[PATCH] Ext4: Don't normalize an falloc request if it can fit in 1 extent.

[PATCH] Ext4: Don't normalize an falloc request if it can fit in 1 extent.

[Greg Harm]

If an fallocate request fits in EXT_UNINIT_MAX_LEN, then set the
EXT4_GET_BLOCKS_NO_NORMALIZE flag. For larger fallocate requests,
let mballoc.c normalize the request.
This fixes a problem where large requests were being split
into non-contiguous extents due to haldar@google.com's
"ext4: do not normalize block requests from fallocate."

Testing: Checked that 8.x MB falloc'ed files are still laid down
next to each other (contiguously).
Checked that the maximum size extent (127.9MB) is allocated as 1
extent.
Checked that a 1GB file is somewhat contiguous (often 5-6
non-contiguous extents now).
Checked that a 120MB file can still be falloc'ed even if there are no
single extents large enough to hold it.

Signed-off-by: Greg Harm <gharm@google.com>
---
 fs/ext4/extents.c |   13 ++++++++++---
 1 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 57cf568..9819216 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -3761,6 +3761,7 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
 	int ret2 = 0;
 	int retries = 0;
 	struct ext4_map_blocks map;
+	int map_blocks_flags;
 	unsigned int credits, blkbits = inode->i_blkbits;
 
 	/*
@@ -3805,9 +3806,15 @@ retry:
 			ret = PTR_ERR(handle);
 			break;
 		}
-		ret = ext4_map_blocks(handle, inode, &map,
-				      EXT4_GET_BLOCKS_CREATE_UNINIT_EXT |
-				      EXT4_GET_BLOCKS_NO_NORMALIZE);
+		map_blocks_flags = EXT4_GET_BLOCKS_CREATE_UNINIT_EXT;
+		/*
+		 * Don't normalize the request if it can fit in one extent so
+		 * that it doesn't get unnecessarily split into multiple
+		 * extents.
+		 */
+		if (len <= EXT_UNINIT_MAX_LEN << blkbits)
+			map_blocks_flags |= EXT4_GET_BLOCKS_NO_NORMALIZE;
+		ret = ext4_map_blocks(handle, inode, &map, map_blocks_flags);
 		if (ret <= 0) {
 #ifdef EXT4FS_DEBUG
 			WARN_ON(ret <= 0);
-- 
1.7.3.1

Re: [PATCH] Ext4: Don't normalize an falloc request if it can fit in 1 extent.

[Andreas Dilger]

On 2011-10-28, at 12:01 PM, Greg Harm <gharm@google.com> wrote:
> If an fallocate request fits in EXT_UNINIT_MAX_LEN, then set the
> EXT4_GET_BLOCKS_NO_NORMALIZE flag. For larger fallocate requests,
> let mballoc.c normalize the request.
> This fixes a problem where large requests were being split
> into non-contiguous extents due to haldar@google.com's
> "ext4: do not normalize block requests from fallocate."
> 
> Testing: Checked that 8.x MB falloc'ed files are still laid down
> next to each other (contiguously).
> Checked that the maximum size extent (127.9MB) is allocated as 1
> extent.
> Checked that a 1GB file is somewhat contiguous (often 5-6
> non-contiguous extents now).
> Checked that a 120MB file can still be falloc'ed even if there are no
> single extents large enough to hold it.

It would be great to add these test cases to xfstests so that this doesn't break in the future. 

> Signed-off-by: Greg Harm <gharm@google.com>
> ---
> fs/ext4/extents.c |   13 ++++++++++---
> 1 files changed, 10 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
> index 57cf568..9819216 100644
> --- a/fs/ext4/extents.c
> +++ b/fs/ext4/extents.c
> @@ -3761,6 +3761,7 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
>    int ret2 = 0;
>    int retries = 0;
>    struct ext4_map_blocks map;
> +    int map_blocks_flags;
>    unsigned int credits, blkbits = inode->i_blkbits;
> 
>    /*
> @@ -3805,9 +3806,15 @@ retry:
>            ret = PTR_ERR(handle);
>            break;
>        }
> -        ret = ext4_map_blocks(handle, inode, &map,
> -                      EXT4_GET_BLOCKS_CREATE_UNINIT_EXT |
> -                      EXT4_GET_BLOCKS_NO_NORMALIZE);
> +        map_blocks_flags = EXT4_GET_BLOCKS_CREATE_UNINIT_EXT;
> +        /*
> +         * Don't normalize the request if it can fit in one extent so
> +         * that it doesn't get unnecessarily split into multiple
> +         * extents.
> +         */
> +        if (len <= EXT_UNINIT_MAX_LEN << blkbits)
> +            map_blocks_flags |= EXT4_GET_BLOCKS_NO_NORMALIZE;
> +        ret = ext4_map_blocks(handle, inode, &map, map_blocks_flags);
>        if (ret <= 0) {
> #ifdef EXT4FS_DEBUG
>            WARN_ON(ret <= 0);
> -- 
> 1.7.3.1
> 

Re: [PATCH] Ext4: Don't normalize an falloc request if it can fit in 1 extent.

[Greg Harm]

On Fri, Oct 28, 2011 at 11:47 AM, Andreas Dilger <aedilger@gmail.com> wrote:
> On 2011-10-28, at 12:01 PM, Greg Harm <gharm@google.com> wrote:
>> If an fallocate request fits in EXT_UNINIT_MAX_LEN, then set the
>> EXT4_GET_BLOCKS_NO_NORMALIZE flag. For larger fallocate requests,
>> let mballoc.c normalize the request.
>> This fixes a problem where large requests were being split
>> into non-contiguous extents due to haldar@google.com's
>> "ext4: do not normalize block requests from fallocate."
>>
>> Testing: Checked that 8.x MB falloc'ed files are still laid down
>> next to each other (contiguously).
>> Checked that the maximum size extent (127.9MB) is allocated as 1
>> extent.
>> Checked that a 1GB file is somewhat contiguous (often 5-6
>> non-contiguous extents now).
>> Checked that a 120MB file can still be falloc'ed even if there are no
>> single extents large enough to hold it.
>
> It would be great to add these test cases to xfstests so that this doesn't break in the future.
Hi Andreas,

I totally agree. I have that on my TODO list (but no current ETA).

Thanks,
Greg

>> Signed-off-by: Greg Harm <gharm@google.com>
>> ---
>> fs/ext4/extents.c |   13 ++++++++++---
>> 1 files changed, 10 insertions(+), 3 deletions(-)
>>
>> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
>> index 57cf568..9819216 100644
>> --- a/fs/ext4/extents.c
>> +++ b/fs/ext4/extents.c
>> @@ -3761,6 +3761,7 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
>>    int ret2 = 0;
>>    int retries = 0;
>>    struct ext4_map_blocks map;
>> +    int map_blocks_flags;
>>    unsigned int credits, blkbits = inode->i_blkbits;
>>
>>    /*
>> @@ -3805,9 +3806,15 @@ retry:
>>            ret = PTR_ERR(handle);
>>            break;
>>        }
>> -        ret = ext4_map_blocks(handle, inode, &map,
>> -                      EXT4_GET_BLOCKS_CREATE_UNINIT_EXT |
>> -                      EXT4_GET_BLOCKS_NO_NORMALIZE);
>> +        map_blocks_flags = EXT4_GET_BLOCKS_CREATE_UNINIT_EXT;
>> +        /*
>> +         * Don't normalize the request if it can fit in one extent so
>> +         * that it doesn't get unnecessarily split into multiple
>> +         * extents.
>> +         */
>> +        if (len <= EXT_UNINIT_MAX_LEN << blkbits)
>> +            map_blocks_flags |= EXT4_GET_BLOCKS_NO_NORMALIZE;
>> +        ret = ext4_map_blocks(handle, inode, &map, map_blocks_flags);
>>        if (ret <= 0) {
>> #ifdef EXT4FS_DEBUG
>>            WARN_ON(ret <= 0);
>> --
>> 1.7.3.1
>>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Ext4: Don't normalize an falloc request if it can fit in 1 extent.

[Ted Ts'o]

On Fri, Oct 28, 2011 at 11:01:57AM -0700, Greg Harm wrote:
> If an fallocate request fits in EXT_UNINIT_MAX_LEN, then set the
> EXT4_GET_BLOCKS_NO_NORMALIZE flag. For larger fallocate requests,
> let mballoc.c normalize the request.
> This fixes a problem where large requests were being split
> into non-contiguous extents due to haldar@google.com's
> "ext4: do not normalize block requests from fallocate."
> 
> Testing: Checked that 8.x MB falloc'ed files are still laid down
> next to each other (contiguously).
> Checked that the maximum size extent (127.9MB) is allocated as 1
> extent.
> Checked that a 1GB file is somewhat contiguous (often 5-6
> non-contiguous extents now).
> Checked that a 120MB file can still be falloc'ed even if there are no
> single extents large enough to hold it.
> 
> Signed-off-by: Greg Harm <gharm@google.com>

Thanks, I had to tweak the patch slightly to make it apply given the
current ext4 tree.

					- Ted

commit 5f11e7c457769808bafe5048a98c5cfd624e85f3
Author: Greg Harm <gharm@google.com>
Date:   Mon Oct 31 18:41:47 2011 -0400

    ext4: Don't normalize an falloc request if it can fit in 1 extent.
    
    If an fallocate request fits in EXT_UNINIT_MAX_LEN, then set the
    EXT4_GET_BLOCKS_NO_NORMALIZE flag. For larger fallocate requests,
    let mballoc.c normalize the request.
    
    This fixes a problem where large requests were being split into
    non-contiguous extents due to commit 556b27abf73: ext4: do not
    normalize block requests from fallocate.
    
    Testing:
    *) Checked that 8.x MB falloc'ed files are still laid down next to
    each other (contiguously).
    *) Checked that the maximum size extent (127.9MB) is allocated as 1
    extent.
    *) Checked that a 1GB file is somewhat contiguous (often 5-6
    non-contiguous extents now).
    *) Checked that a 120MB file can still be falloc'ed even if there are
    no single extents large enough to hold it.
    
    Signed-off-by: Greg Harm <gharm@google.com>
    Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index e585caa..9dfdf8f 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -4336,10 +4336,16 @@ long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
 		trace_ext4_fallocate_exit(inode, offset, max_blocks, ret);
 		return ret;
 	}
-	flags = EXT4_GET_BLOCKS_CREATE_UNINIT_EXT |
-		EXT4_GET_BLOCKS_NO_NORMALIZE;
+	flags = EXT4_GET_BLOCKS_CREATE_UNINIT_EXT;
 	if (mode & FALLOC_FL_KEEP_SIZE)
 		flags |= EXT4_GET_BLOCKS_KEEP_SIZE;
+	/*
+	 * Don't normalize the request if it can fit in one extent so
+	 * that it doesn't get unnecessarily split into multiple
+	 * extents.
+	 */
+	if (len <= EXT_UNINIT_MAX_LEN << blkbits)
+		flags |= EXT4_GET_BLOCKS_NO_NORMALIZE;
 retry:
 	while (ret >= 0 && ret < max_blocks) {
 		map.m_lblk = map.m_lblk + ret;