Re: [PATCH 2/2] mm/page_ref: add tracepoint to track down page reference manipulation

On Mon, Feb 15, 2016 at 11:07:41AM -0500, Steven Rostedt wrote:

On Mon, 15 Feb 2016 12:04:50 +0900
js1304-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org wrote:

quoted

From: Joonsoo Kim <redacted>

CMA allocation should be guaranteed to succeed by definition, but,
unfortunately, it would be failed sometimes. It is hard to track down
the problem, because it is related to page reference manipulation and
we don't have any facility to analyze it.

This patch adds tracepoints to track down page reference manipulation.
With it, we can find exact reason of failure and can fix the problem.
Following is an example of tracepoint output.

<...>-9018  [004]    92.678375: page_ref_set:         pfn=0x17ac9 flags=0x0 count=1 mapcount=0 mapping=(nil) mt=4 val=1
<...>-9018  [004]    92.678378: kernel_stack:
 => get_page_from_freelist (ffffffff81176659)
 => __alloc_pages_nodemask (ffffffff81176d22)
 => alloc_pages_vma (ffffffff811bf675)
 => handle_mm_fault (ffffffff8119e693)
 => __do_page_fault (ffffffff810631ea)
 => trace_do_page_fault (ffffffff81063543)
 => do_async_page_fault (ffffffff8105c40a)
 => async_page_fault (ffffffff817581d8)  
[snip]
<...>-9018  [004]    92.678379: page_ref_mod:         pfn=0x17ac9 flags=0x40048 count=2 mapcount=1 mapping=0xffff880015a78dc1 mt=4 val=1
[snip]
...
...
<...>-9131  [001]    93.174468: test_pages_isolated:  start_pfn=0x17800 end_pfn=0x17c00 fin_pfn=0x17ac9 ret=fail
[snip]
<...>-9018  [004]    93.174843: page_ref_mod_and_test: pfn=0x17ac9 flags=0x40068 count=0 mapcount=0 mapping=0xffff880015a78dc1 mt=4 val=-1 ret=1
 => release_pages (ffffffff8117c9e4)
 => free_pages_and_swap_cache (ffffffff811b0697)
 => tlb_flush_mmu_free (ffffffff81199616)
 => tlb_finish_mmu (ffffffff8119a62c)
 => exit_mmap (ffffffff811a53f7)
 => mmput (ffffffff81073f47)
 => do_exit (ffffffff810794e9)
 => do_group_exit (ffffffff81079def)
 => SyS_exit_group (ffffffff81079e74)
 => entry_SYSCALL_64_fastpath (ffffffff817560b6)  

This output shows that problem comes from exit path. In exit path,
to improve performance, pages are not freed immediately. They are gathered
and processed by batch. During this process, migration cannot be possible
and CMA allocation is failed. This problem is hard to find without this
page reference tracepoint facility.

Enabling this feature bloat kernel text 30 KB in my configuration.

   text    data     bss     dec     hex filename
12127327        2243616 1507328 15878271         f2487f vmlinux_disabled
12157208        2258880 1507328 15923416         f2f8d8 vmlinux_enabled

v2:
o Use static key of each tracepoints to avoid function call overhead
when tracepoints are disabled.
o Print human-readable page flag through show_page_flags()
o Add more description to Kconfig.debug.

Acked-by: Michal Nazarewicz <redacted>
Signed-off-by: Joonsoo Kim <redacted>
---
 include/linux/page_ref.h        |  90 +++++++++++++++++++++++++--
 include/trace/events/page_ref.h | 133 ++++++++++++++++++++++++++++++++++++++++
 mm/Kconfig.debug                |  13 ++++
 mm/Makefile                     |   1 +
 mm/debug_page_ref.c             |  53 ++++++++++++++++
 5 files changed, 285 insertions(+), 5 deletions(-)
 create mode 100644 include/trace/events/page_ref.h
 create mode 100644 mm/debug_page_ref.c

diff --git a/include/linux/page_ref.h b/include/linux/page_ref.h
index 534249c..fd6d9a5 100644
--- a/include/linux/page_ref.h
+++ b/include/linux/page_ref.h

@@ -1,6 +1,54 @@
 #include <linux/atomic.h>
 #include <linux/mm_types.h>
 #include <linux/page-flags.h>
+#include <linux/tracepoint-defs.h>
+
+extern struct tracepoint __tracepoint_page_ref_set;
+extern struct tracepoint __tracepoint_page_ref_mod;
+extern struct tracepoint __tracepoint_page_ref_mod_and_test;
+extern struct tracepoint __tracepoint_page_ref_mod_and_return;
+extern struct tracepoint __tracepoint_page_ref_mod_unless;
+extern struct tracepoint __tracepoint_page_ref_freeze;
+extern struct tracepoint __tracepoint_page_ref_unfreeze;
+
+#ifdef CONFIG_DEBUG_PAGE_REF
+#define page_ref_tracepoint_active(t) static_key_false(&(t).key)

Please don't open code this. Use the following instead:

  trace_page_ref_set_enabled()
  trace_page_ref_mod_enabled()
  trace_page_ref_mod_and_test_enabled()
  trace_page_ref_mod_and_return_enabled()
  trace_page_ref_mod_unless_enabled()
  trace_page_ref_freeze_enabled()
  trace_page_ref_unfreeze_enabled()

They return true when CONFIG_TRACEPOINTS is configured in and the
tracepoint is enabled, and false otherwise.

This implementation is what you proposed before. Please refer below
link and source.

https://lkml.org/lkml/2015/12/9/699
arch/x86/include/asm/msr.h

There is header file dependency problem between mm.h and tracepoint.h.
page_ref.h should be included in mm.h and tracepoint.h cannot
be included in this case.

Thanks.