Linux-EROFS Archive on lore.kernel.org
 help / color / Atom feed
From: Gao Xiang <gaoxiang25@huawei.com>
To: Pratik Shinde <pratikshinde320@gmail.com>
Cc: miaoxie@huawei.com, linux-erofs@lists.ozlabs.org
Subject: Re: [RFCv3] erofs-utils: on-disk extent format for blocks
Date: Thu, 2 Jan 2020 21:33:19 +0800
Message-ID: <20200102133319.GA85503@architecture4> (raw)
In-Reply-To: <CAGu0czRH7KYsxYFbQy3_rMeNCV6Fsg__PRuovdp672ZCu2vjkA@mail.gmail.com>

On Thu, Jan 02, 2020 at 04:32:34PM +0530, Pratik Shinde wrote:
> Hi Gao,
> 
> You are correct. The macro logic can be simplified. I will do that.
> I will work on the kernel part of this change & do some testing on it.
> I will keep you posted about the change and relevant tests I am running on
> it.
> 
> --Pratik.
>

Okay, thanks for your effort. :)

Thanks,
Gao Xiang

> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> On Thu, Jan 2, 2020, 4:17 PM Gao Xiang <gaoxiang25@huawei.com> wrote:
> 
> > Hi Pratik,
> >
> > On Thu, Jan 02, 2020 at 03:17:32PM +0530, Pratik Shinde wrote:
> > > 1)Moved on-disk structures to erofs_fs.h
> > > 2)Some naming changes.
> > >
> > > I think we can keep 'IS_HOLE()' macro, otherwise the code
> > > does not look clean(if used directly/without macro).Its getting
> > > used only in inode.c so it can be kept there.
> > > what do you think ?
> >
> > What I'm little concerned is the relationship between
> > the name of IS_HOLE and its implementation...
> >
> > In other words, why
> >  roundup(start, EROFS_BLKSIZ) == start &&
> >  roundup(end, EROFS_BLKSIZ) == end &&
> >  (end - start) % EROFS_BLKSIZ == 0
> > should be an erofs hole...
> >
> > But that is minor, I reserve my opinion on this for now...
> >
> > This patch generally looks good to me, yet I haven't
> > played with it till now.
> >
> > Could you implement a workable RFC patch for kernel side
> > as well? Since I'm still busying in XZ library and other
> > convert patches for 5.6...
> >
> > I'd like to hear if some other opinions from Chao and Guifu.
> > Since it enhances the on-disk format, we need to think it
> > over (especially its future expandability).
> >
> >
> > To Chao and Guifu,
> > Could you have some extra time looking at this stuff as well?
> >
> >
> > Thanks,
> > Gao Xiang
> >
> > >
> > > Signed-off-by: Pratik Shinde <pratikshinde320@gmail.com>
> > > ---
> > >  include/erofs/internal.h |   9 ++-
> > >  include/erofs_fs.h       |  11 ++++
> > >  lib/inode.c              | 153
> > ++++++++++++++++++++++++++++++++++++++++-------
> > >  3 files changed, 150 insertions(+), 23 deletions(-)
> > >
> > > diff --git a/include/erofs/internal.h b/include/erofs/internal.h
> > > index e13adda..2d7466b 100644
> > > --- a/include/erofs/internal.h
> > > +++ b/include/erofs/internal.h
> > > @@ -63,7 +63,7 @@ struct erofs_sb_info {
> > >  extern struct erofs_sb_info sbi;
> > >
> > >  struct erofs_inode {
> > > -     struct list_head i_hash, i_subdirs, i_xattrs;
> > > +     struct list_head i_hash, i_subdirs, i_xattrs, i_sparse_extents;
> > >
> > >       unsigned int i_count;
> > >       struct erofs_inode *i_parent;
> > > @@ -93,6 +93,7 @@ struct erofs_inode {
> > >
> > >       unsigned int xattr_isize;
> > >       unsigned int extent_isize;
> > > +     unsigned int sparse_extent_isize;
> > >
> > >       erofs_nid_t nid;
> > >       struct erofs_buffer_head *bh;
> > > @@ -139,5 +140,11 @@ static inline const char *erofs_strerror(int err)
> > >       return msg;
> > >  }
> > >
> > > +struct erofs_sparse_extent_node {
> > > +     struct list_head next;
> > > +     erofs_blk_t lblk;
> > > +     erofs_blk_t pblk;
> > > +     u32 len;
> > > +};
> > >  #endif
> > >
> > > diff --git a/include/erofs_fs.h b/include/erofs_fs.h
> > > index bcc4f0c..a63e1c6 100644
> > > --- a/include/erofs_fs.h
> > > +++ b/include/erofs_fs.h
> > > @@ -321,5 +321,16 @@ static inline void
> > erofs_check_ondisk_layout_definitions(void)
> > >                    Z_EROFS_VLE_CLUSTER_TYPE_MAX - 1);
> > >  }
> > >
> > > +/* on-disk sparse extent format */
> > > +struct erofs_sparse_extent {
> > > +     __le32 ee_lblk;
> > > +     __le32 ee_pblk;
> > > +     __le32 ee_len;
> > > +};
> > > +
> > > +struct erofs_sparse_extent_iheader {
> > > +     u32 count;
> > > +};
> > > +
> > >  #endif
> > >
> > > diff --git a/lib/inode.c b/lib/inode.c
> > > index 0e19b11..da20599 100644
> > > --- a/lib/inode.c
> > > +++ b/lib/inode.c
> > > @@ -38,6 +38,97 @@ static unsigned char erofs_type_by_mode[S_IFMT >>
> > S_SHIFT] = {
> > >
> > >  struct list_head inode_hashtable[NR_INODE_HASHTABLE];
> > >
> > > +
> > > +#define IS_HOLE(start, end) (roundup(start, EROFS_BLKSIZ) == start &&
> >       \
> > > +                          roundup(end, EROFS_BLKSIZ) == end &&       \
> > > +                         (end - start) % EROFS_BLKSIZ == 0)
> > > +
> > > +/**
> > > +   read extents of the given file.
> > > +   record the data extents and link them into a chain.
> > > +   exclude holes present in file.
> > > + */
> > > +unsigned int erofs_read_sparse_extents(int fd, struct list_head
> > *extents)
> > > +{
> > > +     erofs_blk_t startblk, endblk, datablk;
> > > +     unsigned int nholes = 0;
> > > +     erofs_off_t data, hole, len = 0, last_data;
> > > +     struct erofs_sparse_extent_node *e_data;
> > > +
> > > +     len = lseek(fd, 0, SEEK_END);
> > > +     if (len < 0)
> > > +             return -errno;
> > > +     if (lseek(fd, 0, SEEK_SET) == -1)
> > > +             return -errno;
> > > +     data = 0;
> > > +     last_data = 0;
> > > +     while (data < len) {
> > > +             hole = lseek(fd, data, SEEK_HOLE);
> > > +             if (hole == len)
> > > +                     break;
> > > +             data = lseek(fd, hole, SEEK_DATA);
> > > +             if (data < 0 || hole > data)
> > > +                     return -EINVAL;
> > > +             if (IS_HOLE(hole, data)) {
> > > +                     startblk = erofs_blknr(hole);
> > > +                     datablk = erofs_blknr(last_data);
> > > +                     endblk = erofs_blknr(data);
> > > +                     last_data = data;
> > > +                     e_data = malloc(sizeof(
> > > +                                      struct erofs_sparse_extent_node));
> > > +                     if (e_data == NULL)
> > > +                             return -ENOMEM;
> > > +                     e_data->lblk = datablk;
> > > +                     e_data->len = (startblk - datablk);
> > > +                     list_add_tail(&e_data->next, extents);
> > > +                     nholes += (endblk - startblk);
> > > +             }
> > > +     }
> > > +     /* rounddown to exclude tail-end data */
> > > +     if (last_data < len && (len - last_data) >= EROFS_BLKSIZ) {
> > > +             e_data = malloc(sizeof(struct erofs_sparse_extent_node));
> > > +             if (e_data == NULL)
> > > +                     return -ENOMEM;
> > > +             startblk = erofs_blknr(last_data);
> > > +             e_data->lblk = startblk;
> > > +             e_data->len = erofs_blknr(rounddown((len - last_data),
> > > +                                       EROFS_BLKSIZ));
> > > +             list_add_tail(&e_data->next, extents);
> > > +     }
> > > +     return nholes;
> > > +}
> > > +
> > > +int erofs_write_sparse_extents(struct erofs_inode *inode, erofs_off_t
> > off)
> > > +{
> > > +     struct erofs_sparse_extent_node *e_node;
> > > +     struct erofs_sparse_extent_iheader *header;
> > > +     char *buf;
> > > +     unsigned int p = 0;
> > > +     int ret;
> > > +
> > > +     buf = malloc(inode->sparse_extent_isize);
> > > +     if (buf == NULL)
> > > +             return -ENOMEM;
> > > +     header = (struct erofs_sparse_extent_iheader *) buf;
> > > +     header->count = 0;
> > > +     p += sizeof(struct erofs_sparse_extent_iheader);
> > > +     list_for_each_entry(e_node, &inode->i_sparse_extents, next) {
> > > +             const struct erofs_sparse_extent ee = {
> > > +                     .ee_lblk = cpu_to_le32(e_node->lblk),
> > > +                     .ee_pblk = cpu_to_le32(e_node->pblk),
> > > +                     .ee_len  = cpu_to_le32(e_node->len)
> > > +             };
> > > +             memcpy(buf + p, &ee, sizeof(struct erofs_sparse_extent));
> > > +             p += sizeof(struct erofs_sparse_extent);
> > > +             header->count++;
> > > +             list_del(&e_node->next);
> > > +             free(e_node);
> > > +     }
> > > +     ret = dev_write(buf, off, inode->sparse_extent_isize);
> > > +     free(buf);
> > > +     return ret;
> > > +}
> > > +
> > >  void erofs_inode_manager_init(void)
> > >  {
> > >       unsigned int i;
> > > @@ -304,8 +395,9 @@ static bool erofs_file_is_compressible(struct
> > erofs_inode *inode)
> > >
> > >  int erofs_write_file(struct erofs_inode *inode)
> > >  {
> > > -     unsigned int nblocks, i;
> > > +     unsigned int nblocks, i, j, nholes;
> > >       int ret, fd;
> > > +     struct erofs_sparse_extent_node *e_node;
> > >
> > >       if (!inode->i_size) {
> > >               inode->datalayout = EROFS_INODE_FLAT_PLAIN;
> > > @@ -322,31 +414,42 @@ int erofs_write_file(struct erofs_inode *inode)
> > >       /* fallback to all data uncompressed */
> > >       inode->datalayout = EROFS_INODE_FLAT_INLINE;
> > >       nblocks = inode->i_size / EROFS_BLKSIZ;
> > > -
> > > -     ret = __allocate_inode_bh_data(inode, nblocks);
> > > -     if (ret)
> > > -             return ret;
> > > -
> > >       fd = open(inode->i_srcpath, O_RDONLY | O_BINARY);
> > >       if (fd < 0)
> > >               return -errno;
> > > -
> > > -     for (i = 0; i < nblocks; ++i) {
> > > -             char buf[EROFS_BLKSIZ];
> > > -
> > > -             ret = read(fd, buf, EROFS_BLKSIZ);
> > > -             if (ret != EROFS_BLKSIZ) {
> > > -                     if (ret < 0)
> > > +     nholes = erofs_read_sparse_extents(fd, &inode->i_sparse_extents);
> > > +     if (nholes < 0) {
> > > +             close(fd);
> > > +             return nholes;
> > > +     }
> > > +     ret = __allocate_inode_bh_data(inode, nblocks - nholes);
> > > +     if (ret) {
> > > +             close(fd);
> > > +             return ret;
> > > +     }
> > > +     i = inode->u.i_blkaddr;
> > > +     inode->sparse_extent_isize = sizeof(struct
> > erofs_sparse_extent_iheader);
> > > +     list_for_each_entry(e_node, &inode->i_sparse_extents, next) {
> > > +             inode->sparse_extent_isize += sizeof(struct
> > erofs_sparse_extent);
> > > +             e_node->pblk = i;
> > > +             ret = lseek(fd, blknr_to_addr(e_node->lblk), SEEK_SET);
> > > +             if (ret < 0)
> > > +                     goto fail;
> > > +             for (j = 0; j < e_node->len; j++) {
> > > +                     char buf[EROFS_BLKSIZ];
> > > +                     ret = read(fd, buf, EROFS_BLKSIZ);
> > > +                     if (ret != EROFS_BLKSIZ) {
> > > +                             if (ret < 0)
> > > +                                     goto fail;
> > > +                             close(fd);
> > > +                             return -EAGAIN;
> > > +                     }
> > > +                     ret = blk_write(buf, e_node->pblk + j, 1);
> > > +                     if (ret)
> > >                               goto fail;
> > > -                     close(fd);
> > > -                     return -EAGAIN;
> > >               }
> > > -
> > > -             ret = blk_write(buf, inode->u.i_blkaddr + i, 1);
> > > -             if (ret)
> > > -                     goto fail;
> > > +             i += e_node->len;
> > >       }
> > > -
> > >       /* read the tail-end data */
> > >       inode->idata_size = inode->i_size % EROFS_BLKSIZ;
> > >       if (inode->idata_size) {
> > > @@ -479,8 +582,14 @@ static bool erofs_bh_flush_write_inode(struct
> > erofs_buffer_head *bh)
> > >               if (ret)
> > >                       return false;
> > >               free(inode->compressmeta);
> > > +             off += inode->extent_isize;
> > >       }
> > >
> > > +     if (inode->sparse_extent_isize) {
> > > +             ret = erofs_write_sparse_extents(inode, off);
> > > +             if (ret)
> > > +                     return false;
> > > +     }
> > >       inode->bh = NULL;
> > >       erofs_iput(inode);
> > >       return erofs_bh_flush_generic_end(bh);
> > > @@ -737,10 +846,11 @@ struct erofs_inode *erofs_new_inode(void)
> > >
> > >       init_list_head(&inode->i_subdirs);
> > >       init_list_head(&inode->i_xattrs);
> > > +     init_list_head(&inode->i_sparse_extents);
> > >
> > >       inode->idata_size = 0;
> > >       inode->xattr_isize = 0;
> > > -     inode->extent_isize = 0;
> > > +     inode->sparse_extent_isize = 0;
> > >
> > >       inode->bh = inode->bh_inline = inode->bh_data = NULL;
> > >       inode->idata = NULL;
> > > @@ -961,4 +1071,3 @@ struct erofs_inode
> > *erofs_mkfs_build_tree_from_path(struct erofs_inode *parent,
> > >
> > >       return erofs_mkfs_build_tree(inode);
> > >  }
> > > -
> > > --
> > > 2.9.3
> > >
> >

      reply index

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-02  9:47 Pratik Shinde
2020-01-02 10:47 ` Gao Xiang
2020-01-02 11:02   ` Pratik Shinde
2020-01-02 13:33     ` Gao Xiang [this message]

Reply instructions:

You may reply publically 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=20200102133319.GA85503@architecture4 \
    --to=gaoxiang25@huawei.com \
    --cc=linux-erofs@lists.ozlabs.org \
    --cc=miaoxie@huawei.com \
    --cc=pratikshinde320@gmail.com \
    /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

Linux-EROFS Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-erofs/0 linux-erofs/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-erofs linux-erofs/ https://lore.kernel.org/linux-erofs \
		linux-erofs@lists.ozlabs.org linux-erofs@ozlabs.org
	public-inbox-index linux-erofs

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.ozlabs.lists.linux-erofs


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git