[tip:x86/mm] x86/mm: Reduce PAE-mode per task pgd allocation overhead from 4K to 32 bytes

[tip-bot for Fenghua Yu <tipbot@zytor.com>]

Commit-ID:  1db491f77b6ed0f32f1d4a3ac40a5be9524f1914
Gitweb:     http://git.kernel.org/tip/1db491f77b6ed0f32f1d4a3ac40a5be9524f1914
Author:     Fenghua Yu <fenghua.yu@intel.com>
AuthorDate: Thu, 15 Jan 2015 20:30:01 -0800
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Thu, 19 Feb 2015 01:28:38 +0100

x86/mm: Reduce PAE-mode per task pgd allocation overhead from 4K to 32 bytes

With more embedded systems emerging using Quark, among other
things, 32-bit kernel matters again. 32-bit machine and kernel
uses PAE paging, which currently wastes at least 4K of memory
per process on Linux where we have to reserve an entire page to
support a single 32-byte PGD structure. It would be a very good
thing if we could eliminate that wastage.

PAE paging is used to access more than 4GB memory on x86-32. And
it is required for NX.

In this patch, we still allocate one page for pgd for a Xen
domain and 64-bit kernel because one page pgd is assumed in
these cases. But we can save memory space by only allocating
32-byte pgd for 32-bit PAE kernel when it is not running as a
Xen domain.

Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Glenn Williamson <glenn.p.williamson@intel.com>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1421382601-46912-1-git-send-email-fenghua.yu@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/mm/pgtable.c | 81 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 78 insertions(+), 3 deletions(-)

diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index 6fb6927..d223e1f 100644
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -271,12 +271,87 @@ static void pgd_prepopulate_pmd(struct mm_struct *mm, pgd_t *pgd, pmd_t *pmds[])
 	}
 }
 
+/*
+ * Xen paravirt assumes pgd table should be in one page. 64 bit kernel also
+ * assumes that pgd should be in one page.
+ *
+ * But kernel with PAE paging that is not running as a Xen domain
+ * only needs to allocate 32 bytes for pgd instead of one page.
+ */
+#ifdef CONFIG_X86_PAE
+
+#include <linux/slab.h>
+
+#define PGD_SIZE	(PTRS_PER_PGD * sizeof(pgd_t))
+#define PGD_ALIGN	32
+
+static struct kmem_cache *pgd_cache;
+
+static int __init pgd_cache_init(void)
+{
+	/*
+	 * When PAE kernel is running as a Xen domain, it does not use
+	 * shared kernel pmd. And this requires a whole page for pgd.
+	 */
+	if (!SHARED_KERNEL_PMD)
+		return 0;
+
+	/*
+	 * when PAE kernel is not running as a Xen domain, it uses
+	 * shared kernel pmd. Shared kernel pmd does not require a whole
+	 * page for pgd. We are able to just allocate a 32-byte for pgd.
+	 * During boot time, we create a 32-byte slab for pgd table allocation.
+	 */
+	pgd_cache = kmem_cache_create("pgd_cache", PGD_SIZE, PGD_ALIGN,
+				      SLAB_PANIC, NULL);
+	if (!pgd_cache)
+		return -ENOMEM;
+
+	return 0;
+}
+core_initcall(pgd_cache_init);
+
+static inline pgd_t *_pgd_alloc(void)
+{
+	/*
+	 * If no SHARED_KERNEL_PMD, PAE kernel is running as a Xen domain.
+	 * We allocate one page for pgd.
+	 */
+	if (!SHARED_KERNEL_PMD)
+		return (pgd_t *)__get_free_page(PGALLOC_GFP);
+
+	/*
+	 * Now PAE kernel is not running as a Xen domain. We can allocate
+	 * a 32-byte slab for pgd to save memory space.
+	 */
+	return kmem_cache_alloc(pgd_cache, PGALLOC_GFP);
+}
+
+static inline void _pgd_free(pgd_t *pgd)
+{
+	if (!SHARED_KERNEL_PMD)
+		free_page((unsigned long)pgd);
+	else
+		kmem_cache_free(pgd_cache, pgd);
+}
+#else
+static inline pgd_t *_pgd_alloc(void)
+{
+	return (pgd_t *)__get_free_page(PGALLOC_GFP);
+}
+
+static inline void _pgd_free(pgd_t *pgd)
+{
+	free_page((unsigned long)pgd);
+}
+#endif /* CONFIG_X86_PAE */
+
 pgd_t *pgd_alloc(struct mm_struct *mm)
 {
 	pgd_t *pgd;
 	pmd_t *pmds[PREALLOCATED_PMDS];
 
-	pgd = (pgd_t *)__get_free_page(PGALLOC_GFP);
+	pgd = _pgd_alloc();
 
 	if (pgd == NULL)
 		goto out;
@@ -306,7 +381,7 @@ pgd_t *pgd_alloc(struct mm_struct *mm)
 out_free_pmds:
 	free_pmds(pmds);
 out_free_pgd:
-	free_page((unsigned long)pgd);
+	_pgd_free(pgd);
 out:
 	return NULL;
 }
@@ -316,7 +391,7 @@ void pgd_free(struct mm_struct *mm, pgd_t *pgd)
 	pgd_mop_up_pmds(mm, pgd);
 	pgd_dtor(pgd);
 	paravirt_pgd_free(mm, pgd);
-	free_page((unsigned long)pgd);
+	_pgd_free(pgd);
 }
 
 /*