[PATCH] ext4: handle the rest of ext4_mb_load_buddy() ENOMEM errors

[PATCH] ext4: handle the rest of ext4_mb_load_buddy() ENOMEM errors

[Konstantin Khlebnikov]

I've got another report about breaking ext4 by ENOMEM error returned from
ext4_mb_load_buddy() caused by memory shortage in memory cgroup.
This time inside ext4_discard_preallocations().

This patch replaces ext4_error() with ext4_warning() where errors returned
from ext4_mb_load_buddy() are not fatal and handled by caller:
* ext4_mb_discard_group_preallocations() - called before generating ENOSPC,
  we'll try to discard other group or return ENOSPC into user-space.
* ext4_trim_all_free() - just stop trimming and return ENOMEM from ioctl.

Some callers cannot handle errors, thus __GFP_NOFAIL is used for them:
* ext4_discard_preallocations()
* ext4_mb_discard_lg_preallocations()

The only unclear case is ext4_group_add_blocks(), probably ext4_std_error()
should handle ENOMEM as warning and don't break filesystem.

Fixes: adb7ef600cc9 ("ext4: use __GFP_NOFAIL in ext4_free_blocks()")
Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
---
 fs/ext4/mballoc.c |   23 ++++++++++++++---------
 1 file changed, 14 insertions(+), 9 deletions(-)

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 5083bce20ac4..b7928cddd539 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -3887,7 +3887,8 @@ ext4_mb_discard_group_preallocations(struct super_block *sb,
 
 	err = ext4_mb_load_buddy(sb, group, &e4b);
 	if (err) {
-		ext4_error(sb, "Error loading buddy information for %u", group);
+		ext4_warning(sb, "Error %d loading buddy information for %u",
+			     err, group);
 		put_bh(bitmap_bh);
 		return 0;
 	}
@@ -4044,10 +4045,11 @@ void ext4_discard_preallocations(struct inode *inode)
 		BUG_ON(pa->pa_type != MB_INODE_PA);
 		group = ext4_get_group_number(sb, pa->pa_pstart);
 
-		err = ext4_mb_load_buddy(sb, group, &e4b);
+		err = ext4_mb_load_buddy_gfp(sb, group, &e4b,
+					     GFP_NOFS|__GFP_NOFAIL);
 		if (err) {
-			ext4_error(sb, "Error loading buddy information for %u",
-					group);
+			ext4_error(sb, "Error %d loading buddy information for %u",
+				   err, group);
 			continue;
 		}
 
@@ -4303,11 +4305,14 @@ ext4_mb_discard_lg_preallocations(struct super_block *sb,
 	spin_unlock(&lg->lg_prealloc_lock);
 
 	list_for_each_entry_safe(pa, tmp, &discard_list, u.pa_tmp_list) {
+		int err;
 
 		group = ext4_get_group_number(sb, pa->pa_pstart);
-		if (ext4_mb_load_buddy(sb, group, &e4b)) {
-			ext4_error(sb, "Error loading buddy information for %u",
-					group);
+		err = ext4_mb_load_buddy_gfp(sb, group, &e4b,
+					     GFP_NOFS|__GFP_NOFAIL);
+		if (err) {
+			ext4_error(sb, "Error %d loading buddy information for %u",
+				   err, group);
 			continue;
 		}
 		ext4_lock_group(sb, group);
@@ -5127,8 +5132,8 @@ ext4_trim_all_free(struct super_block *sb, ext4_group_t group,
 
 	ret = ext4_mb_load_buddy(sb, group, &e4b);
 	if (ret) {
-		ext4_error(sb, "Error in loading buddy "
-				"information for %u", group);
+		ext4_warning(sb, "Error %d loading buddy information for %u",
+			     ret, group);
 		return ret;
 	}
 	bitmap = e4b.bd_bitmap;

Re: [PATCH] ext4: handle the rest of ext4_mb_load_buddy() ENOMEM errors

[Theodore Ts'o]

On Fri, May 19, 2017 at 10:09:54AM +0300, Konstantin Khlebnikov wrote:
> I've got another report about breaking ext4 by ENOMEM error returned from
> ext4_mb_load_buddy() caused by memory shortage in memory cgroup.
> This time inside ext4_discard_preallocations().
> 
> This patch replaces ext4_error() with ext4_warning() where errors returned
> from ext4_mb_load_buddy() are not fatal and handled by caller:
> * ext4_mb_discard_group_preallocations() - called before generating ENOSPC,
>   we'll try to discard other group or return ENOSPC into user-space.
> * ext4_trim_all_free() - just stop trimming and return ENOMEM from ioctl.
> 
> Some callers cannot handle errors, thus __GFP_NOFAIL is used for them:
> * ext4_discard_preallocations()
> * ext4_mb_discard_lg_preallocations()
> 
> The only unclear case is ext4_group_add_blocks(), probably ext4_std_error()
> should handle ENOMEM as warning and don't break filesystem.
> 
> Fixes: adb7ef600cc9 ("ext4: use __GFP_NOFAIL in ext4_free_blocks()")
> Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>

Thanks, applied.

					- Ted