[PATCH v4 14/20] powerpc/mm: Move pte_fragment_alloc() to a common location

From: Christophe Leroy
Date: Tue Sep 18 2018 - 12:58:09 EST


In preparation of next patch which generalises the use of
pte_fragment_alloc() for all, this patch moves the related functions
in a place that is common to all subarches.

The 8xx will need that for supporting 16k pages, as in that mode
page tables still have a size of 4k.

Since pte_fragment with only once fragment is not different
from what is done in the general case, we can easily migrate all
subarchs to pte fragments.

For the time being, it is only code move. We enclose it inside

Signed-off-by: Christophe Leroy <christophe.leroy@xxxxxx>
---
arch/powerpc/mm/Makefile | 4 +-
arch/powerpc/mm/mmu_context.c | 1 -
arch/powerpc/mm/mmu_context_book3s64.c | 67 -------------
arch/powerpc/mm/pgtable-book3s64.c | 85 -----------------
arch/powerpc/mm/pgtable-frag.c | 167 +++++++++++++++++++++++++++++++++
5 files changed, 170 insertions(+), 154 deletions(-)
create mode 100644 arch/powerpc/mm/pgtable-frag.c

diff --git a/arch/powerpc/mm/Makefile b/arch/powerpc/mm/Makefile
index 3c844bdd16c4..bd43b3ee52cb 100644
--- a/arch/powerpc/mm/Makefile
+++ b/arch/powerpc/mm/Makefile
@@ -15,7 +15,9 @@ obj-$(CONFIG_PPC_MMU_NOHASH) += mmu_context_nohash.o tlb_nohash.o \
obj-$(CONFIG_PPC_BOOK3E) += tlb_low_$(BITS)e.o
hash64-$(CONFIG_PPC_NATIVE) := hash_native_64.o
obj-$(CONFIG_PPC_BOOK3E_64) += pgtable-book3e.o
-obj-$(CONFIG_PPC_BOOK3S_64) += pgtable-hash64.o hash_utils_64.o slb_low.o slb.o $(hash64-y) mmu_context_book3s64.o pgtable-book3s64.o
+obj-$(CONFIG_PPC_BOOK3S_64) += pgtable-hash64.o hash_utils_64.o slb_low.o slb.o \
+ $(hash64-y) mmu_context_book3s64.o pgtable-book3s64.o \
+ pgtable-frag.o
obj-$(CONFIG_PPC_RADIX_MMU) += pgtable-radix.o tlb-radix.o
obj-$(CONFIG_PPC_STD_MMU_32) += ppc_mmu_32.o hash_low_32.o mmu_context_hash32.o
obj-$(CONFIG_PPC_STD_MMU) += tlb_hash$(BITS).o
diff --git a/arch/powerpc/mm/mmu_context.c b/arch/powerpc/mm/mmu_context.c
index f84e14f23e50..b89e7dcc14cc 100644
--- a/arch/powerpc/mm/mmu_context.c
+++ b/arch/powerpc/mm/mmu_context.c
@@ -96,4 +96,3 @@ void switch_mm_irqs_off(struct mm_struct *prev, struct mm_struct *next,
*/
switch_mmu_context(prev, next, tsk);
}
-
diff --git a/arch/powerpc/mm/mmu_context_book3s64.c b/arch/powerpc/mm/mmu_context_book3s64.c
index dbd8f762140b..417b0cb67584 100644
--- a/arch/powerpc/mm/mmu_context_book3s64.c
+++ b/arch/powerpc/mm/mmu_context_book3s64.c
@@ -155,50 +155,6 @@ static void destroy_contexts(mm_context_t *ctx)
}
}

-static void pte_frag_destroy(void *pte_frag)
-{
- int count;
- struct page *page;
-
- page = virt_to_page(pte_frag);
- /* drop all the pending references */
- count = ((unsigned long)pte_frag & ~PAGE_MASK) >> PTE_FRAG_SIZE_SHIFT;
- /* We allow PTE_FRAG_NR fragments from a PTE page */
- if (atomic_sub_and_test(PTE_FRAG_NR - count, &page->pt_frag_refcount)) {
- pgtable_page_dtor(page);
- __free_page(page);
- }
-}
-
-static void pmd_frag_destroy(void *pmd_frag)
-{
- int count;
- struct page *page;
-
- page = virt_to_page(pmd_frag);
- /* drop all the pending references */
- count = ((unsigned long)pmd_frag & ~PAGE_MASK) >> PMD_FRAG_SIZE_SHIFT;
- /* We allow PTE_FRAG_NR fragments from a PTE page */
- if (atomic_sub_and_test(PMD_FRAG_NR - count, &page->pt_frag_refcount)) {
- pgtable_pmd_page_dtor(page);
- __free_page(page);
- }
-}
-
-static void destroy_pagetable_cache(struct mm_struct *mm)
-{
- void *frag;
-
- frag = mm->context.pte_frag;
- if (frag)
- pte_frag_destroy(frag);
-
- frag = mm->context.pmd_frag;
- if (frag)
- pmd_frag_destroy(frag);
- return;
-}
-
void destroy_context(struct mm_struct *mm)
{
#ifdef CONFIG_SPAPR_TCE_IOMMU
@@ -212,29 +168,6 @@ void destroy_context(struct mm_struct *mm)
mm->context.id = MMU_NO_CONTEXT;
}

-void arch_exit_mmap(struct mm_struct *mm)
-{
- destroy_pagetable_cache(mm);
-
- if (radix_enabled()) {
- /*
- * Radix doesn't have a valid bit in the process table
- * entries. However we know that at least P9 implementation
- * will avoid caching an entry with an invalid RTS field,
- * and 0 is invalid. So this will do.
- *
- * This runs before the "fullmm" tlb flush in exit_mmap,
- * which does a RIC=2 tlbie to clear the process table
- * entry. See the "fullmm" comments in tlb-radix.c.
- *
- * No barrier required here after the store because
- * this process will do the invalidate, which starts with
- * ptesync.
- */
- process_tb[mm->context.id].prtb0 = 0;
- }
-}
-
#ifdef CONFIG_PPC_RADIX_MMU
void radix__switch_mmu_context(struct mm_struct *prev, struct mm_struct *next)
{
diff --git a/arch/powerpc/mm/pgtable-book3s64.c b/arch/powerpc/mm/pgtable-book3s64.c
index 01d7c0f7c4f0..723cd324fa34 100644
--- a/arch/powerpc/mm/pgtable-book3s64.c
+++ b/arch/powerpc/mm/pgtable-book3s64.c
@@ -317,91 +317,6 @@ void pmd_fragment_free(unsigned long *pmd)
}
}

-static pte_t *get_pte_from_cache(struct mm_struct *mm)
-{
- void *pte_frag, *ret;
-
- spin_lock(&mm->page_table_lock);
- ret = mm->context.pte_frag;
- if (ret) {
- pte_frag = ret + PTE_FRAG_SIZE;
- /*
- * If we have taken up all the fragments mark PTE page NULL
- */
- if (((unsigned long)pte_frag & ~PAGE_MASK) == 0)
- pte_frag = NULL;
- mm->context.pte_frag = pte_frag;
- }
- spin_unlock(&mm->page_table_lock);
- return (pte_t *)ret;
-}
-
-static pte_t *__alloc_for_ptecache(struct mm_struct *mm, int kernel)
-{
- void *ret = NULL;
- struct page *page;
-
- if (!kernel) {
- page = alloc_page(PGALLOC_GFP | __GFP_ACCOUNT);
- if (!page)
- return NULL;
- if (!pgtable_page_ctor(page)) {
- __free_page(page);
- return NULL;
- }
- } else {
- page = alloc_page(PGALLOC_GFP);
- if (!page)
- return NULL;
- }
-
- atomic_set(&page->pt_frag_refcount, 1);
-
- ret = page_address(page);
- /*
- * if we support only one fragment just return the
- * allocated page.
- */
- if (PTE_FRAG_NR == 1)
- return ret;
- spin_lock(&mm->page_table_lock);
- /*
- * If we find pgtable_page set, we return
- * the allocated page with single fragement
- * count.
- */
- if (likely(!mm->context.pte_frag)) {
- atomic_set(&page->pt_frag_refcount, PTE_FRAG_NR);
- mm->context.pte_frag = ret + PTE_FRAG_SIZE;
- }
- spin_unlock(&mm->page_table_lock);
-
- return (pte_t *)ret;
-}
-
-pte_t *pte_fragment_alloc(struct mm_struct *mm, unsigned long vmaddr, int kernel)
-{
- pte_t *pte;
-
- pte = get_pte_from_cache(mm);
- if (pte)
- return pte;
-
- return __alloc_for_ptecache(mm, kernel);
-}
-
-void pte_fragment_free(unsigned long *table, int kernel)
-{
- struct page *page = virt_to_page(table);
-
- BUG_ON(atomic_read(&page->pt_frag_refcount) <= 0);
- if (atomic_dec_and_test(&page->pt_frag_refcount)) {
- if (!kernel)
- pgtable_page_dtor(page);
- __free_page(page);
- }
-}
-
static inline void pgtable_free(void *table, int index)
{
switch (index) {
diff --git a/arch/powerpc/mm/pgtable-frag.c b/arch/powerpc/mm/pgtable-frag.c
new file mode 100644
index 000000000000..bc924822dcd6
--- /dev/null
+++ b/arch/powerpc/mm/pgtable-frag.c
@@ -0,0 +1,167 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Handling Page Tables through page fragments
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/gfp.h>
+#include <linux/mm.h>
+#include <linux/percpu.h>
+#include <linux/hardirq.h>
+#include <linux/hugetlb.h>
+#include <asm/pgalloc.h>
+#include <asm/tlbflush.h>
+#include <asm/tlb.h>
+
+static void pte_frag_destroy(void *pte_frag)
+{
+ int count;
+ struct page *page;
+
+ page = virt_to_page(pte_frag);
+ /* drop all the pending references */
+ count = ((unsigned long)pte_frag & ~PAGE_MASK) >> PTE_FRAG_SIZE_SHIFT;
+ /* We allow PTE_FRAG_NR fragments from a PTE page */
+ if (atomic_sub_and_test(PTE_FRAG_NR - count, &page->pt_frag_refcount)) {
+ pgtable_page_dtor(page);
+ __free_page(page);
+ }
+}
+
+static void pmd_frag_destroy(void *pmd_frag)
+{
+ int count;
+ struct page *page;
+
+ page = virt_to_page(pmd_frag);
+ /* drop all the pending references */
+ count = ((unsigned long)pmd_frag & ~PAGE_MASK) >> PMD_FRAG_SIZE_SHIFT;
+ /* We allow PTE_FRAG_NR fragments from a PTE page */
+ if (atomic_sub_and_test(PMD_FRAG_NR - count, &page->pt_frag_refcount)) {
+ pgtable_pmd_page_dtor(page);
+ __free_page(page);
+ }
+}
+
+static void destroy_pagetable_cache(struct mm_struct *mm)
+{
+ void *frag;
+
+ frag = mm->context.pte_frag;
+ if (frag)
+ pte_frag_destroy(frag);
+
+ frag = mm->context.pmd_frag;
+ if (frag)
+ pmd_frag_destroy(frag);
+}
+
+void arch_exit_mmap(struct mm_struct *mm)
+{
+ destroy_pagetable_cache(mm);
+
+ if (radix_enabled()) {
+ /*
+ * Radix doesn't have a valid bit in the process table
+ * entries. However we know that at least P9 implementation
+ * will avoid caching an entry with an invalid RTS field,
+ * and 0 is invalid. So this will do.
+ *
+ * This runs before the "fullmm" tlb flush in exit_mmap,
+ * which does a RIC=2 tlbie to clear the process table
+ * entry. See the "fullmm" comments in tlb-radix.c.
+ *
+ * No barrier required here after the store because
+ * this process will do the invalidate, which starts with
+ * ptesync.
+ */
+ process_tb[mm->context.id].prtb0 = 0;
+ }
+}
+
+static pte_t *get_pte_from_cache(struct mm_struct *mm)
+{
+ void *pte_frag, *ret;
+
+ spin_lock(&mm->page_table_lock);
+ ret = mm->context.pte_frag;
+ if (ret) {
+ pte_frag = ret + PTE_FRAG_SIZE;
+ /*
+ * If we have taken up all the fragments mark PTE page NULL
+ */
+ if (((unsigned long)pte_frag & ~PAGE_MASK) == 0)
+ pte_frag = NULL;
+ mm->context.pte_frag = pte_frag;
+ }
+ spin_unlock(&mm->page_table_lock);
+ return (pte_t *)ret;
+}
+
+static pte_t *__alloc_for_ptecache(struct mm_struct *mm, int kernel)
+{
+ void *ret = NULL;
+ struct page *page;
+
+ if (!kernel) {
+ page = alloc_page(PGALLOC_GFP | __GFP_ACCOUNT);
+ if (!page)
+ return NULL;
+ if (!pgtable_page_ctor(page)) {
+ __free_page(page);
+ return NULL;
+ }
+ } else {
+ page = alloc_page(PGALLOC_GFP);
+ if (!page)
+ return NULL;
+ }
+
+ atomic_set(&page->pt_frag_refcount, 1);
+
+ ret = page_address(page);
+ /*
+ * if we support only one fragment just return the
+ * allocated page.
+ */
+ if (PTE_FRAG_NR == 1)
+ return ret;
+ spin_lock(&mm->page_table_lock);
+ /*
+ * If we find pgtable_page set, we return
+ * the allocated page with single fragement
+ * count.
+ */
+ if (likely(!mm->context.pte_frag)) {
+ atomic_set(&page->pt_frag_refcount, PTE_FRAG_NR);
+ mm->context.pte_frag = ret + PTE_FRAG_SIZE;
+ }
+ spin_unlock(&mm->page_table_lock);
+
+ return (pte_t *)ret;
+}
+
+pte_t *pte_fragment_alloc(struct mm_struct *mm, unsigned long vmaddr, int kernel)
+{
+ pte_t *pte;
+
+ pte = get_pte_from_cache(mm);
+ if (pte)
+ return pte;
+
+ return __alloc_for_ptecache(mm, kernel);
+}
+
+void pte_fragment_free(unsigned long *table, int kernel)
+{
+ struct page *page = virt_to_page(table);
+
+ BUG_ON(atomic_read(&page->pt_frag_refcount) <= 0);
+ if (atomic_dec_and_test(&page->pt_frag_refcount)) {
+ if (!kernel)
+ pgtable_page_dtor(page);
+ __free_page(page);
+ }
+}
--
2.13.3