Skip to content

Commit

Permalink
mm: hwpoison: change PageHWPoison behavior on hugetlb pages
Browse files Browse the repository at this point in the history 
[ Upstream commit b37ff71cc626a0c1b5e098ff9a0b723815f6aaeb ]

We'd like to narrow down the error region in memory error on hugetlb
pages.  However, currently we set PageHWPoison flags on all subpages in
the error hugepage and add # of subpages to num_hwpoison_pages, which
doesn't fit our purpose.

So this patch changes the behavior and we only set PageHWPoison on the
head page then increase num_hwpoison_pages only by 1.  This is a
preparation for narrow-down part which comes in later patches.

Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Naoya Horiguchi <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: "Aneesh Kumar K.V" <[email protected]>
Cc: Anshuman Khandual <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Signed-off-by: Sasha Levin <[email protected]>
  • Loading branch information
Naoya Horiguchi authored and Sasha Levin committed Jun 30, 2021
1 parent ffcf4e9 commit 3c5c213
Show file tree
Hide file tree
Showing 2 changed files with 24 additions and 72 deletions.
9 changes: 0 additions & 9 deletions include/linux/swapops.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -196,15 +196,6 @@ static inline void num_poisoned_pages_dec(void)
atomic_long_dec(&num_poisoned_pages);
}

static inline void num_poisoned_pages_add(long num)
{
atomic_long_add(num, &num_poisoned_pages);
}

static inline void num_poisoned_pages_sub(long num)
{
atomic_long_sub(num, &num_poisoned_pages);
}
#else

static inline swp_entry_t make_hwpoison_entry(struct page *page)
Expand Down
87 changes: 24 additions & 63 deletions mm/memory-failure.c
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1010,22 +1010,6 @@ static int hwpoison_user_mappings(struct page *p, unsigned long pfn,
return ret;
}

static void set_page_hwpoison_huge_page(struct page *hpage)
{
int i;
int nr_pages = 1 << compound_order(hpage);
for (i = 0; i < nr_pages; i++)
SetPageHWPoison(hpage + i);
}

static void clear_page_hwpoison_huge_page(struct page *hpage)
{
int i;
int nr_pages = 1 << compound_order(hpage);
for (i = 0; i < nr_pages; i++)
ClearPageHWPoison(hpage + i);
}

/**
* memory_failure - Handle memory failure of a page.
* @pfn: Page Number of the corrupted page
Expand All @@ -1051,7 +1035,6 @@ int memory_failure(unsigned long pfn, int trapno, int flags)
struct page *hpage;
struct page *orig_head;
int res;
unsigned int nr_pages;
unsigned long page_flags;

if (!sysctl_memory_failure_recovery)
Expand All @@ -1065,24 +1048,23 @@ int memory_failure(unsigned long pfn, int trapno, int flags)

p = pfn_to_page(pfn);
orig_head = hpage = compound_head(p);

/* tmporary check code, to be updated in later patches */
if (PageHuge(p)) {
if (TestSetPageHWPoison(hpage)) {
pr_err("Memory failure: %#lx: already hardware poisoned\n", pfn);
return 0;
}
goto tmp;
}
if (TestSetPageHWPoison(p)) {
pr_err("Memory failure: %#lx: already hardware poisoned\n",
pfn);
return 0;
}

/*
* Currently errors on hugetlbfs pages are measured in hugepage units,
* so nr_pages should be 1 << compound_order. OTOH when errors are on
* transparent hugepages, they are supposed to be split and error
* measurement is done in normal page units. So nr_pages should be one
* in this case.
*/
if (PageHuge(p))
nr_pages = 1 << compound_order(hpage);
else /* normal page or thp */
nr_pages = 1;
num_poisoned_pages_add(nr_pages);
tmp:
num_poisoned_pages_inc();

/*
* We need/can do nothing about count=0 pages.
Expand Down Expand Up @@ -1110,12 +1092,11 @@ int memory_failure(unsigned long pfn, int trapno, int flags)
if (PageHWPoison(hpage)) {
if ((hwpoison_filter(p) && TestClearPageHWPoison(p))
|| (p != hpage && TestSetPageHWPoison(hpage))) {
num_poisoned_pages_sub(nr_pages);
num_poisoned_pages_dec();
unlock_page(hpage);
return 0;
}
}
set_page_hwpoison_huge_page(hpage);
res = dequeue_hwpoisoned_huge_page(hpage);
action_result(pfn, MF_MSG_FREE_HUGE,
res ? MF_IGNORED : MF_DELAYED);
Expand All @@ -1138,7 +1119,7 @@ int memory_failure(unsigned long pfn, int trapno, int flags)
pr_err("Memory failure: %#lx: thp split failed\n",
pfn);
if (TestClearPageHWPoison(p))
num_poisoned_pages_sub(nr_pages);
num_poisoned_pages_dec();
put_hwpoison_page(p);
return -EBUSY;
}
Expand Down Expand Up @@ -1202,14 +1183,14 @@ int memory_failure(unsigned long pfn, int trapno, int flags)
*/
if (!PageHWPoison(p)) {
pr_err("Memory failure: %#lx: just unpoisoned\n", pfn);
num_poisoned_pages_sub(nr_pages);
num_poisoned_pages_dec();
unlock_page(hpage);
put_hwpoison_page(hpage);
return 0;
}
if (hwpoison_filter(p)) {
if (TestClearPageHWPoison(p))
num_poisoned_pages_sub(nr_pages);
num_poisoned_pages_dec();
unlock_page(hpage);
put_hwpoison_page(hpage);
return 0;
Expand All @@ -1228,14 +1209,6 @@ int memory_failure(unsigned long pfn, int trapno, int flags)
put_hwpoison_page(hpage);
return 0;
}
/*
* Set PG_hwpoison on all pages in an error hugepage,
* because containment is done in hugepage unit for now.
* Since we have done TestSetPageHWPoison() for the head page with
* page lock held, we can safely set PG_hwpoison bits on tail pages.
*/
if (PageHuge(p))
set_page_hwpoison_huge_page(hpage);

/*
* It's very difficult to mess with pages currently under IO
Expand Down Expand Up @@ -1407,7 +1380,6 @@ int unpoison_memory(unsigned long pfn)
struct page *page;
struct page *p;
int freeit = 0;
unsigned int nr_pages;
static DEFINE_RATELIMIT_STATE(unpoison_rs, DEFAULT_RATELIMIT_INTERVAL,
DEFAULT_RATELIMIT_BURST);

Expand Down Expand Up @@ -1452,8 +1424,6 @@ int unpoison_memory(unsigned long pfn)
return 0;
}

nr_pages = 1 << compound_order(page);

if (!get_hwpoison_page(p)) {
/*
* Since HWPoisoned hugepage should have non-zero refcount,
Expand Down Expand Up @@ -1483,10 +1453,8 @@ int unpoison_memory(unsigned long pfn)
if (TestClearPageHWPoison(page)) {
unpoison_pr_info("Unpoison: Software-unpoisoned page %#lx\n",
pfn, &unpoison_rs);
num_poisoned_pages_sub(nr_pages);
num_poisoned_pages_dec();
freeit = 1;
if (PageHuge(page))
clear_page_hwpoison_huge_page(page);
}
unlock_page(page);

Expand Down Expand Up @@ -1612,14 +1580,10 @@ static int soft_offline_huge_page(struct page *page, int flags)
ret = -EIO;
} else {
/* overcommit hugetlb page will be freed to buddy */
if (PageHuge(page)) {
set_page_hwpoison_huge_page(hpage);
SetPageHWPoison(page);
if (PageHuge(page))
dequeue_hwpoisoned_huge_page(hpage);
num_poisoned_pages_add(1 << compound_order(hpage));
} else {
SetPageHWPoison(page);
num_poisoned_pages_inc();
}
num_poisoned_pages_inc();
}
return ret;
}
Expand Down Expand Up @@ -1728,15 +1692,12 @@ static int soft_offline_in_use_page(struct page *page, int flags)

static void soft_offline_free_page(struct page *page)
{
if (PageHuge(page)) {
struct page *hpage = compound_head(page);
struct page *head = compound_head(page);

set_page_hwpoison_huge_page(hpage);
if (!dequeue_hwpoisoned_huge_page(hpage))
num_poisoned_pages_add(1 << compound_order(hpage));
} else {
if (!TestSetPageHWPoison(page))
num_poisoned_pages_inc();
if (!TestSetPageHWPoison(head)) {
num_poisoned_pages_inc();
if (PageHuge(head))
dequeue_hwpoisoned_huge_page(head);
}
}

Expand Down

0 comments on commit 3c5c213

Please sign in to comment.