public inbox for linux-ext4@vger.kernel.org
 help / color / mirror / Atom feed
From: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
To: Mingming Cao <cmm@us.ibm.com>, Theodore Tso <tytso@mit.edu>
Cc: ext4 development <linux-ext4@vger.kernel.org>
Subject: Re: Patches for the patchqueue
Date: Thu, 5 Jun 2008 15:25:53 +0530	[thread overview]
Message-ID: <20080605095510.GH8942@skywalker> (raw)
In-Reply-To: <20080605095350.GG8942@skywalker>

[-- Attachment #1: Type: text/plain, Size: 1 bytes --]



[-- Attachment #2: 0001-delalloc-enospc.patch --]
[-- Type: text/x-diff, Size: 6892 bytes --]

ext4: delalloc  block reservation fix

From: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>

a) We need to decrement the meta data blocks that got allocated
from percpu s_freeblocks_counter

b) We need to protect the reservation block counter so that
reserve and release space doesn't race each other.

c) don't check for free space in ext4_mb_new_blocks with delalloc
We already reserved the space.

e) Don't release space for block allocation from fallocate space.
We don't  reserve space for them

f) clear the delay bit in ext4_da_get_block_write instead of __block_write_full_page
so that we clear the delay bit for every successfull block allocation. We may fail
while marking inode dirty in ext4_da_get_block_write after allocating block. So
it is better to clear the delay bit in ext4_da_get_block_write rather than
__block_write_full_page

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
---

 fs/ext4/balloc.c  |    8 ++++++++
 fs/ext4/ext4_i.h  |    2 ++
 fs/ext4/inode.c   |   46 ++++++++++++++++++++++++++++++++--------------
 fs/ext4/mballoc.c |    7 ++++++-
 fs/ext4/super.c   |    2 ++
 5 files changed, 50 insertions(+), 15 deletions(-)


diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c
index 9b9057c..4932d29 100644
--- a/fs/ext4/balloc.c
+++ b/fs/ext4/balloc.c
@@ -1971,6 +1971,14 @@ static ext4_fsblk_t do_blk_alloc(handle_t *handle, struct inode *inode,
 		ar.flags = 0;
 	ret = ext4_mb_new_blocks(handle, &ar, errp);
 	*count = ar.len;
+	/*
+	 * Account for the allocated meta blocks
+	 */
+	if (!(*errp) && meta) {
+		spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
+		EXT4_I(inode)->i_allocated_meta_blocks += ar.len;
+		spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
+	}
 	return ret;
 }
 
diff --git a/fs/ext4/ext4_i.h b/fs/ext4/ext4_i.h
index fea6a5d..ef7409f 100644
--- a/fs/ext4/ext4_i.h
+++ b/fs/ext4/ext4_i.h
@@ -167,7 +167,9 @@ struct ext4_inode_info {
 	/* allocation reservation info for delalloc */
 	unsigned long i_reserved_data_blocks;
 	unsigned long i_reserved_meta_blocks;
+	unsigned long i_allocated_meta_blocks;
 	unsigned short i_delalloc_reserved_flag;
+	spinlock_t i_block_reservation_lock;
 };
 
 #endif	/* _EXT4_I */
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 2d00d05..417ed2c 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1426,11 +1426,12 @@ static int ext4_da_reserve_space(struct inode *inode, int nrblocks)
        struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
        unsigned long md_needed, mdblocks, total = 0;
 
-       /*
-        * calculate the amount of metadata blocks to reserve
-	* in order to allocate nrblocks
-	* worse case is one extent per block
-	*/
+	/*
+	 * recalculate the amount of metadata blocks to reserve
+	 * in order to allocate nrblocks
+	 * worse case is one extent per block
+	 */
+	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
 	total = EXT4_I(inode)->i_reserved_data_blocks + nrblocks;
 	mdblocks = ext4_ext_calc_metadata_amount(inode, total);
 	BUG_ON(mdblocks < EXT4_I(inode)->i_reserved_meta_blocks);
@@ -1438,42 +1439,51 @@ static int ext4_da_reserve_space(struct inode *inode, int nrblocks)
 	md_needed = mdblocks - EXT4_I(inode)->i_reserved_meta_blocks;
 	total = md_needed + nrblocks;
 
-	if (ext4_has_free_blocks(sbi, total) < total)
+	if (ext4_has_free_blocks(sbi, total) < total) {
+		spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
 		return -ENOSPC;
+	}
 
 	/* reduce fs free blocks counter */
 	percpu_counter_sub(&sbi->s_freeblocks_counter, total);
 
 	EXT4_I(inode)->i_reserved_data_blocks += nrblocks;
-	EXT4_I(inode)->i_reserved_meta_blocks += md_needed;
+	EXT4_I(inode)->i_reserved_meta_blocks = mdblocks;
 
+	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
 	return 0;       /* success */
 }
 
 void ext4_da_release_space(struct inode *inode, int used, int to_free)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
-	int total, mdb, release;
+	int total, mdb, mdb_free, release;
 
-	/* calculate the number of metablocks still need to be reserved */
+	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
+	/* recalculate the number of metablocks still need to be reserved */
 	total = EXT4_I(inode)->i_reserved_data_blocks - used - to_free;
 	mdb = ext4_ext_calc_metadata_amount(inode, total);
 
 	/* figure out how many metablocks to release */
 	BUG_ON(mdb > EXT4_I(inode)->i_reserved_meta_blocks);
-	mdb = EXT4_I(inode)->i_reserved_meta_blocks - mdb;
+	mdb_free = EXT4_I(inode)->i_reserved_meta_blocks - mdb;
 
-	release = to_free + mdb;
+	/* Account for allocated meta_blocks */
+	mdb_free -= EXT4_I(inode)->i_allocated_meta_blocks;
+
+	release = to_free + mdb_free;
 
 	/* update fs free blocks counter for truncate case */
 	percpu_counter_add(&sbi->s_freeblocks_counter, release);
 
 	/* update per-inode reservations */
 	BUG_ON(used + to_free > EXT4_I(inode)->i_reserved_data_blocks);
-	EXT4_I(inode)->i_reserved_data_blocks -= used + to_free;
+	EXT4_I(inode)->i_reserved_data_blocks -= (used + to_free);
 
 	BUG_ON(mdb > EXT4_I(inode)->i_reserved_meta_blocks);
-	EXT4_I(inode)->i_reserved_meta_blocks -= mdb;
+	EXT4_I(inode)->i_reserved_meta_blocks = mdb;
+	EXT4_I(inode)->i_allocated_meta_blocks = 0;
+	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
 }
 
 static void ext4_da_page_release_reservation(struct page *page,
@@ -1555,7 +1565,15 @@ static int ext4_da_get_block_write(struct inode *inode, sector_t iblock,
 		bh_result->b_size = (ret << inode->i_blkbits);
 
 		/* release reserved-but-unused meta blocks */
-		ext4_da_release_space(inode, ret, 0);
+		if (buffer_delay(bh_result)) {
+			ext4_da_release_space(inode, ret, 0);
+			/*
+			 * clear the delay bit now that we allocated
+			 * blocks. If it is not a single block request
+			 * we clear the delay bit in mpage_put_bnr_to_bhs
+			 */
+			clear_buffer_delay(bh_result);
+		}
 
 		/*
 		 * Update on-disk size along with block allocation
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 334e585..ec44d52 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -4049,7 +4049,12 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
 					    &(ar->len), errp);
 		return block;
 	}
-	ar->len = ext4_has_free_blocks(sbi, ar->len);
+	if (!EXT4_I(ar->inode)->i_delalloc_reserved_flag) {
+		/*
+		 * With delalloc we already reserved the blocks
+		 */
+		ar->len = ext4_has_free_blocks(sbi, ar->len);
+	}
 
 	if (ar->len == 0) {
 		*errp = -ENOSPC;
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index a990475..a33a0cf 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -575,7 +575,9 @@ static struct inode *ext4_alloc_inode(struct super_block *sb)
 	spin_lock_init(&ei->i_prealloc_lock);
 	ei->i_reserved_data_blocks = 0;
 	ei->i_reserved_meta_blocks = 0;
+	ei->i_allocated_meta_blocks = 0;
 	ei->i_delalloc_reserved_flag = 0;
+	spin_lock_init(&(ei->i_block_reservation_lock));
 	return &ei->vfs_inode;
 }
 

  reply	other threads:[~2008-06-05  9:59 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-06-05  9:50 Patches for the patchqueue Aneesh Kumar K.V
2008-06-05  9:51 ` Aneesh Kumar K.V
2008-06-05  9:53   ` Aneesh Kumar K.V
2008-06-05  9:55     ` Aneesh Kumar K.V [this message]
2008-06-05 18:02       ` Mingming Cao
2008-06-05 18:59         ` Aneesh Kumar K.V
2008-06-05 18:51     ` Andreas Dilger
  -- strict thread matches above, loose matches on Subject: below --
2008-06-06 18:24 Aneesh Kumar K.V

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=20080605095510.GH8942@skywalker \
    --to=aneesh.kumar@linux.vnet.ibm.com \
    --cc=cmm@us.ibm.com \
    --cc=linux-ext4@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox