Skip to contentSkip to conversation

ext4: no need to remove extent if len is 0 in ext4_es_remove_extent()

PATCH In review Eryu Guan 3 participants 7 messages Feb 22, 2013 – Feb 24, 2013 ext4 file system

lore
Share & tools
lore

Eryu Guan

author
#0 lore raw

PATCH 1/1 ext4: no need to remove extent if len is 0 in ext4_es_remove_extent()

len is 0 means no extent needs to be removed, so return immediately. Otherwise it could trigger the following BUG_ON()

	436	end = offset + len - 1;
	437	BUG_ON(end < offset);

This could be reproduced by a simple truncate(1) command by an unprivileged user

	truncate -s $(($((2**32 - 1)) * 4096)) /mnt/ext4/testfile

The same is true for __es_insert_extent().

Patched kernel passed xfstests regression test.

Also remove comments about EXT4_I(inode)->i_es_lock, this rwlock isn't hold by callers.

Cc Zheng Liu <[email protected]>
Cc "Theodore Ts'o" <[email protected]>
---
 fs/ext4/extents_status.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/ext4/extents_status.c +6 -2
index 564d981..3ac09ca 100644
@@ -328,6 +328,9 @@ static int __es_insert_extent(struct ext4_es_tree *tree, ext4_lblk_t offset,
 	struct extent_status *es;
 	ext4_lblk_t end = offset + len - 1;
 
+	if (!len)
+		return 0;
+
 	BUG_ON(end < offset);
 	es = tree->cache_es;
 	if (es && offset == (extent_status_end(es) + 1)) {
@@ -386,7 +389,6 @@ out:
 
 /*
  * ext4_es_insert_extent() adds a space to a delayed extent tree.
- * Caller holds inode->i_es_lock.
  *
  * ext4_es_insert_extent is called by ext4_da_write_begin and
  * ext4_es_remove_extent.
@@ -415,7 +417,6 @@ int ext4_es_insert_extent(struct inode *inode, ext4_lblk_t offset,
 
 /*
  * ext4_es_remove_extent() removes a space from a delayed extent tree.
- * Caller holds inode->i_es_lock.
  *
  * Return 0 on success, error code on failure.
  */
@@ -433,6 +434,9 @@ int ext4_es_remove_extent(struct inode *inode, ext4_lblk_t offset,
 	es_debug("remove [%u/%u) from extent status tree of inode %lu\n",
 		 offset, len, inode->i_ino);
 
+	if (!len)
+		return err;
+
 	end = offset + len - 1;
 	BUG_ON(end < offset);
 	write_lock(&EXT4_I(inode)->i_es_lock);
Signature · 1 line
1.8.1.2
Reviewed-by on v1 1/1
#1 lore raw
Quoted 13 lines · Eryu Guan

len is 0 means no extent needs to be removed, so return immediately. Otherwise it could trigger the following BUG_ON()

	436	end = offset + len - 1;
	437	BUG_ON(end < offset);

This could be reproduced by a simple truncate(1) command by an unprivileged user

	truncate -s $(($((2**32 - 1)) * 4096)) /mnt/ext4/testfile

The same is true for __es_insert_extent().

Quoted 8 lines

Patched kernel passed xfstests regression test.

Also remove comments about EXT4_I(inode)->i_es_lock, this rwlock isn't hold by callers.

Cc Zheng Liu <[email protected]>
Cc "Theodore Ts'o" <[email protected]>

Thanks for fixing it.

Regards,
- Zheng

Quoted 51 lines
---
 fs/ext4/extents_status.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
fs/ext4/extents_status.c +6 -2
index 564d981..3ac09ca 100644
@@ -328,6 +328,9 @@ static int __es_insert_extent(struct ext4_es_tree *tree, ext4_lblk_t offset,
 	struct extent_status *es;
 	ext4_lblk_t end = offset + len - 1;
 
+	if (!len)
+		return 0;
+
 	BUG_ON(end < offset);
 	es = tree->cache_es;
 	if (es && offset == (extent_status_end(es) + 1)) {
@@ -386,7 +389,6 @@ out:
 
 /*
  * ext4_es_insert_extent() adds a space to a delayed extent tree.
- * Caller holds inode->i_es_lock.
  *
  * ext4_es_insert_extent is called by ext4_da_write_begin and
  * ext4_es_remove_extent.
@@ -415,7 +417,6 @@ int ext4_es_insert_extent(struct inode *inode, ext4_lblk_t offset,
 
 /*
  * ext4_es_remove_extent() removes a space from a delayed extent tree.
- * Caller holds inode->i_es_lock.
  *
  * Return 0 on success, error code on failure.
  */
@@ -433,6 +434,9 @@ int ext4_es_remove_extent(struct inode *inode, ext4_lblk_t offset,
 	es_debug("remove [%u/%u) from extent status tree of inode %lu\n",
 		 offset, len, inode->i_ino);
 
+	if (!len)
+		return err;
+
 	end = offset + len - 1;
 	BUG_ON(end < offset);
 	write_lock(&EXT4_I(inode)->i_es_lock);
Signature · 6 lines
1.8.1.2

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
maintainer Reviewed-by Zheng Liu
#2 lore raw

This patch didn't apply since it was apparently against an older version of the extents status patches. Here is the version after I fixed it up so it would apply into the current ext4 tree. Zheng, can you do a quick sanity check to make sure I didn't screw up anything? Thanks!

Eryu, thanks for testing and submitting a bug fix!!

- Ted

Quoted 1 line

From 7d46e5051453b2c4dfac4e31ae1afb30064cc404 Mon Sep 17 00:00:00 2001

From: Eryu Guan <[email protected]>
Date: Fri, 22 Feb 2013 12:54:36 -0500
Subject: [PATCH] ext4: no need to remove extent if len is 0 in
ext4_es_remove_extent()

len is 0 means no extent needs to be removed, so return immediately. Otherwise it could trigger the following BUG_ON() in ext4_es_remove_extent()

	end = lblk + len - 1;
	BUG_ON(end < lblk);

This could be reproduced by a simple truncate(1) command by an unprivileged user

	truncate -s $(($((2**32 - 1)) * 4096)) /mnt/ext4/testfile

The same is true for __es_insert_extent().

Patched kernel passed xfstests regression test.

---
 fs/ext4/extents_status.c | 9 +++++++++
 1 file changed, 9 insertions(+)
fs/ext4/extents_status.c +9 -0
index 9f1380e..2be245b 100644
@@ -392,6 +392,9 @@ static int __es_insert_extent(struct inode *inode, struct extent_status *newes)
 	struct rb_node *parent = NULL;
 	struct extent_status *es;
 
+	if (!len)
+		return 0;
+
 	while (*p) {
 		parent = *p;
 		es = rb_entry(parent, struct extent_status, rb_node);
@@ -456,6 +459,9 @@ int ext4_es_insert_extent(struct inode *inode, ext4_lblk_t lblk,
 	es_debug("add [%u/%u) %llu %llx to extent status tree of inode %lu\n",
 		 lblk, len, pblk, status, inode->i_ino);
 
+	if (!len)
+		return 0;
+
 	BUG_ON(end < lblk);
 
 	newes.es_lblk = lblk;
@@ -649,6 +655,9 @@ int ext4_es_remove_extent(struct inode *inode, ext4_lblk_t lblk,
 	es_debug("remove [%u/%u) from extent status tree of inode %lu\n",
 		 lblk, len, inode->i_ino);
 
+	if (!len)
+		return err;
+
 	end = lblk + len - 1;
 	BUG_ON(end < lblk);
 
Signature · 1 line
1.7.12.rc0.22.gcdd159b
author
#3 lore raw
Quoted 3 lines · Theodore Ts'o

This patch didn't apply since it was apparently against an older version of the extents status patches. Here is the version after I fixed it up so it would apply into the current ext4 tree. Zheng, can

Thanks Ted! I was making the patch on top of Linus' tree. Linus' tree vs ext4 tree which one is preferred for submitting patch?

Thanks,
Eryu

Quoted 73 lines

you do a quick sanity check to make sure I didn't screw up anything? Thanks!

Eryu, thanks for testing and submitting a bug fix!!

- Ted

From 7d46e5051453b2c4dfac4e31ae1afb30064cc404 Mon Sep 17 00:00:00 2001
From: Eryu Guan <[email protected]>
Date: Fri, 22 Feb 2013 12:54:36 -0500
Subject: [PATCH] ext4: no need to remove extent if len is 0 in
ext4_es_remove_extent()

len is 0 means no extent needs to be removed, so return immediately. Otherwise it could trigger the following BUG_ON() in ext4_es_remove_extent()

        end = lblk + len - 1;
        BUG_ON(end < lblk);

This could be reproduced by a simple truncate(1) command by an unprivileged user

        truncate -s $(($((2**32 - 1)) * 4096)) /mnt/ext4/testfile

The same is true for __es_insert_extent().

Patched kernel passed xfstests regression test.

---
 fs/ext4/extents_status.c | 9 +++++++++
 1 file changed, 9 insertions(+)
fs/ext4/extents_status.c +9 -1
index 9f1380e..2be245b 100644
@@ -392,6 +392,9 @@ static int __es_insert_extent(struct inode *inode, struct extent_status *newes)
        struct rb_node *parent = NULL;
        struct extent_status *es;
 
+       if (!len)
+               return 0;
+
        while (*p) {
                parent = *p;
                es = rb_entry(parent, struct extent_status, rb_node);
@@ -456,6 +459,9 @@ int ext4_es_insert_extent(struct inode *inode, ext4_lblk_t lblk,
        es_debug("add [%u/%u) %llu %llx to extent status tree of inode %lu\n",
                 lblk, len, pblk, status, inode->i_ino);
 
+       if (!len)
+               return 0;
+
        BUG_ON(end < lblk);
 
        newes.es_lblk = lblk;
@@ -649,6 +655,9 @@ int ext4_es_remove_extent(struct inode *inode, ext4_lblk_t lblk,
        es_debug("remove [%u/%u) from extent status tree of inode %lu\n",
                 lblk, len, inode->i_ino);
 
+       if (!len)
+               return err;
+
        end = lblk + len - 1;
        BUG_ON(end < lblk);
 
--

1.7.12.rc0.22.gcdd159b

maintainer
#4 lore raw
Quoted 7 lines · Eryu Guan
Quoted 3 lines · Theodore Ts'o

This patch didn't apply since it was apparently against an older version of the extents status patches. Here is the version after I fixed it up so it would apply into the current ext4 tree. Zheng, can

Thanks Ted! I was making the patch on top of Linus' tree. Linus' tree vs ext4 tree which one is preferred for submitting patch?

The ext4 tree in general is the one which is preferred; the dev branch is the tip of what we hope to push to Linus. At the moment, it's in final testing. The three branch pointers which are important on the ext4 tree are origin, master, and dev. The origin branch is where we have branched off of Linus's tree. At the moment, ext4/origin is pointing at v3.8-rc3. The ext4/master branch is always between origin and dev (inclusive). The dev branch is a rewinding branch, which means that everything between master and dev may be get modified (i.e., to add a Reviewed-by: or to fix up some comments, etc.), or may get dropped (if it turns out we discover the patch is not ready for prime time). The dev branch is also what gets included into linux-next.

The master branch represents those patches which have been "finalized", which means once we bump the master branch, all of the commits between origin and master (inclusive) are guaranteed not to change. So for people who are building on top of master, it's safe for them to use git. For people who are building on top of dev, if you want to make changes, it's recommended you use a tool like quilt, guilt, or stgit.

Speaking of quilt/guilt, the set of patches between master and dev can be found here:

	http://repo.or.cz/w/ext4-patch-queue.git
	git://repo.or.cz/ext4-patch-queue.git

For those people who are interested, or who want to more easily cherry pick specific patches out of the ext4 patch queue, the ext4/dev branch (usually, assuming I've remembered to update the ext4 patch queue tree) can be reconstructed as follows:

git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git ext4 mkdir -p ext4/.git/patches
cd ext4/.git/patches
git clone git://repo.or.cz/ext4-patch-queue.git dev
cd dev
sh timestamps
cd ../../..
git branch dev $(head -1 .git/patches/dev/series | sed -e 's/# BASE //') git checkout dev
guilt push stable-boundary
guilt pop

(This assumes you are using guilt version v0.35, found at
git://repo.or.cz/guilt.git; note that the tip of the guilt tree has incompatible changes in how they parse patches, so I haven't upgraded to the tip of guilt tree yet.)

Anyway, most people will send me patches against Linus's tree, and that's fine; if there are problems, I can usually fix up the patches. But it's most convenient for me if people send against either the ext4/master, or most preferably, the ext4/dev branch.

BTW, I've updated the ext4 wiki to include the above information.

Thanks,

- Ted

author
#5 lore raw
Quoted 58 lines · Theodore Ts'o
Quoted 7 lines · Eryu Guan
Quoted 3 lines · Theodore Ts'o

This patch didn't apply since it was apparently against an older version of the extents status patches. Here is the version after I fixed it up so it would apply into the current ext4 tree. Zheng, can

Thanks Ted! I was making the patch on top of Linus' tree. Linus' tree vs ext4 tree which one is preferred for submitting patch?

The ext4 tree in general is the one which is preferred; the dev branch is the tip of what we hope to push to Linus. At the moment, it's in final testing. The three branch pointers which are important on the ext4 tree are origin, master, and dev. The origin branch is where we have branched off of Linus's tree. At the moment, ext4/origin is pointing at v3.8-rc3. The ext4/master branch is always between origin and dev (inclusive). The dev branch is a rewinding branch, which means that everything between master and dev may be get modified (i.e., to add a Reviewed-by: or to fix up some comments, etc.), or may get dropped (if it turns out we discover the patch is not ready for prime time). The dev branch is also what gets included into linux-next.

The master branch represents those patches which have been "finalized", which means once we bump the master branch, all of the commits between origin and master (inclusive) are guaranteed not to change. So for people who are building on top of master, it's safe for them to use git. For people who are building on top of dev, if you want to make changes, it's recommended you use a tool like quilt, guilt, or stgit.

Speaking of quilt/guilt, the set of patches between master and dev can be found here:

        http://repo.or.cz/w/ext4-patch-queue.git
        git://repo.or.cz/ext4-patch-queue.git

For those people who are interested, or who want to more easily cherry pick specific patches out of the ext4 patch queue, the ext4/dev branch (usually, assuming I've remembered to update the ext4 patch queue tree) can be reconstructed as follows:

git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git ext4 mkdir -p ext4/.git/patches
cd ext4/.git/patches
git clone git://repo.or.cz/ext4-patch-queue.git dev
cd dev
sh timestamps
cd ../../..
git branch dev $(head -1 .git/patches/dev/series | sed -e 's/# BASE //') git checkout dev
guilt push stable-boundary
guilt pop

(This assumes you are using guilt version v0.35, found at
git://repo.or.cz/guilt.git; note that the tip of the guilt tree has incompatible changes in how they parse patches, so I haven't upgraded to the tip of guilt tree yet.)

Quoted 6 lines

Anyway, most people will send me patches against Linus's tree, and that's fine; if there are problems, I can usually fix up the patches. But it's most convenient for me if people send against either the ext4/master, or most preferably, the ext4/dev branch.

BTW, I've updated the ext4 wiki to include the above information.

Thanks for your excellent explanation! I think I'd like to try ext4/master first :)

Eryu

Quoted 4 lines

Thanks,

- Ted

Reviewed-by
#6 lore raw
In reply to Theodore Ts'o

Hi Ted,

One minor comment below for the note.

On 02/23/2013 01:55 AM, Theodore Ts'o wrote:
[snip]

Quoted 34 lines

From 7d46e5051453b2c4dfac4e31ae1afb30064cc404 Mon Sep 17 00:00:00 2001
From: Eryu Guan <[email protected]>
Date: Fri, 22 Feb 2013 12:54:36 -0500
Subject: [PATCH] ext4: no need to remove extent if len is 0 in
ext4_es_remove_extent()

len is 0 means no extent needs to be removed, so return immediately. Otherwise it could trigger the following BUG_ON() in ext4_es_remove_extent()

	end = lblk + len - 1;
	BUG_ON(end < lblk);

This could be reproduced by a simple truncate(1) command by an unprivileged user

	truncate -s $(($((2**32 - 1)) * 4096)) /mnt/ext4/testfile

The same is true for __es_insert_extent().

Patched kernel passed xfstests regression test.

---
 fs/ext4/extents_status.c | 9 +++++++++
 1 file changed, 9 insertions(+)
fs/ext4/extents_status.c +0 -0
index 9f1380e..2be245b 100644

fs/ext4/extents_status.c static int __es_insert_extent(struct inode *inode, struct extent_status *newes) · line 397

Quoted 6 lines
fs/ext4/extents_status.c +3 -0
@@ -392,6 +392,9 @@ static int __es_insert_extent(struct inode *inode, struct extent_status *newes)
 	struct rb_node *parent = NULL;
 	struct extent_status *es;
 
+	if (!len)
+		return 0;
+

This will cause a compile error because we don't define a 'len' variable. But I have noticed that you have fixed it in latest dev branch. So just for the note.

Otherwise the patch looks good.

Thanks for fixing it,
- Zheng

Quoted 24 lines
fs/ext4/extents_status.c +6 -0
 	while (*p) {
 		parent = *p;
 		es = rb_entry(parent, struct extent_status, rb_node);
@@ -456,6 +459,9 @@ int ext4_es_insert_extent(struct inode *inode, ext4_lblk_t lblk,
 	es_debug("add [%u/%u) %llu %llx to extent status tree of inode %lu\n",
 		 lblk, len, pblk, status, inode->i_ino);
 
+	if (!len)
+		return 0;
+
 	BUG_ON(end < lblk);
 
 	newes.es_lblk = lblk;
@@ -649,6 +655,9 @@ int ext4_es_remove_extent(struct inode *inode, ext4_lblk_t lblk,
 	es_debug("remove [%u/%u) from extent status tree of inode %lu\n",
 		 lblk, len, inode->i_ino);
 
+	if (!len)
+		return err;
+
 	end = lblk + len - 1;
 	BUG_ON(end < lblk);