Eric Lee / linux-smarc-t335x-v3.2

Download as
Browse Code ยป

Commit 4eaf3f64397c3db3c5785eee508270d62a9fabd9

Authored by Haicheng Li
Committed by Linus Torvalds
1 parent 1f522509c7
Exists in master and in 4 other branches v3.2_AM335xPSP_04.06.00.11, v3.2_NEWT335xPSP_04.06.00.11, v3.2_SBC_SMARTMEN, v3.2_SMARCT335xPSP_04.06.00.11

mem-hotplug: fix potential race while building zonelist for new populated zone 

Add global mutex zonelists_mutex to fix the possible race:

     CPU0                                  CPU1                    CPU2
(1) zone->present_pages += online_pages;
(2)                                       build_all_zonelists();
(3)                                                               alloc_page();
(4)                                                               free_page();
(5) build_all_zonelists();
(6)   __build_all_zonelists();
(7)     zone->pageset = alloc_percpu();

In step (3,4), zone->pageset still points to boot_pageset, so bad
things may happen if 2+ nodes are in this state. Even if only 1 node
is accessing the boot_pageset, (3) may still consume too much memory
to fail the memory allocations in step (7).

Besides, atomic operation ensures alloc_percpu() in step (7) will never fail
since there is a new fresh memory block added in step(6).

[haicheng.li@linux.intel.com: hold zonelists_mutex when build_all_zonelists]
Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Reviewed-by: Andi Kleen <andi.kleen@intel.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Showing 4 changed files with 22 additions and 10 deletions Side-by-side Diff

include/linux/mmzone.h
Diff comments   View file @ 4eaf3f6
... ... @@ -650,6 +650,7 @@
650 650  
651 651 #include <linux/memory_hotplug.h>
652 652  
  653 +extern struct mutex zonelists_mutex;
653 654 void get_zone_counts(unsigned long *active, unsigned long *inactive,
654 655 unsigned long *free);
655 656 void build_all_zonelists(void *data);
... ... @@ -357,8 +357,11 @@
357 357 return -ENOMEM;
358 358 }
359 359  
360   - if (pgdat->node_zonelists->_zonerefs->zone == NULL)
  360 + if (pgdat->node_zonelists->_zonerefs->zone == NULL) {
  361 + mutex_lock(&zonelists_mutex);
361 362 build_all_zonelists(NULL);
  363 + mutex_unlock(&zonelists_mutex);
  364 + }
362 365 #endif
363 366  
364 367 cpu_maps_update_begin();
... ... @@ -389,11 +389,6 @@
389 389 int nid;
390 390 int ret;
391 391 struct memory_notify arg;
392   - /*
393   - * mutex to protect zone->pageset when it's still shared
394   - * in onlined_pages()
395   - */
396   - static DEFINE_MUTEX(zone_pageset_mutex);
397 392  
398 393 arg.start_pfn = pfn;
399 394 arg.nr_pages = nr_pages;
400 395  
... ... @@ -420,14 +415,14 @@
420 415 * This means the page allocator ignores this zone.
421 416 * So, zonelist must be updated after online.
422 417 */
423   - mutex_lock(&zone_pageset_mutex);
  418 + mutex_lock(&zonelists_mutex);
424 419 if (!populated_zone(zone))
425 420 need_zonelists_rebuild = 1;
426 421  
427 422 ret = walk_system_ram_range(pfn, nr_pages, &onlined_pages,
428 423 online_pages_range);
429 424 if (ret) {
430   - mutex_unlock(&zone_pageset_mutex);
  425 + mutex_unlock(&zonelists_mutex);
431 426 printk(KERN_DEBUG "online_pages %lx at %lx failed\n",
432 427 nr_pages, pfn);
433 428 memory_notify(MEM_CANCEL_ONLINE, &arg);
... ... @@ -441,7 +436,7 @@
441 436 else
442 437 zone_pcp_update(zone);
443 438  
444   - mutex_unlock(&zone_pageset_mutex);
  439 + mutex_unlock(&zonelists_mutex);
445 440 setup_per_zone_wmarks();
446 441 calculate_zone_inactive_ratio(zone);
447 442 if (onlined_pages) {
... ... @@ -2571,8 +2571,11 @@
2571 2571 strncpy((char*)table->data, saved_string,
2572 2572 NUMA_ZONELIST_ORDER_LEN);
2573 2573 user_zonelist_order = oldval;
2574   - } else if (oldval != user_zonelist_order)
  2574 + } else if (oldval != user_zonelist_order) {
  2575 + mutex_lock(&zonelists_mutex);
2575 2576 build_all_zonelists(NULL);
  2577 + mutex_unlock(&zonelists_mutex);
  2578 + }
2576 2579 }
2577 2580 out:
2578 2581 mutex_unlock(&zl_order_mutex);
... ... @@ -2924,6 +2927,12 @@
2924 2927 static DEFINE_PER_CPU(struct per_cpu_pageset, boot_pageset);
2925 2928 static void setup_zone_pageset(struct zone *zone);
2926 2929  
  2930 +/*
  2931 + * Global mutex to protect against size modification of zonelists
  2932 + * as well as to serialize pageset setup for the new populated zone.
  2933 + */
  2934 +DEFINE_MUTEX(zonelists_mutex);
  2935 +
2927 2936 /* return values int ....just for stop_machine() */
2928 2937 static __init_refok int __build_all_zonelists(void *data)
2929 2938 {
... ... @@ -2967,6 +2976,10 @@
2967 2976 return 0;
2968 2977 }
2969 2978  
  2979 +/*
  2980 + * Called with zonelists_mutex held always
  2981 + * unless system_state == SYSTEM_BOOTING.
  2982 + */
2970 2983 void build_all_zonelists(void *data)
2971 2984 {
2972 2985 set_zonelist_order();