perf: Optimize the !vmalloc backed buffer

Reduce code and data by using the knowledge that for
!PERF_USE_VMALLOC data_order is always 0.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
LKML-Reference: <20100521090710.795019386@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 7bd17f0488f8b79b4054180a089d49d284349157..09cd9c1abfda940b1cf477267e8d7db61eae9703 100644 (file)
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -588,8 +588,8 @@ struct perf_mmap_data {
        struct rcu_head                 rcu_head;
 #ifdef CONFIG_PERF_USE_VMALLOC
        struct work_struct              work;
+       int                             page_order;     /* allocation order  */
 #endif
-       int                             data_order;     /* allocation order  */
        int                             nr_pages;       /* nr of data pages  */
        int                             writable;       /* are we writable   */
        int                             nr_locked;      /* nr pages mlocked  */

diff --git a/kernel/perf_event.c b/kernel/perf_event.c
index b67549a086269bc8c5551524ffea7137cb2047ed..953ce46d7b2fd51eedc45ddd863b617d723c287a 100644 (file)
--- a/kernel/perf_event.c
+++ b/kernel/perf_event.c
@@ -2297,11 +2297,6 @@ unlock:
        rcu_read_unlock();
 }
 
-static unsigned long perf_data_size(struct perf_mmap_data *data)
-{
-       return data->nr_pages << (PAGE_SHIFT + data->data_order);
-}
-
 #ifndef CONFIG_PERF_USE_VMALLOC
 
 /*
@@ -2359,7 +2354,6 @@ perf_mmap_data_alloc(struct perf_event *event, int nr_pages)
                        goto fail_data_pages;
        }
 
-       data->data_order = 0;
        data->nr_pages = nr_pages;
 
        return data;
@@ -2395,6 +2389,11 @@ static void perf_mmap_data_free(struct perf_mmap_data *data)
        kfree(data);
 }
 
+static inline int page_order(struct perf_mmap_data *data)
+{
+       return 0;
+}
+
 #else
 
 /*
@@ -2403,10 +2402,15 @@ static void perf_mmap_data_free(struct perf_mmap_data *data)
  * Required for architectures that have d-cache aliasing issues.
  */
 
+static inline int page_order(struct perf_mmap_data *data)
+{
+       return data->page_order;
+}
+
 static struct page *
 perf_mmap_to_page(struct perf_mmap_data *data, unsigned long pgoff)
 {
-       if (pgoff > (1UL << data->data_order))
+       if (pgoff > (1UL << page_order(data)))
                return NULL;
 
        return vmalloc_to_page((void *)data->user_page + pgoff * PAGE_SIZE);
@@ -2426,7 +2430,7 @@ static void perf_mmap_data_free_work(struct work_struct *work)
        int i, nr;
 
        data = container_of(work, struct perf_mmap_data, work);
-       nr = 1 << data->data_order;
+       nr = 1 << page_order(data);
 
        base = data->user_page;
        for (i = 0; i < nr + 1; i++)
@@ -2465,7 +2469,7 @@ perf_mmap_data_alloc(struct perf_event *event, int nr_pages)
 
        data->user_page = all_buf;
        data->data_pages[0] = all_buf + PAGE_SIZE;
-       data->data_order = ilog2(nr_pages);
+       data->page_order = ilog2(nr_pages);
        data->nr_pages = 1;
 
        return data;
@@ -2479,6 +2483,11 @@ fail:
 
 #endif
 
+static unsigned long perf_data_size(struct perf_mmap_data *data)
+{
+       return data->nr_pages << (PAGE_SHIFT + page_order(data));
+}
+
 static int perf_mmap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 {
        struct perf_event *event = vma->vm_file->private_data;
@@ -2979,10 +2988,12 @@ void perf_output_copy(struct perf_output_handle *handle,
                handle->addr += size;
                handle->size -= size;
                if (!handle->size) {
+                       struct perf_mmap_data *data = handle->data;
+
                        handle->page++;
-                       handle->page &= handle->data->nr_pages - 1;
-                       handle->addr = handle->data->data_pages[handle->page];
-                       handle->size = PAGE_SIZE << handle->data->data_order;
+                       handle->page &= data->nr_pages - 1;
+                       handle->addr = data->data_pages[handle->page];
+                       handle->size = PAGE_SIZE << page_order(data);
                }
        } while (len);
 }
@@ -3050,12 +3061,12 @@ int perf_output_begin(struct perf_output_handle *handle,
        if (head - local_read(&data->wakeup) > data->watermark)
                local_add(data->watermark, &data->wakeup);
 
-       handle->page = handle->offset >> (PAGE_SHIFT + data->data_order);
+       handle->page = handle->offset >> (PAGE_SHIFT + page_order(data));
        handle->page &= data->nr_pages - 1;
-       handle->size = handle->offset & ((PAGE_SIZE << data->data_order) - 1);
+       handle->size = handle->offset & ((PAGE_SIZE << page_order(data)) - 1);
        handle->addr = data->data_pages[handle->page];
        handle->addr += handle->size;
-       handle->size = (PAGE_SIZE << data->data_order) - handle->size;
+       handle->size = (PAGE_SIZE << page_order(data)) - handle->size;
 
        if (have_lost) {
                lost_event.header.type = PERF_RECORD_LOST;