All of lore.kernel.org
 help / color / mirror / Atom feed
From: Ritesh Harjani <riteshh@linux.ibm.com>
To: linux-ext4@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org, Jan Kara <jack@suse.com>,
	tytso@mit.edu, "Aneesh Kumar K . V" <aneesh.kumar@linux.ibm.com>,
	linux-kernel@vger.kernel.org,
	Ritesh Harjani <riteshh@linux.ibm.com>
Subject: [RFCv4 1/6] ext4: mballoc: Refactor ext4_mb_good_group()
Date: Sun, 10 May 2020 12:38:21 +0530	[thread overview]
Message-ID: <aa1b1da1feb9b38174822410a4049b7c7bdfa0e7.1589086820.git.riteshh@linux.ibm.com> (raw)
In-Reply-To: <cover.1589086820.git.riteshh@linux.ibm.com>

ext4_mb_good_group() definition was changed some time back
and now it even initializes the buddy cache (via ext4_mb_init_group()),
if in case the EXT4_MB_GRP_NEED_INIT() is true for a group.
Note that ext4_mb_init_group() could sleep and so should not be called
under a spinlock held.
This is fine as of now because ext4_mb_good_group() is called before
loading the buddy bitmap without ext4_lock_group() held
and again called after loading the bitmap, only this time with
ext4_lock_group() held.
But still this whole thing is confusing.

So this patch refactors out ext4_mb_good_group_nolock() which should be
called when without holding ext4_lock_group().
Also in further patches we hold the spinlock (ext4_lock_group()) while
doing any calculations which involves grp->bb_free or grp->bb_fragments.

Signed-off-by: Ritesh Harjani <riteshh@linux.ibm.com>
---
 fs/ext4/mballoc.c | 80 ++++++++++++++++++++++++++++++-----------------
 1 file changed, 52 insertions(+), 28 deletions(-)

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 33a69424942c..da11a4a738bd 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -2066,15 +2066,16 @@ void ext4_mb_scan_aligned(struct ext4_allocation_context *ac,
 }
 
 /*
- * This is now called BEFORE we load the buddy bitmap.
+ * This is also called BEFORE we load the buddy bitmap.
  * Returns either 1 or 0 indicating that the group is either suitable
- * for the allocation or not. In addition it can also return negative
- * error code when something goes wrong.
+ * for the allocation or not. This should be called with ext4_lock_group()
+ * held to protect grp->bb_free from getting changed due to another
+ * allocation/discard is happening in parallel.
  */
-static int ext4_mb_good_group(struct ext4_allocation_context *ac,
+static bool ext4_mb_good_group(struct ext4_allocation_context *ac,
 				ext4_group_t group, int cr)
 {
-	unsigned free, fragments;
+	ext4_grpblk_t free, fragments;
 	int flex_size = ext4_flex_bg_size(EXT4_SB(ac->ac_sb));
 	struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group);
 
@@ -2082,23 +2083,16 @@ static int ext4_mb_good_group(struct ext4_allocation_context *ac,
 
 	free = grp->bb_free;
 	if (free == 0)
-		return 0;
+		return false;
 	if (cr <= 2 && free < ac->ac_g_ex.fe_len)
-		return 0;
+		return false;
 
 	if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(grp)))
-		return 0;
-
-	/* We only do this if the grp has never been initialized */
-	if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) {
-		int ret = ext4_mb_init_group(ac->ac_sb, group, GFP_NOFS);
-		if (ret)
-			return ret;
-	}
+		return false;
 
 	fragments = grp->bb_fragments;
 	if (fragments == 0)
-		return 0;
+		return false;
 
 	switch (cr) {
 	case 0:
@@ -2108,31 +2102,63 @@ static int ext4_mb_good_group(struct ext4_allocation_context *ac,
 		if ((ac->ac_flags & EXT4_MB_HINT_DATA) &&
 		    (flex_size >= EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME) &&
 		    ((group % flex_size) == 0))
-			return 0;
+			return false;
 
 		if ((ac->ac_2order > ac->ac_sb->s_blocksize_bits+1) ||
 		    (free / fragments) >= ac->ac_g_ex.fe_len)
-			return 1;
+			return true;
 
 		if (grp->bb_largest_free_order < ac->ac_2order)
-			return 0;
+			return false;
 
-		return 1;
+		return true;
 	case 1:
 		if ((free / fragments) >= ac->ac_g_ex.fe_len)
-			return 1;
+			return true;
 		break;
 	case 2:
 		if (free >= ac->ac_g_ex.fe_len)
-			return 1;
+			return true;
 		break;
 	case 3:
-		return 1;
+		return true;
 	default:
 		BUG();
 	}
 
-	return 0;
+	return false;
+}
+
+/*
+ * This could return negative error code if something goes wrong
+ * during ext4_mb_init_group(). This should not be called with
+ * ext4_lock_group() held.
+ */
+static int ext4_mb_good_group_nolock(struct ext4_allocation_context *ac,
+				     ext4_group_t group, int cr)
+{
+	struct ext4_group_info *grp = ext4_get_group_info(ac->ac_sb, group);
+	ext4_grpblk_t free;
+	int ret = 0;
+
+	free = grp->bb_free;
+	if (free == 0)
+		goto out;
+	if (cr <= 2 && free < ac->ac_g_ex.fe_len)
+		goto out;
+	if (unlikely(EXT4_MB_GRP_BBITMAP_CORRUPT(grp)))
+		goto out;
+
+	/* We only do this if the grp has never been initialized */
+	if (unlikely(EXT4_MB_GRP_NEED_INIT(grp))) {
+		ret = ext4_mb_init_group(ac->ac_sb, group, GFP_NOFS);
+		if (ret)
+			return ret;
+	}
+
+	ret = ext4_mb_good_group(ac, group, cr);
+out:
+	return ret;
 }
 
 static noinline_for_stack int
@@ -2220,7 +2246,7 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 				group = 0;
 
 			/* This now checks without needing the buddy page */
-			ret = ext4_mb_good_group(ac, group, cr);
+			ret = ext4_mb_good_group_nolock(ac, group, cr);
 			if (ret <= 0) {
 				if (!first_err)
 					first_err = ret;
@@ -2238,11 +2264,9 @@ ext4_mb_regular_allocator(struct ext4_allocation_context *ac)
 			 * block group
 			 */
 			ret = ext4_mb_good_group(ac, group, cr);
-			if (ret <= 0) {
+			if (ret == 0) {
 				ext4_unlock_group(sb, group);
 				ext4_mb_unload_buddy(&e4b);
-				if (!first_err)
-					first_err = ret;
 				continue;
 			}
 
-- 
2.21.0


  reply	other threads:[~2020-05-10  7:08 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-05-10  7:08 [RFCv4 0/6] Improve ext4 handling of ENOSPC with multi-threaded use-case Ritesh Harjani
2020-05-10  7:08 ` Ritesh Harjani [this message]
2020-05-10  7:08 ` [RFCv4 2/6] ext4: mballoc: Use ext4_lock_group() around calculations involving bb_free Ritesh Harjani
2020-05-10  7:08 ` [RFCv4 3/6] ext4: mballoc: Optimize ext4_mb_good_group_nolock further if grp needs init Ritesh Harjani
2020-05-10  7:08 ` [RFCv4 4/6] ext4: mballoc: Add blocks to PA list under same spinlock after allocating blocks Ritesh Harjani
2020-05-10  7:08 ` [RFCv4 5/6] ext4: mballoc: Refactor ext4_mb_discard_preallocations() Ritesh Harjani
2020-05-10  7:08 ` [RFCv4 6/6] ext4: mballoc: Introduce pcpu seqcnt for freeing PA to improve ENOSPC handling Ritesh Harjani

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=aa1b1da1feb9b38174822410a4049b7c7bdfa0e7.1589086820.git.riteshh@linux.ibm.com \
    --to=riteshh@linux.ibm.com \
    --cc=aneesh.kumar@linux.ibm.com \
    --cc=jack@suse.com \
    --cc=linux-ext4@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=tytso@mit.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.