aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorMel Gorman <mel@csn.ul.ie>2009-09-21 20:02:41 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2009-09-22 10:17:34 -0400
commit4b4f278c030aa4b6ee0915f396e9a9478d92d610 (patch)
tree10825c2d197977bc9080093a6ecbd3ce80723876
parent38a398572fa2d8124f7479e40db581b5b72719c9 (diff)
tracing, page-allocator: add trace events for page allocation and page freeing
This patch adds trace events for the allocation and freeing of pages, including the freeing of pagevecs. Using the events, it will be known what struct page and pfns are being allocated and freed and what the call site was in many cases. The page alloc tracepoints be used as an indicator as to whether the workload was heavily dependant on the page allocator or not. You can make a guess based on vmstat but you can't get a per-process breakdown. Depending on the call path, the call_site for page allocation may be __get_free_pages() instead of a useful callsite. Instead of passing down a return address similar to slab debugging, the user should enable the stacktrace and seg-addr options to get a proper stack trace. The pagevec free tracepoint has a different usecase. It can be used to get a idea of how many pages are being dumped off the LRU and whether it is kswapd doing the work or a process doing direct reclaim. Signed-off-by: Mel Gorman <mel@csn.ul.ie> Acked-by: Rik van Riel <riel@redhat.com> Reviewed-by: Ingo Molnar <mingo@elte.hu> Cc: Larry Woodman <lwoodman@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Li Ming Chun <macli@brc.ubc.ca> Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-rw-r--r--include/trace/events/kmem.h74
-rw-r--r--mm/page_alloc.c7
2 files changed, 80 insertions, 1 deletions
diff --git a/include/trace/events/kmem.h b/include/trace/events/kmem.h
index 1493c541f9c4..0d358a0d45c1 100644
--- a/include/trace/events/kmem.h
+++ b/include/trace/events/kmem.h
@@ -225,6 +225,80 @@ TRACE_EVENT(kmem_cache_free,
225 225
226 TP_printk("call_site=%lx ptr=%p", __entry->call_site, __entry->ptr) 226 TP_printk("call_site=%lx ptr=%p", __entry->call_site, __entry->ptr)
227); 227);
228
229TRACE_EVENT(mm_page_free_direct,
230
231 TP_PROTO(struct page *page, unsigned int order),
232
233 TP_ARGS(page, order),
234
235 TP_STRUCT__entry(
236 __field( struct page *, page )
237 __field( unsigned int, order )
238 ),
239
240 TP_fast_assign(
241 __entry->page = page;
242 __entry->order = order;
243 ),
244
245 TP_printk("page=%p pfn=%lu order=%d",
246 __entry->page,
247 page_to_pfn(__entry->page),
248 __entry->order)
249);
250
251TRACE_EVENT(mm_pagevec_free,
252
253 TP_PROTO(struct page *page, int cold),
254
255 TP_ARGS(page, cold),
256
257 TP_STRUCT__entry(
258 __field( struct page *, page )
259 __field( int, cold )
260 ),
261
262 TP_fast_assign(
263 __entry->page = page;
264 __entry->cold = cold;
265 ),
266
267 TP_printk("page=%p pfn=%lu order=0 cold=%d",
268 __entry->page,
269 page_to_pfn(__entry->page),
270 __entry->cold)
271);
272
273TRACE_EVENT(mm_page_alloc,
274
275 TP_PROTO(struct page *page, unsigned int order,
276 gfp_t gfp_flags, int migratetype),
277
278 TP_ARGS(page, order, gfp_flags, migratetype),
279
280 TP_STRUCT__entry(
281 __field( struct page *, page )
282 __field( unsigned int, order )
283 __field( gfp_t, gfp_flags )
284 __field( int, migratetype )
285 ),
286
287 TP_fast_assign(
288 __entry->page = page;
289 __entry->order = order;
290 __entry->gfp_flags = gfp_flags;
291 __entry->migratetype = migratetype;
292 ),
293
294 TP_printk("page=%p pfn=%lu order=%d migratetype=%d gfp_flags=%s",
295 __entry->page,
296 page_to_pfn(__entry->page),
297 __entry->order,
298 __entry->migratetype,
299 show_gfp_flags(__entry->gfp_flags))
300);
301
228#endif /* _TRACE_KMEM_H */ 302#endif /* _TRACE_KMEM_H */
229 303
230/* This part must be outside protection */ 304/* This part must be outside protection */
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 913a8ebd3a8e..80f954d82d77 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1076,6 +1076,7 @@ static void free_hot_cold_page(struct page *page, int cold)
1076 1076
1077void free_hot_page(struct page *page) 1077void free_hot_page(struct page *page)
1078{ 1078{
1079 trace_mm_page_free_direct(page, 0);
1079 free_hot_cold_page(page, 0); 1080 free_hot_cold_page(page, 0);
1080} 1081}
1081 1082
@@ -1920,6 +1921,7 @@ __alloc_pages_nodemask(gfp_t gfp_mask, unsigned int order,
1920 zonelist, high_zoneidx, nodemask, 1921 zonelist, high_zoneidx, nodemask,
1921 preferred_zone, migratetype); 1922 preferred_zone, migratetype);
1922 1923
1924 trace_mm_page_alloc(page, order, gfp_mask, migratetype);
1923 return page; 1925 return page;
1924} 1926}
1925EXPORT_SYMBOL(__alloc_pages_nodemask); 1927EXPORT_SYMBOL(__alloc_pages_nodemask);
@@ -1954,13 +1956,16 @@ void __pagevec_free(struct pagevec *pvec)
1954{ 1956{
1955 int i = pagevec_count(pvec); 1957 int i = pagevec_count(pvec);
1956 1958
1957 while (--i >= 0) 1959 while (--i >= 0) {
1960 trace_mm_pagevec_free(pvec->pages[i], pvec->cold);
1958 free_hot_cold_page(pvec->pages[i], pvec->cold); 1961 free_hot_cold_page(pvec->pages[i], pvec->cold);
1962 }
1959} 1963}
1960 1964
1961void __free_pages(struct page *page, unsigned int order) 1965void __free_pages(struct page *page, unsigned int order)
1962{ 1966{
1963 if (put_page_testzero(page)) { 1967 if (put_page_testzero(page)) {
1968 trace_mm_page_free_direct(page, order);
1964 if (order == 0) 1969 if (order == 0)
1965 free_hot_page(page); 1970 free_hot_page(page);
1966 else 1971 else