From: Jan Kara Subject: Re: [PATCH V5 3/5]ext4: quota handling for delayed allocation Date: Tue, 13 Jan 2009 16:09:48 +0100 Message-ID: <20090113150948.GD10064@duck.suse.cz> References: <1231216855.9267.27.camel@mingming-laptop> <20090106093528.GE10705@duck.suse.cz> <20090106093936.GF10705@duck.suse.cz> <1231807629.6752.30.camel@mingming-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Andrew Morton , tytso , linux-ext4 , linux-fsdevel To: Mingming Cao Return-path: Received: from styx.suse.cz ([82.119.242.94]:44274 "EHLO mail.suse.cz" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751432AbZAMPJu (ORCPT ); Tue, 13 Jan 2009 10:09:50 -0500 Content-Disposition: inline In-Reply-To: <1231807629.6752.30.camel@mingming-laptop> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Mon 12-01-09 16:47:09, Mingming Cao wrote: > =E5=9C=A8 2009-01-06=E4=BA=8C=E7=9A=84 10:39 +0100=EF=BC=8CJan Kara=E5= =86=99=E9=81=93=EF=BC=9A > > On Tue 06-01-09 10:35:29, Jan Kara wrote: > > > On Mon 05-01-09 20:40:55, Mingming Cao wrote: > > > > ext4: quota reservation for delayed allocation > > > >=20 > > > > Uses quota reservation/claim/release to handle quota properly f= or delayed > > > > allocation in the three steps: 1) quotas are reserved when data= being copied > > > > to cache when block allocation is defered 2) when new blocks ar= e allocated. > > > > reserved quotas are converted to the real allocated quota, 2) o= ver-booked > > > > quotas for metadata blocks are released back. > > > >=20 > > > > Signed-off-by: Mingming Cao > > > The patch looks fine as far as I can tell :). But I'm not to fa= miliar > > > with the code so I won't add my Acked-by... > > On a second thought: > >=20 > > > > +unsigned long long ext4_get_reserved_space(struct inode *inode= ) > > > > +{ > > > > + unsigned long long total; > > > > + > > > > + spin_lock(&EXT4_I(inode)->i_block_reservation_lock); > > > > + total =3D EXT4_I(inode)->i_reserved_data_blocks + > > > > + EXT4_I(inode)->i_reserved_meta_blocks; > > > > + spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); > > > > + > > > > + return total; > > > > +} > > Shouldn't we return here (total << inode->i_blksize)? Quota expec= ts > > bytes as output... > >=20 >=20 > This is updated ext4 part to return number of bytes reserved as outpu= t. OK, it looks fine now. Acked-by: Jan Kara Honza >=20 > Thanks, > Mingming > ext4: quota reservation for delayed allocation >=20 > Uses quota reservation/claim/release to handle quota properly for del= ayed > allocation in the three steps: 1) quotas are reserved when data being= copied > to cache when block allocation is defered 2) when new blocks are allo= cated. > reserved quotas are converted to the real allocated quota, 2) over-bo= oked > quotas for metadata blocks are released back. >=20 >=20 > Signed-off-by: Mingming Cao > --- > fs/ext4/ext4.h | 1 + > fs/ext4/inode.c | 36 +++++++++++++++++++++++++++++++++--- > fs/ext4/mballoc.c | 44 ++++++++++++++++++++++++++-----------------= - > fs/ext4/super.c | 4 ++++ > 4 files changed, 64 insertions(+), 21 deletions(-) >=20 > Index: linux-2.6.29-rc1/fs/ext4/inode.c > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > --- linux-2.6.29-rc1.orig/fs/ext4/inode.c 2009-01-10 15:43:05.0000000= 00 -0800 > +++ linux-2.6.29-rc1/fs/ext4/inode.c 2009-01-12 16:31:34.000000000 -0= 800 > @@ -973,6 +973,17 @@ out: > return err; > } > =20 > +qsize_t ext4_get_reserved_space(struct inode *inode) > +{ > + unsigned long long total; > + > + spin_lock(&EXT4_I(inode)->i_block_reservation_lock); > + total =3D EXT4_I(inode)->i_reserved_data_blocks + > + EXT4_I(inode)->i_reserved_meta_blocks; > + spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); > + > + return (qsize_t)total << inode->i_blkbits; > +} > /* > * Calculate the number of metadata blocks need to reserve > * to allocate @blocks for non extent file based file > @@ -1034,8 +1045,14 @@ static void ext4_da_update_reserve_space > /* update per-inode reservations */ > BUG_ON(used > EXT4_I(inode)->i_reserved_data_blocks); > EXT4_I(inode)->i_reserved_data_blocks -=3D used; > - > spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); > + > + /* > + * free those over-booking quota for metadata blocks > + */ > + > + if (mdb_free) > + vfs_dq_release_reservation_block(inode, mdb_free); > } > =20 > /* > @@ -1547,8 +1564,8 @@ static int ext4_journalled_write_end(str > static int ext4_da_reserve_space(struct inode *inode, int nrblocks) > { > int retries =3D 0; > - struct ext4_sb_info *sbi =3D EXT4_SB(inode->i_sb); > - unsigned long md_needed, mdblocks, total =3D 0; > + struct ext4_sb_info *sbi =3D EXT4_SB(inode->i_sb); > + unsigned long md_needed, mdblocks, total =3D 0; > =20 > /* > * recalculate the amount of metadata blocks to reserve > @@ -1564,12 +1581,23 @@ repeat: > md_needed =3D mdblocks - EXT4_I(inode)->i_reserved_meta_blocks; > total =3D md_needed + nrblocks; > =20 > + /* > + * Make quota reservation here to prevent quota overflow > + * later. Real quota accounting is done at pages writeout > + * time. > + */ > + if (vfs_dq_reserve_block(inode, total)) { > + spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); > + return -EDQUOT; > + } > + > if (ext4_claim_free_blocks(sbi, total)) { > spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); > if (ext4_should_retry_alloc(inode->i_sb, &retries)) { > yield(); > goto repeat; > } > + vfs_dq_release_reservation_block(inode, total); > return -ENOSPC; > } > EXT4_I(inode)->i_reserved_data_blocks +=3D nrblocks; > @@ -1623,6 +1651,8 @@ static void ext4_da_release_space(struct > BUG_ON(mdb > EXT4_I(inode)->i_reserved_meta_blocks); > EXT4_I(inode)->i_reserved_meta_blocks =3D mdb; > spin_unlock(&EXT4_I(inode)->i_block_reservation_lock); > + > + vfs_dq_release_reservation_block(inode, release); > } > =20 > static void ext4_da_page_release_reservation(struct page *page, > Index: linux-2.6.29-rc1/fs/ext4/super.c > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > --- linux-2.6.29-rc1.orig/fs/ext4/super.c 2009-01-10 15:43:05.0000000= 00 -0800 > +++ linux-2.6.29-rc1/fs/ext4/super.c 2009-01-12 16:28:44.000000000 -0= 800 > @@ -945,6 +945,10 @@ static struct dquot_operations ext4_quot > .initialize =3D ext4_dquot_initialize, > .drop =3D ext4_dquot_drop, > .alloc_space =3D dquot_alloc_space, > + .reserve_space =3D dquot_reserve_space, > + .claim_space =3D dquot_claim_space, > + .release_rsv =3D dquot_release_reserved_space, > + .get_reserved_space =3D ext4_get_reserved_space, > .alloc_inode =3D dquot_alloc_inode, > .free_space =3D dquot_free_space, > .free_inode =3D dquot_free_inode, > Index: linux-2.6.29-rc1/fs/ext4/mballoc.c > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > --- linux-2.6.29-rc1.orig/fs/ext4/mballoc.c 2009-01-10 15:43:05.00000= 0000 -0800 > +++ linux-2.6.29-rc1/fs/ext4/mballoc.c 2009-01-12 16:28:44.000000000 = -0800 > @@ -3086,9 +3086,12 @@ ext4_mb_mark_diskspace_used(struct ext4_ > if (!(ac->ac_flags & EXT4_MB_DELALLOC_RESERVED)) > /* release all the reserved blocks if non delalloc */ > percpu_counter_sub(&sbi->s_dirtyblocks_counter, reserv_blks); > - else > + else { > percpu_counter_sub(&sbi->s_dirtyblocks_counter, > ac->ac_b_ex.fe_len); > + /* convert reserved quota blocks to real quota blocks */ > + vfs_dq_claim_block(ac->ac_inode, ac->ac_b_ex.fe_len); > + } > =20 > if (sbi->s_log_groups_per_flex) { > ext4_group_t flex_group =3D ext4_flex_group(sbi, > @@ -4533,7 +4536,7 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t > struct ext4_sb_info *sbi; > struct super_block *sb; > ext4_fsblk_t block =3D 0; > - unsigned int inquota; > + unsigned int inquota =3D 0; > unsigned int reserv_blks =3D 0; > =20 > sb =3D ar->inode->i_sb; > @@ -4551,9 +4554,17 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t > (unsigned long long) ar->pleft, > (unsigned long long) ar->pright); > =20 > - if (!EXT4_I(ar->inode)->i_delalloc_reserved_flag) { > - /* > - * With delalloc we already reserved the blocks > + /* > + * For delayed allocation, we could skip the ENOSPC and > + * EDQUOT check, as blocks and quotas have been already > + * reserved when data being copied into pagecache. > + */ > + if (EXT4_I(ar->inode)->i_delalloc_reserved_flag) > + ar->flags |=3D EXT4_MB_DELALLOC_RESERVED; > + else { > + /* Without delayed allocation we need to verify > + * there is enough free blocks to do block allocation > + * and verify allocation doesn't exceed the quota limits. > */ > while (ar->len && ext4_claim_free_blocks(sbi, ar->len)) { > /* let others to free the space */ > @@ -4565,19 +4576,16 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t > return 0; > } > reserv_blks =3D ar->len; > + while (ar->len && DQUOT_ALLOC_BLOCK(ar->inode, ar->len)) { > + ar->flags |=3D EXT4_MB_HINT_NOPREALLOC; > + ar->len--; > + } > + inquota =3D ar->len; > + if (ar->len =3D=3D 0) { > + *errp =3D -EDQUOT; > + goto out3; > + } > } > - while (ar->len && DQUOT_ALLOC_BLOCK(ar->inode, ar->len)) { > - ar->flags |=3D EXT4_MB_HINT_NOPREALLOC; > - ar->len--; > - } > - if (ar->len =3D=3D 0) { > - *errp =3D -EDQUOT; > - goto out3; > - } > - inquota =3D ar->len; > - > - if (EXT4_I(ar->inode)->i_delalloc_reserved_flag) > - ar->flags |=3D EXT4_MB_DELALLOC_RESERVED; > =20 > ac =3D kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS); > if (!ac) { > @@ -4643,7 +4651,7 @@ repeat: > out2: > kmem_cache_free(ext4_ac_cachep, ac); > out1: > - if (ar->len < inquota) > + if (inquota && ar->len < inquota) > DQUOT_FREE_BLOCK(ar->inode, inquota - ar->len); > out3: > if (!ar->len) { > Index: linux-2.6.29-rc1/fs/ext4/ext4.h > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > --- linux-2.6.29-rc1.orig/fs/ext4/ext4.h 2009-01-10 15:43:05.00000000= 0 -0800 > +++ linux-2.6.29-rc1/fs/ext4/ext4.h 2009-01-12 16:28:44.000000000 -08= 00 > @@ -1098,6 +1098,7 @@ extern int ext4_chunk_trans_blocks(struc > extern int ext4_block_truncate_page(handle_t *handle, > struct address_space *mapping, loff_t from); > extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct page= *page); > +extern qsize_t ext4_get_reserved_space(struct inode *inode); > =20 > /* ioctl.c */ > extern long ext4_ioctl(struct file *, unsigned int, unsigned long); --=20 Jan Kara SUSE Labs, CR -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html