History

base/memory, hotplug: fix a kernel oops in show_valid_zones()

Reading a sysfs "memoryN/valid_zones" file leads to the following oops when the first page of a range is not backed by struct page. show_valid_zones() assumes that 'start_pfn' is always valid for page_zone(). BUG: unable to handle kernel paging request at ffffea017a000000 IP: show_valid_zones+0x6f/0x160 This issue may happen on x86-64 systems with 64GiB or more memory since their memory block size is bumped up to 2GiB. [1] An example of such systems is desribed below. 0x3240000000 is only aligned by 1GiB and this memory block starts from 0x3200000000, which is not backed by struct page. BIOS-e820: [mem 0x0000003240000000-0x000000603fffffff] usable Since test_pages_in_a_zone() already checks holes, fix this issue by extending this function to return 'valid_start' and 'valid_end' for a given range. show_valid_zones() then proceeds with the valid range. [1] 'Commit bdee237c0343 ("x86: mm: Use 2GB memory block size on large-memory x86-64 systems")' Link: http://lkml.kernel.org/r/20170127222149.30893-3-toshi.kani@hpe.com Signed-off-by: Toshi Kani <toshi.kani@hpe.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Zhang Zhen <zhenzhang.zhang@huawei.com> Cc: Reza Arbab <arbab@linux.vnet.ibm.com> Cc: David Rientjes <rientjes@google.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: <stable@vger.kernel.org> [4.4+]

Toshi Kani committed 9 years ago

Commit a96dfdd1 parent deb88a23 files · +23 −12Browse files

Unified · Split

3 files
  1. drivers/base/memory.c +6 −6
  2. include/linux/memory_hotplug.h +2 −1
  3. mm/memory_hotplug.c +15 −5
drivers/base/memory.c+6−6
@@ -389,33 +389,33 @@ static ssize_t show_valid_zones(struct device *dev, · 388 unmodified lines
389{389{
390 struct memory_block *mem = to_memory_block(dev);390 struct memory_block *mem = to_memory_block(dev);
391 unsigned long start_pfn, end_pfn;391 unsigned long start_pfn, end_pfn;
392 unsigned long valid_start, valid_end, valid_pages;
392 unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;393 unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
393 struct page *first_page;
394 struct zone *zone;394 struct zone *zone;
395 int zone_shift = 0;395 int zone_shift = 0;
396 396
397 start_pfn = section_nr_to_pfn(mem->start_section_nr);397 start_pfn = section_nr_to_pfn(mem->start_section_nr);
398 end_pfn = start_pfn + nr_pages;398 end_pfn = start_pfn + nr_pages;
399 first_page = pfn_to_page(start_pfn);
400 399
401 /* The block contains more than one zone can not be offlined. */400 /* The block contains more than one zone can not be offlined. */
402 if (!test_pages_in_a_zone(start_pfn, end_pfn))401 if (!test_pages_in_a_zone(start_pfn, end_pfn, &valid_start, &valid_end))
403 return sprintf(buf, "none\n");402 return sprintf(buf, "none\n");
404 403
405 zone = page_zone(first_page);404 zone = page_zone(pfn_to_page(valid_start));
405 valid_pages = valid_end - valid_start;
406 406
407 /* MMOP_ONLINE_KEEP */407 /* MMOP_ONLINE_KEEP */
408 sprintf(buf, "%s", zone->name);408 sprintf(buf, "%s", zone->name);
409 409
410 /* MMOP_ONLINE_KERNEL */410 /* MMOP_ONLINE_KERNEL */
411 zone_can_shift(start_pfn, nr_pages, ZONE_NORMAL, &zone_shift);411 zone_can_shift(valid_start, valid_pages, ZONE_NORMAL, &zone_shift);
412 if (zone_shift) {412 if (zone_shift) {
413 strcat(buf, " ");413 strcat(buf, " ");
414 strcat(buf, (zone + zone_shift)->name);414 strcat(buf, (zone + zone_shift)->name);
415 }415 }
416 416
417 /* MMOP_ONLINE_MOVABLE */417 /* MMOP_ONLINE_MOVABLE */
418 zone_can_shift(start_pfn, nr_pages, ZONE_MOVABLE, &zone_shift);418 zone_can_shift(valid_start, valid_pages, ZONE_MOVABLE, &zone_shift);
419 if (zone_shift) {419 if (zone_shift) {
420 strcat(buf, " ");420 strcat(buf, " ");
421 strcat(buf, (zone + zone_shift)->name);421 strcat(buf, (zone + zone_shift)->name);
include/linux/memory_hotplug.h+2−1
@@ -85,7 +85,8 @@ extern int zone_grow_waitqueues(struct zone *zone, unsigned long nr_pages); · 84 unmodified lines
85extern int add_one_highpage(struct page *page, int pfn, int bad_ppro);85extern int add_one_highpage(struct page *page, int pfn, int bad_ppro);
86/* VM interface that may be used by firmware interface */86/* VM interface that may be used by firmware interface */
87extern int online_pages(unsigned long, unsigned long, int);87extern int online_pages(unsigned long, unsigned long, int);
88extern int test_pages_in_a_zone(unsigned long, unsigned long);88extern int test_pages_in_a_zone(unsigned long start_pfn, unsigned long end_pfn,
89 unsigned long *valid_start, unsigned long *valid_end);
89extern void __offline_isolated_pages(unsigned long, unsigned long);90extern void __offline_isolated_pages(unsigned long, unsigned long);
90 91
91typedef void (*online_page_callback_t)(struct page *page);92typedef void (*online_page_callback_t)(struct page *page);
mm/memory_hotplug.c+15−5
@@ -1484,10 +1484,13 @@ bool is_mem_section_removable(unsigned long start_pfn, unsigned long nr_pages) · 1483 unmodified lines
1484 1484
1485/*1485/*
1486 * Confirm all pages in a range [start, end) belong to the same zone.1486 * Confirm all pages in a range [start, end) belong to the same zone.
1487 * When true, return its valid [start, end).
1487 */1488 */
1488int test_pages_in_a_zone(unsigned long start_pfn, unsigned long end_pfn)1489int test_pages_in_a_zone(unsigned long start_pfn, unsigned long end_pfn,
1490 unsigned long *valid_start, unsigned long *valid_end)
1489{1491{
1490 unsigned long pfn, sec_end_pfn;1492 unsigned long pfn, sec_end_pfn;
1493 unsigned long start, end;
1491 struct zone *zone = NULL;1494 struct zone *zone = NULL;
1492 struct page *page;1495 struct page *page;
1493 int i;1496 int i;
@@ -1509,14 +1512,20 @@ int test_pages_in_a_zone(unsigned long start_pfn, unsigned long end_pfn) · 15 unmodified lines
1509 page = pfn_to_page(pfn + i);1512 page = pfn_to_page(pfn + i);
1510 if (zone && page_zone(page) != zone)1513 if (zone && page_zone(page) != zone)
1511 return 0;1514 return 0;
1515 if (!zone)
1516 start = pfn + i;
1512 zone = page_zone(page);1517 zone = page_zone(page);
1518 end = pfn + MAX_ORDER_NR_PAGES;
1513 }1519 }
1514 }1520 }
1515 1521
1516 if (zone)1522 if (zone) {
1523 *valid_start = start;
1524 *valid_end = end;
1517 return 1;1525 return 1;
1518 else1526 } else {
1519 return 0;1527 return 0;
1528 }
1520}1529}
1521 1530
1522/*1531/*
@@ -1843,6 +1852,7 @@ static int __ref __offline_pages(unsigned long start_pfn, · 320 unmodified lines
1843 long offlined_pages;1852 long offlined_pages;
1844 int ret, drain, retry_max, node;1853 int ret, drain, retry_max, node;
1845 unsigned long flags;1854 unsigned long flags;
1855 unsigned long valid_start, valid_end;
1846 struct zone *zone;1856 struct zone *zone;
1847 struct memory_notify arg;1857 struct memory_notify arg;
1848 1858
@@ -1853,10 +1863,10 @@ static int __ref __offline_pages(unsigned long start_pfn, · 4 unmodified lines
1853 return -EINVAL;1863 return -EINVAL;
1854 /* This makes hotplug much easier...and readable.1864 /* This makes hotplug much easier...and readable.
1855 we assume this for now. .*/1865 we assume this for now. .*/
1856 if (!test_pages_in_a_zone(start_pfn, end_pfn))1866 if (!test_pages_in_a_zone(start_pfn, end_pfn, &valid_start, &valid_end))
1857 return -EINVAL;1867 return -EINVAL;
1858 1868
1859 zone = page_zone(pfn_to_page(start_pfn));1869 zone = page_zone(pfn_to_page(valid_start));
1860 node = zone_to_nid(zone);1870 node = zone_to_nid(zone);
1861 nr_pages = end_pfn - start_pfn;1871 nr_pages = end_pfn - start_pfn;
1862 1872