big allocation status

big allocation status

[Tao Ma]

Hi Ted,

Do you have any new branch or patch set that can be used for testing
about big allocation? The latest patch set in the mailist is what you
have sent at the beginning of May. Do you have an updated one? We will
try to test it soon and it would grateful if we are on the same page.

Great thanks.

Regards,
Tao

Re: big allocation status

[Ted Ts'o]

On Fri, Jun 03, 2011 at 09:57:23AM +0800, Tao Ma wrote:
> Hi Ted,
> 
> Do you have any new branch or patch set that can be used for testing
> about big allocation? The latest patch set in the mailist is what you
> have sent at the beginning of May. Do you have an updated one? We will
> try to test it soon and it would grateful if we are on the same page.

There is a patch set available at:

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

The bigalloc patches are after the stable-boundary, and will apply
against 3.0-rc1.

Note that I've fixed them up so that they apply on top of the punch
changes; however, the punch ioctl will not work correctly if bigalloc
is enabled, since that hasn't been fixed up yet.

Also, I'm still working on the e2fsprogs support.

						- Ted

Re: big allocation status

[Tao Ma]

On 06/04/2011 11:38 AM, Ted Ts'o wrote:
> On Fri, Jun 03, 2011 at 09:57:23AM +0800, Tao Ma wrote:
>> Hi Ted,
>>
>> Do you have any new branch or patch set that can be used for testing
>> about big allocation? The latest patch set in the mailist is what you
>> have sent at the beginning of May. Do you have an updated one? We will
>> try to test it soon and it would grateful if we are on the same page.
> 
> There is a patch set available at:
> 
> git://repo.or.cz/ext4-patch-queue.git
> 
> The bigalloc patches are after the stable-boundary, and will apply
> against 3.0-rc1.
Great thanks, we will schedule some test maybe next week.
> 
> Note that I've fixed them up so that they apply on top of the punch
> changes; however, the punch ioctl will not work correctly if bigalloc
> is enabled, since that hasn't been fixed up yet.
Fine, our test cases will not use punch ioctl AFAIK.
> 
> Also, I'm still working on the e2fsprogs support.
Cool, but do you have a mkfs.ext4 that we can let the kernel part work
happily? Or we have to add the feature by hand by editing the disk
format. ;)

Regards,
Tao

Re: big allocation status

[Ted Ts'o]

On Sat, Jun 04, 2011 at 04:34:09PM +0800, Tao Ma wrote:
> Cool, but do you have a mkfs.ext4 that we can let the kernel part work
> happily? Or we have to add the feature by hand by editing the disk
> format. ;)

Grab the bigalloc-hack branch on the e2fsprogs git tree on
git.kernel.org.  This will allow you to create bigalloc file systems
via "mke2fs -t ext4 -O bigalloc".  Don't base anything on this branch;
the code is incredibly hacky, and not how the final bigalloc
support will be done.

						- Ted

Re: big allocation status

[Ted Ts'o]

On Fri, Jun 03, 2011 at 11:38:05PM -0400, Ted Ts'o wrote:
> On Fri, Jun 03, 2011 at 09:57:23AM +0800, Tao Ma wrote:
> > Hi Ted,
> > 
> > Do you have any new branch or patch set that can be used for testing
> > about big allocation? The latest patch set in the mailist is what you
> > have sent at the beginning of May. Do you have an updated one? We will
> > try to test it soon and it would grateful if we are on the same page.
> 
> There is a patch set available at:
> 
> git://repo.or.cz/ext4-patch-queue.git
> 
> The bigalloc patches are after the stable-boundary, and will apply
> against 3.0-rc1.

I've updated the bigalloc patch set to fix a number of bugs that I and
a colleague of mine at Google have found.

This patch set known have problems with i_blocks getting out of sync
if delalloc is enabled (particularly with sparse files and truncates),
and as mentioned before, doesn't work correctly with the punch ioctl.
These are bugs that we plan to fix before anything gets pushed into
upstream, of course.

	      					- Ted

Re: big allocation status

[Tao Ma]

On 06/07/2011 07:04 AM, Ted Ts'o wrote:
> On Fri, Jun 03, 2011 at 11:38:05PM -0400, Ted Ts'o wrote:
>> On Fri, Jun 03, 2011 at 09:57:23AM +0800, Tao Ma wrote:
>>> Hi Ted,
>>>
>>> Do you have any new branch or patch set that can be used for testing
>>> about big allocation? The latest patch set in the mailist is what you
>>> have sent at the beginning of May. Do you have an updated one? We will
>>> try to test it soon and it would grateful if we are on the same page.
>>
>> There is a patch set available at:
>>
>> git://repo.or.cz/ext4-patch-queue.git
>>
>> The bigalloc patches are after the stable-boundary, and will apply
>> against 3.0-rc1.
> 
> I've updated the bigalloc patch set to fix a number of bugs that I and
> a colleague of mine at Google have found.
> 
> This patch set known have problems with i_blocks getting out of sync
> if delalloc is enabled (particularly with sparse files and truncates),
> and as mentioned before, doesn't work correctly with the punch ioctl.
> These are bugs that we plan to fix before anything gets pushed into
> upstream, of course.
cool, I will base my test on it. Thanks.

Regards,
Tao

Re: big allocation status

[Tao Ma]

Hi Ted,
	We began our test this week. And we found an error in dmesg:

[10484.043482] EXT4-fs (sda10): delayed block allocation failed for
inode 2289279 at logical offset 0 with max blocks 2 with error -28
[10484.043484] EXT4-fs (sda10): This should not happen!! Data will be lost
[10484.043484]
[10484.043486] Total free blocks count 0
[10484.043487] Free/Dirty block details
[10484.043488] free_blocks=0
[10484.043489] dirty_blocks=1495584
[10484.043489] Block reservation details
[10484.043490] i_reserved_data_blocks=2
[10484.043491] i_reserved_meta_blocks=1

yes, df shows that the file system is used 100% now, so I am not sure
whether this is a bug or not.

btw, this is a postmark test.

Regards,
Tao


On 06/07/2011 07:04 AM, Ted Ts'o wrote:
> On Fri, Jun 03, 2011 at 11:38:05PM -0400, Ted Ts'o wrote:
>> On Fri, Jun 03, 2011 at 09:57:23AM +0800, Tao Ma wrote:
>>> Hi Ted,
>>>
>>> Do you have any new branch or patch set that can be used for testing
>>> about big allocation? The latest patch set in the mailist is what you
>>> have sent at the beginning of May. Do you have an updated one? We will
>>> try to test it soon and it would grateful if we are on the same page.
>>
>> There is a patch set available at:
>>
>> git://repo.or.cz/ext4-patch-queue.git
>>
>> The bigalloc patches are after the stable-boundary, and will apply
>> against 3.0-rc1.
> 
> I've updated the bigalloc patch set to fix a number of bugs that I and
> a colleague of mine at Google have found.
> 
> This patch set known have problems with i_blocks getting out of sync
> if delalloc is enabled (particularly with sparse files and truncates),
> and as mentioned before, doesn't work correctly with the punch ioctl.
> These are bugs that we plan to fix before anything gets pushed into
> upstream, of course.
> 
> 	      					- Ted
> --
> 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