Skip to contentSkip to conversation

mm/madvise: handle MADV_{HWPOISON,SOFT_OFFLINE} from madvise_unlock()

PATCH In review SeongJae Park 2 participants 4 messages Feb 11, 2025 – Feb 12, 2025

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mm/madvise: handle MADV_{HWPOISON,SOFT_OFFLINE} from madvise_unlock()

madvise_lock() does nothing for MADV_HWPOSION and MADV_SOFT_OFFLINE behavior, but madvise_unlock() does mmap_lock unlocking regardless of the behavior. Commit 948a0a9ea070 ("mm/madvise: split out mmap locking operations for madvise()") in mm-unstable, which introduced the wrong function didn't cause a real problem because do_madvise() was not calling madvise_unlock() for the behavior.

Later, commit f19c9d7b57cf ("mm/madvise: split out madvise() behavior execution") in mm-unstable made do_madvise() to call madvise_unlock() even for the two behaviors. As a result, the kernel tries to unlock unlocked mmap_lock.

Fix the issue by handling the two behaviors in madvise_unlock(). For the two behaviors, do nothing but just return. Also remove an unnecessary blank line in madvise_lock().

Technically speaking this patch fixes commit f19c9d7b57cf ("mm/madvise:
split out madvise() behavior execution"). But since the broken commit is not in the mainline yet, squashing this fix into commit 948a0a9ea070 ("mm/madvise: split out mmap locking operations for madvise()") would make more sense, so adding Fixes: tag with it.

Fixes 948a0a9ea070 ("mm/madvise: split out mmap locking operations for madvise()") # mm-unstable
---
 mm/madvise.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
mm/madvise.c +5 -1
index b5ef8e03d8b0..b8969457f3ef 100644
@@ -1577,7 +1577,6 @@ int madvise_set_anon_name(struct mm_struct *mm, unsigned long start,
 
 static int madvise_lock(struct mm_struct *mm, int behavior)
 {
-
 #ifdef CONFIG_MEMORY_FAILURE
 	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
 		return 0;
@@ -1595,6 +1594,11 @@ static int madvise_lock(struct mm_struct *mm, int behavior)
 
 static void madvise_unlock(struct mm_struct *mm, int behavior)
 {
+#ifdef CONFIG_MEMORY_FAILURE
+	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
+		return;
+#endif
+
 	if (madvise_need_mmap_write(behavior))
 		mmap_write_unlock(mm);
 	else

base-commit: 8bf30f9d23eb5040d37e6e712789cee8e71e1577

Signature · 1 line
2.39.5
#1 lore raw

+cc Naresh, Arnd for another reports/discussion of the same issue [0] while lore/lei is broken.

Hi,

Lore breaking means I missed this :) thankfully you cc'd me (_this_ is why I am so adament about people following get_maintainer.pl procedure btw) so I was able to now notice + reply :)

This is totally my bad for missing this on review, so mea culpa.

[0]:https://lwn.net/ml/linux-mm/CA+G9fYt5QwJ4_F8fJj7jx9_0Le9kOVSeG38ox9qnKqwsrDdvHQ@mail.gmail.com/

Quoted 43 lines · SeongJae Park

madvise_lock() does nothing for MADV_HWPOSION and MADV_SOFT_OFFLINE behavior, but madvise_unlock() does mmap_lock unlocking regardless of the behavior. Commit 948a0a9ea070 ("mm/madvise: split out mmap locking operations for madvise()") in mm-unstable, which introduced the wrong function didn't cause a real problem because do_madvise() was not calling madvise_unlock() for the behavior.

Later, commit f19c9d7b57cf ("mm/madvise: split out madvise() behavior execution") in mm-unstable made do_madvise() to call madvise_unlock() even for the two behaviors. As a result, the kernel tries to unlock unlocked mmap_lock.

Fix the issue by handling the two behaviors in madvise_unlock(). For the two behaviors, do nothing but just return. Also remove an unnecessary blank line in madvise_lock().

Technically speaking this patch fixes commit f19c9d7b57cf ("mm/madvise:
split out madvise() behavior execution"). But since the broken commit is not in the mainline yet, squashing this fix into commit 948a0a9ea070 ("mm/madvise: split out mmap locking operations for madvise()") would make more sense, so adding Fixes: tag with it.

Fixes 948a0a9ea070 ("mm/madvise: split out mmap locking operations for madvise()") # mm-unstable
---
 mm/madvise.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
mm/madvise.c +0 -1
index b5ef8e03d8b0..b8969457f3ef 100644
@@ -1577,7 +1577,6 @@ int madvise_set_anon_name(struct mm_struct *mm, unsigned long start,
 
 static int madvise_lock(struct mm_struct *mm, int behavior)
 {
-
 #ifdef CONFIG_MEMORY_FAILURE
 	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
 		return 0;

mm/madvise.c static int madvise_lock(struct mm_struct *mm, int behavior) · line 1600

Quoted 7 lines
mm/madvise.c +4 -0
@@ -1595,6 +1594,11 @@ static int madvise_lock(struct mm_struct *mm, int behavior)
 
 static void madvise_unlock(struct mm_struct *mm, int behavior)
 {
+#ifdef CONFIG_MEMORY_FAILURE
+	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
+		return;
+#endif

I agree this fixes the issue but this is horrible. let's abstract this please rather than doing the same crap that already existed, only now twice.

Quoted 8 lines
mm/madvise.c +1 -0
+
 	if (madvise_need_mmap_write(behavior))
 		mmap_write_unlock(mm);
 	else

base-commit: 8bf30f9d23eb5040d37e6e712789cee8e71e1577 --
2.39.5

I attach a fix-patch concept for something I think that'd be nicer, do with it what thy wilt! :P sorry I don't mean to be 'one of those' maintainers who copy/pastes code + demands somebody do it (by no means do I do so), but since this is so small I feel it's kind of quicker for me to do it this way.

Obviously take it or leave it/adapt it/etc. This is compile-tested only...

----8<----
From 9fce3e47bf0fff2a2291be66002af346cdbca665 Mon Sep 17 00:00:00 2001
From: Lorenzo Stoakes <[email protected]>
Date: Tue, 11 Feb 2025 10:44:26 +0000
Subject: [PATCH] mm/madvise: fix madvise_[un]lock() issue

We are asymmetric in our locking/unlocking in the case of memory failure madvise() behaviour options, correct this and abstract the memory failure check.

---
 mm/madvise.c | 29 +++++++++++++++++++++++------
 1 file changed, 23 insertions(+), 6 deletions(-)
mm/madvise.c +23 -6
index b5ef8e03d8b0..1a7af59c3aa9 100644
@@ -1575,14 +1575,29 @@ int madvise_set_anon_name(struct mm_struct *mm, unsigned long start,
 }
 #endif /* CONFIG_ANON_VMA_NAME */
 
-static int madvise_lock(struct mm_struct *mm, int behavior)
-{
-
 #ifdef CONFIG_MEMORY_FAILURE
-	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
-		return 0;
+static bool is_memory_failure(int behavior)
+{
+	switch (behavior) {
+	case MADV_HWPOISON:
+	case MADV_SOFT_OFFLINE:
+		return true;
+	default:
+		return false;
+	}
+}
+#else
+static bool is_memory_failure(int behavior)
+{
+	return false;
+}
 #endif
 
+static int madvise_lock(struct mm_struct *mm, int behavior)
+{
+	if (is_memory_failure(behavior))
+	    return 0;
+
 	if (madvise_need_mmap_write(behavior)) {
 		if (mmap_write_lock_killable(mm))
 			return -EINTR;
@@ -1590,11 +1605,13 @@ static int madvise_lock(struct mm_struct *mm, int behavior)
 		mmap_read_lock(mm);
 	}
 	return 0;
-
 }
 
 static void madvise_unlock(struct mm_struct *mm, int behavior)
 {
+	if (is_memory_failure(behavior))
+	    return;
+
 	if (madvise_need_mmap_write(behavior))
 		mmap_write_unlock(mm);
 	else

--
2.48.1

author Reviewed-by Tested-by
#2 lore raw
Quoted 10 lines · On Tue, 11 Feb 2025 10:51:41 +0000 Lorenzo Stoakes…

+cc Naresh, Arnd for another reports/discussion of the same issue [0] while lore/lei is broken.

Hi,

Lore breaking means I missed this :) thankfully you cc'd me (_this_ is why I am so adament about people following get_maintainer.pl procedure btw) so I was able to now notice + reply :)

This is totally my bad for missing this on review, so mea culpa.

No worry, your reviews are always very helpful!

Quoted 7 lines

[0]:https://lwn.net/ml/linux-mm/CA+G9fYt5QwJ4_F8fJj7jx9_0Le9kOVSeG38ox9qnKqwsrDdvHQ@mail.gmail.com/

Quoted 3 lines · SeongJae Park

madvise_lock() does nothing for MADV_HWPOSION and MADV_SOFT_OFFLINE behavior, but madvise_unlock() does mmap_lock unlocking regardless of the behavior.

[...]

Quoted 17 lines
Quoted 17 lines
 mm/madvise.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
mm/madvise.c +0 -1
index b5ef8e03d8b0..b8969457f3ef 100644
@@ -1577,7 +1577,6 @@ int madvise_set_anon_name(struct mm_struct *mm, unsigned long start,
 
 static int madvise_lock(struct mm_struct *mm, int behavior)
 {
-
 #ifdef CONFIG_MEMORY_FAILURE
 	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
 		return 0;
@@ -1595,6 +1594,11 @@ static int madvise_lock(struct mm_struct *mm, int behavior)
 
Quoted 9 lines

mm/madvise.c static int madvise_lock(struct mm_struct *mm, int behavior)

Quoted 6 lines
mm/madvise.c +4 -0
 static void madvise_unlock(struct mm_struct *mm, int behavior)
 {
+#ifdef CONFIG_MEMORY_FAILURE
+	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
+		return;
+#endif

I agree this fixes the issue but this is horrible. let's abstract this please rather than doing the same crap that already existed, only now twice.

I agree abstracting this is a better idea.

Quoted 10 lines
Quoted 9 lines
mm/madvise.c +1 -0
+
 	if (madvise_need_mmap_write(behavior))
 		mmap_write_unlock(mm);
 	else

base-commit: 8bf30f9d23eb5040d37e6e712789cee8e71e1577 --
2.39.5

Quoted 7 lines

I attach a fix-patch concept for something I think that'd be nicer, do with it what thy wilt! :P sorry I don't mean to be 'one of those' maintainers who copy/pastes code + demands somebody do it (by no means do I do so), but since this is so small I feel it's kind of quicker for me to do it this way.

Obviously take it or leave it/adapt it/etc. This is compile-tested only...

I further ran the repro program and confirmed this fixes the issue :)

Quoted 7 lines

----8<----
From 9fce3e47bf0fff2a2291be66002af346cdbca665 Mon Sep 17 00:00:00 2001
From: Lorenzo Stoakes <[email protected]>
Date: Tue, 11 Feb 2025 10:44:26 +0000
Subject: [PATCH] mm/madvise: fix madvise_[un]lock() issue

Quoted 5 lines

We are asymmetric in our locking/unlocking in the case of memory failure madvise() behaviour options, correct this and abstract the memory failure check.

Thanks,
SJ

[...]

#3 lore raw
Quoted 7 lines · SeongJae Park
Quoted 5 lines · On Tue, 11 Feb 2025 10:51:41 +0000 Lorenzo Stoakes…

+cc Naresh, Arnd for another reports/discussion of the same issue [0] while lore/lei is broken.

Hi,

Quoted 7 lines
Quoted 5 lines

Lore breaking means I missed this :) thankfully you cc'd me (_this_ is why I am so adament about people following get_maintainer.pl procedure btw) so I was able to now notice + reply :)

This is totally my bad for missing this on review, so mea culpa.

No worry, your reviews are always very helpful!

Very kind of you to say :>)

Quoted 36 lines
Quoted 8 lines

[0]:https://lwn.net/ml/linux-mm/CA+G9fYt5QwJ4_F8fJj7jx9_0Le9kOVSeG38ox9qnKqwsrDdvHQ@mail.gmail.com/

Quoted 3 lines · SeongJae Park

madvise_lock() does nothing for MADV_HWPOSION and MADV_SOFT_OFFLINE behavior, but madvise_unlock() does mmap_lock unlocking regardless of the behavior.

[...]

Quoted 17 lines
Quoted 17 lines
 mm/madvise.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
mm/madvise.c +0 -1
index b5ef8e03d8b0..b8969457f3ef 100644
@@ -1577,7 +1577,6 @@ int madvise_set_anon_name(struct mm_struct *mm, unsigned long start,
 
 static int madvise_lock(struct mm_struct *mm, int behavior)
 {
-
 #ifdef CONFIG_MEMORY_FAILURE
 	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
 		return 0;
@@ -1595,6 +1594,11 @@ static int madvise_lock(struct mm_struct *mm, int behavior)
 
Quoted 9 lines

mm/madvise.c static int madvise_lock(struct mm_struct *mm, int behavior)

Quoted 6 lines
mm/madvise.c +4 -0
 static void madvise_unlock(struct mm_struct *mm, int behavior)
 {
+#ifdef CONFIG_MEMORY_FAILURE
+	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
+		return;
+#endif

I agree this fixes the issue but this is horrible. let's abstract this please rather than doing the same crap that already existed, only now twice.

Quoted 1 line

I agree abstracting this is a better idea.

Thanks!

Quoted 17 lines
Quoted 17 lines
Quoted 10 lines
mm/madvise.c +1 -0
+
 	if (madvise_need_mmap_write(behavior))
 		mmap_write_unlock(mm);
 	else

base-commit: 8bf30f9d23eb5040d37e6e712789cee8e71e1577 --
2.39.5

I attach a fix-patch concept for something I think that'd be nicer, do with it what thy wilt! :P sorry I don't mean to be 'one of those' maintainers who copy/pastes code + demands somebody do it (by no means do I do so), but since this is so small I feel it's kind of quicker for me to do it this way.

Quoted 3 lines
Quoted 1 line

Obviously take it or leave it/adapt it/etc. This is compile-tested only...

I further ran the repro program and confirmed this fixes the issue :)

Great, perfect!

Quoted 8 lines
Quoted 8 lines

----8<----
From 9fce3e47bf0fff2a2291be66002af346cdbca665 Mon Sep 17 00:00:00 2001
From: Lorenzo Stoakes <[email protected]>
Date: Tue, 11 Feb 2025 10:44:26 +0000
Subject: [PATCH] mm/madvise: fix madvise_[un]lock() issue

Quoted 8 lines
Quoted 5 lines

We are asymmetric in our locking/unlocking in the case of memory failure madvise() behaviour options, correct this and abstract the memory failure check.

Thanks :)

I accidentally seem to have included some whitespace errors which Andrew kindly fixed up :>) so apologies for that!

Quoted 6 lines

Thanks,
SJ

[...]

Cheers, Lorenzo