Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1304232 > unrolled thread

[PATCH v1] mm: soft-offline: exit with failure for non anonymous thp

Started byNaoya Horiguchi <n-horiguchi@ah.jp.nec.com>
First post2016-01-08 08:30 +0100
Last post2016-01-13 01:00 +0100
Articles 6 — 2 participants

Back to article view | Back to linux.kernel

Contents

  [PATCH v1] mm: soft-offline: exit with failure for non anonymous thp Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> - 2016-01-08 08:30 +0100
    Re: [PATCH v1] mm: soft-offline: exit with failure for non  anonymous thp Andrew Morton <akpm@linux-foundation.org> - 2016-01-08 21:40 +0100
      Re: [PATCH v1] mm: soft-offline: exit with failure for non anonymous thp Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> - 2016-01-12 04:20 +0100
        [PATCH v2 1/2] mm: soft-offline: clean up soft_offline_page() Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> - 2016-01-12 04:20 +0100
        [PATCH v2 2/2] mm: soft-offline: exit with failure for non anonymous thp Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> - 2016-01-12 04:20 +0100
          Re: [PATCH v2 2/2] mm: soft-offline: exit with failure for non  anonymous thp Andrew Morton <akpm@linux-foundation.org> - 2016-01-13 01:00 +0100

#1304232 — [PATCH v1] mm: soft-offline: exit with failure for non anonymous thp

FromNaoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date2016-01-08 08:30 +0100
Subject[PATCH v1] mm: soft-offline: exit with failure for non anonymous thp
Message-ID<qOzEK-3rg-19@gated-at.bofh.it>
Currently memory_failure() doesn't handle non anonymous thp case, because we
can hardly expect the error handling to be successful, and it can just hit
some corner case which results in BUG_ON or something severe like that.
This is also a case for soft offline code, so let's make it in the same way.

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
 mm/memory-failure.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diff --git v4.4-rc8/mm/memory-failure.c v4.4-rc8_patched/mm/memory-failure.c
index 750b789..30e9085 100644
--- v4.4-rc8/mm/memory-failure.c
+++ v4.4-rc8_patched/mm/memory-failure.c
@@ -1751,9 +1751,11 @@ int soft_offline_page(struct page *page, int flags)
 		return -EBUSY;
 	}
 	if (!PageHuge(page) && PageTransHuge(hpage)) {
-		if (PageAnon(hpage) && unlikely(split_huge_page(hpage))) {
-			pr_info("soft offline: %#lx: failed to split THP\n",
-				pfn);
+		if (!PageAnon(hpage) || unlikely(split_huge_page(hpage))) {
+			if (!PageAnon(hpage))
+				pr_info("soft offline: %#lx: non anonymous thp\n", pfn);
+			else
+				pr_info("soft offline: %#lx: thp split failed\n", pfn);
 			if (flags & MF_COUNT_INCREASED)
 				put_hwpoison_page(page);
 			return -EBUSY;
-- 
1.7.1

[toc] | [next] | [standalone]

#1304998 — Re: [PATCH v1] mm: soft-offline: exit with failure for non anonymous thp

FromAndrew Morton <akpm@linux-foundation.org>
Date2016-01-08 21:40 +0100
SubjectRe: [PATCH v1] mm: soft-offline: exit with failure for non anonymous thp
Message-ID<qOLZg-3xd-9@gated-at.bofh.it>
In reply to#1304232
On Fri,  8 Jan 2016 16:24:02 +0900 Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> wrote:

> Currently memory_failure() doesn't handle non anonymous thp case, because we
> can hardly expect the error handling to be successful, and it can just hit
> some corner case which results in BUG_ON or something severe like that.
> This is also a case for soft offline code, so let's make it in the same way.
> 
> ...
>
> --- v4.4-rc8/mm/memory-failure.c
> +++ v4.4-rc8_patched/mm/memory-failure.c
> @@ -1751,9 +1751,11 @@ int soft_offline_page(struct page *page, int flags)
>  		return -EBUSY;
>  	}
>  	if (!PageHuge(page) && PageTransHuge(hpage)) {
> -		if (PageAnon(hpage) && unlikely(split_huge_page(hpage))) {
> -			pr_info("soft offline: %#lx: failed to split THP\n",
> -				pfn);
> +		if (!PageAnon(hpage) || unlikely(split_huge_page(hpage))) {
> +			if (!PageAnon(hpage))
> +				pr_info("soft offline: %#lx: non anonymous thp\n", pfn);
> +			else
> +				pr_info("soft offline: %#lx: thp split failed\n", pfn);
>  			if (flags & MF_COUNT_INCREASED)
>  				put_hwpoison_page(page);
>  			return -EBUSY;

Kirill's
http://ozlabs.org/~akpm/mmots/broken-out/thp-mm-split_huge_page-caller-need-to-lock-page.patch
mucks with this code as well.  Could you please redo this patch against
linux-next?

Thanks.

[toc] | [prev] | [next] | [standalone]

#1306957

FromNaoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date2016-01-12 04:20 +0100
Message-ID<qPXEZ-3xb-9@gated-at.bofh.it>
In reply to#1304998
On Fri, Jan 08, 2016 at 12:33:00PM -0800, Andrew Morton wrote:
> On Fri,  8 Jan 2016 16:24:02 +0900 Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> wrote:
>
> > Currently memory_failure() doesn't handle non anonymous thp case, because we
> > can hardly expect the error handling to be successful, and it can just hit
> > some corner case which results in BUG_ON or something severe like that.
> > This is also a case for soft offline code, so let's make it in the same way.
> >
> > ...
> >
> > --- v4.4-rc8/mm/memory-failure.c
> > +++ v4.4-rc8_patched/mm/memory-failure.c
> > @@ -1751,9 +1751,11 @@ int soft_offline_page(struct page *page, int flags)
> >         return -EBUSY;
> >     }
> >     if (!PageHuge(page) && PageTransHuge(hpage)) {
> > -       if (PageAnon(hpage) && unlikely(split_huge_page(hpage))) {
> > -           pr_info("soft offline: %#lx: failed to split THP\n",
> > -               pfn);
> > +       if (!PageAnon(hpage) || unlikely(split_huge_page(hpage))) {
> > +           if (!PageAnon(hpage))
> > +               pr_info("soft offline: %#lx: non anonymous thp\n", pfn);
> > +           else
> > +               pr_info("soft offline: %#lx: thp split failed\n", pfn);
> >             if (flags & MF_COUNT_INCREASED)
> >                 put_hwpoison_page(page);
> >             return -EBUSY;
>
> Kirill's
> http://ozlabs.org/~akpm/mmots/broken-out/thp-mm-split_huge_page-caller-need-to-lock-page.patch
> mucks with this code as well.  Could you please redo this patch against
> linux-next?

OK, I did it, which will come after this email.

Thanks,
Naoya Horiguchi

[toc] | [prev] | [next] | [standalone]

#1306958 — [PATCH v2 1/2] mm: soft-offline: clean up soft_offline_page()

FromNaoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date2016-01-12 04:20 +0100
Subject[PATCH v2 1/2] mm: soft-offline: clean up soft_offline_page()
Message-ID<qPXEZ-3xb-11@gated-at.bofh.it>
In reply to#1306957
soft_offline_page() has some deeply indented code, that's the sign of demand
for cleanup. So let's do this. No functionality change.

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
 mm/memory-failure.c | 78 ++++++++++++++++++++++++++++++++---------------------
 1 file changed, 47 insertions(+), 31 deletions(-)

diff --git next-20160111/mm/memory-failure.c next-20160111_patched/mm/memory-failure.c
index 1b99403..2015c9a 100644
--- next-20160111/mm/memory-failure.c
+++ next-20160111_patched/mm/memory-failure.c
@@ -1684,6 +1684,49 @@ static int __soft_offline_page(struct page *page, int flags)
 	return ret;
 }
 
+static int soft_offline_in_use_page(struct page *page, int flags)
+{
+	int ret;
+	struct page *hpage = compound_head(page);
+
+	if (!PageHuge(page) && PageTransHuge(hpage)) {
+		lock_page(hpage);
+		ret = split_huge_page(hpage);
+		unlock_page(hpage);
+		if (unlikely(ret || PageTransCompound(page) ||
+			     !PageAnon(page))) {
+			pr_info("soft offline: %#lx: failed to split THP\n",
+				page_to_pfn(page));
+			if (flags & MF_COUNT_INCREASED)
+				put_hwpoison_page(hpage);
+			return -EBUSY;
+		}
+		get_hwpoison_page(page);
+		put_hwpoison_page(hpage);
+	}
+
+	if (PageHuge(page))
+		ret = soft_offline_huge_page(page, flags);
+	else
+		ret = __soft_offline_page(page, flags);
+
+	return ret;
+}
+
+static void soft_offline_free_page(struct page *page)
+{
+	if (PageHuge(page)) {
+		struct page *hpage = 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();
+	}
+}
+
 /**
  * soft_offline_page - Soft offline a page.
  * @page: page to offline
@@ -1710,7 +1753,6 @@ int soft_offline_page(struct page *page, int flags)
 {
 	int ret;
 	unsigned long pfn = page_to_pfn(page);
-	struct page *hpage = compound_head(page);
 
 	if (PageHWPoison(page)) {
 		pr_info("soft offline: %#lx page already poisoned\n", pfn);
@@ -1723,36 +1765,10 @@ int soft_offline_page(struct page *page, int flags)
 	ret = get_any_page(page, pfn, flags);
 	put_online_mems();
 
-	if (ret > 0) { /* for in-use pages */
-		if (!PageHuge(page) && PageTransHuge(hpage)) {
-			lock_page(hpage);
-			ret = split_huge_page(hpage);
-			unlock_page(hpage);
-			if (unlikely(ret || PageTransCompound(page) ||
-					!PageAnon(page))) {
-				pr_info("soft offline: %#lx: failed to split THP\n",
-					pfn);
-				if (flags & MF_COUNT_INCREASED)
-					put_hwpoison_page(hpage);
-				return -EBUSY;
-			}
-			get_hwpoison_page(page);
-			put_hwpoison_page(hpage);
-		}
+	if (ret > 0)
+		ret = soft_offline_in_use_page(page, flags);
+	else if (ret == 0)
+		soft_offline_free_page(page);
 
-		if (PageHuge(page))
-			ret = soft_offline_huge_page(page, flags);
-		else
-			ret = __soft_offline_page(page, flags);
-	} else if (ret == 0) { /* for free pages */
-		if (PageHuge(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();
-		}
-	}
 	return ret;
 }
-- 
2.7.0

[toc] | [prev] | [next] | [standalone]

#1306959 — [PATCH v2 2/2] mm: soft-offline: exit with failure for non anonymous thp

FromNaoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date2016-01-12 04:20 +0100
Subject[PATCH v2 2/2] mm: soft-offline: exit with failure for non anonymous thp
Message-ID<qPXEZ-3xb-15@gated-at.bofh.it>
In reply to#1306957
Currently memory_failure() doesn't handle non anonymous thp case, because we
can hardly expect the error handling to be successful, and it can just hit
some corner case which results in BUG_ON or something severe like that.
This is also the case for soft offline code, so let's make it in the same way.

Orignal code has a MF_COUNT_INCREASED check before put_hwpoison_page(), but
it's unnecessary because get_any_page() is already called when running on
this code, which takes a refcount of the target page regardress of the flag.
So this patch also removes it.

Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
ChangeLog v1->v2:
- rebased to next-20160111
---
 mm/memory-failure.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git next-20160111/mm/memory-failure.c next-20160111_patched/mm/memory-failure.c
index 2015c9a..6a2f290 100644
--- next-20160111/mm/memory-failure.c
+++ next-20160111_patched/mm/memory-failure.c
@@ -1691,16 +1691,16 @@ static int soft_offline_in_use_page(struct page *page, int flags)
 
 	if (!PageHuge(page) && PageTransHuge(hpage)) {
 		lock_page(hpage);
-		ret = split_huge_page(hpage);
-		unlock_page(hpage);
-		if (unlikely(ret || PageTransCompound(page) ||
-			     !PageAnon(page))) {
-			pr_info("soft offline: %#lx: failed to split THP\n",
-				page_to_pfn(page));
-			if (flags & MF_COUNT_INCREASED)
-				put_hwpoison_page(hpage);
+		if (!PageAnon(hpage) || unlikely(split_huge_page(hpage))) {
+			unlock_page(hpage);
+			if (!PageAnon(hpage))
+				pr_info("soft offline: %#lx: non anonymous thp\n", pfn);
+			else
+				pr_info("soft offline: %#lx: thp split failed\n", pfn);
+			put_hwpoison_page(hpage);
 			return -EBUSY;
 		}
+		unlock_page(hpage);
 		get_hwpoison_page(page);
 		put_hwpoison_page(hpage);
 	}
-- 
2.7.0

[toc] | [prev] | [next] | [standalone]

#1307926 — Re: [PATCH v2 2/2] mm: soft-offline: exit with failure for non anonymous thp

FromAndrew Morton <akpm@linux-foundation.org>
Date2016-01-13 01:00 +0100
SubjectRe: [PATCH v2 2/2] mm: soft-offline: exit with failure for non anonymous thp
Message-ID<qQh10-88E-13@gated-at.bofh.it>
In reply to#1306959
On Tue, 12 Jan 2016 12:10:45 +0900 Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> wrote:

> Currently memory_failure() doesn't handle non anonymous thp case, because we
> can hardly expect the error handling to be successful, and it can just hit
> some corner case which results in BUG_ON or something severe like that.
> This is also the case for soft offline code, so let's make it in the same way.
> 
> Orignal code has a MF_COUNT_INCREASED check before put_hwpoison_page(), but
> it's unnecessary because get_any_page() is already called when running on
> this code, which takes a refcount of the target page regardress of the flag.
> So this patch also removes it.
> 
> ...
>
> --- next-20160111/mm/memory-failure.c
> +++ next-20160111_patched/mm/memory-failure.c
> @@ -1691,16 +1691,16 @@ static int soft_offline_in_use_page(struct page *page, int flags)
>  
>  	if (!PageHuge(page) && PageTransHuge(hpage)) {
>  		lock_page(hpage);
> -		ret = split_huge_page(hpage);
> -		unlock_page(hpage);
> -		if (unlikely(ret || PageTransCompound(page) ||
> -			     !PageAnon(page))) {
> -			pr_info("soft offline: %#lx: failed to split THP\n",
> -				page_to_pfn(page));
> -			if (flags & MF_COUNT_INCREASED)
> -				put_hwpoison_page(hpage);
> +		if (!PageAnon(hpage) || unlikely(split_huge_page(hpage))) {
> +			unlock_page(hpage);
> +			if (!PageAnon(hpage))
> +				pr_info("soft offline: %#lx: non anonymous thp\n", pfn);
> +			else
> +				pr_info("soft offline: %#lx: thp split failed\n", pfn);
> +			put_hwpoison_page(hpage);
>  			return -EBUSY;
>  		}

hm, what happened there.

mm/memory-failure.c: In function 'soft_offline_in_use_page':
mm/memory-failure.c:1697: error: 'pfn' undeclared (first use in this function)
mm/memory-failure.c:1697: error: (Each undeclared identifier is reported only once
mm/memory-failure.c:1697: error: for each function it appears in.)

--- a/mm/memory-failure.c~mm-soft-offline-exit-with-failure-for-non-anonymous-thp-fix
+++ a/mm/memory-failure.c
@@ -1694,9 +1694,9 @@ static int soft_offline_in_use_page(stru
 		if (!PageAnon(hpage) || unlikely(split_huge_page(hpage))) {
 			unlock_page(hpage);
 			if (!PageAnon(hpage))
-				pr_info("soft offline: %#lx: non anonymous thp\n", pfn);
+				pr_info("soft offline: %#lx: non anonymous thp\n", page_to_pfn(page));
 			else
-				pr_info("soft offline: %#lx: thp split failed\n", pfn);
+				pr_info("soft offline: %#lx: thp split failed\n", page_to_pfn(page));
 			put_hwpoison_page(hpage);
 			return -EBUSY;
 		}
_

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web