From: Sandeep K Sinha Subject: Re: can't get e4defrag to work Date: Mon, 26 Jan 2009 17:39:47 +0530 Message-ID: <37d33d830901260409y2cb01512u19646c477fa0e365@mail.gmail.com> References: <7a45233b0901210704w5038d8b8g5c657e3a4746f34d@mail.gmail.com> <49797056.7030607@rs.jp.nec.com> <87f94c370901230751m107afe09w6df76e258c57eae3@mail.gmail.com> <497D6843.50905@rs.jp.nec.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Greg Freemyer , Derkjan de Haan , Takashi Sato , linux-ext4@vger.kernel.org To: Akira Fujita Return-path: Received: from wa-out-1112.google.com ([209.85.146.183]:30188 "EHLO wa-out-1112.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751401AbZAZMJt (ORCPT ); Mon, 26 Jan 2009 07:09:49 -0500 Received: by wa-out-1112.google.com with SMTP id v33so495762wah.21 for ; Mon, 26 Jan 2009 04:09:48 -0800 (PST) In-Reply-To: <497D6843.50905@rs.jp.nec.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Mon, Jan 26, 2009 at 1:07 PM, Akira Fujita wrote: > Hi Greg, > > Greg Freemyer wrote: >> >> Will the new defrag patchset be based on the ioctl's that Ted >> suggested last month? > > Yes, but the new defrag that I will release corresponds to (3) of > Ted's suggestion. > (http://marc.info/?l=linux-ext4&m=122880166227883&w=3) > > For (1) and (2), probably it will be necessary to change > the allocation method and add new ioctls, so I will address them later. > > Have you already worked on these features? > Well, for option (2), I did try it on ext2/3 and it works fine for me. Same on ext4 was a bit problematic as a lot of interfaces are changing to the mballoc from balloc. And I have read lot of changes pipelined for block allocation in ext4. So, just waiting to get some information on that so that I can do the same for ext4 as well. This is a sample code that I tried and it works fine. It you wish, I can produce a formal version for the same. Here are the changes: #diff a/fs/ext2/ioctl.c b/fs/ext2/ioctl.c 19a20,35 > #define EXT2_BG_ALLOC 7 > > struct ext2fs_bg_range { > unsigned int bg_start; > unsigned int bg_end; > }; > > struct ext2fs_bg_ioc { > > struct ext2fs_bg_range bg_range[10]; > unsigned long fs_bg_range_count; > unsigned long blk_req_count; > }; > > extern ext2_fsblk_t fs_bg_ext2_new_blocks(struct inode *inode, struct ext2fs_bg_range bg[], > unsigned long fs_bg_range_count, unsigned long *count, int *errp); 25a42 > struct ext2fs_bg_ioc bg_ioc; 27,28c44,45 < int ret; > int ret ,err; < > case EXT2_BG_ALLOC: > copy_from_user((struct ext2fs_bg_ioc *)&bg_ioc, (const void __user *)arg, sizeof(struct ext2fs_bg_ioc)); > ret = fs_bg_ext2_new_blocks(inode, bg_ioc.bg_range, bg_ioc.fs_bg_range_count, > &bg_ioc.blk_req_count, &err); > > break; 162a198,200 > case EXT2_IOC32_BG_ALLOC_BLOCKS: > cmd = EXT2_IOC_BG_ALLOC_BLOCKS; > break; diff a/fs/ext2/balloc.c b/fs/ext2/balloc.c 16a17 > #include 1543a1545,1732 > struct ext2fs_bg_range { > unsigned int bg_start; > unsigned int bg_end; > }; > > ext2_fsblk_t fs_bg_ext2_new_blocks(struct inode *inode, struct ext2fs_bg_range bg_range[], > unsigned int fs_bg_range_count, unsigned long *count, int *errp) > { > struct buffer_head *bitmap_bh = NULL; > struct buffer_head *gdp_bh; > int group_no = 0; > int goal_group = 0; > int fs_bg; > int bg_start = 0, bg_end = 0; > ext2_grpblk_t grp_alloc_blk; /* blockgroup-relative allocated block*/ > ext2_fsblk_t ret_block; /* filesyetem-wide allocated block */ > ext2_fsblk_t goal; /* goal block for allocation */ > int bgi; /* blockgroup iteration index */ > int performed_allocation = 0; > ext2_grpblk_t free_blocks; /* number of free blocks in a group */ > struct super_block *sb; > struct ext2_group_desc *gdp; > struct ext2_super_block *es; > struct ext2_sb_info *sbi; > struct ext2_reserve_window_node *my_rsv = NULL; > unsigned short windowsz = 0; > unsigned long ngroups; > unsigned long num = *count; > > sb = inode->i_sb; > *errp = -ENOSPC; > if (!sb) { > printk("ext2_new_blocks: nonexistent device"); > return 0; > } > > /* > * Check quota for allocation of this block. > */ > if (DQUOT_ALLOC_BLOCK(inode, num)) { > *errp = -EDQUOT; > return 0; > } > > sbi = EXT2_SB(sb); > es = EXT2_SB(sb)->s_es; > ext2_debug("goal=%lu.\n", goal); > > if (!ext2_has_free_blocks(sbi)) { > *errp = -ENOSPC; > goto out; > } > > goal = ext2_group_first_block_no(sb,EXT2_I(inode)->i_block_group); > > for(fs_bg = 0; fs_bg < fs_bg_range_count; fs_bg++) > { > bg_start = bg_range[fs_bg].bg_start; > bg_end = bg_range[fs_bg].bg_end; > > group_no = bg_start; > goal_group = bg_start; > > retry_alloc: > > ngroups = (bg_end - bg_start) + 1; > smp_rmb(); > > /* > * Now search the rest of the groups. We assume that > * group_no and gdp correctly point to the last group visited. > */ > > for (bgi = 0; bgi < ngroups; bgi++, group_no++) { > if (group_no > bg_end) /* ngroups */ > group_no = bg_start; /* previous value 0 */ > gdp = ext2_get_group_desc(sb, group_no, &gdp_bh); > if (!gdp) > goto io_error; > > free_blocks = le16_to_cpu(gdp->bg_free_blocks_count); > /* > * skip this group if the number of > * free blocks is less than half of the reservation > * window size. > */ > if (free_blocks <= (windowsz/2)) > continue; > > brelse(bitmap_bh); > bitmap_bh = read_block_bitmap(sb, group_no); > if (!bitmap_bh) > goto io_error; > /* > * try to allocate block(s) from this group, without a goal(-1). > */ > grp_alloc_blk = ext2_try_to_allocate_with_rsv(sb, group_no, > bitmap_bh, -1, my_rsv, &num); > if (grp_alloc_blk >= 0) > goto allocated; > } > } > /* > * We may end up a bogus ealier ENOSPC error due to > * filesystem is "full" of reservations, but > * there maybe indeed free blocks avaliable on disk > * In this case, we just forget about the reservations > * just do block allocation as without reservations. > */ > if (my_rsv) { > my_rsv = NULL; > windowsz = 0; > group_no = goal_group; > goto retry_alloc; > } > /* No space left on the device */ > *errp = -ENOSPC; > goto out; > > allocated: > > ext2_debug("using block group %d(%d)\n", > group_no, gdp->bg_free_blocks_count); > > ret_block = grp_alloc_blk + ext2_group_first_block_no(sb, group_no); > > if (in_range(le32_to_cpu(gdp->bg_block_bitmap), ret_block, num) || > in_range(le32_to_cpu(gdp->bg_inode_bitmap), ret_block, num) || > in_range(ret_block, le32_to_cpu(gdp->bg_inode_table), > EXT2_SB(sb)->s_itb_per_group) || > in_range(ret_block + num - 1, le32_to_cpu(gdp->bg_inode_table), > EXT2_SB(sb)->s_itb_per_group)) { > ext2_error(sb, "ext2_new_blocks", > "Allocating block in system zone - " > "blocks from "E2FSBLK", length %lu", > ret_block, num); > /* > * ext2_try_to_allocate marked the blocks we allocated as in > * use. So we may want to selectively mark some of the blocks > * as free > */ > goto retry_alloc; > } > > performed_allocation = 1; > > if (ret_block + num - 1 >= le32_to_cpu(es->s_blocks_count)) { > ext2_error(sb, "ext2_new_blocks", > "block("E2FSBLK") >= blocks count(%d) - " > "block_group = %d, es == %p ", ret_block, > le32_to_cpu(es->s_blocks_count), group_no, es); > goto out; > } > > group_adjust_blocks(sb, group_no, gdp, gdp_bh, -num); > percpu_counter_sub(&sbi->s_freeblocks_counter, num); > > mark_buffer_dirty(bitmap_bh); > if (sb->s_flags & MS_SYNCHRONOUS) > sync_dirty_buffer(bitmap_bh); > > *errp = 0; > brelse(bitmap_bh); > DQUOT_FREE_BLOCK(inode, *count-num); > *count = num; > return ret_block; > io_error: > *errp = -EIO; > out: > /* > * Undo the block allocation > */ > if (!performed_allocation) > DQUOT_FREE_BLOCK(inode, *count); > brelse(bitmap_bh); > return 0; > } > EXPORT_SYMBOL(fs_bg_ext2_new_blocks); This works fine. I would surely like to see a similar ABI in the linux kernel soon. > Regards, > Akira Fujita > > --Separator@a-fujita@rs.jp.nec.com: > greg.freemyer@gmail.com > haanjdj@gmail.com > sandeepksinha@gmail.com > linux-ext4@vger.kernel.org > -- Regards, Sandeep. "To learn is to change. Education is a process that changes the learner."