All of lore.kernel.org
 help / color / mirror / Atom feed
From: Yongqiang Yang <xiaoqiangnk@gmail.com>
To: Tao Ma <tm@tao.ma>
Cc: linux-ext4@vger.kernel.org, jack@suse.cz, jeff.liu@oracle.com,
	achender@linux.vnet.ibm.com, adityakali@google.com
Subject: Re: [RFC PATCH V2 5/6] ext4: let ext4 maintian delayed extent trees
Date: Fri, 30 Sep 2011 10:08:50 +0800	[thread overview]
Message-ID: <CAGBYx2YjSJ5pks4gM+9M0c=okTRXvsjbiQCb78NkO8QP_Kx3Kw@mail.gmail.com> (raw)
In-Reply-To: <4E848124.6070500@tao.ma>

On Thu, Sep 29, 2011 at 10:31 PM, Tao Ma <tm@tao.ma> wrote:
> On 09/29/2011 01:08 PM, Yongqiang Yang wrote:
>> This patch let ext4 maintain delayed extent trees.
>>
>> Signed-off-by: Yongqiang Yang <xiaoqiangnk@gmail.com>
>> ---
>>  fs/ext4/ext4.h     |    1 +
>>  fs/ext4/extents.c  |    2 ++
>>  fs/ext4/indirect.c |    3 +++
>>  fs/ext4/inode.c    |   28 ++++++++++++++++++++++++++--
>>  fs/ext4/super.c    |   12 +++++++++++-
>>  5 files changed, 43 insertions(+), 3 deletions(-)
>>
>> diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
>> index d3c6b97..177ec0a 100644
>> --- a/fs/ext4/ext4.h
>> +++ b/fs/ext4/ext4.h
>> @@ -519,6 +519,7 @@ struct ext4_new_group_data {
>>  #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT                0x0020
>>       /* Don't normalize allocation size (used for fallocate) */
>>  #define EXT4_GET_BLOCKS_NO_NORMALIZE         0x0040
>> +#define EXT4_GET_BLOCKS_DEALLOC                      0x0080
>>
>>  /*
>>   * Flags used by ext4_free_blocks
>> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
>> index 9124cd2..bdbb984 100644
>> --- a/fs/ext4/extents.c
>> +++ b/fs/ext4/extents.c
>> @@ -3688,6 +3688,8 @@ void ext4_ext_truncate(struct inode *inode)
>>
>>       last_block = (inode->i_size + sb->s_blocksize - 1)
>>                       >> EXT4_BLOCK_SIZE_BITS(sb);
>> +     err = ext4_de_remove_space(inode, last_block,
>> +                                EXT_MAX_BLOCKS - last_block);
>>       err = ext4_ext_remove_space(inode, last_block);
>>
>>       /* In a multi-transaction truncate, we only make the final
>> diff --git a/fs/ext4/indirect.c b/fs/ext4/indirect.c
>> index 0962642..25cdb5b 100644
>> --- a/fs/ext4/indirect.c
>> +++ b/fs/ext4/indirect.c
>> @@ -22,6 +22,7 @@
>>
>>  #include <linux/module.h>
>>  #include "ext4_jbd2.h"
>> +#include "ext4_extents.h"
>>  #include "truncate.h"
>>
>>  #include <trace/events/ext4.h>
>> @@ -1383,6 +1384,8 @@ void ext4_ind_truncate(struct inode *inode)
>>       down_write(&ei->i_data_sem);
>>
>>       ext4_discard_preallocations(inode);
>> +     ext4_de_remove_space(inode, last_block,
>> +                          EXT_MAX_BLOCKS - last_block);
>>
>>       /*
>>        * The orphan list entry will now protect us from any crash which
>> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
>> index f86b149..0f9f108 100644
>> --- a/fs/ext4/inode.c
>> +++ b/fs/ext4/inode.c
>> @@ -442,7 +442,15 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode,
>>       up_read((&EXT4_I(inode)->i_data_sem));
>>
>>       if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
>> -             int ret = check_block_validity(inode, map);
>> +             int ret;
>> +             if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
>> +                     /* delayed alloc may be allocated by fallocate,
>> +                      * we need to handle delayed extent here.
>> +                      */
>> +                     down_write((&EXT4_I(inode)->i_data_sem));
>> +                     goto delayed_mapped;
>> +             }
>> +             ret = check_block_validity(inode, map);
> I am not quite sure of this. So do you mean when we write_begin the
> extent isn't allocated, while in the time of writepage, the extent is
> fallocted, right? If this is the case, where do we update the reserve_space?
> I mean in ext4_da_get_block_prep, we call ext4_da_reserve_space, and if
> there is no fallocate, we will call ext4_da_update_reserve_space in
> ext4_ext_handle_uninitialized_extents. So in your case, the 2nd
> ext4_da_update_reserve_space wouldn't be called. I am not sure whether
> there will be some problem or not.
Hi  Tao,

What if fallocated blocks have been allocated by direct I/O?  I placed
a BUG() here, it was indeed triggered by xfstests 127.

The comment is not clear, sorry for that.

Yongqiang.
>
> Thanks
> Tao
>>               if (ret != 0)
>>                       return ret;
>>       }
>> @@ -517,8 +525,18 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode,
>>                       (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE))
>>                       ext4_da_update_reserve_space(inode, retval, 1);
>>       }
>> -     if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
>> +     if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
>>               ext4_clear_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED);
>> +delayed_mapped:
>> +             if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
>> +                     int ret;
>> +                     /* delayed allocation blocks has been allocated */
>> +                     ret = ext4_de_remove_space(inode, map->m_lblk,
>> +                                                map->m_len);
>> +                     if (ret < 0)
>> +                             retval = ret;
>> +             }
>> +     }
>>
>>       up_write((&EXT4_I(inode)->i_data_sem));
>>       if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
>> @@ -1630,6 +1648,12 @@ static int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
>>                       /* not enough space to reserve */
>>                       return ret;
>>
>> +             down_write((&EXT4_I(inode)->i_data_sem));
>> +             ret = ext4_de_add_space(inode, map.m_lblk, map.m_len);
>> +             up_write((&EXT4_I(inode)->i_data_sem));
>> +             if (ret)
>> +                     return ret;
>> +
>>               map_bh(bh, inode->i_sb, invalid_block);
>>               set_buffer_new(bh);
>>               set_buffer_delay(bh);
>> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
>> index 247fcdd..a248551 100644
>> --- a/fs/ext4/super.c
>> +++ b/fs/ext4/super.c
>> @@ -49,6 +49,7 @@
>>  #include "xattr.h"
>>  #include "acl.h"
>>  #include "mballoc.h"
>> +#include "ext4_extents.h"
>>
>>  #define CREATE_TRACE_POINTS
>>  #include <trace/events/ext4.h>
>> @@ -967,6 +968,7 @@ void ext4_clear_inode(struct inode *inode)
>>       end_writeback(inode);
>>       dquot_drop(inode);
>>       ext4_discard_preallocations(inode);
>> +     ext4_de_remove_space(inode, 0, EXT_MAX_BLOCKS);
>>       if (EXT4_I(inode)->jinode) {
>>               jbd2_journal_release_jbd_inode(EXT4_JOURNAL(inode),
>>                                              EXT4_I(inode)->jinode);
>> @@ -4976,9 +4978,14 @@ static int __init ext4_init_fs(void)
>>               init_waitqueue_head(&ext4__ioend_wq[i]);
>>       }
>>
>> -     err = ext4_init_pageio();
>> +     err = ext4_init_de();
>>       if (err)
>>               return err;
>> +
>> +     err = ext4_init_pageio();
>> +     if (err)
>> +             goto out8;
>> +
>>       err = ext4_init_system_zone();
>>       if (err)
>>               goto out7;
>> @@ -5030,6 +5037,9 @@ out6:
>>       ext4_exit_system_zone();
>>  out7:
>>       ext4_exit_pageio();
>> +out8:
>> +     ext4_exit_de();
>> +
>>       return err;
>>  }
>>
>
>



-- 
Best Wishes
Yongqiang Yang
--
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

  reply	other threads:[~2011-09-30  2:08 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-09-29  5:08 [RFC V2] ext4: implementation of delayed extent tree Yongqiang Yang
2011-09-29  5:08 ` [RFC PATCH V2 1/6] ext4: add two structures supporting " Yongqiang Yang
2011-09-29  5:08 ` [RFC PATCH V2 2/6] ext4: add a delayed extents tree in inode info Yongqiang Yang
2011-09-29  6:56   ` Tao Ma
2011-09-29  7:09     ` Yongqiang Yang
2011-09-29  5:08 ` [RFC PATCH V2 3/6] ext4: add operations on delayed extent tree Yongqiang Yang
2011-09-29  8:05   ` Tao Ma
2011-09-29  8:36     ` Yongqiang Yang
2011-09-29  9:40       ` Tao Ma
2011-09-29 13:12         ` Yongqiang Yang
2011-09-29 14:10           ` Tao Ma
2011-09-29  5:08 ` [RFC PATCH V2 4/6] ext4: initialize " Yongqiang Yang
2011-09-29  5:08 ` [RFC PATCH V2 5/6] ext4: let ext4 maintian delayed extent trees Yongqiang Yang
2011-09-29  8:45   ` Amir Goldstein
2011-09-29 14:31   ` Tao Ma
2011-09-30  2:08     ` Yongqiang Yang [this message]
2011-09-29  5:08 ` [RFC PATCH V2 6/6] ext4: reimplement fiemap on delayed extent tree Yongqiang Yang
2011-09-29 15:28   ` Jeff liu
2011-09-30  0:54     ` Yongqiang Yang
2011-10-03 16:00   ` Lukas Czerner

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='CAGBYx2YjSJ5pks4gM+9M0c=okTRXvsjbiQCb78NkO8QP_Kx3Kw@mail.gmail.com' \
    --to=xiaoqiangnk@gmail.com \
    --cc=achender@linux.vnet.ibm.com \
    --cc=adityakali@google.com \
    --cc=jack@suse.cz \
    --cc=jeff.liu@oracle.com \
    --cc=linux-ext4@vger.kernel.org \
    --cc=tm@tao.ma \
    /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.