mode 100644
index 000000000..f0c47e6a7
--- /dev/null
+++ b/arch/hexagon/include/asm/pgalloc.h
@@ -0,0 +1,96 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Page table support for the Hexagon architecture
+ *
+ * Copyright (c) 2010-2011, The Linux Foundation. All rights reserved.
+ */
+
+#ifndef _ASM_PGALLOC_H
+#define _ASM_PGALLOC_H
+
+#include <asm/mem-layout.h>
+#include <asm/atomic.h>
+
+#include <asm-generic/pgalloc.h>
+
+extern unsigned long long kmap_generation;
+
+/*
+ * Page table creation interface
+ */
+static inline pgd_t *pgd_alloc(struct mm_struct *mm)
+{
+ pgd_t *pgd;
+
+ pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
+
+ /*
+ * There may be better ways to do this, but to ensure
+ * that new address spaces always contain the kernel
+ * base mapping, and to ensure that the user area is
+ * initially marked invalid, initialize the new map
+ * map with a copy of the kernel's persistent map.
+ */
+
+ memcpy(pgd, swapper_pg_dir, PTRS_PER_PGD*sizeof(pgd_t));
+ mm->context.generation = kmap_generation;
+
+ /* Physical version is what is passed to virtual machine on switch */
+ mm->context.ptbase = __pa(pgd);
+
+ return pgd;
+}
+
+static inline void pmd_populate(struct mm_struct *mm, pmd_t *pmd,
+ pgtable_t pte)
+{
+ /*
+ * Conveniently, zero in 3 LSB means indirect 4K page table.
+ * Not so convenient when you're trying to vary the page size.
+ */
+ set_pmd(pmd, __pmd(((unsigned long)page_to_pfn(pte) << PAGE_SHIFT) |
+ HEXAGON_L1_PTE_SIZE));
+}
+
+/*
+ * Other architectures seem to have ways of making all processes
+ * share the same pmd's for their kernel mappings, but the v0.3
+ * Hexagon VM spec has a "monolithic" L1 table for user and kernel
+ * segments. We track "generations" of the kernel map to minimize
+ * overhead, and update the "slave" copies of the kernel mappings
+ * as part of switch_mm. However, we still need to update the
+ * kernel map of the active thread who's calling pmd_populate_kernel...
+ */
+static inline void pmd_populate_kernel(struct mm_struct *mm, pmd_t *pmd,
+ pte_t *pte)
+{
+ extern spinlock_t kmap_gen_lock;
+ pmd_t *ppmd;
+ int pmdindex;
+
+ spin_lock(&kmap_gen_lock);
+ kmap_generation++;
+ mm->context.generation = kmap_generation;
+ current->active_mm->context.generation = kmap_generation;
+ spin_unlock(&kmap_gen_lock);
+
+ set_pmd(pmd, __pmd(((unsigned long)__pa(pte)) | HEXAGON_L1_PTE_SIZE));
+
+ /*
+ * Now the "slave" copy of the current thread.
+ * This is pointer arithmetic, not byte addresses!
+ */
+ pmdindex = (pgd_t *)pmd - mm->pgd;
+ ppmd = (pmd_t *)current->active_mm->pgd + pmdindex;
+ set_pmd(ppmd, __pmd(((unsigned long)__pa(pte)) | HEXAGON_L1_PTE_SIZE));
+ if (pmdindex > max_kernel_seg)
+ max_kernel_seg = pmdindex;
+}
+
+#define __pte_free_tlb(tlb, pte, addr) \
+do { \
+ pgtable_pte_page_dtor((pte)); \
+ tlb_remove_page((tlb), (pte)); \
+} while (0)
+
+#endif