[PATCH 03/15] ring_buffer: Align buffer_page struct allocations only to fit the flags.

[David Sharp <dhsharp@google.com>]

buffer_page structs need to be aligned to 4 byte boundaries because the page
flags are stored in the two least-significant bits of the pointers in the page
list. Aligning to cache lines is sufficient, but doesn't seem to be necessary.
Reducing the alignement to only 4 bytes may improve cache efficiency.

Testing with Autotest's tracing_microbenchmark, there was no significant change
in overhead with this change.

Signed-off-by: David Sharp <dhsharp@google.com>
---
 kernel/trace/ring_buffer.c |   20 ++++++++++++--------
 1 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 8ef7cc4..957a8b8 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -644,8 +644,14 @@ EXPORT_SYMBOL_GPL(ring_buffer_normalize_time_stamp);
 #define RB_PAGE_HEAD		1UL
 #define RB_PAGE_UPDATE		2UL
 
-
 #define RB_FLAG_MASK		3UL
+#define RB_PAGE_ALIGNMENT	(RB_FLAG_MASK+1)
+
+/* Ensure alignment of struct buffer_page */
+static __attribute__((unused)) void check_buffer_page_alignment(void)
+{
+        BUILD_BUG_ON(__alignof__(struct buffer_page) % RB_PAGE_ALIGNMENT != 0);
+}
 
 /* PAGE_MOVED is not part of the mask */
 #define RB_PAGE_MOVED		4UL
@@ -1004,8 +1010,8 @@ static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
 	WARN_ON(!nr_pages);
 
 	for (i = 0; i < nr_pages; i++) {
-		bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()),
-				    GFP_KERNEL, cpu_to_node(cpu_buffer->cpu));
+		bpage = kzalloc_node(sizeof(*bpage), GFP_KERNEL,
+				     cpu_to_node(cpu_buffer->cpu));
 		if (!bpage)
 			goto free_pages;
 
@@ -1059,8 +1065,7 @@ rb_allocate_cpu_buffer(struct ring_buffer *buffer, int cpu)
 	lockdep_set_class(&cpu_buffer->reader_lock, buffer->reader_lock_key);
 	cpu_buffer->lock = (arch_spinlock_t)__ARCH_SPIN_LOCK_UNLOCKED;
 
-	bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()),
-			    GFP_KERNEL, cpu_to_node(cpu));
+	bpage = kzalloc_node(sizeof(*bpage), GFP_KERNEL, cpu_to_node(cpu));
 	if (!bpage)
 		goto fail_free_buffer;
 
@@ -1375,9 +1380,8 @@ int ring_buffer_resize(struct ring_buffer *buffer, unsigned long size)
 
 	for_each_buffer_cpu(buffer, cpu) {
 		for (i = 0; i < new_pages; i++) {
-			bpage = kzalloc_node(ALIGN(sizeof(*bpage),
-						  cache_line_size()),
-					    GFP_KERNEL, cpu_to_node(cpu));
+			bpage = kzalloc_node(sizeof(*bpage), GFP_KERNEL,
+					     cpu_to_node(cpu));
 			if (!bpage)
 				goto free_pages;
 			list_add(&bpage->list, &pages);
-- 
1.7.3.1