All of lore.kernel.org
 help / color / mirror / Atom feed
From: Gao Xiang <hsiangkao@linux.alibaba.com>
To: hongnanLi <hongnan.li@linux.alibaba.com>
Cc: Chao Yu <chao@kernel.org>,
	linux-erofs@lists.ozlabs.org, xiang@kernel.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] erofs: update ctx->pos for every emitted dirent
Date: Mon, 20 Jun 2022 20:19:44 +0800	[thread overview]
Message-ID: <YrBl4CMZUiO6YqNM@B-P7TQMD6M-0146.local> (raw)
In-Reply-To: <70fe93a3-7af5-b563-dcb7-3f7be81348ed@linux.alibaba.com>

Hi Hongnan,

On Mon, Jun 20, 2022 at 05:37:07PM +0800, hongnanLi wrote:
> on 2022/6/19 8:19, Chao Yu wrote:
> > On 2022/6/9 11:40, Hongnan Li wrote:
> > > erofs_readdir update ctx->pos after filling a batch of dentries
> > > and it may cause dir/files duplication for NFS readdirplus which
> > > depends on ctx->pos to fill dir correctly. So update ctx->pos for
> > > every emitted dirent in erofs_fill_dentries to fix it.
> > > 
> > > Fixes: 3e917cc305c6 ("erofs: make filesystem exportable")
> > > Signed-off-by: Hongnan Li <hongnan.li@linux.alibaba.com>
> > > ---
> > >   fs/erofs/dir.c | 20 ++++++++++----------
> > >   1 file changed, 10 insertions(+), 10 deletions(-)
> > > 
> > > diff --git a/fs/erofs/dir.c b/fs/erofs/dir.c
> > > index 18e59821c597..94ef5287237a 100644
> > > --- a/fs/erofs/dir.c
> > > +++ b/fs/erofs/dir.c
> > > @@ -22,10 +22,9 @@ static void debug_one_dentry(unsigned char
> > > d_type, const char *de_name,
> > >   }
> > >   static int erofs_fill_dentries(struct inode *dir, struct
> > > dir_context *ctx,
> > > -                   void *dentry_blk, unsigned int *ofs,
> > > +                   void *dentry_blk, struct erofs_dirent *de,
> > >                      unsigned int nameoff, unsigned int maxsize)
> > >   {
> > > -    struct erofs_dirent *de = dentry_blk + *ofs;
> > >       const struct erofs_dirent *end = dentry_blk + nameoff;
> > >       while (de < end) {
> > > @@ -59,9 +58,8 @@ static int erofs_fill_dentries(struct inode *dir,
> > > struct dir_context *ctx,
> > >               /* stopped by some reason */
> > >               return 1;
> > >           ++de;
> > > -        *ofs += sizeof(struct erofs_dirent);
> > > +        ctx->pos += sizeof(struct erofs_dirent);
> > >       }
> > > -    *ofs = maxsize;
> > >       return 0;
> > >   }
> > > @@ -95,7 +93,7 @@ static int erofs_readdir(struct file *f, struct
> > > dir_context *ctx)
> > >                     "invalid de[0].nameoff %u @ nid %llu",
> > >                     nameoff, EROFS_I(dir)->nid);
> > >               err = -EFSCORRUPTED;
> > > -            goto skip_this;
> > > +            break;
> > >           }
> > >           maxsize = min_t(unsigned int,
> > > @@ -106,17 +104,19 @@ static int erofs_readdir(struct file *f,
> > > struct dir_context *ctx)
> > >               initial = false;
> > >               ofs = roundup(ofs, sizeof(struct erofs_dirent));
> > > -            if (ofs >= nameoff)
> > > +            if (ofs >= nameoff) {
> > > +                ctx->pos = blknr_to_addr(i) + ofs;
> > >                   goto skip_this;
> > > +            }
> > >           }
> > > -        err = erofs_fill_dentries(dir, ctx, de, &ofs,
> > > -                      nameoff, maxsize);
> > > -skip_this:
> > >           ctx->pos = blknr_to_addr(i) + ofs;
> > 
> > Why updating ctx->pos before erofs_fill_dentries()?
> > 
> > Thanks,
> 
> It’s to ensure the ctx->pos is correct and up to date in
> erofs_fill_dentries() so that we can update ctx->pos instead of ofs for
> every emitted dirent.
> 

How about this, since blknr_to_addr(i) + maxsize should be the start of
the next dir block.

	if (initial) {
		ofs = roundup(ofs, sizeof(struct erofs_dirent));
		ctx->pos = blknr_to_addr(i) + ofs;
		if (ofs >= nameoff)
			goto skip_this;
	}
	err = erofs_fill_dentries(dir, ctx, de, (void *)de + ofs,
				  nameoff, maxsize);
	if (err)
		break;
	ctx->pos = blknr_to_addr(i) + maxsize;

Thanks,
Gao Xiang

WARNING: multiple messages have this Message-ID (diff)
From: Gao Xiang <hsiangkao@linux.alibaba.com>
To: hongnanLi <hongnan.li@linux.alibaba.com>
Cc: linux-erofs@lists.ozlabs.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] erofs: update ctx->pos for every emitted dirent
Date: Mon, 20 Jun 2022 20:19:44 +0800	[thread overview]
Message-ID: <YrBl4CMZUiO6YqNM@B-P7TQMD6M-0146.local> (raw)
In-Reply-To: <70fe93a3-7af5-b563-dcb7-3f7be81348ed@linux.alibaba.com>

Hi Hongnan,

On Mon, Jun 20, 2022 at 05:37:07PM +0800, hongnanLi wrote:
> on 2022/6/19 8:19, Chao Yu wrote:
> > On 2022/6/9 11:40, Hongnan Li wrote:
> > > erofs_readdir update ctx->pos after filling a batch of dentries
> > > and it may cause dir/files duplication for NFS readdirplus which
> > > depends on ctx->pos to fill dir correctly. So update ctx->pos for
> > > every emitted dirent in erofs_fill_dentries to fix it.
> > > 
> > > Fixes: 3e917cc305c6 ("erofs: make filesystem exportable")
> > > Signed-off-by: Hongnan Li <hongnan.li@linux.alibaba.com>
> > > ---
> > >   fs/erofs/dir.c | 20 ++++++++++----------
> > >   1 file changed, 10 insertions(+), 10 deletions(-)
> > > 
> > > diff --git a/fs/erofs/dir.c b/fs/erofs/dir.c
> > > index 18e59821c597..94ef5287237a 100644
> > > --- a/fs/erofs/dir.c
> > > +++ b/fs/erofs/dir.c
> > > @@ -22,10 +22,9 @@ static void debug_one_dentry(unsigned char
> > > d_type, const char *de_name,
> > >   }
> > >   static int erofs_fill_dentries(struct inode *dir, struct
> > > dir_context *ctx,
> > > -                   void *dentry_blk, unsigned int *ofs,
> > > +                   void *dentry_blk, struct erofs_dirent *de,
> > >                      unsigned int nameoff, unsigned int maxsize)
> > >   {
> > > -    struct erofs_dirent *de = dentry_blk + *ofs;
> > >       const struct erofs_dirent *end = dentry_blk + nameoff;
> > >       while (de < end) {
> > > @@ -59,9 +58,8 @@ static int erofs_fill_dentries(struct inode *dir,
> > > struct dir_context *ctx,
> > >               /* stopped by some reason */
> > >               return 1;
> > >           ++de;
> > > -        *ofs += sizeof(struct erofs_dirent);
> > > +        ctx->pos += sizeof(struct erofs_dirent);
> > >       }
> > > -    *ofs = maxsize;
> > >       return 0;
> > >   }
> > > @@ -95,7 +93,7 @@ static int erofs_readdir(struct file *f, struct
> > > dir_context *ctx)
> > >                     "invalid de[0].nameoff %u @ nid %llu",
> > >                     nameoff, EROFS_I(dir)->nid);
> > >               err = -EFSCORRUPTED;
> > > -            goto skip_this;
> > > +            break;
> > >           }
> > >           maxsize = min_t(unsigned int,
> > > @@ -106,17 +104,19 @@ static int erofs_readdir(struct file *f,
> > > struct dir_context *ctx)
> > >               initial = false;
> > >               ofs = roundup(ofs, sizeof(struct erofs_dirent));
> > > -            if (ofs >= nameoff)
> > > +            if (ofs >= nameoff) {
> > > +                ctx->pos = blknr_to_addr(i) + ofs;
> > >                   goto skip_this;
> > > +            }
> > >           }
> > > -        err = erofs_fill_dentries(dir, ctx, de, &ofs,
> > > -                      nameoff, maxsize);
> > > -skip_this:
> > >           ctx->pos = blknr_to_addr(i) + ofs;
> > 
> > Why updating ctx->pos before erofs_fill_dentries()?
> > 
> > Thanks,
> 
> It’s to ensure the ctx->pos is correct and up to date in
> erofs_fill_dentries() so that we can update ctx->pos instead of ofs for
> every emitted dirent.
> 

How about this, since blknr_to_addr(i) + maxsize should be the start of
the next dir block.

	if (initial) {
		ofs = roundup(ofs, sizeof(struct erofs_dirent));
		ctx->pos = blknr_to_addr(i) + ofs;
		if (ofs >= nameoff)
			goto skip_this;
	}
	err = erofs_fill_dentries(dir, ctx, de, (void *)de + ofs,
				  nameoff, maxsize);
	if (err)
		break;
	ctx->pos = blknr_to_addr(i) + maxsize;

Thanks,
Gao Xiang

  reply	other threads:[~2022-06-20 12:19 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-05-27  7:25 [PATCH] erofs: update ctx->pos for every emitted dirent Hongnan Li
2022-05-28 12:56 ` Gao Xiang
2022-05-28 12:56   ` Gao Xiang
2022-05-29  6:26 ` Chao Yu
2022-06-09  3:40 ` [PATCH v2] " Hongnan Li
2022-06-19  0:19   ` Chao Yu
2022-06-20  9:37     ` hongnanLi
2022-06-20 12:19       ` Gao Xiang [this message]
2022-06-20 12:19         ` Gao Xiang
2022-06-24  9:31         ` hongnanLi
2022-06-24  9:31           ` hongnanLi
2022-06-29  8:15 ` [PATCH v3] " Hongnan Li
2022-07-22  8:16   ` JeffleXu

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=YrBl4CMZUiO6YqNM@B-P7TQMD6M-0146.local \
    --to=hsiangkao@linux.alibaba.com \
    --cc=chao@kernel.org \
    --cc=hongnan.li@linux.alibaba.com \
    --cc=linux-erofs@lists.ozlabs.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=xiang@kernel.org \
    /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.