[tip:x86/mm] x86/mm/cpa: Better use CLFLUSHOPT

[tip-bot for Peter Zijlstra]

Commit-ID:  c38116bb940ae37f51fccd315b420ee5961dcb76
Gitweb:     https://git.kernel.org/tip/c38116bb940ae37f51fccd315b420ee5961dcb76
Author:     Peter Zijlstra <peterz@xxxxxxxxxxxxx>
AuthorDate: Mon, 3 Dec 2018 18:03:52 +0100
Committer:  Ingo Molnar <mingo@xxxxxxxxxx>
CommitDate: Mon, 17 Dec 2018 18:54:29 +0100

x86/mm/cpa: Better use CLFLUSHOPT

Currently we issue an MFENCE before and after flushing a range. This
means that if we flush a bunch of single page ranges -- like with the
cpa array, we issue a whole bunch of superfluous MFENCEs.

Reorgainze the code a little to avoid this.

[ mingo: capitalize instructions, tweak changelog and comments. ]

Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
Cc: Andy Lutomirski <luto@xxxxxxxxxx>
Cc: Borislav Petkov <bp@xxxxxxxxx>
Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
Cc: H. Peter Anvin <hpa@xxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Rik van Riel <riel@xxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Tom.StDenis@xxxxxxx
Cc: dave.hansen@xxxxxxxxx
Link: http://lkml.kernel.org/r/20181203171043.626999883@xxxxxxxxxxxxx
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
---
 arch/x86/mm/pageattr.c | 29 +++++++++++++++++------------
 1 file changed, 17 insertions(+), 12 deletions(-)

diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c
index 85ef53b86fa0..7d05149995dc 100644
--- a/arch/x86/mm/pageattr.c
+++ b/arch/x86/mm/pageattr.c
@@ -251,15 +251,7 @@ static unsigned long __cpa_addr(struct cpa_data *cpa, unsigned long idx)
  * Flushing functions
  */
 
-/**
- * clflush_cache_range - flush a cache range with clflush
- * @vaddr:	virtual start address
- * @size:	number of bytes to flush
- *
- * clflushopt is an unordered instruction which needs fencing with mfence or
- * sfence to avoid ordering issues.
- */
-void clflush_cache_range(void *vaddr, unsigned int size)
+static void clflush_cache_range_opt(void *vaddr, unsigned int size)
 {
 	const unsigned long clflush_size = boot_cpu_data.x86_clflush_size;
 	void *p = (void *)((unsigned long)vaddr & ~(clflush_size - 1));
@@ -268,11 +260,22 @@ void clflush_cache_range(void *vaddr, unsigned int size)
 	if (p >= vend)
 		return;
 
-	mb();
-
 	for (; p < vend; p += clflush_size)
 		clflushopt(p);
+}
 
+/**
+ * clflush_cache_range - flush a cache range with clflush
+ * @vaddr:	virtual start address
+ * @size:	number of bytes to flush
+ *
+ * CLFLUSHOPT is an unordered instruction which needs fencing with MFENCE or
+ * SFENCE to avoid ordering issues.
+ */
+void clflush_cache_range(void *vaddr, unsigned int size)
+{
+	mb();
+	clflush_cache_range_opt(vaddr, size);
 	mb();
 }
 EXPORT_SYMBOL_GPL(clflush_cache_range);
@@ -333,6 +336,7 @@ static void cpa_flush(struct cpa_data *data, int cache)
 	if (!cache)
 		return;
 
+	mb();
 	for (i = 0; i < cpa->numpages; i++) {
 		unsigned long addr = __cpa_addr(cpa, i);
 		unsigned int level;
@@ -343,8 +347,9 @@ static void cpa_flush(struct cpa_data *data, int cache)
 		 * Only flush present addresses:
 		 */
 		if (pte && (pte_val(*pte) & _PAGE_PRESENT))
-			clflush_cache_range((void *)addr, PAGE_SIZE);
+			clflush_cache_range_opt((void *)addr, PAGE_SIZE);
 	}
+	mb();
 }
 
 static bool overlaps(unsigned long r1_start, unsigned long r1_end,