[v10,23/25] xfs: Add delay ready attr remove routines
diff mbox series

Message ID 20200625233018.14585-24-allison.henderson@oracle.com
State Superseded
Headers show
Series
  • xfs: Delay Ready Attributes
Related show

Commit Message

Allison Collins June 25, 2020, 11:30 p.m. UTC
This patch modifies the attr remove routines to be delay ready. This
means they no longer roll or commit transactions, but instead return
-EAGAIN to have the calling routine roll and refresh the transaction. In
this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
uses a sort of state machine like switch to keep track of where it was
when EAGAIN was returned. xfs_attr_node_removename has also been
modified to use the switch, and a new version of xfs_attr_remove_args
consists of a simple loop to refresh the transaction until the operation
is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
transaction where ever the existing code used to.

Calls to xfs_attr_rmtval_remove are replaced with the delay ready
version __xfs_attr_rmtval_remove. We will rename
__xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
done.

xfs_attr_rmtval_remove itself is still in use by the set routines (used
during a rename).  For reasons of perserving existing function, we
modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
set.  Similar to how xfs_attr_remove_args does here.  Once we transition
the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
used and will be removed.

This patch also adds a new struct xfs_delattr_context, which we will use
to keep track of the current state of an attribute operation. The new
xfs_delattr_state enum is used to track various operations that are in
progress so that we know not to repeat them, and resume where we left
off before EAGAIN was returned to cycle out the transaction. Other
members take the place of local variables that need to retain their
values across multiple function recalls.  See xfs_attr.h for a more
detailed diagram of the states.

Signed-off-by: Allison Collins <allison.henderson@oracle.com>
---
 fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
 fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
 fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
 fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
 fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
 fs/xfs/xfs_attr_inactive.c      |   2 +-
 6 files changed, 208 insertions(+), 66 deletions(-)

Comments

Brian Foster July 9, 2020, 1:24 p.m. UTC | #1
On Thu, Jun 25, 2020 at 04:30:16PM -0700, Allison Collins wrote:
> This patch modifies the attr remove routines to be delay ready. This
> means they no longer roll or commit transactions, but instead return
> -EAGAIN to have the calling routine roll and refresh the transaction. In
> this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
> uses a sort of state machine like switch to keep track of where it was
> when EAGAIN was returned. xfs_attr_node_removename has also been
> modified to use the switch, and a new version of xfs_attr_remove_args
> consists of a simple loop to refresh the transaction until the operation
> is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
> transaction where ever the existing code used to.
> 
> Calls to xfs_attr_rmtval_remove are replaced with the delay ready
> version __xfs_attr_rmtval_remove. We will rename
> __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
> done.
> 
> xfs_attr_rmtval_remove itself is still in use by the set routines (used
> during a rename).  For reasons of perserving existing function, we
> modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
> set.  Similar to how xfs_attr_remove_args does here.  Once we transition
> the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
> used and will be removed.
> 
> This patch also adds a new struct xfs_delattr_context, which we will use
> to keep track of the current state of an attribute operation. The new
> xfs_delattr_state enum is used to track various operations that are in
> progress so that we know not to repeat them, and resume where we left
> off before EAGAIN was returned to cycle out the transaction. Other
> members take the place of local variables that need to retain their
> values across multiple function recalls.  See xfs_attr.h for a more
> detailed diagram of the states.
> 
> Signed-off-by: Allison Collins <allison.henderson@oracle.com>
> ---
>  fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
>  fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
>  fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
>  fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
>  fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
>  fs/xfs/xfs_attr_inactive.c      |   2 +-
>  6 files changed, 208 insertions(+), 66 deletions(-)
> 
> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
> index 4b78c86..5c460f4 100644
> --- a/fs/xfs/libxfs/xfs_attr.c
> +++ b/fs/xfs/libxfs/xfs_attr.c
> @@ -53,7 +53,7 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
>   */
>  STATIC int xfs_attr_node_get(xfs_da_args_t *args);
>  STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
> -STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
> +STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
>  STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
>  				 struct xfs_da_state **state);
>  STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
> @@ -264,6 +264,32 @@ xfs_attr_set_shortform(
>  }
>  
>  /*
> + * Checks to see if a delayed attribute transaction should be rolled.  If so,
> + * also checks for a defer finish.  Transaction is finished and rolled as
> + * needed, and returns true of false if the delayed operation should continue.
> + */
> +bool
> +xfs_attr_roll_again(

The function name suggests this is more of a status checking function
than one that does actual work. I'd suggest something more like
xfs_attr_trans_roll() based on the implementation.

> +	struct xfs_delattr_context	*dac,
> +	int				*error)
> +{
> +	struct xfs_da_args              *args = dac->da_args;
> +
> +	if (*error != -EAGAIN)
> +		return false;
> +
> +	if (dac->flags & XFS_DAC_DEFER_FINISH) {
> +		dac->flags &= ~XFS_DAC_DEFER_FINISH;
> +		*error = xfs_defer_finish(&args->trans);
> +		if (*error)
> +			return false;
> +	}
> +

I also find the semantics of this function a little confusing. How would
a caller distinguish between error == -EAGAIN as passed in from the
caller vs. error being set by one of the transaction processing calls?

> +	*error = xfs_trans_roll_inode(&args->trans, args->dp);
> +	return *error == 0;
> +}
> +
> +/*
>   * Set the attribute specified in @args.
>   */
>  int
> @@ -364,23 +390,47 @@ xfs_has_attr(
>   */
>  int
>  xfs_attr_remove_args(
> -	struct xfs_da_args      *args)
> +	struct xfs_da_args	*args)
>  {
> -	struct xfs_inode	*dp = args->dp;
> -	int			error;
> +	int				error = 0;
> +	struct xfs_delattr_context	dac = {
> +		.da_args	= args,
> +	};
> +
> +	do {
> +		error = xfs_attr_remove_iter(&dac);
> +	} while (xfs_attr_roll_again(&dac, &error));
> +
> +	return error;
> +}
> +
> +/*
> + * Remove the attribute specified in @args.
> + *
> + * This function may return -EAGAIN to signal that the transaction needs to be
> + * rolled.  Callers should continue calling this function until they receive a
> + * return value other than -EAGAIN.
> + */
> +int
> +xfs_attr_remove_iter(
> +	struct xfs_delattr_context	*dac)
> +{
> +	struct xfs_da_args		*args = dac->da_args;
> +	struct xfs_inode		*dp = args->dp;
> +
> +	if (dac->dela_state == XFS_DAS_RM_SHRINK)
> +		goto node;

Thoughts on my comments on this on the previous version?

>  
>  	if (!xfs_inode_hasattr(dp)) {
> -		error = -ENOATTR;
> +		return -ENOATTR;
>  	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
>  		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
> -		error = xfs_attr_shortform_remove(args);
> +		return xfs_attr_shortform_remove(args);
>  	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
> -		error = xfs_attr_leaf_removename(args);
> -	} else {
> -		error = xfs_attr_node_removename(args);
> +		return xfs_attr_leaf_removename(args);
>  	}
> -
> -	return error;
> +node:
> +	return  xfs_attr_node_removename(dac);
>  }
>  
>  /*
...
> @@ -1268,17 +1348,14 @@ xfs_attr_node_removename(
>  		error = xfs_da3_join(state);
>  		if (error)
>  			goto out;
> -		error = xfs_defer_finish(&args->trans);
> -		if (error)
> -			goto out;
> -		/*
> -		 * Commit the Btree join operation and start a new trans.
> -		 */
> -		error = xfs_trans_roll_inode(&args->trans, dp);
> -		if (error)
> -			goto out;
> +
> +		dac->flags |= XFS_DAC_DEFER_FINISH;
> +		dac->dela_state = XFS_DAS_RM_SHRINK;
> +		return -EAGAIN;
>  	}
>  
> +das_rm_shrink:
> +

I think I also mentioned the tail of this function might sit better in
the caller..

Brian

>  	/*
>  	 * If the result is small enough, push it all into the inode.
>  	 */
> diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
> index 3e97a93..6c58792 100644
> --- a/fs/xfs/libxfs/xfs_attr.h
> +++ b/fs/xfs/libxfs/xfs_attr.h
> @@ -74,6 +74,75 @@ struct xfs_attr_list_context {
>  };
>  
>  
> +/*
> + * ========================================================================
> + * Structure used to pass context around among the delayed routines.
> + * ========================================================================
> + */
> +
> +/*
> + * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
> + * states indicate places where the function would return -EAGAIN, and then
> + * immediately resume from after being recalled by the calling function. States
> + * marked as a "subroutine state" indicate that they belong to a subroutine, and
> + * so the calling function needs to pass them back to that subroutine to allow
> + * it to finish where it left off. But they otherwise do not have a role in the
> + * calling function other than just passing through.
> + *
> + * xfs_attr_remove_iter()
> + *	  XFS_DAS_RM_SHRINK ─┐
> + *	  (subroutine state) │
> + *	                     └─>xfs_attr_node_removename()
> + *	                                      │
> + *	                                      v
> + *	                                   need to
> + *	                                shrink tree? ─n─┐
> + *	                                      │         │
> + *	                                      y         │
> + *	                                      │         │
> + *	                                      v         │
> + *	                              XFS_DAS_RM_SHRINK │
> + *	                                      │         │
> + *	                                      v         │
> + *	                                     done <─────┘
> + *
> + */
> +
> +/*
> + * Enum values for xfs_delattr_context.da_state
> + *
> + * These values are used by delayed attribute operations to keep track  of where
> + * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
> + * calling function to roll the transaction, and then recall the subroutine to
> + * finish the operation.  The enum is then used by the subroutine to jump back
> + * to where it was and resume executing where it left off.
> + */
> +enum xfs_delattr_state {
> +				      /* Zero is uninitalized */
> +	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
> +};
> +
> +/*
> + * Defines for xfs_delattr_context.flags
> + */
> +#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
> +#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
> +
> +/*
> + * Context used for keeping track of delayed attribute operations
> + */
> +struct xfs_delattr_context {
> +	struct xfs_da_args      *da_args;
> +
> +	/* Used in xfs_attr_node_removename to roll through removing blocks */
> +	struct xfs_da_state     *da_state;
> +	struct xfs_da_state_blk *blk;
> +
> +	/* Used to keep track of current state of delayed operation */
> +	unsigned int            flags;
> +	enum xfs_delattr_state  dela_state;
> +};
> +
>  /*========================================================================
>   * Function prototypes for the kernel.
>   *========================================================================*/
> @@ -91,6 +160,10 @@ int xfs_attr_set(struct xfs_da_args *args);
>  int xfs_attr_set_args(struct xfs_da_args *args);
>  int xfs_has_attr(struct xfs_da_args *args);
>  int xfs_attr_remove_args(struct xfs_da_args *args);
> +int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
> +bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
>  bool xfs_attr_namecheck(const void *name, size_t length);
> +void xfs_delattr_context_init(struct xfs_delattr_context *dac,
> +			      struct xfs_da_args *args);
>  
>  #endif	/* __XFS_ATTR_H__ */
> diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
> index 351351c..20521bf 100644
> --- a/fs/xfs/libxfs/xfs_attr_leaf.c
> +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
> @@ -19,8 +19,8 @@
>  #include "xfs_bmap_btree.h"
>  #include "xfs_bmap.h"
>  #include "xfs_attr_sf.h"
> -#include "xfs_attr_remote.h"
>  #include "xfs_attr.h"
> +#include "xfs_attr_remote.h"
>  #include "xfs_attr_leaf.h"
>  #include "xfs_error.h"
>  #include "xfs_trace.h"
> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
> index 85dca51..20e4605 100644
> --- a/fs/xfs/libxfs/xfs_attr_remote.c
> +++ b/fs/xfs/libxfs/xfs_attr_remote.c
> @@ -676,12 +676,14 @@ xfs_attr_rmtval_invalidate(
>   */
>  int
>  xfs_attr_rmtval_remove(
> -	struct xfs_da_args      *args)
> +	struct xfs_da_args		*args)
>  {
> -	xfs_dablk_t		lblkno;
> -	int			blkcnt;
> -	int			error = 0;
> -	int			retval = 0;
> +	xfs_dablk_t			lblkno;
> +	int				blkcnt;
> +	int				error = 0;
> +	struct xfs_delattr_context	dac  = {
> +		.da_args	= args,
> +	};
>  
>  	trace_xfs_attr_rmtval_remove(args);
>  
> @@ -691,19 +693,10 @@ xfs_attr_rmtval_remove(
>  	lblkno = args->rmtblkno;
>  	blkcnt = args->rmtblkcnt;
>  	do {
> -		retval = __xfs_attr_rmtval_remove(args);
> -		if (retval && retval != EAGAIN)
> -			return retval;
> -
> -		/*
> -		 * Close out trans and start the next one in the chain.
> -		 */
> -		error = xfs_trans_roll_inode(&args->trans, args->dp);
> -		if (error)
> -			return error;
> -	} while (retval == -EAGAIN);
> +		error = __xfs_attr_rmtval_remove(&dac);
> +	} while (xfs_attr_roll_again(&dac, &error));
>  
> -	return 0;
> +	return error;
>  }
>  
>  /*
> @@ -713,9 +706,10 @@ xfs_attr_rmtval_remove(
>   */
>  int
>  __xfs_attr_rmtval_remove(
> -	struct xfs_da_args	*args)
> +	struct xfs_delattr_context	*dac)
>  {
> -	int			error, done;
> +	struct xfs_da_args		*args = dac->da_args;
> +	int				error, done;
>  
>  	/*
>  	 * Unmap value blocks for this attr.
> @@ -725,12 +719,10 @@ __xfs_attr_rmtval_remove(
>  	if (error)
>  		return error;
>  
> -	error = xfs_defer_finish(&args->trans);
> -	if (error)
> -		return error;
> -
> -	if (!done)
> +	if (!done) {
> +		dac->flags |= XFS_DAC_DEFER_FINISH;
>  		return -EAGAIN;
> +	}
>  
>  	return error;
>  }
> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
> index 9eee615..002fd30 100644
> --- a/fs/xfs/libxfs/xfs_attr_remote.h
> +++ b/fs/xfs/libxfs/xfs_attr_remote.h
> @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args);
>  int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
>  		xfs_buf_flags_t incore_flags);
>  int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
> -int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
> +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
>  #endif /* __XFS_ATTR_REMOTE_H__ */
> diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
> index bfad669..aaa7e66 100644
> --- a/fs/xfs/xfs_attr_inactive.c
> +++ b/fs/xfs/xfs_attr_inactive.c
> @@ -15,10 +15,10 @@
>  #include "xfs_da_format.h"
>  #include "xfs_da_btree.h"
>  #include "xfs_inode.h"
> +#include "xfs_attr.h"
>  #include "xfs_attr_remote.h"
>  #include "xfs_trans.h"
>  #include "xfs_bmap.h"
> -#include "xfs_attr.h"
>  #include "xfs_attr_leaf.h"
>  #include "xfs_quota.h"
>  #include "xfs_dir2.h"
> -- 
> 2.7.4
>
Darrick J. Wong July 9, 2020, 6:04 p.m. UTC | #2
On Thu, Jul 09, 2020 at 09:24:56AM -0400, Brian Foster wrote:
> On Thu, Jun 25, 2020 at 04:30:16PM -0700, Allison Collins wrote:
> > This patch modifies the attr remove routines to be delay ready. This
> > means they no longer roll or commit transactions, but instead return
> > -EAGAIN to have the calling routine roll and refresh the transaction. In
> > this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
> > uses a sort of state machine like switch to keep track of where it was
> > when EAGAIN was returned. xfs_attr_node_removename has also been
> > modified to use the switch, and a new version of xfs_attr_remove_args
> > consists of a simple loop to refresh the transaction until the operation
> > is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
> > transaction where ever the existing code used to.
> > 
> > Calls to xfs_attr_rmtval_remove are replaced with the delay ready
> > version __xfs_attr_rmtval_remove. We will rename
> > __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
> > done.
> > 
> > xfs_attr_rmtval_remove itself is still in use by the set routines (used
> > during a rename).  For reasons of perserving existing function, we
> > modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
> > set.  Similar to how xfs_attr_remove_args does here.  Once we transition
> > the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
> > used and will be removed.
> > 
> > This patch also adds a new struct xfs_delattr_context, which we will use
> > to keep track of the current state of an attribute operation. The new
> > xfs_delattr_state enum is used to track various operations that are in
> > progress so that we know not to repeat them, and resume where we left
> > off before EAGAIN was returned to cycle out the transaction. Other
> > members take the place of local variables that need to retain their
> > values across multiple function recalls.  See xfs_attr.h for a more
> > detailed diagram of the states.
> > 
> > Signed-off-by: Allison Collins <allison.henderson@oracle.com>
> > ---
> >  fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
> >  fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
> >  fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
> >  fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
> >  fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
> >  fs/xfs/xfs_attr_inactive.c      |   2 +-
> >  6 files changed, 208 insertions(+), 66 deletions(-)
> > 
> > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
> > index 4b78c86..5c460f4 100644
> > --- a/fs/xfs/libxfs/xfs_attr.c
> > +++ b/fs/xfs/libxfs/xfs_attr.c
> > @@ -53,7 +53,7 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
> >   */
> >  STATIC int xfs_attr_node_get(xfs_da_args_t *args);
> >  STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
> > -STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
> > +STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
> >  STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
> >  				 struct xfs_da_state **state);
> >  STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
> > @@ -264,6 +264,32 @@ xfs_attr_set_shortform(
> >  }
> >  
> >  /*
> > + * Checks to see if a delayed attribute transaction should be rolled.  If so,
> > + * also checks for a defer finish.  Transaction is finished and rolled as
> > + * needed, and returns true of false if the delayed operation should continue.
> > + */
> > +bool
> > +xfs_attr_roll_again(
> 
> The function name suggests this is more of a status checking function
> than one that does actual work. I'd suggest something more like
> xfs_attr_trans_roll() based on the implementation.
> 
> > +	struct xfs_delattr_context	*dac,
> > +	int				*error)
> > +{
> > +	struct xfs_da_args              *args = dac->da_args;
> > +
> > +	if (*error != -EAGAIN)
> > +		return false;
> > +
> > +	if (dac->flags & XFS_DAC_DEFER_FINISH) {
> > +		dac->flags &= ~XFS_DAC_DEFER_FINISH;
> > +		*error = xfs_defer_finish(&args->trans);
> > +		if (*error)
> > +			return false;
> > +	}
> > +
> 
> I also find the semantics of this function a little confusing. How would
> a caller distinguish between error == -EAGAIN as passed in from the
> caller vs. error being set by one of the transaction processing calls?

Do any of them do that?

> > +	*error = xfs_trans_roll_inode(&args->trans, args->dp);

I keep wondering why it's necessary to call xfs_defer_finish immediately
followed by xfs_trans_roll_inode.  Maybe we've already resolved this
question, but if we did I don't remember the answer, and the answer
should be captured as a comment because I never remember. :P

Oh right, we /did/ resolve this question.  The defer_finish exists to
finish all the deferred ops so that we return to the caller with a fresh
transaction and no dfops.

Comment, please...

	if (dac->flags & XFS_DAC_DEFER_FINISH) {
		/*
		 * The caller wants us to finish all the deferred ops so
		 * that we avoid pinning the log tail with a large
		 * number of deferred ops.
		 */
		dac->flags &= ~XFS_DAC_DEFER_FINISH;
		*error = xfs_defer_finish(&args->trans);
		if (*error)
			return false;
	}

Oh, another question: if we joined the inode to the transaction with
ilockflags == 0 (i.e. the transaction does not automatically unlock the
inode) then why do we need the separate _trans_roll_inode call?  Won't
the _defer_finish call return with the inode joined to a clean
transaction?

(Or did we resolve this question already as well, and all we need is a
comment to remind your forgetful maintainer?)

--D

> > +	return *error == 0;
> > +}
> > +
> > +/*
> >   * Set the attribute specified in @args.
> >   */
> >  int
> > @@ -364,23 +390,47 @@ xfs_has_attr(
> >   */
> >  int
> >  xfs_attr_remove_args(
> > -	struct xfs_da_args      *args)
> > +	struct xfs_da_args	*args)
> >  {
> > -	struct xfs_inode	*dp = args->dp;
> > -	int			error;
> > +	int				error = 0;
> > +	struct xfs_delattr_context	dac = {
> > +		.da_args	= args,
> > +	};
> > +
> > +	do {
> > +		error = xfs_attr_remove_iter(&dac);
> > +	} while (xfs_attr_roll_again(&dac, &error));
> > +
> > +	return error;
> > +}
> > +
> > +/*
> > + * Remove the attribute specified in @args.
> > + *
> > + * This function may return -EAGAIN to signal that the transaction needs to be
> > + * rolled.  Callers should continue calling this function until they receive a
> > + * return value other than -EAGAIN.
> > + */
> > +int
> > +xfs_attr_remove_iter(
> > +	struct xfs_delattr_context	*dac)
> > +{
> > +	struct xfs_da_args		*args = dac->da_args;
> > +	struct xfs_inode		*dp = args->dp;
> > +
> > +	if (dac->dela_state == XFS_DAS_RM_SHRINK)
> > +		goto node;
> 
> Thoughts on my comments on this on the previous version?
> 
> >  
> >  	if (!xfs_inode_hasattr(dp)) {
> > -		error = -ENOATTR;
> > +		return -ENOATTR;
> >  	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
> >  		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
> > -		error = xfs_attr_shortform_remove(args);
> > +		return xfs_attr_shortform_remove(args);
> >  	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
> > -		error = xfs_attr_leaf_removename(args);
> > -	} else {
> > -		error = xfs_attr_node_removename(args);
> > +		return xfs_attr_leaf_removename(args);
> >  	}
> > -
> > -	return error;
> > +node:
> > +	return  xfs_attr_node_removename(dac);
> >  }
> >  
> >  /*
> ...
> > @@ -1268,17 +1348,14 @@ xfs_attr_node_removename(
> >  		error = xfs_da3_join(state);
> >  		if (error)
> >  			goto out;
> > -		error = xfs_defer_finish(&args->trans);
> > -		if (error)
> > -			goto out;
> > -		/*
> > -		 * Commit the Btree join operation and start a new trans.
> > -		 */
> > -		error = xfs_trans_roll_inode(&args->trans, dp);
> > -		if (error)
> > -			goto out;
> > +
> > +		dac->flags |= XFS_DAC_DEFER_FINISH;
> > +		dac->dela_state = XFS_DAS_RM_SHRINK;
> > +		return -EAGAIN;
> >  	}
> >  
> > +das_rm_shrink:
> > +
> 
> I think I also mentioned the tail of this function might sit better in
> the caller..
> 
> Brian
> 
> >  	/*
> >  	 * If the result is small enough, push it all into the inode.
> >  	 */
> > diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
> > index 3e97a93..6c58792 100644
> > --- a/fs/xfs/libxfs/xfs_attr.h
> > +++ b/fs/xfs/libxfs/xfs_attr.h
> > @@ -74,6 +74,75 @@ struct xfs_attr_list_context {
> >  };
> >  
> >  
> > +/*
> > + * ========================================================================
> > + * Structure used to pass context around among the delayed routines.
> > + * ========================================================================
> > + */
> > +
> > +/*
> > + * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
> > + * states indicate places where the function would return -EAGAIN, and then
> > + * immediately resume from after being recalled by the calling function. States
> > + * marked as a "subroutine state" indicate that they belong to a subroutine, and
> > + * so the calling function needs to pass them back to that subroutine to allow
> > + * it to finish where it left off. But they otherwise do not have a role in the
> > + * calling function other than just passing through.
> > + *
> > + * xfs_attr_remove_iter()
> > + *	  XFS_DAS_RM_SHRINK ─┐
> > + *	  (subroutine state) │
> > + *	                     └─>xfs_attr_node_removename()
> > + *	                                      │
> > + *	                                      v
> > + *	                                   need to
> > + *	                                shrink tree? ─n─┐
> > + *	                                      │         │
> > + *	                                      y         │
> > + *	                                      │         │
> > + *	                                      v         │
> > + *	                              XFS_DAS_RM_SHRINK │
> > + *	                                      │         │
> > + *	                                      v         │
> > + *	                                     done <─────┘
> > + *
> > + */
> > +
> > +/*
> > + * Enum values for xfs_delattr_context.da_state
> > + *
> > + * These values are used by delayed attribute operations to keep track  of where
> > + * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
> > + * calling function to roll the transaction, and then recall the subroutine to
> > + * finish the operation.  The enum is then used by the subroutine to jump back
> > + * to where it was and resume executing where it left off.
> > + */
> > +enum xfs_delattr_state {
> > +				      /* Zero is uninitalized */
> > +	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
> > +};
> > +
> > +/*
> > + * Defines for xfs_delattr_context.flags
> > + */
> > +#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
> > +#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
> > +
> > +/*
> > + * Context used for keeping track of delayed attribute operations
> > + */
> > +struct xfs_delattr_context {
> > +	struct xfs_da_args      *da_args;
> > +
> > +	/* Used in xfs_attr_node_removename to roll through removing blocks */
> > +	struct xfs_da_state     *da_state;
> > +	struct xfs_da_state_blk *blk;
> > +
> > +	/* Used to keep track of current state of delayed operation */
> > +	unsigned int            flags;
> > +	enum xfs_delattr_state  dela_state;
> > +};
> > +
> >  /*========================================================================
> >   * Function prototypes for the kernel.
> >   *========================================================================*/
> > @@ -91,6 +160,10 @@ int xfs_attr_set(struct xfs_da_args *args);
> >  int xfs_attr_set_args(struct xfs_da_args *args);
> >  int xfs_has_attr(struct xfs_da_args *args);
> >  int xfs_attr_remove_args(struct xfs_da_args *args);
> > +int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
> > +bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
> >  bool xfs_attr_namecheck(const void *name, size_t length);
> > +void xfs_delattr_context_init(struct xfs_delattr_context *dac,
> > +			      struct xfs_da_args *args);
> >  
> >  #endif	/* __XFS_ATTR_H__ */
> > diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
> > index 351351c..20521bf 100644
> > --- a/fs/xfs/libxfs/xfs_attr_leaf.c
> > +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
> > @@ -19,8 +19,8 @@
> >  #include "xfs_bmap_btree.h"
> >  #include "xfs_bmap.h"
> >  #include "xfs_attr_sf.h"
> > -#include "xfs_attr_remote.h"
> >  #include "xfs_attr.h"
> > +#include "xfs_attr_remote.h"
> >  #include "xfs_attr_leaf.h"
> >  #include "xfs_error.h"
> >  #include "xfs_trace.h"
> > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
> > index 85dca51..20e4605 100644
> > --- a/fs/xfs/libxfs/xfs_attr_remote.c
> > +++ b/fs/xfs/libxfs/xfs_attr_remote.c
> > @@ -676,12 +676,14 @@ xfs_attr_rmtval_invalidate(
> >   */
> >  int
> >  xfs_attr_rmtval_remove(
> > -	struct xfs_da_args      *args)
> > +	struct xfs_da_args		*args)
> >  {
> > -	xfs_dablk_t		lblkno;
> > -	int			blkcnt;
> > -	int			error = 0;
> > -	int			retval = 0;
> > +	xfs_dablk_t			lblkno;
> > +	int				blkcnt;
> > +	int				error = 0;
> > +	struct xfs_delattr_context	dac  = {
> > +		.da_args	= args,
> > +	};
> >  
> >  	trace_xfs_attr_rmtval_remove(args);
> >  
> > @@ -691,19 +693,10 @@ xfs_attr_rmtval_remove(
> >  	lblkno = args->rmtblkno;
> >  	blkcnt = args->rmtblkcnt;
> >  	do {
> > -		retval = __xfs_attr_rmtval_remove(args);
> > -		if (retval && retval != EAGAIN)
> > -			return retval;
> > -
> > -		/*
> > -		 * Close out trans and start the next one in the chain.
> > -		 */
> > -		error = xfs_trans_roll_inode(&args->trans, args->dp);
> > -		if (error)
> > -			return error;
> > -	} while (retval == -EAGAIN);
> > +		error = __xfs_attr_rmtval_remove(&dac);
> > +	} while (xfs_attr_roll_again(&dac, &error));
> >  
> > -	return 0;
> > +	return error;
> >  }
> >  
> >  /*
> > @@ -713,9 +706,10 @@ xfs_attr_rmtval_remove(
> >   */
> >  int
> >  __xfs_attr_rmtval_remove(
> > -	struct xfs_da_args	*args)
> > +	struct xfs_delattr_context	*dac)
> >  {
> > -	int			error, done;
> > +	struct xfs_da_args		*args = dac->da_args;
> > +	int				error, done;
> >  
> >  	/*
> >  	 * Unmap value blocks for this attr.
> > @@ -725,12 +719,10 @@ __xfs_attr_rmtval_remove(
> >  	if (error)
> >  		return error;
> >  
> > -	error = xfs_defer_finish(&args->trans);
> > -	if (error)
> > -		return error;
> > -
> > -	if (!done)
> > +	if (!done) {
> > +		dac->flags |= XFS_DAC_DEFER_FINISH;
> >  		return -EAGAIN;
> > +	}
> >  
> >  	return error;
> >  }
> > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
> > index 9eee615..002fd30 100644
> > --- a/fs/xfs/libxfs/xfs_attr_remote.h
> > +++ b/fs/xfs/libxfs/xfs_attr_remote.h
> > @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args);
> >  int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
> >  		xfs_buf_flags_t incore_flags);
> >  int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
> > -int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
> > +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
> >  #endif /* __XFS_ATTR_REMOTE_H__ */
> > diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
> > index bfad669..aaa7e66 100644
> > --- a/fs/xfs/xfs_attr_inactive.c
> > +++ b/fs/xfs/xfs_attr_inactive.c
> > @@ -15,10 +15,10 @@
> >  #include "xfs_da_format.h"
> >  #include "xfs_da_btree.h"
> >  #include "xfs_inode.h"
> > +#include "xfs_attr.h"
> >  #include "xfs_attr_remote.h"
> >  #include "xfs_trans.h"
> >  #include "xfs_bmap.h"
> > -#include "xfs_attr.h"
> >  #include "xfs_attr_leaf.h"
> >  #include "xfs_quota.h"
> >  #include "xfs_dir2.h"
> > -- 
> > 2.7.4
> > 
>
Allison Collins July 9, 2020, 10:02 p.m. UTC | #3
On 7/9/20 6:24 AM, Brian Foster wrote:
> On Thu, Jun 25, 2020 at 04:30:16PM -0700, Allison Collins wrote:
>> This patch modifies the attr remove routines to be delay ready. This
>> means they no longer roll or commit transactions, but instead return
>> -EAGAIN to have the calling routine roll and refresh the transaction. In
>> this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
>> uses a sort of state machine like switch to keep track of where it was
>> when EAGAIN was returned. xfs_attr_node_removename has also been
>> modified to use the switch, and a new version of xfs_attr_remove_args
>> consists of a simple loop to refresh the transaction until the operation
>> is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
>> transaction where ever the existing code used to.
>>
>> Calls to xfs_attr_rmtval_remove are replaced with the delay ready
>> version __xfs_attr_rmtval_remove. We will rename
>> __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
>> done.
>>
>> xfs_attr_rmtval_remove itself is still in use by the set routines (used
>> during a rename).  For reasons of perserving existing function, we
>> modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
>> set.  Similar to how xfs_attr_remove_args does here.  Once we transition
>> the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
>> used and will be removed.
>>
>> This patch also adds a new struct xfs_delattr_context, which we will use
>> to keep track of the current state of an attribute operation. The new
>> xfs_delattr_state enum is used to track various operations that are in
>> progress so that we know not to repeat them, and resume where we left
>> off before EAGAIN was returned to cycle out the transaction. Other
>> members take the place of local variables that need to retain their
>> values across multiple function recalls.  See xfs_attr.h for a more
>> detailed diagram of the states.
>>
>> Signed-off-by: Allison Collins <allison.henderson@oracle.com>
>> ---
>>   fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
>>   fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
>>   fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
>>   fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
>>   fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
>>   fs/xfs/xfs_attr_inactive.c      |   2 +-
>>   6 files changed, 208 insertions(+), 66 deletions(-)
>>
>> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
>> index 4b78c86..5c460f4 100644
>> --- a/fs/xfs/libxfs/xfs_attr.c
>> +++ b/fs/xfs/libxfs/xfs_attr.c
>> @@ -53,7 +53,7 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
>>    */
>>   STATIC int xfs_attr_node_get(xfs_da_args_t *args);
>>   STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
>> -STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
>> +STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
>>   STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
>>   				 struct xfs_da_state **state);
>>   STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
>> @@ -264,6 +264,32 @@ xfs_attr_set_shortform(
>>   }
>>   
>>   /*
>> + * Checks to see if a delayed attribute transaction should be rolled.  If so,
>> + * also checks for a defer finish.  Transaction is finished and rolled as
>> + * needed, and returns true of false if the delayed operation should continue.
>> + */
>> +bool
>> +xfs_attr_roll_again(
> 
> The function name suggests this is more of a status checking function
> than one that does actual work. I'd suggest something more like
> xfs_attr_trans_roll() based on the implementation.
Sure, I believe Darrick had asked for this helper in the last review, so 
as long as folks are in agreement with the name, that is fine.

> 
>> +	struct xfs_delattr_context	*dac,
>> +	int				*error)
>> +{
>> +	struct xfs_da_args              *args = dac->da_args;
>> +
>> +	if (*error != -EAGAIN)
>> +		return false;
>> +
>> +	if (dac->flags & XFS_DAC_DEFER_FINISH) {
>> +		dac->flags &= ~XFS_DAC_DEFER_FINISH;
>> +		*error = xfs_defer_finish(&args->trans);
>> +		if (*error)
>> +			return false;
>> +	}
>> +
> 
> I also find the semantics of this function a little confusing. How would
> a caller distinguish between error == -EAGAIN as passed in from the
> caller vs. error being set by one of the transaction processing calls?
I believe the helper was just meant to be a straight hoist from the v9 
version.  The error parameter here is an input for use in this scope, 
it's not for the caller to do anything with.  Other than to pipe it back 
in.  The caller just needs to know whether or not to continue.

That signature is a bit odd, though this pattern does appear in a few 
places around the code, so it does help consolidate a bit.  I suppose 
it's just a question of which people prefer.

> 
>> +	*error = xfs_trans_roll_inode(&args->trans, args->dp);
>> +	return *error == 0;
>> +}
>> +
>> +/*
>>    * Set the attribute specified in @args.
>>    */
>>   int
>> @@ -364,23 +390,47 @@ xfs_has_attr(
>>    */
>>   int
>>   xfs_attr_remove_args(
>> -	struct xfs_da_args      *args)
>> +	struct xfs_da_args	*args)
>>   {
>> -	struct xfs_inode	*dp = args->dp;
>> -	int			error;
>> +	int				error = 0;
>> +	struct xfs_delattr_context	dac = {
>> +		.da_args	= args,
>> +	};
>> +
>> +	do {
>> +		error = xfs_attr_remove_iter(&dac);
>> +	} while (xfs_attr_roll_again(&dac, &error));
>> +
>> +	return error;
>> +}
>> +
>> +/*
>> + * Remove the attribute specified in @args.
>> + *
>> + * This function may return -EAGAIN to signal that the transaction needs to be
>> + * rolled.  Callers should continue calling this function until they receive a
>> + * return value other than -EAGAIN.
>> + */
>> +int
>> +xfs_attr_remove_iter(
>> +	struct xfs_delattr_context	*dac)
>> +{
>> +	struct xfs_da_args		*args = dac->da_args;
>> +	struct xfs_inode		*dp = args->dp;
>> +
>> +	if (dac->dela_state == XFS_DAS_RM_SHRINK)
>> +		goto node;
> 
> Thoughts on my comments on this on the previous version?
Apologies, I must have missed responding to it, I had a few family 
emergencies at the time and must have lost track.  I do believe we need 
the goto because in the process of shrinking, it is possible that we 
turn back into a block which would cause us to fall into the wrong 
if/else logic with out it. I will run the test and double check though. 
I'm pretty sure I recall running into that though.

> 
>>   
>>   	if (!xfs_inode_hasattr(dp)) {
>> -		error = -ENOATTR;
>> +		return -ENOATTR;
>>   	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
>>   		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
>> -		error = xfs_attr_shortform_remove(args);
>> +		return xfs_attr_shortform_remove(args);
>>   	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
>> -		error = xfs_attr_leaf_removename(args);
>> -	} else {
>> -		error = xfs_attr_node_removename(args);
>> +		return xfs_attr_leaf_removename(args);
>>   	}
>> -
>> -	return error;
>> +node:
>> +	return  xfs_attr_node_removename(dac);
>>   }
>>   
>>   /*
> ...
>> @@ -1268,17 +1348,14 @@ xfs_attr_node_removename(
>>   		error = xfs_da3_join(state);
>>   		if (error)
>>   			goto out;
>> -		error = xfs_defer_finish(&args->trans);
>> -		if (error)
>> -			goto out;
>> -		/*
>> -		 * Commit the Btree join operation and start a new trans.
>> -		 */
>> -		error = xfs_trans_roll_inode(&args->trans, dp);
>> -		if (error)
>> -			goto out;
>> +
>> +		dac->flags |= XFS_DAC_DEFER_FINISH;
>> +		dac->dela_state = XFS_DAS_RM_SHRINK;
>> +		return -EAGAIN;
>>   	}
>>   
>> +das_rm_shrink:
>> +
> 
> I think I also mentioned the tail of this function might sit better in
> the caller..
I see, yes I'll see if I could pull that up.  If I missed anything else, 
please call it out.  Thank you for your reviews!

Allison

> 
> Brian
> 
>>   	/*
>>   	 * If the result is small enough, push it all into the inode.
>>   	 */
>> diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
>> index 3e97a93..6c58792 100644
>> --- a/fs/xfs/libxfs/xfs_attr.h
>> +++ b/fs/xfs/libxfs/xfs_attr.h
>> @@ -74,6 +74,75 @@ struct xfs_attr_list_context {
>>   };
>>   
>>   
>> +/*
>> + * ========================================================================
>> + * Structure used to pass context around among the delayed routines.
>> + * ========================================================================
>> + */
>> +
>> +/*
>> + * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
>> + * states indicate places where the function would return -EAGAIN, and then
>> + * immediately resume from after being recalled by the calling function. States
>> + * marked as a "subroutine state" indicate that they belong to a subroutine, and
>> + * so the calling function needs to pass them back to that subroutine to allow
>> + * it to finish where it left off. But they otherwise do not have a role in the
>> + * calling function other than just passing through.
>> + *
>> + * xfs_attr_remove_iter()
>> + *	  XFS_DAS_RM_SHRINK ─┐
>> + *	  (subroutine state) │
>> + *	                     └─>xfs_attr_node_removename()
>> + *	                                      │
>> + *	                                      v
>> + *	                                   need to
>> + *	                                shrink tree? ─n─┐
>> + *	                                      │         │
>> + *	                                      y         │
>> + *	                                      │         │
>> + *	                                      v         │
>> + *	                              XFS_DAS_RM_SHRINK │
>> + *	                                      │         │
>> + *	                                      v         │
>> + *	                                     done <─────┘
>> + *
>> + */
>> +
>> +/*
>> + * Enum values for xfs_delattr_context.da_state
>> + *
>> + * These values are used by delayed attribute operations to keep track  of where
>> + * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
>> + * calling function to roll the transaction, and then recall the subroutine to
>> + * finish the operation.  The enum is then used by the subroutine to jump back
>> + * to where it was and resume executing where it left off.
>> + */
>> +enum xfs_delattr_state {
>> +				      /* Zero is uninitalized */
>> +	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
>> +};
>> +
>> +/*
>> + * Defines for xfs_delattr_context.flags
>> + */
>> +#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
>> +#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
>> +
>> +/*
>> + * Context used for keeping track of delayed attribute operations
>> + */
>> +struct xfs_delattr_context {
>> +	struct xfs_da_args      *da_args;
>> +
>> +	/* Used in xfs_attr_node_removename to roll through removing blocks */
>> +	struct xfs_da_state     *da_state;
>> +	struct xfs_da_state_blk *blk;
>> +
>> +	/* Used to keep track of current state of delayed operation */
>> +	unsigned int            flags;
>> +	enum xfs_delattr_state  dela_state;
>> +};
>> +
>>   /*========================================================================
>>    * Function prototypes for the kernel.
>>    *========================================================================*/
>> @@ -91,6 +160,10 @@ int xfs_attr_set(struct xfs_da_args *args);
>>   int xfs_attr_set_args(struct xfs_da_args *args);
>>   int xfs_has_attr(struct xfs_da_args *args);
>>   int xfs_attr_remove_args(struct xfs_da_args *args);
>> +int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
>> +bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
>>   bool xfs_attr_namecheck(const void *name, size_t length);
>> +void xfs_delattr_context_init(struct xfs_delattr_context *dac,
>> +			      struct xfs_da_args *args);
>>   
>>   #endif	/* __XFS_ATTR_H__ */
>> diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
>> index 351351c..20521bf 100644
>> --- a/fs/xfs/libxfs/xfs_attr_leaf.c
>> +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
>> @@ -19,8 +19,8 @@
>>   #include "xfs_bmap_btree.h"
>>   #include "xfs_bmap.h"
>>   #include "xfs_attr_sf.h"
>> -#include "xfs_attr_remote.h"
>>   #include "xfs_attr.h"
>> +#include "xfs_attr_remote.h"
>>   #include "xfs_attr_leaf.h"
>>   #include "xfs_error.h"
>>   #include "xfs_trace.h"
>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
>> index 85dca51..20e4605 100644
>> --- a/fs/xfs/libxfs/xfs_attr_remote.c
>> +++ b/fs/xfs/libxfs/xfs_attr_remote.c
>> @@ -676,12 +676,14 @@ xfs_attr_rmtval_invalidate(
>>    */
>>   int
>>   xfs_attr_rmtval_remove(
>> -	struct xfs_da_args      *args)
>> +	struct xfs_da_args		*args)
>>   {
>> -	xfs_dablk_t		lblkno;
>> -	int			blkcnt;
>> -	int			error = 0;
>> -	int			retval = 0;
>> +	xfs_dablk_t			lblkno;
>> +	int				blkcnt;
>> +	int				error = 0;
>> +	struct xfs_delattr_context	dac  = {
>> +		.da_args	= args,
>> +	};
>>   
>>   	trace_xfs_attr_rmtval_remove(args);
>>   
>> @@ -691,19 +693,10 @@ xfs_attr_rmtval_remove(
>>   	lblkno = args->rmtblkno;
>>   	blkcnt = args->rmtblkcnt;
>>   	do {
>> -		retval = __xfs_attr_rmtval_remove(args);
>> -		if (retval && retval != EAGAIN)
>> -			return retval;
>> -
>> -		/*
>> -		 * Close out trans and start the next one in the chain.
>> -		 */
>> -		error = xfs_trans_roll_inode(&args->trans, args->dp);
>> -		if (error)
>> -			return error;
>> -	} while (retval == -EAGAIN);
>> +		error = __xfs_attr_rmtval_remove(&dac);
>> +	} while (xfs_attr_roll_again(&dac, &error));
>>   
>> -	return 0;
>> +	return error;
>>   }
>>   
>>   /*
>> @@ -713,9 +706,10 @@ xfs_attr_rmtval_remove(
>>    */
>>   int
>>   __xfs_attr_rmtval_remove(
>> -	struct xfs_da_args	*args)
>> +	struct xfs_delattr_context	*dac)
>>   {
>> -	int			error, done;
>> +	struct xfs_da_args		*args = dac->da_args;
>> +	int				error, done;
>>   
>>   	/*
>>   	 * Unmap value blocks for this attr.
>> @@ -725,12 +719,10 @@ __xfs_attr_rmtval_remove(
>>   	if (error)
>>   		return error;
>>   
>> -	error = xfs_defer_finish(&args->trans);
>> -	if (error)
>> -		return error;
>> -
>> -	if (!done)
>> +	if (!done) {
>> +		dac->flags |= XFS_DAC_DEFER_FINISH;
>>   		return -EAGAIN;
>> +	}
>>   
>>   	return error;
>>   }
>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
>> index 9eee615..002fd30 100644
>> --- a/fs/xfs/libxfs/xfs_attr_remote.h
>> +++ b/fs/xfs/libxfs/xfs_attr_remote.h
>> @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args);
>>   int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
>>   		xfs_buf_flags_t incore_flags);
>>   int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
>> -int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
>> +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
>>   #endif /* __XFS_ATTR_REMOTE_H__ */
>> diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
>> index bfad669..aaa7e66 100644
>> --- a/fs/xfs/xfs_attr_inactive.c
>> +++ b/fs/xfs/xfs_attr_inactive.c
>> @@ -15,10 +15,10 @@
>>   #include "xfs_da_format.h"
>>   #include "xfs_da_btree.h"
>>   #include "xfs_inode.h"
>> +#include "xfs_attr.h"
>>   #include "xfs_attr_remote.h"
>>   #include "xfs_trans.h"
>>   #include "xfs_bmap.h"
>> -#include "xfs_attr.h"
>>   #include "xfs_attr_leaf.h"
>>   #include "xfs_quota.h"
>>   #include "xfs_dir2.h"
>> -- 
>> 2.7.4
>>
>
Allison Collins July 10, 2020, 6:22 a.m. UTC | #4
On 7/9/20 11:04 AM, Darrick J. Wong wrote:
> On Thu, Jul 09, 2020 at 09:24:56AM -0400, Brian Foster wrote:
>> On Thu, Jun 25, 2020 at 04:30:16PM -0700, Allison Collins wrote:
>>> This patch modifies the attr remove routines to be delay ready. This
>>> means they no longer roll or commit transactions, but instead return
>>> -EAGAIN to have the calling routine roll and refresh the transaction. In
>>> this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
>>> uses a sort of state machine like switch to keep track of where it was
>>> when EAGAIN was returned. xfs_attr_node_removename has also been
>>> modified to use the switch, and a new version of xfs_attr_remove_args
>>> consists of a simple loop to refresh the transaction until the operation
>>> is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
>>> transaction where ever the existing code used to.
>>>
>>> Calls to xfs_attr_rmtval_remove are replaced with the delay ready
>>> version __xfs_attr_rmtval_remove. We will rename
>>> __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
>>> done.
>>>
>>> xfs_attr_rmtval_remove itself is still in use by the set routines (used
>>> during a rename).  For reasons of perserving existing function, we
>>> modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
>>> set.  Similar to how xfs_attr_remove_args does here.  Once we transition
>>> the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
>>> used and will be removed.
>>>
>>> This patch also adds a new struct xfs_delattr_context, which we will use
>>> to keep track of the current state of an attribute operation. The new
>>> xfs_delattr_state enum is used to track various operations that are in
>>> progress so that we know not to repeat them, and resume where we left
>>> off before EAGAIN was returned to cycle out the transaction. Other
>>> members take the place of local variables that need to retain their
>>> values across multiple function recalls.  See xfs_attr.h for a more
>>> detailed diagram of the states.
>>>
>>> Signed-off-by: Allison Collins <allison.henderson@oracle.com>
>>> ---
>>>   fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
>>>   fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
>>>   fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
>>>   fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
>>>   fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
>>>   fs/xfs/xfs_attr_inactive.c      |   2 +-
>>>   6 files changed, 208 insertions(+), 66 deletions(-)
>>>
>>> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
>>> index 4b78c86..5c460f4 100644
>>> --- a/fs/xfs/libxfs/xfs_attr.c
>>> +++ b/fs/xfs/libxfs/xfs_attr.c
>>> @@ -53,7 +53,7 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
>>>    */
>>>   STATIC int xfs_attr_node_get(xfs_da_args_t *args);
>>>   STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
>>> -STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
>>> +STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
>>>   STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
>>>   				 struct xfs_da_state **state);
>>>   STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
>>> @@ -264,6 +264,32 @@ xfs_attr_set_shortform(
>>>   }
>>>   
>>>   /*
>>> + * Checks to see if a delayed attribute transaction should be rolled.  If so,
>>> + * also checks for a defer finish.  Transaction is finished and rolled as
>>> + * needed, and returns true of false if the delayed operation should continue.
>>> + */
>>> +bool
>>> +xfs_attr_roll_again(
>>
>> The function name suggests this is more of a status checking function
>> than one that does actual work. I'd suggest something more like
>> xfs_attr_trans_roll() based on the implementation.
>>
>>> +	struct xfs_delattr_context	*dac,
>>> +	int				*error)
>>> +{
>>> +	struct xfs_da_args              *args = dac->da_args;
>>> +
>>> +	if (*error != -EAGAIN)
>>> +		return false;
>>> +
>>> +	if (dac->flags & XFS_DAC_DEFER_FINISH) {
>>> +		dac->flags &= ~XFS_DAC_DEFER_FINISH;
>>> +		*error = xfs_defer_finish(&args->trans);
>>> +		if (*error)
>>> +			return false;
>>> +	}
>>> +
>>
>> I also find the semantics of this function a little confusing. How would
>> a caller distinguish between error == -EAGAIN as passed in from the
>> caller vs. error being set by one of the transaction processing calls?
> 
> Do any of them do that?
> 
>>> +	*error = xfs_trans_roll_inode(&args->trans, args->dp);
> 
> I keep wondering why it's necessary to call xfs_defer_finish immediately
> followed by xfs_trans_roll_inode.  Maybe we've already resolved this
> question, but if we did I don't remember the answer, and the answer
> should be captured as a comment because I never remember. :P
> 
> Oh right, we /did/ resolve this question.  The defer_finish exists to
> finish all the deferred ops so that we return to the caller with a fresh
> transaction and no dfops.
> 
> Comment, please...
> 
> 	if (dac->flags & XFS_DAC_DEFER_FINISH) {
> 		/*
> 		 * The caller wants us to finish all the deferred ops so
> 		 * that we avoid pinning the log tail with a large
> 		 * number of deferred ops.
> 		 */
Sure, will add.

> 		dac->flags &= ~XFS_DAC_DEFER_FINISH;
> 		*error = xfs_defer_finish(&args->trans);
> 		if (*error)
> 			return false;
> 	}
> 
> Oh, another question: if we joined the inode to the transaction with
> ilockflags == 0 (i.e. the transaction does not automatically unlock the
> inode) then why do we need the separate _trans_roll_inode call?  Won't
> the _defer_finish call return with the inode joined to a clean
> transaction?
> 
> (Or did we resolve this question already as well, and all we need is a
> comment to remind your forgetful maintainer?)
Mostly I think it was just because people wanted this to be a straight 
hoist as much as possible.  Reason being that the calling functions roll 
the transaction whether or not the helpers call defer finish, so this 
would be about equivalent.

In the case of the remove path, we probably could get away with skipping 
the roll if we call defer finish, but then later when we call this from 
the set routines, I think there's a race with the AIL such that we need 
to roll the transaction and then join the leaf to the new transaction. 
So then you'd need to hoist the roll back out, or plumb a flag in to 
roll anyway, but I think that would make the semantics even more goofy. 
I sort of get the impression that trying to stick to a "refactor only" 
goal helps people to be a little more comfortable with the changes :-)

Allison

> 
> --D
> 
>>> +	return *error == 0;
>>> +}
>>> +
>>> +/*
>>>    * Set the attribute specified in @args.
>>>    */
>>>   int
>>> @@ -364,23 +390,47 @@ xfs_has_attr(
>>>    */
>>>   int
>>>   xfs_attr_remove_args(
>>> -	struct xfs_da_args      *args)
>>> +	struct xfs_da_args	*args)
>>>   {
>>> -	struct xfs_inode	*dp = args->dp;
>>> -	int			error;
>>> +	int				error = 0;
>>> +	struct xfs_delattr_context	dac = {
>>> +		.da_args	= args,
>>> +	};
>>> +
>>> +	do {
>>> +		error = xfs_attr_remove_iter(&dac);
>>> +	} while (xfs_attr_roll_again(&dac, &error));
>>> +
>>> +	return error;
>>> +}
>>> +
>>> +/*
>>> + * Remove the attribute specified in @args.
>>> + *
>>> + * This function may return -EAGAIN to signal that the transaction needs to be
>>> + * rolled.  Callers should continue calling this function until they receive a
>>> + * return value other than -EAGAIN.
>>> + */
>>> +int
>>> +xfs_attr_remove_iter(
>>> +	struct xfs_delattr_context	*dac)
>>> +{
>>> +	struct xfs_da_args		*args = dac->da_args;
>>> +	struct xfs_inode		*dp = args->dp;
>>> +
>>> +	if (dac->dela_state == XFS_DAS_RM_SHRINK)
>>> +		goto node;
>>
>> Thoughts on my comments on this on the previous version?
>>
>>>   
>>>   	if (!xfs_inode_hasattr(dp)) {
>>> -		error = -ENOATTR;
>>> +		return -ENOATTR;
>>>   	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
>>>   		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
>>> -		error = xfs_attr_shortform_remove(args);
>>> +		return xfs_attr_shortform_remove(args);
>>>   	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
>>> -		error = xfs_attr_leaf_removename(args);
>>> -	} else {
>>> -		error = xfs_attr_node_removename(args);
>>> +		return xfs_attr_leaf_removename(args);
>>>   	}
>>> -
>>> -	return error;
>>> +node:
>>> +	return  xfs_attr_node_removename(dac);
>>>   }
>>>   
>>>   /*
>> ...
>>> @@ -1268,17 +1348,14 @@ xfs_attr_node_removename(
>>>   		error = xfs_da3_join(state);
>>>   		if (error)
>>>   			goto out;
>>> -		error = xfs_defer_finish(&args->trans);
>>> -		if (error)
>>> -			goto out;
>>> -		/*
>>> -		 * Commit the Btree join operation and start a new trans.
>>> -		 */
>>> -		error = xfs_trans_roll_inode(&args->trans, dp);
>>> -		if (error)
>>> -			goto out;
>>> +
>>> +		dac->flags |= XFS_DAC_DEFER_FINISH;
>>> +		dac->dela_state = XFS_DAS_RM_SHRINK;
>>> +		return -EAGAIN;
>>>   	}
>>>   
>>> +das_rm_shrink:
>>> +
>>
>> I think I also mentioned the tail of this function might sit better in
>> the caller..
>>
>> Brian
>>
>>>   	/*
>>>   	 * If the result is small enough, push it all into the inode.
>>>   	 */
>>> diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
>>> index 3e97a93..6c58792 100644
>>> --- a/fs/xfs/libxfs/xfs_attr.h
>>> +++ b/fs/xfs/libxfs/xfs_attr.h
>>> @@ -74,6 +74,75 @@ struct xfs_attr_list_context {
>>>   };
>>>   
>>>   
>>> +/*
>>> + * ========================================================================
>>> + * Structure used to pass context around among the delayed routines.
>>> + * ========================================================================
>>> + */
>>> +
>>> +/*
>>> + * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
>>> + * states indicate places where the function would return -EAGAIN, and then
>>> + * immediately resume from after being recalled by the calling function. States
>>> + * marked as a "subroutine state" indicate that they belong to a subroutine, and
>>> + * so the calling function needs to pass them back to that subroutine to allow
>>> + * it to finish where it left off. But they otherwise do not have a role in the
>>> + * calling function other than just passing through.
>>> + *
>>> + * xfs_attr_remove_iter()
>>> + *	  XFS_DAS_RM_SHRINK ─┐
>>> + *	  (subroutine state) │
>>> + *	                     └─>xfs_attr_node_removename()
>>> + *	                                      │
>>> + *	                                      v
>>> + *	                                   need to
>>> + *	                                shrink tree? ─n─┐
>>> + *	                                      │         │
>>> + *	                                      y         │
>>> + *	                                      │         │
>>> + *	                                      v         │
>>> + *	                              XFS_DAS_RM_SHRINK │
>>> + *	                                      │         │
>>> + *	                                      v         │
>>> + *	                                     done <─────┘
>>> + *
>>> + */
>>> +
>>> +/*
>>> + * Enum values for xfs_delattr_context.da_state
>>> + *
>>> + * These values are used by delayed attribute operations to keep track  of where
>>> + * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
>>> + * calling function to roll the transaction, and then recall the subroutine to
>>> + * finish the operation.  The enum is then used by the subroutine to jump back
>>> + * to where it was and resume executing where it left off.
>>> + */
>>> +enum xfs_delattr_state {
>>> +				      /* Zero is uninitalized */
>>> +	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
>>> +};
>>> +
>>> +/*
>>> + * Defines for xfs_delattr_context.flags
>>> + */
>>> +#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
>>> +#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
>>> +
>>> +/*
>>> + * Context used for keeping track of delayed attribute operations
>>> + */
>>> +struct xfs_delattr_context {
>>> +	struct xfs_da_args      *da_args;
>>> +
>>> +	/* Used in xfs_attr_node_removename to roll through removing blocks */
>>> +	struct xfs_da_state     *da_state;
>>> +	struct xfs_da_state_blk *blk;
>>> +
>>> +	/* Used to keep track of current state of delayed operation */
>>> +	unsigned int            flags;
>>> +	enum xfs_delattr_state  dela_state;
>>> +};
>>> +
>>>   /*========================================================================
>>>    * Function prototypes for the kernel.
>>>    *========================================================================*/
>>> @@ -91,6 +160,10 @@ int xfs_attr_set(struct xfs_da_args *args);
>>>   int xfs_attr_set_args(struct xfs_da_args *args);
>>>   int xfs_has_attr(struct xfs_da_args *args);
>>>   int xfs_attr_remove_args(struct xfs_da_args *args);
>>> +int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
>>> +bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
>>>   bool xfs_attr_namecheck(const void *name, size_t length);
>>> +void xfs_delattr_context_init(struct xfs_delattr_context *dac,
>>> +			      struct xfs_da_args *args);
>>>   
>>>   #endif	/* __XFS_ATTR_H__ */
>>> diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
>>> index 351351c..20521bf 100644
>>> --- a/fs/xfs/libxfs/xfs_attr_leaf.c
>>> +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
>>> @@ -19,8 +19,8 @@
>>>   #include "xfs_bmap_btree.h"
>>>   #include "xfs_bmap.h"
>>>   #include "xfs_attr_sf.h"
>>> -#include "xfs_attr_remote.h"
>>>   #include "xfs_attr.h"
>>> +#include "xfs_attr_remote.h"
>>>   #include "xfs_attr_leaf.h"
>>>   #include "xfs_error.h"
>>>   #include "xfs_trace.h"
>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
>>> index 85dca51..20e4605 100644
>>> --- a/fs/xfs/libxfs/xfs_attr_remote.c
>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.c
>>> @@ -676,12 +676,14 @@ xfs_attr_rmtval_invalidate(
>>>    */
>>>   int
>>>   xfs_attr_rmtval_remove(
>>> -	struct xfs_da_args      *args)
>>> +	struct xfs_da_args		*args)
>>>   {
>>> -	xfs_dablk_t		lblkno;
>>> -	int			blkcnt;
>>> -	int			error = 0;
>>> -	int			retval = 0;
>>> +	xfs_dablk_t			lblkno;
>>> +	int				blkcnt;
>>> +	int				error = 0;
>>> +	struct xfs_delattr_context	dac  = {
>>> +		.da_args	= args,
>>> +	};
>>>   
>>>   	trace_xfs_attr_rmtval_remove(args);
>>>   
>>> @@ -691,19 +693,10 @@ xfs_attr_rmtval_remove(
>>>   	lblkno = args->rmtblkno;
>>>   	blkcnt = args->rmtblkcnt;
>>>   	do {
>>> -		retval = __xfs_attr_rmtval_remove(args);
>>> -		if (retval && retval != EAGAIN)
>>> -			return retval;
>>> -
>>> -		/*
>>> -		 * Close out trans and start the next one in the chain.
>>> -		 */
>>> -		error = xfs_trans_roll_inode(&args->trans, args->dp);
>>> -		if (error)
>>> -			return error;
>>> -	} while (retval == -EAGAIN);
>>> +		error = __xfs_attr_rmtval_remove(&dac);
>>> +	} while (xfs_attr_roll_again(&dac, &error));
>>>   
>>> -	return 0;
>>> +	return error;
>>>   }
>>>   
>>>   /*
>>> @@ -713,9 +706,10 @@ xfs_attr_rmtval_remove(
>>>    */
>>>   int
>>>   __xfs_attr_rmtval_remove(
>>> -	struct xfs_da_args	*args)
>>> +	struct xfs_delattr_context	*dac)
>>>   {
>>> -	int			error, done;
>>> +	struct xfs_da_args		*args = dac->da_args;
>>> +	int				error, done;
>>>   
>>>   	/*
>>>   	 * Unmap value blocks for this attr.
>>> @@ -725,12 +719,10 @@ __xfs_attr_rmtval_remove(
>>>   	if (error)
>>>   		return error;
>>>   
>>> -	error = xfs_defer_finish(&args->trans);
>>> -	if (error)
>>> -		return error;
>>> -
>>> -	if (!done)
>>> +	if (!done) {
>>> +		dac->flags |= XFS_DAC_DEFER_FINISH;
>>>   		return -EAGAIN;
>>> +	}
>>>   
>>>   	return error;
>>>   }
>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
>>> index 9eee615..002fd30 100644
>>> --- a/fs/xfs/libxfs/xfs_attr_remote.h
>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.h
>>> @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args);
>>>   int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
>>>   		xfs_buf_flags_t incore_flags);
>>>   int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
>>> -int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
>>> +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
>>>   #endif /* __XFS_ATTR_REMOTE_H__ */
>>> diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
>>> index bfad669..aaa7e66 100644
>>> --- a/fs/xfs/xfs_attr_inactive.c
>>> +++ b/fs/xfs/xfs_attr_inactive.c
>>> @@ -15,10 +15,10 @@
>>>   #include "xfs_da_format.h"
>>>   #include "xfs_da_btree.h"
>>>   #include "xfs_inode.h"
>>> +#include "xfs_attr.h"
>>>   #include "xfs_attr_remote.h"
>>>   #include "xfs_trans.h"
>>>   #include "xfs_bmap.h"
>>> -#include "xfs_attr.h"
>>>   #include "xfs_attr_leaf.h"
>>>   #include "xfs_quota.h"
>>>   #include "xfs_dir2.h"
>>> -- 
>>> 2.7.4
>>>
>>
Brian Foster July 13, 2020, 1:36 p.m. UTC | #5
On Thu, Jul 09, 2020 at 11:04:02AM -0700, Darrick J. Wong wrote:
> On Thu, Jul 09, 2020 at 09:24:56AM -0400, Brian Foster wrote:
> > On Thu, Jun 25, 2020 at 04:30:16PM -0700, Allison Collins wrote:
> > > This patch modifies the attr remove routines to be delay ready. This
> > > means they no longer roll or commit transactions, but instead return
> > > -EAGAIN to have the calling routine roll and refresh the transaction. In
> > > this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
> > > uses a sort of state machine like switch to keep track of where it was
> > > when EAGAIN was returned. xfs_attr_node_removename has also been
> > > modified to use the switch, and a new version of xfs_attr_remove_args
> > > consists of a simple loop to refresh the transaction until the operation
> > > is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
> > > transaction where ever the existing code used to.
> > > 
> > > Calls to xfs_attr_rmtval_remove are replaced with the delay ready
> > > version __xfs_attr_rmtval_remove. We will rename
> > > __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
> > > done.
> > > 
> > > xfs_attr_rmtval_remove itself is still in use by the set routines (used
> > > during a rename).  For reasons of perserving existing function, we
> > > modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
> > > set.  Similar to how xfs_attr_remove_args does here.  Once we transition
> > > the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
> > > used and will be removed.
> > > 
> > > This patch also adds a new struct xfs_delattr_context, which we will use
> > > to keep track of the current state of an attribute operation. The new
> > > xfs_delattr_state enum is used to track various operations that are in
> > > progress so that we know not to repeat them, and resume where we left
> > > off before EAGAIN was returned to cycle out the transaction. Other
> > > members take the place of local variables that need to retain their
> > > values across multiple function recalls.  See xfs_attr.h for a more
> > > detailed diagram of the states.
> > > 
> > > Signed-off-by: Allison Collins <allison.henderson@oracle.com>
> > > ---
> > >  fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
> > >  fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
> > >  fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
> > >  fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
> > >  fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
> > >  fs/xfs/xfs_attr_inactive.c      |   2 +-
> > >  6 files changed, 208 insertions(+), 66 deletions(-)
> > > 
> > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
> > > index 4b78c86..5c460f4 100644
> > > --- a/fs/xfs/libxfs/xfs_attr.c
> > > +++ b/fs/xfs/libxfs/xfs_attr.c
> > > @@ -53,7 +53,7 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
> > >   */
> > >  STATIC int xfs_attr_node_get(xfs_da_args_t *args);
> > >  STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
> > > -STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
> > > +STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
> > >  STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
> > >  				 struct xfs_da_state **state);
> > >  STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
> > > @@ -264,6 +264,32 @@ xfs_attr_set_shortform(
> > >  }
> > >  
> > >  /*
> > > + * Checks to see if a delayed attribute transaction should be rolled.  If so,
> > > + * also checks for a defer finish.  Transaction is finished and rolled as
> > > + * needed, and returns true of false if the delayed operation should continue.
> > > + */
> > > +bool
> > > +xfs_attr_roll_again(
> > 
> > The function name suggests this is more of a status checking function
> > than one that does actual work. I'd suggest something more like
> > xfs_attr_trans_roll() based on the implementation.
> > 
> > > +	struct xfs_delattr_context	*dac,
> > > +	int				*error)
> > > +{
> > > +	struct xfs_da_args              *args = dac->da_args;
> > > +
> > > +	if (*error != -EAGAIN)
> > > +		return false;
> > > +
> > > +	if (dac->flags & XFS_DAC_DEFER_FINISH) {
> > > +		dac->flags &= ~XFS_DAC_DEFER_FINISH;
> > > +		*error = xfs_defer_finish(&args->trans);
> > > +		if (*error)
> > > +			return false;
> > > +	}
> > > +
> > 
> > I also find the semantics of this function a little confusing. How would
> > a caller distinguish between error == -EAGAIN as passed in from the
> > caller vs. error being set by one of the transaction processing calls?
> 
> Do any of them do that?
> 

Not sure, but I doubt it. I didn't really look because the nature of the
feedback wasn't necessarily to suggest there was a bug, but rather that
the logic of the helper seems overloaded/convoluted. This has two
callers so I don't see why we wouldn't let the caller context determine
when to roll (i.e. error == -EAGAIN), implement the helper as a
deliberate transaction roll and make it return an error code that
properly flows back up the callchain if any part of the roll actually
fails.

> > > +	*error = xfs_trans_roll_inode(&args->trans, args->dp);
> 
> I keep wondering why it's necessary to call xfs_defer_finish immediately
> followed by xfs_trans_roll_inode.  Maybe we've already resolved this
> question, but if we did I don't remember the answer, and the answer
> should be captured as a comment because I never remember. :P
> 
> Oh right, we /did/ resolve this question.  The defer_finish exists to
> finish all the deferred ops so that we return to the caller with a fresh
> transaction and no dfops.
> 

There's also a subtle relationship that completing deferred operations
might be required for crash recovery integrity in certain cases. I've
just discovered a similar issue with my recent insert range changes that
broke things (patch coming later..) because rolling the transaction
alone defers the intents for the rmapbt changes that were associated
with the changes in the primary transaction rather than immediately
logging them. I suspect we might want to think about ways to improve
this interface because currently that is rather easy to miss.

Brian

> Comment, please...
> 
> 	if (dac->flags & XFS_DAC_DEFER_FINISH) {
> 		/*
> 		 * The caller wants us to finish all the deferred ops so
> 		 * that we avoid pinning the log tail with a large
> 		 * number of deferred ops.
> 		 */
> 		dac->flags &= ~XFS_DAC_DEFER_FINISH;
> 		*error = xfs_defer_finish(&args->trans);
> 		if (*error)
> 			return false;
> 	}
> 
> Oh, another question: if we joined the inode to the transaction with
> ilockflags == 0 (i.e. the transaction does not automatically unlock the
> inode) then why do we need the separate _trans_roll_inode call?  Won't
> the _defer_finish call return with the inode joined to a clean
> transaction?
> 
> (Or did we resolve this question already as well, and all we need is a
> comment to remind your forgetful maintainer?)
> 
> --D
> 
> > > +	return *error == 0;
> > > +}
> > > +
> > > +/*
> > >   * Set the attribute specified in @args.
> > >   */
> > >  int
> > > @@ -364,23 +390,47 @@ xfs_has_attr(
> > >   */
> > >  int
> > >  xfs_attr_remove_args(
> > > -	struct xfs_da_args      *args)
> > > +	struct xfs_da_args	*args)
> > >  {
> > > -	struct xfs_inode	*dp = args->dp;
> > > -	int			error;
> > > +	int				error = 0;
> > > +	struct xfs_delattr_context	dac = {
> > > +		.da_args	= args,
> > > +	};
> > > +
> > > +	do {
> > > +		error = xfs_attr_remove_iter(&dac);
> > > +	} while (xfs_attr_roll_again(&dac, &error));
> > > +
> > > +	return error;
> > > +}
> > > +
> > > +/*
> > > + * Remove the attribute specified in @args.
> > > + *
> > > + * This function may return -EAGAIN to signal that the transaction needs to be
> > > + * rolled.  Callers should continue calling this function until they receive a
> > > + * return value other than -EAGAIN.
> > > + */
> > > +int
> > > +xfs_attr_remove_iter(
> > > +	struct xfs_delattr_context	*dac)
> > > +{
> > > +	struct xfs_da_args		*args = dac->da_args;
> > > +	struct xfs_inode		*dp = args->dp;
> > > +
> > > +	if (dac->dela_state == XFS_DAS_RM_SHRINK)
> > > +		goto node;
> > 
> > Thoughts on my comments on this on the previous version?
> > 
> > >  
> > >  	if (!xfs_inode_hasattr(dp)) {
> > > -		error = -ENOATTR;
> > > +		return -ENOATTR;
> > >  	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
> > >  		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
> > > -		error = xfs_attr_shortform_remove(args);
> > > +		return xfs_attr_shortform_remove(args);
> > >  	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
> > > -		error = xfs_attr_leaf_removename(args);
> > > -	} else {
> > > -		error = xfs_attr_node_removename(args);
> > > +		return xfs_attr_leaf_removename(args);
> > >  	}
> > > -
> > > -	return error;
> > > +node:
> > > +	return  xfs_attr_node_removename(dac);
> > >  }
> > >  
> > >  /*
> > ...
> > > @@ -1268,17 +1348,14 @@ xfs_attr_node_removename(
> > >  		error = xfs_da3_join(state);
> > >  		if (error)
> > >  			goto out;
> > > -		error = xfs_defer_finish(&args->trans);
> > > -		if (error)
> > > -			goto out;
> > > -		/*
> > > -		 * Commit the Btree join operation and start a new trans.
> > > -		 */
> > > -		error = xfs_trans_roll_inode(&args->trans, dp);
> > > -		if (error)
> > > -			goto out;
> > > +
> > > +		dac->flags |= XFS_DAC_DEFER_FINISH;
> > > +		dac->dela_state = XFS_DAS_RM_SHRINK;
> > > +		return -EAGAIN;
> > >  	}
> > >  
> > > +das_rm_shrink:
> > > +
> > 
> > I think I also mentioned the tail of this function might sit better in
> > the caller..
> > 
> > Brian
> > 
> > >  	/*
> > >  	 * If the result is small enough, push it all into the inode.
> > >  	 */
> > > diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
> > > index 3e97a93..6c58792 100644
> > > --- a/fs/xfs/libxfs/xfs_attr.h
> > > +++ b/fs/xfs/libxfs/xfs_attr.h
> > > @@ -74,6 +74,75 @@ struct xfs_attr_list_context {
> > >  };
> > >  
> > >  
> > > +/*
> > > + * ========================================================================
> > > + * Structure used to pass context around among the delayed routines.
> > > + * ========================================================================
> > > + */
> > > +
> > > +/*
> > > + * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
> > > + * states indicate places where the function would return -EAGAIN, and then
> > > + * immediately resume from after being recalled by the calling function. States
> > > + * marked as a "subroutine state" indicate that they belong to a subroutine, and
> > > + * so the calling function needs to pass them back to that subroutine to allow
> > > + * it to finish where it left off. But they otherwise do not have a role in the
> > > + * calling function other than just passing through.
> > > + *
> > > + * xfs_attr_remove_iter()
> > > + *	  XFS_DAS_RM_SHRINK ─┐
> > > + *	  (subroutine state) │
> > > + *	                     └─>xfs_attr_node_removename()
> > > + *	                                      │
> > > + *	                                      v
> > > + *	                                   need to
> > > + *	                                shrink tree? ─n─┐
> > > + *	                                      │         │
> > > + *	                                      y         │
> > > + *	                                      │         │
> > > + *	                                      v         │
> > > + *	                              XFS_DAS_RM_SHRINK │
> > > + *	                                      │         │
> > > + *	                                      v         │
> > > + *	                                     done <─────┘
> > > + *
> > > + */
> > > +
> > > +/*
> > > + * Enum values for xfs_delattr_context.da_state
> > > + *
> > > + * These values are used by delayed attribute operations to keep track  of where
> > > + * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
> > > + * calling function to roll the transaction, and then recall the subroutine to
> > > + * finish the operation.  The enum is then used by the subroutine to jump back
> > > + * to where it was and resume executing where it left off.
> > > + */
> > > +enum xfs_delattr_state {
> > > +				      /* Zero is uninitalized */
> > > +	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
> > > +};
> > > +
> > > +/*
> > > + * Defines for xfs_delattr_context.flags
> > > + */
> > > +#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
> > > +#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
> > > +
> > > +/*
> > > + * Context used for keeping track of delayed attribute operations
> > > + */
> > > +struct xfs_delattr_context {
> > > +	struct xfs_da_args      *da_args;
> > > +
> > > +	/* Used in xfs_attr_node_removename to roll through removing blocks */
> > > +	struct xfs_da_state     *da_state;
> > > +	struct xfs_da_state_blk *blk;
> > > +
> > > +	/* Used to keep track of current state of delayed operation */
> > > +	unsigned int            flags;
> > > +	enum xfs_delattr_state  dela_state;
> > > +};
> > > +
> > >  /*========================================================================
> > >   * Function prototypes for the kernel.
> > >   *========================================================================*/
> > > @@ -91,6 +160,10 @@ int xfs_attr_set(struct xfs_da_args *args);
> > >  int xfs_attr_set_args(struct xfs_da_args *args);
> > >  int xfs_has_attr(struct xfs_da_args *args);
> > >  int xfs_attr_remove_args(struct xfs_da_args *args);
> > > +int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
> > > +bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
> > >  bool xfs_attr_namecheck(const void *name, size_t length);
> > > +void xfs_delattr_context_init(struct xfs_delattr_context *dac,
> > > +			      struct xfs_da_args *args);
> > >  
> > >  #endif	/* __XFS_ATTR_H__ */
> > > diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
> > > index 351351c..20521bf 100644
> > > --- a/fs/xfs/libxfs/xfs_attr_leaf.c
> > > +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
> > > @@ -19,8 +19,8 @@
> > >  #include "xfs_bmap_btree.h"
> > >  #include "xfs_bmap.h"
> > >  #include "xfs_attr_sf.h"
> > > -#include "xfs_attr_remote.h"
> > >  #include "xfs_attr.h"
> > > +#include "xfs_attr_remote.h"
> > >  #include "xfs_attr_leaf.h"
> > >  #include "xfs_error.h"
> > >  #include "xfs_trace.h"
> > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
> > > index 85dca51..20e4605 100644
> > > --- a/fs/xfs/libxfs/xfs_attr_remote.c
> > > +++ b/fs/xfs/libxfs/xfs_attr_remote.c
> > > @@ -676,12 +676,14 @@ xfs_attr_rmtval_invalidate(
> > >   */
> > >  int
> > >  xfs_attr_rmtval_remove(
> > > -	struct xfs_da_args      *args)
> > > +	struct xfs_da_args		*args)
> > >  {
> > > -	xfs_dablk_t		lblkno;
> > > -	int			blkcnt;
> > > -	int			error = 0;
> > > -	int			retval = 0;
> > > +	xfs_dablk_t			lblkno;
> > > +	int				blkcnt;
> > > +	int				error = 0;
> > > +	struct xfs_delattr_context	dac  = {
> > > +		.da_args	= args,
> > > +	};
> > >  
> > >  	trace_xfs_attr_rmtval_remove(args);
> > >  
> > > @@ -691,19 +693,10 @@ xfs_attr_rmtval_remove(
> > >  	lblkno = args->rmtblkno;
> > >  	blkcnt = args->rmtblkcnt;
> > >  	do {
> > > -		retval = __xfs_attr_rmtval_remove(args);
> > > -		if (retval && retval != EAGAIN)
> > > -			return retval;
> > > -
> > > -		/*
> > > -		 * Close out trans and start the next one in the chain.
> > > -		 */
> > > -		error = xfs_trans_roll_inode(&args->trans, args->dp);
> > > -		if (error)
> > > -			return error;
> > > -	} while (retval == -EAGAIN);
> > > +		error = __xfs_attr_rmtval_remove(&dac);
> > > +	} while (xfs_attr_roll_again(&dac, &error));
> > >  
> > > -	return 0;
> > > +	return error;
> > >  }
> > >  
> > >  /*
> > > @@ -713,9 +706,10 @@ xfs_attr_rmtval_remove(
> > >   */
> > >  int
> > >  __xfs_attr_rmtval_remove(
> > > -	struct xfs_da_args	*args)
> > > +	struct xfs_delattr_context	*dac)
> > >  {
> > > -	int			error, done;
> > > +	struct xfs_da_args		*args = dac->da_args;
> > > +	int				error, done;
> > >  
> > >  	/*
> > >  	 * Unmap value blocks for this attr.
> > > @@ -725,12 +719,10 @@ __xfs_attr_rmtval_remove(
> > >  	if (error)
> > >  		return error;
> > >  
> > > -	error = xfs_defer_finish(&args->trans);
> > > -	if (error)
> > > -		return error;
> > > -
> > > -	if (!done)
> > > +	if (!done) {
> > > +		dac->flags |= XFS_DAC_DEFER_FINISH;
> > >  		return -EAGAIN;
> > > +	}
> > >  
> > >  	return error;
> > >  }
> > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
> > > index 9eee615..002fd30 100644
> > > --- a/fs/xfs/libxfs/xfs_attr_remote.h
> > > +++ b/fs/xfs/libxfs/xfs_attr_remote.h
> > > @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args);
> > >  int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
> > >  		xfs_buf_flags_t incore_flags);
> > >  int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
> > > -int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
> > > +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
> > >  #endif /* __XFS_ATTR_REMOTE_H__ */
> > > diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
> > > index bfad669..aaa7e66 100644
> > > --- a/fs/xfs/xfs_attr_inactive.c
> > > +++ b/fs/xfs/xfs_attr_inactive.c
> > > @@ -15,10 +15,10 @@
> > >  #include "xfs_da_format.h"
> > >  #include "xfs_da_btree.h"
> > >  #include "xfs_inode.h"
> > > +#include "xfs_attr.h"
> > >  #include "xfs_attr_remote.h"
> > >  #include "xfs_trans.h"
> > >  #include "xfs_bmap.h"
> > > -#include "xfs_attr.h"
> > >  #include "xfs_attr_leaf.h"
> > >  #include "xfs_quota.h"
> > >  #include "xfs_dir2.h"
> > > -- 
> > > 2.7.4
> > > 
> > 
>
Brian Foster July 13, 2020, 1:58 p.m. UTC | #6
On Thu, Jul 09, 2020 at 03:02:08PM -0700, Allison Collins wrote:
> 
> 
> On 7/9/20 6:24 AM, Brian Foster wrote:
> > On Thu, Jun 25, 2020 at 04:30:16PM -0700, Allison Collins wrote:
> > > This patch modifies the attr remove routines to be delay ready. This
> > > means they no longer roll or commit transactions, but instead return
> > > -EAGAIN to have the calling routine roll and refresh the transaction. In
> > > this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
> > > uses a sort of state machine like switch to keep track of where it was
> > > when EAGAIN was returned. xfs_attr_node_removename has also been
> > > modified to use the switch, and a new version of xfs_attr_remove_args
> > > consists of a simple loop to refresh the transaction until the operation
> > > is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
> > > transaction where ever the existing code used to.
> > > 
> > > Calls to xfs_attr_rmtval_remove are replaced with the delay ready
> > > version __xfs_attr_rmtval_remove. We will rename
> > > __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
> > > done.
> > > 
> > > xfs_attr_rmtval_remove itself is still in use by the set routines (used
> > > during a rename).  For reasons of perserving existing function, we
> > > modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
> > > set.  Similar to how xfs_attr_remove_args does here.  Once we transition
> > > the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
> > > used and will be removed.
> > > 
> > > This patch also adds a new struct xfs_delattr_context, which we will use
> > > to keep track of the current state of an attribute operation. The new
> > > xfs_delattr_state enum is used to track various operations that are in
> > > progress so that we know not to repeat them, and resume where we left
> > > off before EAGAIN was returned to cycle out the transaction. Other
> > > members take the place of local variables that need to retain their
> > > values across multiple function recalls.  See xfs_attr.h for a more
> > > detailed diagram of the states.
> > > 
> > > Signed-off-by: Allison Collins <allison.henderson@oracle.com>
> > > ---
> > >   fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
> > >   fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
> > >   fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
> > >   fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
> > >   fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
> > >   fs/xfs/xfs_attr_inactive.c      |   2 +-
> > >   6 files changed, 208 insertions(+), 66 deletions(-)
> > > 
> > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
> > > index 4b78c86..5c460f4 100644
> > > --- a/fs/xfs/libxfs/xfs_attr.c
> > > +++ b/fs/xfs/libxfs/xfs_attr.c
> > > @@ -53,7 +53,7 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
> > >    */
> > >   STATIC int xfs_attr_node_get(xfs_da_args_t *args);
> > >   STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
> > > -STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
> > > +STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
> > >   STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
> > >   				 struct xfs_da_state **state);
> > >   STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
> > > @@ -264,6 +264,32 @@ xfs_attr_set_shortform(
> > >   }
> > >   /*
> > > + * Checks to see if a delayed attribute transaction should be rolled.  If so,
> > > + * also checks for a defer finish.  Transaction is finished and rolled as
> > > + * needed, and returns true of false if the delayed operation should continue.
> > > + */
> > > +bool
> > > +xfs_attr_roll_again(
> > 
> > The function name suggests this is more of a status checking function
> > than one that does actual work. I'd suggest something more like
> > xfs_attr_trans_roll() based on the implementation.
> Sure, I believe Darrick had asked for this helper in the last review, so as
> long as folks are in agreement with the name, that is fine.
> 
> > 
> > > +	struct xfs_delattr_context	*dac,
> > > +	int				*error)
> > > +{
> > > +	struct xfs_da_args              *args = dac->da_args;
> > > +
> > > +	if (*error != -EAGAIN)
> > > +		return false;
> > > +
> > > +	if (dac->flags & XFS_DAC_DEFER_FINISH) {
> > > +		dac->flags &= ~XFS_DAC_DEFER_FINISH;
> > > +		*error = xfs_defer_finish(&args->trans);
> > > +		if (*error)
> > > +			return false;
> > > +	}
> > > +
> > 
> > I also find the semantics of this function a little confusing. How would
> > a caller distinguish between error == -EAGAIN as passed in from the
> > caller vs. error being set by one of the transaction processing calls?
> I believe the helper was just meant to be a straight hoist from the v9
> version.  The error parameter here is an input for use in this scope, it's
> not for the caller to do anything with.  Other than to pipe it back in.  The
> caller just needs to know whether or not to continue.
> 

I don't mind having the helper and afaict it looks correct, but I find
the combination of error checking and functional change just obfuscate
the logic in the caller. I could see the reason for this kind of
implementation if it were called all over the place and we wanted to
reduce boilerplate error checking code or something, but this only has a
handful of callers. IMO this would be much cleaner as:

	int xfs_attr_trans_roll(struct xfs_delattr_context *);

... and let the caller care about when to use it and what to do with
resulting errors.

> That signature is a bit odd, though this pattern does appear in a few places
> around the code, so it does help consolidate a bit.  I suppose it's just a
> question of which people prefer.
> 
> > 
> > > +	*error = xfs_trans_roll_inode(&args->trans, args->dp);
> > > +	return *error == 0;
> > > +}
> > > +
> > > +/*
> > >    * Set the attribute specified in @args.
> > >    */
> > >   int
> > > @@ -364,23 +390,47 @@ xfs_has_attr(
> > >    */
> > >   int
> > >   xfs_attr_remove_args(
> > > -	struct xfs_da_args      *args)
> > > +	struct xfs_da_args	*args)
> > >   {
> > > -	struct xfs_inode	*dp = args->dp;
> > > -	int			error;
> > > +	int				error = 0;
> > > +	struct xfs_delattr_context	dac = {
> > > +		.da_args	= args,
> > > +	};
> > > +
> > > +	do {
> > > +		error = xfs_attr_remove_iter(&dac);
> > > +	} while (xfs_attr_roll_again(&dac, &error));
> > > +
> > > +	return error;
> > > +}
> > > +
> > > +/*
> > > + * Remove the attribute specified in @args.
> > > + *
> > > + * This function may return -EAGAIN to signal that the transaction needs to be
> > > + * rolled.  Callers should continue calling this function until they receive a
> > > + * return value other than -EAGAIN.
> > > + */
> > > +int
> > > +xfs_attr_remove_iter(
> > > +	struct xfs_delattr_context	*dac)
> > > +{
> > > +	struct xfs_da_args		*args = dac->da_args;
> > > +	struct xfs_inode		*dp = args->dp;
> > > +
> > > +	if (dac->dela_state == XFS_DAS_RM_SHRINK)
> > > +		goto node;
> > 
> > Thoughts on my comments on this on the previous version?
> Apologies, I must have missed responding to it, I had a few family
> emergencies at the time and must have lost track.  I do believe we need the
> goto because in the process of shrinking, it is possible that we turn back
> into a block which would cause us to fall into the wrong if/else logic with
> out it. I will run the test and double check though. I'm pretty sure I
> recall running into that though.
> 

Ok. No worries, I just saw unaddressed feedback and wanted to make sure
it wasn't lost...

Brian

> > 
> > >   	if (!xfs_inode_hasattr(dp)) {
> > > -		error = -ENOATTR;
> > > +		return -ENOATTR;
> > >   	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
> > >   		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
> > > -		error = xfs_attr_shortform_remove(args);
> > > +		return xfs_attr_shortform_remove(args);
> > >   	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
> > > -		error = xfs_attr_leaf_removename(args);
> > > -	} else {
> > > -		error = xfs_attr_node_removename(args);
> > > +		return xfs_attr_leaf_removename(args);
> > >   	}
> > > -
> > > -	return error;
> > > +node:
> > > +	return  xfs_attr_node_removename(dac);
> > >   }
> > >   /*
> > ...
> > > @@ -1268,17 +1348,14 @@ xfs_attr_node_removename(
> > >   		error = xfs_da3_join(state);
> > >   		if (error)
> > >   			goto out;
> > > -		error = xfs_defer_finish(&args->trans);
> > > -		if (error)
> > > -			goto out;
> > > -		/*
> > > -		 * Commit the Btree join operation and start a new trans.
> > > -		 */
> > > -		error = xfs_trans_roll_inode(&args->trans, dp);
> > > -		if (error)
> > > -			goto out;
> > > +
> > > +		dac->flags |= XFS_DAC_DEFER_FINISH;
> > > +		dac->dela_state = XFS_DAS_RM_SHRINK;
> > > +		return -EAGAIN;
> > >   	}
> > > +das_rm_shrink:
> > > +
> > 
> > I think I also mentioned the tail of this function might sit better in
> > the caller..
> I see, yes I'll see if I could pull that up.  If I missed anything else,
> please call it out.  Thank you for your reviews!
> 
> Allison
> 
> > 
> > Brian
> > 
> > >   	/*
> > >   	 * If the result is small enough, push it all into the inode.
> > >   	 */
> > > diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
> > > index 3e97a93..6c58792 100644
> > > --- a/fs/xfs/libxfs/xfs_attr.h
> > > +++ b/fs/xfs/libxfs/xfs_attr.h
> > > @@ -74,6 +74,75 @@ struct xfs_attr_list_context {
> > >   };
> > > +/*
> > > + * ========================================================================
> > > + * Structure used to pass context around among the delayed routines.
> > > + * ========================================================================
> > > + */
> > > +
> > > +/*
> > > + * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
> > > + * states indicate places where the function would return -EAGAIN, and then
> > > + * immediately resume from after being recalled by the calling function. States
> > > + * marked as a "subroutine state" indicate that they belong to a subroutine, and
> > > + * so the calling function needs to pass them back to that subroutine to allow
> > > + * it to finish where it left off. But they otherwise do not have a role in the
> > > + * calling function other than just passing through.
> > > + *
> > > + * xfs_attr_remove_iter()
> > > + *	  XFS_DAS_RM_SHRINK ─┐
> > > + *	  (subroutine state) │
> > > + *	                     └─>xfs_attr_node_removename()
> > > + *	                                      │
> > > + *	                                      v
> > > + *	                                   need to
> > > + *	                                shrink tree? ─n─┐
> > > + *	                                      │         │
> > > + *	                                      y         │
> > > + *	                                      │         │
> > > + *	                                      v         │
> > > + *	                              XFS_DAS_RM_SHRINK │
> > > + *	                                      │         │
> > > + *	                                      v         │
> > > + *	                                     done <─────┘
> > > + *
> > > + */
> > > +
> > > +/*
> > > + * Enum values for xfs_delattr_context.da_state
> > > + *
> > > + * These values are used by delayed attribute operations to keep track  of where
> > > + * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
> > > + * calling function to roll the transaction, and then recall the subroutine to
> > > + * finish the operation.  The enum is then used by the subroutine to jump back
> > > + * to where it was and resume executing where it left off.
> > > + */
> > > +enum xfs_delattr_state {
> > > +				      /* Zero is uninitalized */
> > > +	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
> > > +};
> > > +
> > > +/*
> > > + * Defines for xfs_delattr_context.flags
> > > + */
> > > +#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
> > > +#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
> > > +
> > > +/*
> > > + * Context used for keeping track of delayed attribute operations
> > > + */
> > > +struct xfs_delattr_context {
> > > +	struct xfs_da_args      *da_args;
> > > +
> > > +	/* Used in xfs_attr_node_removename to roll through removing blocks */
> > > +	struct xfs_da_state     *da_state;
> > > +	struct xfs_da_state_blk *blk;
> > > +
> > > +	/* Used to keep track of current state of delayed operation */
> > > +	unsigned int            flags;
> > > +	enum xfs_delattr_state  dela_state;
> > > +};
> > > +
> > >   /*========================================================================
> > >    * Function prototypes for the kernel.
> > >    *========================================================================*/
> > > @@ -91,6 +160,10 @@ int xfs_attr_set(struct xfs_da_args *args);
> > >   int xfs_attr_set_args(struct xfs_da_args *args);
> > >   int xfs_has_attr(struct xfs_da_args *args);
> > >   int xfs_attr_remove_args(struct xfs_da_args *args);
> > > +int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
> > > +bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
> > >   bool xfs_attr_namecheck(const void *name, size_t length);
> > > +void xfs_delattr_context_init(struct xfs_delattr_context *dac,
> > > +			      struct xfs_da_args *args);
> > >   #endif	/* __XFS_ATTR_H__ */
> > > diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
> > > index 351351c..20521bf 100644
> > > --- a/fs/xfs/libxfs/xfs_attr_leaf.c
> > > +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
> > > @@ -19,8 +19,8 @@
> > >   #include "xfs_bmap_btree.h"
> > >   #include "xfs_bmap.h"
> > >   #include "xfs_attr_sf.h"
> > > -#include "xfs_attr_remote.h"
> > >   #include "xfs_attr.h"
> > > +#include "xfs_attr_remote.h"
> > >   #include "xfs_attr_leaf.h"
> > >   #include "xfs_error.h"
> > >   #include "xfs_trace.h"
> > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
> > > index 85dca51..20e4605 100644
> > > --- a/fs/xfs/libxfs/xfs_attr_remote.c
> > > +++ b/fs/xfs/libxfs/xfs_attr_remote.c
> > > @@ -676,12 +676,14 @@ xfs_attr_rmtval_invalidate(
> > >    */
> > >   int
> > >   xfs_attr_rmtval_remove(
> > > -	struct xfs_da_args      *args)
> > > +	struct xfs_da_args		*args)
> > >   {
> > > -	xfs_dablk_t		lblkno;
> > > -	int			blkcnt;
> > > -	int			error = 0;
> > > -	int			retval = 0;
> > > +	xfs_dablk_t			lblkno;
> > > +	int				blkcnt;
> > > +	int				error = 0;
> > > +	struct xfs_delattr_context	dac  = {
> > > +		.da_args	= args,
> > > +	};
> > >   	trace_xfs_attr_rmtval_remove(args);
> > > @@ -691,19 +693,10 @@ xfs_attr_rmtval_remove(
> > >   	lblkno = args->rmtblkno;
> > >   	blkcnt = args->rmtblkcnt;
> > >   	do {
> > > -		retval = __xfs_attr_rmtval_remove(args);
> > > -		if (retval && retval != EAGAIN)
> > > -			return retval;
> > > -
> > > -		/*
> > > -		 * Close out trans and start the next one in the chain.
> > > -		 */
> > > -		error = xfs_trans_roll_inode(&args->trans, args->dp);
> > > -		if (error)
> > > -			return error;
> > > -	} while (retval == -EAGAIN);
> > > +		error = __xfs_attr_rmtval_remove(&dac);
> > > +	} while (xfs_attr_roll_again(&dac, &error));
> > > -	return 0;
> > > +	return error;
> > >   }
> > >   /*
> > > @@ -713,9 +706,10 @@ xfs_attr_rmtval_remove(
> > >    */
> > >   int
> > >   __xfs_attr_rmtval_remove(
> > > -	struct xfs_da_args	*args)
> > > +	struct xfs_delattr_context	*dac)
> > >   {
> > > -	int			error, done;
> > > +	struct xfs_da_args		*args = dac->da_args;
> > > +	int				error, done;
> > >   	/*
> > >   	 * Unmap value blocks for this attr.
> > > @@ -725,12 +719,10 @@ __xfs_attr_rmtval_remove(
> > >   	if (error)
> > >   		return error;
> > > -	error = xfs_defer_finish(&args->trans);
> > > -	if (error)
> > > -		return error;
> > > -
> > > -	if (!done)
> > > +	if (!done) {
> > > +		dac->flags |= XFS_DAC_DEFER_FINISH;
> > >   		return -EAGAIN;
> > > +	}
> > >   	return error;
> > >   }
> > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
> > > index 9eee615..002fd30 100644
> > > --- a/fs/xfs/libxfs/xfs_attr_remote.h
> > > +++ b/fs/xfs/libxfs/xfs_attr_remote.h
> > > @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args);
> > >   int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
> > >   		xfs_buf_flags_t incore_flags);
> > >   int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
> > > -int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
> > > +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
> > >   #endif /* __XFS_ATTR_REMOTE_H__ */
> > > diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
> > > index bfad669..aaa7e66 100644
> > > --- a/fs/xfs/xfs_attr_inactive.c
> > > +++ b/fs/xfs/xfs_attr_inactive.c
> > > @@ -15,10 +15,10 @@
> > >   #include "xfs_da_format.h"
> > >   #include "xfs_da_btree.h"
> > >   #include "xfs_inode.h"
> > > +#include "xfs_attr.h"
> > >   #include "xfs_attr_remote.h"
> > >   #include "xfs_trans.h"
> > >   #include "xfs_bmap.h"
> > > -#include "xfs_attr.h"
> > >   #include "xfs_attr_leaf.h"
> > >   #include "xfs_quota.h"
> > >   #include "xfs_dir2.h"
> > > -- 
> > > 2.7.4
> > > 
> > 
>
Allison Collins July 15, 2020, 5:04 a.m. UTC | #7
On 7/13/20 6:58 AM, Brian Foster wrote:
> On Thu, Jul 09, 2020 at 03:02:08PM -0700, Allison Collins wrote:
>>
>>
>> On 7/9/20 6:24 AM, Brian Foster wrote:
>>> On Thu, Jun 25, 2020 at 04:30:16PM -0700, Allison Collins wrote:
>>>> This patch modifies the attr remove routines to be delay ready. This
>>>> means they no longer roll or commit transactions, but instead return
>>>> -EAGAIN to have the calling routine roll and refresh the transaction. In
>>>> this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which
>>>> uses a sort of state machine like switch to keep track of where it was
>>>> when EAGAIN was returned. xfs_attr_node_removename has also been
>>>> modified to use the switch, and a new version of xfs_attr_remove_args
>>>> consists of a simple loop to refresh the transaction until the operation
>>>> is completed.  A new XFS_DAC_DEFER_FINISH flag is used to finish the
>>>> transaction where ever the existing code used to.
>>>>
>>>> Calls to xfs_attr_rmtval_remove are replaced with the delay ready
>>>> version __xfs_attr_rmtval_remove. We will rename
>>>> __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are
>>>> done.
>>>>
>>>> xfs_attr_rmtval_remove itself is still in use by the set routines (used
>>>> during a rename).  For reasons of perserving existing function, we
>>>> modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is
>>>> set.  Similar to how xfs_attr_remove_args does here.  Once we transition
>>>> the set routines to be delay ready, xfs_attr_rmtval_remove is no longer
>>>> used and will be removed.
>>>>
>>>> This patch also adds a new struct xfs_delattr_context, which we will use
>>>> to keep track of the current state of an attribute operation. The new
>>>> xfs_delattr_state enum is used to track various operations that are in
>>>> progress so that we know not to repeat them, and resume where we left
>>>> off before EAGAIN was returned to cycle out the transaction. Other
>>>> members take the place of local variables that need to retain their
>>>> values across multiple function recalls.  See xfs_attr.h for a more
>>>> detailed diagram of the states.
>>>>
>>>> Signed-off-by: Allison Collins <allison.henderson@oracle.com>
>>>> ---
>>>>    fs/xfs/libxfs/xfs_attr.c        | 155 ++++++++++++++++++++++++++++++----------
>>>>    fs/xfs/libxfs/xfs_attr.h        |  73 +++++++++++++++++++
>>>>    fs/xfs/libxfs/xfs_attr_leaf.c   |   2 +-
>>>>    fs/xfs/libxfs/xfs_attr_remote.c |  40 +++++------
>>>>    fs/xfs/libxfs/xfs_attr_remote.h |   2 +-
>>>>    fs/xfs/xfs_attr_inactive.c      |   2 +-
>>>>    6 files changed, 208 insertions(+), 66 deletions(-)
>>>>
>>>> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
>>>> index 4b78c86..5c460f4 100644
>>>> --- a/fs/xfs/libxfs/xfs_attr.c
>>>> +++ b/fs/xfs/libxfs/xfs_attr.c
>>>> @@ -53,7 +53,7 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
>>>>     */
>>>>    STATIC int xfs_attr_node_get(xfs_da_args_t *args);
>>>>    STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
>>>> -STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
>>>> +STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
>>>>    STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
>>>>    				 struct xfs_da_state **state);
>>>>    STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
>>>> @@ -264,6 +264,32 @@ xfs_attr_set_shortform(
>>>>    }
>>>>    /*
>>>> + * Checks to see if a delayed attribute transaction should be rolled.  If so,
>>>> + * also checks for a defer finish.  Transaction is finished and rolled as
>>>> + * needed, and returns true of false if the delayed operation should continue.
>>>> + */
>>>> +bool
>>>> +xfs_attr_roll_again(
>>>
>>> The function name suggests this is more of a status checking function
>>> than one that does actual work. I'd suggest something more like
>>> xfs_attr_trans_roll() based on the implementation.
>> Sure, I believe Darrick had asked for this helper in the last review, so as
>> long as folks are in agreement with the name, that is fine.
>>
>>>
>>>> +	struct xfs_delattr_context	*dac,
>>>> +	int				*error)
>>>> +{
>>>> +	struct xfs_da_args              *args = dac->da_args;
>>>> +
>>>> +	if (*error != -EAGAIN)
>>>> +		return false;
>>>> +
>>>> +	if (dac->flags & XFS_DAC_DEFER_FINISH) {
>>>> +		dac->flags &= ~XFS_DAC_DEFER_FINISH;
>>>> +		*error = xfs_defer_finish(&args->trans);
>>>> +		if (*error)
>>>> +			return false;
>>>> +	}
>>>> +
>>>
>>> I also find the semantics of this function a little confusing. How would
>>> a caller distinguish between error == -EAGAIN as passed in from the
>>> caller vs. error being set by one of the transaction processing calls?
>> I believe the helper was just meant to be a straight hoist from the v9
>> version.  The error parameter here is an input for use in this scope, it's
>> not for the caller to do anything with.  Other than to pipe it back in.  The
>> caller just needs to know whether or not to continue.
>>
> 
> I don't mind having the helper and afaict it looks correct, but I find
> the combination of error checking and functional change just obfuscate
> the logic in the caller. I could see the reason for this kind of
> implementation if it were called all over the place and we wanted to
> reduce boilerplate error checking code or something, but this only has a
> handful of callers. IMO this would be much cleaner as:
> 
> 	int xfs_attr_trans_roll(struct xfs_delattr_context *);
> 
> ... and let the caller care about when to use it and what to do with
> resulting errors.

Sure I am fine with that as long as everyone else is.  Unless I hear 
otherwise, I will pull the error handling into the caller.

> 
>> That signature is a bit odd, though this pattern does appear in a few places
>> around the code, so it does help consolidate a bit.  I suppose it's just a
>> question of which people prefer.
>>
>>>
>>>> +	*error = xfs_trans_roll_inode(&args->trans, args->dp);
>>>> +	return *error == 0;
>>>> +}
>>>> +
>>>> +/*
>>>>     * Set the attribute specified in @args.
>>>>     */
>>>>    int
>>>> @@ -364,23 +390,47 @@ xfs_has_attr(
>>>>     */
>>>>    int
>>>>    xfs_attr_remove_args(
>>>> -	struct xfs_da_args      *args)
>>>> +	struct xfs_da_args	*args)
>>>>    {
>>>> -	struct xfs_inode	*dp = args->dp;
>>>> -	int			error;
>>>> +	int				error = 0;
>>>> +	struct xfs_delattr_context	dac = {
>>>> +		.da_args	= args,
>>>> +	};
>>>> +
>>>> +	do {
>>>> +		error = xfs_attr_remove_iter(&dac);
>>>> +	} while (xfs_attr_roll_again(&dac, &error));
>>>> +
>>>> +	return error;
>>>> +}
>>>> +
>>>> +/*
>>>> + * Remove the attribute specified in @args.
>>>> + *
>>>> + * This function may return -EAGAIN to signal that the transaction needs to be
>>>> + * rolled.  Callers should continue calling this function until they receive a
>>>> + * return value other than -EAGAIN.
>>>> + */
>>>> +int
>>>> +xfs_attr_remove_iter(
>>>> +	struct xfs_delattr_context	*dac)
>>>> +{
>>>> +	struct xfs_da_args		*args = dac->da_args;
>>>> +	struct xfs_inode		*dp = args->dp;
>>>> +
>>>> +	if (dac->dela_state == XFS_DAS_RM_SHRINK)
>>>> +		goto node;
>>>
>>> Thoughts on my comments on this on the previous version?
>> Apologies, I must have missed responding to it, I had a few family
>> emergencies at the time and must have lost track.  I do believe we need the
>> goto because in the process of shrinking, it is possible that we turn back
>> into a block which would cause us to fall into the wrong if/else logic with
>> out it. I will run the test and double check though. I'm pretty sure I
>> recall running into that though.
>>
> 
> Ok. No worries, I just saw unaddressed feedback and wanted to make sure
> it wasn't lost...
> 
> Brian

Yep, I checked this part, we do need the goto, or we do indeed get stuck 
in the wrong code path on re-entry.  In any case, will adjust the helper 
function mentioned above.  Thanks for the reviews!

Allison

> 
>>>
>>>>    	if (!xfs_inode_hasattr(dp)) {
>>>> -		error = -ENOATTR;
>>>> +		return -ENOATTR;
>>>>    	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
>>>>    		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
>>>> -		error = xfs_attr_shortform_remove(args);
>>>> +		return xfs_attr_shortform_remove(args);
>>>>    	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
>>>> -		error = xfs_attr_leaf_removename(args);
>>>> -	} else {
>>>> -		error = xfs_attr_node_removename(args);
>>>> +		return xfs_attr_leaf_removename(args);
>>>>    	}
>>>> -
>>>> -	return error;
>>>> +node:
>>>> +	return  xfs_attr_node_removename(dac);
>>>>    }
>>>>    /*
>>> ...
>>>> @@ -1268,17 +1348,14 @@ xfs_attr_node_removename(
>>>>    		error = xfs_da3_join(state);
>>>>    		if (error)
>>>>    			goto out;
>>>> -		error = xfs_defer_finish(&args->trans);
>>>> -		if (error)
>>>> -			goto out;
>>>> -		/*
>>>> -		 * Commit the Btree join operation and start a new trans.
>>>> -		 */
>>>> -		error = xfs_trans_roll_inode(&args->trans, dp);
>>>> -		if (error)
>>>> -			goto out;
>>>> +
>>>> +		dac->flags |= XFS_DAC_DEFER_FINISH;
>>>> +		dac->dela_state = XFS_DAS_RM_SHRINK;
>>>> +		return -EAGAIN;
>>>>    	}
>>>> +das_rm_shrink:
>>>> +
>>>
>>> I think I also mentioned the tail of this function might sit better in
>>> the caller..
>> I see, yes I'll see if I could pull that up.  If I missed anything else,
>> please call it out.  Thank you for your reviews!
>>
>> Allison
>>
>>>
>>> Brian
>>>
>>>>    	/*
>>>>    	 * If the result is small enough, push it all into the inode.
>>>>    	 */
>>>> diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
>>>> index 3e97a93..6c58792 100644
>>>> --- a/fs/xfs/libxfs/xfs_attr.h
>>>> +++ b/fs/xfs/libxfs/xfs_attr.h
>>>> @@ -74,6 +74,75 @@ struct xfs_attr_list_context {
>>>>    };
>>>> +/*
>>>> + * ========================================================================
>>>> + * Structure used to pass context around among the delayed routines.
>>>> + * ========================================================================
>>>> + */
>>>> +
>>>> +/*
>>>> + * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
>>>> + * states indicate places where the function would return -EAGAIN, and then
>>>> + * immediately resume from after being recalled by the calling function. States
>>>> + * marked as a "subroutine state" indicate that they belong to a subroutine, and
>>>> + * so the calling function needs to pass them back to that subroutine to allow
>>>> + * it to finish where it left off. But they otherwise do not have a role in the
>>>> + * calling function other than just passing through.
>>>> + *
>>>> + * xfs_attr_remove_iter()
>>>> + *	  XFS_DAS_RM_SHRINK ─┐
>>>> + *	  (subroutine state) │
>>>> + *	                     └─>xfs_attr_node_removename()
>>>> + *	                                      │
>>>> + *	                                      v
>>>> + *	                                   need to
>>>> + *	                                shrink tree? ─n─┐
>>>> + *	                                      │         │
>>>> + *	                                      y         │
>>>> + *	                                      │         │
>>>> + *	                                      v         │
>>>> + *	                              XFS_DAS_RM_SHRINK │
>>>> + *	                                      │         │
>>>> + *	                                      v         │
>>>> + *	                                     done <─────┘
>>>> + *
>>>> + */
>>>> +
>>>> +/*
>>>> + * Enum values for xfs_delattr_context.da_state
>>>> + *
>>>> + * These values are used by delayed attribute operations to keep track  of where
>>>> + * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
>>>> + * calling function to roll the transaction, and then recall the subroutine to
>>>> + * finish the operation.  The enum is then used by the subroutine to jump back
>>>> + * to where it was and resume executing where it left off.
>>>> + */
>>>> +enum xfs_delattr_state {
>>>> +				      /* Zero is uninitalized */
>>>> +	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
>>>> +};
>>>> +
>>>> +/*
>>>> + * Defines for xfs_delattr_context.flags
>>>> + */
>>>> +#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
>>>> +#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
>>>> +
>>>> +/*
>>>> + * Context used for keeping track of delayed attribute operations
>>>> + */
>>>> +struct xfs_delattr_context {
>>>> +	struct xfs_da_args      *da_args;
>>>> +
>>>> +	/* Used in xfs_attr_node_removename to roll through removing blocks */
>>>> +	struct xfs_da_state     *da_state;
>>>> +	struct xfs_da_state_blk *blk;
>>>> +
>>>> +	/* Used to keep track of current state of delayed operation */
>>>> +	unsigned int            flags;
>>>> +	enum xfs_delattr_state  dela_state;
>>>> +};
>>>> +
>>>>    /*========================================================================
>>>>     * Function prototypes for the kernel.
>>>>     *========================================================================*/
>>>> @@ -91,6 +160,10 @@ int xfs_attr_set(struct xfs_da_args *args);
>>>>    int xfs_attr_set_args(struct xfs_da_args *args);
>>>>    int xfs_has_attr(struct xfs_da_args *args);
>>>>    int xfs_attr_remove_args(struct xfs_da_args *args);
>>>> +int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
>>>> +bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
>>>>    bool xfs_attr_namecheck(const void *name, size_t length);
>>>> +void xfs_delattr_context_init(struct xfs_delattr_context *dac,
>>>> +			      struct xfs_da_args *args);
>>>>    #endif	/* __XFS_ATTR_H__ */
>>>> diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
>>>> index 351351c..20521bf 100644
>>>> --- a/fs/xfs/libxfs/xfs_attr_leaf.c
>>>> +++ b/fs/xfs/libxfs/xfs_attr_leaf.c
>>>> @@ -19,8 +19,8 @@
>>>>    #include "xfs_bmap_btree.h"
>>>>    #include "xfs_bmap.h"
>>>>    #include "xfs_attr_sf.h"
>>>> -#include "xfs_attr_remote.h"
>>>>    #include "xfs_attr.h"
>>>> +#include "xfs_attr_remote.h"
>>>>    #include "xfs_attr_leaf.h"
>>>>    #include "xfs_error.h"
>>>>    #include "xfs_trace.h"
>>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
>>>> index 85dca51..20e4605 100644
>>>> --- a/fs/xfs/libxfs/xfs_attr_remote.c
>>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.c
>>>> @@ -676,12 +676,14 @@ xfs_attr_rmtval_invalidate(
>>>>     */
>>>>    int
>>>>    xfs_attr_rmtval_remove(
>>>> -	struct xfs_da_args      *args)
>>>> +	struct xfs_da_args		*args)
>>>>    {
>>>> -	xfs_dablk_t		lblkno;
>>>> -	int			blkcnt;
>>>> -	int			error = 0;
>>>> -	int			retval = 0;
>>>> +	xfs_dablk_t			lblkno;
>>>> +	int				blkcnt;
>>>> +	int				error = 0;
>>>> +	struct xfs_delattr_context	dac  = {
>>>> +		.da_args	= args,
>>>> +	};
>>>>    	trace_xfs_attr_rmtval_remove(args);
>>>> @@ -691,19 +693,10 @@ xfs_attr_rmtval_remove(
>>>>    	lblkno = args->rmtblkno;
>>>>    	blkcnt = args->rmtblkcnt;
>>>>    	do {
>>>> -		retval = __xfs_attr_rmtval_remove(args);
>>>> -		if (retval && retval != EAGAIN)
>>>> -			return retval;
>>>> -
>>>> -		/*
>>>> -		 * Close out trans and start the next one in the chain.
>>>> -		 */
>>>> -		error = xfs_trans_roll_inode(&args->trans, args->dp);
>>>> -		if (error)
>>>> -			return error;
>>>> -	} while (retval == -EAGAIN);
>>>> +		error = __xfs_attr_rmtval_remove(&dac);
>>>> +	} while (xfs_attr_roll_again(&dac, &error));
>>>> -	return 0;
>>>> +	return error;
>>>>    }
>>>>    /*
>>>> @@ -713,9 +706,10 @@ xfs_attr_rmtval_remove(
>>>>     */
>>>>    int
>>>>    __xfs_attr_rmtval_remove(
>>>> -	struct xfs_da_args	*args)
>>>> +	struct xfs_delattr_context	*dac)
>>>>    {
>>>> -	int			error, done;
>>>> +	struct xfs_da_args		*args = dac->da_args;
>>>> +	int				error, done;
>>>>    	/*
>>>>    	 * Unmap value blocks for this attr.
>>>> @@ -725,12 +719,10 @@ __xfs_attr_rmtval_remove(
>>>>    	if (error)
>>>>    		return error;
>>>> -	error = xfs_defer_finish(&args->trans);
>>>> -	if (error)
>>>> -		return error;
>>>> -
>>>> -	if (!done)
>>>> +	if (!done) {
>>>> +		dac->flags |= XFS_DAC_DEFER_FINISH;
>>>>    		return -EAGAIN;
>>>> +	}
>>>>    	return error;
>>>>    }
>>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
>>>> index 9eee615..002fd30 100644
>>>> --- a/fs/xfs/libxfs/xfs_attr_remote.h
>>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.h
>>>> @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args);
>>>>    int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
>>>>    		xfs_buf_flags_t incore_flags);
>>>>    int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
>>>> -int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
>>>> +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
>>>>    #endif /* __XFS_ATTR_REMOTE_H__ */
>>>> diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
>>>> index bfad669..aaa7e66 100644
>>>> --- a/fs/xfs/xfs_attr_inactive.c
>>>> +++ b/fs/xfs/xfs_attr_inactive.c
>>>> @@ -15,10 +15,10 @@
>>>>    #include "xfs_da_format.h"
>>>>    #include "xfs_da_btree.h"
>>>>    #include "xfs_inode.h"
>>>> +#include "xfs_attr.h"
>>>>    #include "xfs_attr_remote.h"
>>>>    #include "xfs_trans.h"
>>>>    #include "xfs_bmap.h"
>>>> -#include "xfs_attr.h"
>>>>    #include "xfs_attr_leaf.h"
>>>>    #include "xfs_quota.h"
>>>>    #include "xfs_dir2.h"
>>>> -- 
>>>> 2.7.4
>>>>
>>>
>>
>

Patch
diff mbox series

diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c
index 4b78c86..5c460f4 100644
--- a/fs/xfs/libxfs/xfs_attr.c
+++ b/fs/xfs/libxfs/xfs_attr.c
@@ -53,7 +53,7 @@  STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
  */
 STATIC int xfs_attr_node_get(xfs_da_args_t *args);
 STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
-STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
+STATIC int xfs_attr_node_removename(struct xfs_delattr_context *dac);
 STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
 				 struct xfs_da_state **state);
 STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
@@ -264,6 +264,32 @@  xfs_attr_set_shortform(
 }
 
 /*
+ * Checks to see if a delayed attribute transaction should be rolled.  If so,
+ * also checks for a defer finish.  Transaction is finished and rolled as
+ * needed, and returns true of false if the delayed operation should continue.
+ */
+bool
+xfs_attr_roll_again(
+	struct xfs_delattr_context	*dac,
+	int				*error)
+{
+	struct xfs_da_args              *args = dac->da_args;
+
+	if (*error != -EAGAIN)
+		return false;
+
+	if (dac->flags & XFS_DAC_DEFER_FINISH) {
+		dac->flags &= ~XFS_DAC_DEFER_FINISH;
+		*error = xfs_defer_finish(&args->trans);
+		if (*error)
+			return false;
+	}
+
+	*error = xfs_trans_roll_inode(&args->trans, args->dp);
+	return *error == 0;
+}
+
+/*
  * Set the attribute specified in @args.
  */
 int
@@ -364,23 +390,47 @@  xfs_has_attr(
  */
 int
 xfs_attr_remove_args(
-	struct xfs_da_args      *args)
+	struct xfs_da_args	*args)
 {
-	struct xfs_inode	*dp = args->dp;
-	int			error;
+	int				error = 0;
+	struct xfs_delattr_context	dac = {
+		.da_args	= args,
+	};
+
+	do {
+		error = xfs_attr_remove_iter(&dac);
+	} while (xfs_attr_roll_again(&dac, &error));
+
+	return error;
+}
+
+/*
+ * Remove the attribute specified in @args.
+ *
+ * This function may return -EAGAIN to signal that the transaction needs to be
+ * rolled.  Callers should continue calling this function until they receive a
+ * return value other than -EAGAIN.
+ */
+int
+xfs_attr_remove_iter(
+	struct xfs_delattr_context	*dac)
+{
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_inode		*dp = args->dp;
+
+	if (dac->dela_state == XFS_DAS_RM_SHRINK)
+		goto node;
 
 	if (!xfs_inode_hasattr(dp)) {
-		error = -ENOATTR;
+		return -ENOATTR;
 	} else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) {
 		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
-		error = xfs_attr_shortform_remove(args);
+		return xfs_attr_shortform_remove(args);
 	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
-		error = xfs_attr_leaf_removename(args);
-	} else {
-		error = xfs_attr_node_removename(args);
+		return xfs_attr_leaf_removename(args);
 	}
-
-	return error;
+node:
+	return  xfs_attr_node_removename(dac);
 }
 
 /*
@@ -1177,11 +1227,12 @@  xfs_attr_leaf_mark_incomplete(
  */
 STATIC
 int xfs_attr_node_removename_setup(
-	struct xfs_da_args	*args,
-	struct xfs_da_state	**state)
+	struct xfs_delattr_context	*dac,
+	struct xfs_da_state		**state)
 {
-	int			error;
-	struct xfs_da_state_blk	*blk;
+	struct xfs_da_args		*args = dac->da_args;
+	int				error;
+	struct xfs_da_state_blk		*blk;
 
 	error = xfs_attr_node_hasname(args, state);
 	if (error != -EEXIST)
@@ -1191,6 +1242,13 @@  int xfs_attr_node_removename_setup(
 	ASSERT(blk->bp != NULL);
 	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
 
+	/*
+	 * Store blk and state in the context incase we need to cycle out the
+	 * transaction
+	 */
+	dac->blk = blk;
+	dac->da_state = *state;
+
 	if (args->rmtblkno > 0) {
 		error = xfs_attr_leaf_mark_incomplete(args, *state);
 		if (error)
@@ -1203,13 +1261,16 @@  int xfs_attr_node_removename_setup(
 }
 
 STATIC int
-xfs_attr_node_remove_rmt(
-	struct xfs_da_args	*args,
-	struct xfs_da_state	*state)
+xfs_attr_node_remove_rmt (
+	struct xfs_delattr_context	*dac,
+	struct xfs_da_state		*state)
 {
-	int			error = 0;
+	int				error = 0;
 
-	error = xfs_attr_rmtval_remove(args);
+	/*
+	 * May return -EAGAIN to request that the caller recall this function
+	 */
+	error = __xfs_attr_rmtval_remove(dac);
 	if (error)
 		return error;
 
@@ -1226,21 +1287,35 @@  xfs_attr_node_remove_rmt(
  * This will involve walking down the Btree, and may involve joining
  * leaf nodes and even joining intermediate nodes up to and including
  * the root node (a special case of an intermediate node).
+ *
+ * This routine is meant to function as either an inline or delayed operation,
+ * and may return -EAGAIN when the transaction needs to be rolled.  Calling
+ * functions will need to handle this, and recall the function until a
+ * successful error code is returned.
  */
 STATIC int
 xfs_attr_node_removename(
-	struct xfs_da_args	*args)
+	struct xfs_delattr_context	*dac)
 {
-	struct xfs_da_state	*state;
-	struct xfs_da_state_blk	*blk;
-	int			retval, error;
-	struct xfs_inode	*dp = args->dp;
+	struct xfs_da_args		*args = dac->da_args;
+	struct xfs_da_state		*state;
+	struct xfs_da_state_blk		*blk;
+	int				retval, error;
+	struct xfs_inode		*dp = args->dp;
 
 	trace_xfs_attr_node_removename(args);
+	state = dac->da_state;
+	blk = dac->blk;
 
-	error = xfs_attr_node_removename_setup(args, &state);
-	if (error)
-		goto out;
+	if (dac->dela_state == XFS_DAS_RM_SHRINK)
+		goto das_rm_shrink;
+
+	if ((dac->flags & XFS_DAC_NODE_RMVNAME_INIT) == 0) {
+		dac->flags |= XFS_DAC_NODE_RMVNAME_INIT;
+		error = xfs_attr_node_removename_setup(dac, &state);
+		if (error)
+			goto out;
+	}
 
 	/*
 	 * If there is an out-of-line value, de-allocate the blocks.
@@ -1248,8 +1323,13 @@  xfs_attr_node_removename(
 	 * overflow the maximum size of a transaction and/or hit a deadlock.
 	 */
 	if (args->rmtblkno > 0) {
-		error = xfs_attr_node_remove_rmt(args, state);
-		if (error)
+		/*
+		 * May return -EAGAIN. Remove blocks until args->rmtblkno == 0
+		 */
+		error = xfs_attr_node_remove_rmt(dac, state);
+		if (error == -EAGAIN)
+			return error;
+		else if (error)
 			goto out;
 	}
 
@@ -1268,17 +1348,14 @@  xfs_attr_node_removename(
 		error = xfs_da3_join(state);
 		if (error)
 			goto out;
-		error = xfs_defer_finish(&args->trans);
-		if (error)
-			goto out;
-		/*
-		 * Commit the Btree join operation and start a new trans.
-		 */
-		error = xfs_trans_roll_inode(&args->trans, dp);
-		if (error)
-			goto out;
+
+		dac->flags |= XFS_DAC_DEFER_FINISH;
+		dac->dela_state = XFS_DAS_RM_SHRINK;
+		return -EAGAIN;
 	}
 
+das_rm_shrink:
+
 	/*
 	 * If the result is small enough, push it all into the inode.
 	 */
diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h
index 3e97a93..6c58792 100644
--- a/fs/xfs/libxfs/xfs_attr.h
+++ b/fs/xfs/libxfs/xfs_attr.h
@@ -74,6 +74,75 @@  struct xfs_attr_list_context {
 };
 
 
+/*
+ * ========================================================================
+ * Structure used to pass context around among the delayed routines.
+ * ========================================================================
+ */
+
+/*
+ * Below is a state machine diagram for attr remove operations. The  XFS_DAS_*
+ * states indicate places where the function would return -EAGAIN, and then
+ * immediately resume from after being recalled by the calling function. States
+ * marked as a "subroutine state" indicate that they belong to a subroutine, and
+ * so the calling function needs to pass them back to that subroutine to allow
+ * it to finish where it left off. But they otherwise do not have a role in the
+ * calling function other than just passing through.
+ *
+ * xfs_attr_remove_iter()
+ *	  XFS_DAS_RM_SHRINK ─┐
+ *	  (subroutine state) │
+ *	                     └─>xfs_attr_node_removename()
+ *	                                      │
+ *	                                      v
+ *	                                   need to
+ *	                                shrink tree? ─n─┐
+ *	                                      │         │
+ *	                                      y         │
+ *	                                      │         │
+ *	                                      v         │
+ *	                              XFS_DAS_RM_SHRINK │
+ *	                                      │         │
+ *	                                      v         │
+ *	                                     done <─────┘
+ *
+ */
+
+/*
+ * Enum values for xfs_delattr_context.da_state
+ *
+ * These values are used by delayed attribute operations to keep track  of where
+ * they were before they returned -EAGAIN.  A return code of -EAGAIN signals the
+ * calling function to roll the transaction, and then recall the subroutine to
+ * finish the operation.  The enum is then used by the subroutine to jump back
+ * to where it was and resume executing where it left off.
+ */
+enum xfs_delattr_state {
+				      /* Zero is uninitalized */
+	XFS_DAS_RM_SHRINK	= 1,  /* We are shrinking the tree */
+};
+
+/*
+ * Defines for xfs_delattr_context.flags
+ */
+#define XFS_DAC_DEFER_FINISH		0x01 /* finish the transaction */
+#define XFS_DAC_NODE_RMVNAME_INIT	0x02 /* xfs_attr_node_removename init */
+
+/*
+ * Context used for keeping track of delayed attribute operations
+ */
+struct xfs_delattr_context {
+	struct xfs_da_args      *da_args;
+
+	/* Used in xfs_attr_node_removename to roll through removing blocks */
+	struct xfs_da_state     *da_state;
+	struct xfs_da_state_blk *blk;
+
+	/* Used to keep track of current state of delayed operation */
+	unsigned int            flags;
+	enum xfs_delattr_state  dela_state;
+};
+
 /*========================================================================
  * Function prototypes for the kernel.
  *========================================================================*/
@@ -91,6 +160,10 @@  int xfs_attr_set(struct xfs_da_args *args);
 int xfs_attr_set_args(struct xfs_da_args *args);
 int xfs_has_attr(struct xfs_da_args *args);
 int xfs_attr_remove_args(struct xfs_da_args *args);
+int xfs_attr_remove_iter(struct xfs_delattr_context *dac);
+bool xfs_attr_roll_again(struct xfs_delattr_context *dac, int *error);
 bool xfs_attr_namecheck(const void *name, size_t length);
+void xfs_delattr_context_init(struct xfs_delattr_context *dac,
+			      struct xfs_da_args *args);
 
 #endif	/* __XFS_ATTR_H__ */
diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
index 351351c..20521bf 100644
--- a/fs/xfs/libxfs/xfs_attr_leaf.c
+++ b/fs/xfs/libxfs/xfs_attr_leaf.c
@@ -19,8 +19,8 @@ 
 #include "xfs_bmap_btree.h"
 #include "xfs_bmap.h"
 #include "xfs_attr_sf.h"
-#include "xfs_attr_remote.h"
 #include "xfs_attr.h"
+#include "xfs_attr_remote.h"
 #include "xfs_attr_leaf.h"
 #include "xfs_error.h"
 #include "xfs_trace.h"
diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c
index 85dca51..20e4605 100644
--- a/fs/xfs/libxfs/xfs_attr_remote.c
+++ b/fs/xfs/libxfs/xfs_attr_remote.c
@@ -676,12 +676,14 @@  xfs_attr_rmtval_invalidate(
  */
 int
 xfs_attr_rmtval_remove(
-	struct xfs_da_args      *args)
+	struct xfs_da_args		*args)
 {
-	xfs_dablk_t		lblkno;
-	int			blkcnt;
-	int			error = 0;
-	int			retval = 0;
+	xfs_dablk_t			lblkno;
+	int				blkcnt;
+	int				error = 0;
+	struct xfs_delattr_context	dac  = {
+		.da_args	= args,
+	};
 
 	trace_xfs_attr_rmtval_remove(args);
 
@@ -691,19 +693,10 @@  xfs_attr_rmtval_remove(
 	lblkno = args->rmtblkno;
 	blkcnt = args->rmtblkcnt;
 	do {
-		retval = __xfs_attr_rmtval_remove(args);
-		if (retval && retval != EAGAIN)
-			return retval;
-
-		/*
-		 * Close out trans and start the next one in the chain.
-		 */
-		error = xfs_trans_roll_inode(&args->trans, args->dp);
-		if (error)
-			return error;
-	} while (retval == -EAGAIN);
+		error = __xfs_attr_rmtval_remove(&dac);
+	} while (xfs_attr_roll_again(&dac, &error));
 
-	return 0;
+	return error;
 }
 
 /*
@@ -713,9 +706,10 @@  xfs_attr_rmtval_remove(
  */
 int
 __xfs_attr_rmtval_remove(
-	struct xfs_da_args	*args)
+	struct xfs_delattr_context	*dac)
 {
-	int			error, done;
+	struct xfs_da_args		*args = dac->da_args;
+	int				error, done;
 
 	/*
 	 * Unmap value blocks for this attr.
@@ -725,12 +719,10 @@  __xfs_attr_rmtval_remove(
 	if (error)
 		return error;
 
-	error = xfs_defer_finish(&args->trans);
-	if (error)
-		return error;
-
-	if (!done)
+	if (!done) {
+		dac->flags |= XFS_DAC_DEFER_FINISH;
 		return -EAGAIN;
+	}
 
 	return error;
 }
diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h
index 9eee615..002fd30 100644
--- a/fs/xfs/libxfs/xfs_attr_remote.h
+++ b/fs/xfs/libxfs/xfs_attr_remote.h
@@ -14,5 +14,5 @@  int xfs_attr_rmtval_remove(struct xfs_da_args *args);
 int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map,
 		xfs_buf_flags_t incore_flags);
 int xfs_attr_rmtval_invalidate(struct xfs_da_args *args);
-int __xfs_attr_rmtval_remove(struct xfs_da_args *args);
+int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac);
 #endif /* __XFS_ATTR_REMOTE_H__ */
diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c
index bfad669..aaa7e66 100644
--- a/fs/xfs/xfs_attr_inactive.c
+++ b/fs/xfs/xfs_attr_inactive.c
@@ -15,10 +15,10 @@ 
 #include "xfs_da_format.h"
 #include "xfs_da_btree.h"
 #include "xfs_inode.h"
+#include "xfs_attr.h"
 #include "xfs_attr_remote.h"
 #include "xfs_trans.h"
 #include "xfs_bmap.h"
-#include "xfs_attr.h"
 #include "xfs_attr_leaf.h"
 #include "xfs_quota.h"
 #include "xfs_dir2.h"