| Message ID | 20200403221229.4995-20-allison.henderson@oracle.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | xfs: Delay Ready Attributes | expand |
On Fri, Apr 03, 2020 at 03:12:28PM -0700, Allison Collins wrote: > This patch modifies the attr set 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_set_args has become xfs_attr_set_iter, which uses a > state machine like switch to keep track of where it was when EAGAIN was > returned. > > Two new helper functions have been added: xfs_attr_rmtval_set_init and > xfs_attr_rmtval_set_blk. They provide a subset of logic similar to > xfs_attr_rmtval_set, but they store the current block in the delay attr > context to allow the caller to roll the transaction between allocations. > This helps to simplify and consolidate code used by > xfs_attr_leaf_addname and xfs_attr_node_addname. xfs_attr_set_args has > now become a simple loop to refresh the transaction until the operation > is completed. Lastly, xfs_attr_rmtval_remove is no longer used, and is > removed. > > Below is a state machine diagram for attr set 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_set_iter() > │ > v > need to upgrade > from sf to leaf? ──n─┐ > │ │ > y │ > │ │ > V │ > XFS_DAS_ADD_LEAF │ > │ │ > v │ > ┌──────n── fork has <──────┘ > │ only 1 blk? > │ │ > │ y > │ │ > │ v > │ xfs_attr_leaf_try_add() > │ │ > │ v > │ had enough > ├──────n── space? > │ │ > │ y > │ │ > │ v > │ XFS_DAS_FOUND_LBLK ──┐ > │ │ > │ XFS_DAS_FLIP_LFLAG ──┤ > │ (subroutine state) │ > │ │ > │ XFS_DAS_ALLOC_LEAF ──┤ > │ (subroutine state) │ > │ └─>xfs_attr_leaf_addname() > │ │ > │ v > │ ┌─────n── need to > │ │ alloc blks? > │ │ │ > │ │ y > │ │ │ > │ │ v > │ │ ┌─>XFS_DAS_ALLOC_LEAF > │ │ │ │ > │ │ │ v > │ │ └──y── need to alloc > │ │ more blocks? > │ │ │ > │ │ n > │ │ │ > │ │ v > │ │ was this > │ └────────> a rename? ──n─┐ > │ │ │ > │ y │ > │ │ │ > │ v │ > │ flip incomplete │ > │ flag │ > │ │ │ > │ v │ > │ XFS_DAS_FLIP_LFLAG │ > │ │ │ > │ v │ > │ remove │ > │ XFS_DAS_RM_LBLK ─> old name │ > │ ^ │ │ > │ │ v │ > │ └──────y── more to │ > │ remove │ > │ │ │ > │ n │ > │ │ │ > │ v │ > │ done <──────┘ > └────> XFS_DAS_LEAF_TO_NODE ─┐ > │ > XFS_DAS_FOUND_NBLK ──┤ > (subroutine state) │ > │ > XFS_DAS_ALLOC_NODE ──┤ > (subroutine state) │ > │ > XFS_DAS_FLIP_NFLAG ──┤ > (subroutine state) │ > │ > └─>xfs_attr_node_addname() > │ > v > find space to store > attr. Split if needed > │ > v > XFS_DAS_FOUND_NBLK > │ > v > ┌─────n── need to > │ alloc blks? > │ │ > │ y > │ │ > │ v > │ ┌─>XFS_DAS_ALLOC_NODE > │ │ │ > │ │ v > │ └──y── need to alloc > │ more blocks? > │ │ > │ n > │ │ > │ v > │ was this > └────────> a rename? ──n─┐ > │ │ > y │ > │ │ > v │ > flip incomplete │ > flag │ > │ │ > v │ > XFS_DAS_FLIP_NFLAG │ > │ │ > v │ > remove │ > XFS_DAS_RM_NBLK ─> old name │ > ^ │ │ > │ v │ > └──────y── more to │ > remove │ > │ │ > n │ > │ │ > v │ > done <──────┘ > > Signed-off-by: Allison Collins <allison.henderson@oracle.com> > --- Only a cursory pass given the previous feedback... > fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- > fs/xfs/libxfs/xfs_attr.h | 16 ++ > fs/xfs/libxfs/xfs_attr_leaf.c | 1 + > fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- > fs/xfs/libxfs/xfs_attr_remote.h | 4 + > fs/xfs/xfs_attr_inactive.c | 1 + > fs/xfs/xfs_trace.h | 1 - > 7 files changed, 351 insertions(+), 167 deletions(-) > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > index f700976..c160b7a 100644 > --- a/fs/xfs/libxfs/xfs_attr.c > +++ b/fs/xfs/libxfs/xfs_attr.c > @@ -44,7 +44,7 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args); > * Internal routines when attribute list is one block. > */ > STATIC int xfs_attr_leaf_get(xfs_da_args_t *args); > -STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args); > +STATIC int xfs_attr_leaf_addname(struct xfs_delattr_context *dac); > STATIC int xfs_attr_leaf_removename(struct xfs_delattr_context *dac); > STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > > @@ -52,12 +52,13 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > * Internal routines when attribute list is more than one block. > */ > 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_addname(struct xfs_delattr_context *dac); > 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); > STATIC int xfs_attr_refillstate(xfs_da_state_t *state); > +STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp); > > STATIC void > xfs_delattr_context_init( > @@ -227,8 +228,11 @@ xfs_attr_is_shortform( > > /* > * Attempts to set an attr in shortform, or converts the tree to leaf form if > - * there is not enough room. If the attr is set, the transaction is committed > - * and set to NULL. > + * there is not enough room. This function is meant to operate as a helper > + * routine to the delayed attribute functions. It returns -EAGAIN to indicate > + * that the calling function should roll the transaction, and then proceed to > + * add the attr in leaf form. This subroutine does not expect to be recalled > + * again like the other delayed attr routines do. > */ > STATIC int > xfs_attr_set_shortform( > @@ -236,16 +240,16 @@ xfs_attr_set_shortform( > struct xfs_buf **leaf_bp) > { > struct xfs_inode *dp = args->dp; > - int error, error2 = 0; > + int error = 0; > > /* > * Try to add the attr to the attribute list in the inode. > */ > error = xfs_attr_try_sf_addname(dp, args); > + > + /* Should only be 0, -EEXIST or ENOSPC */ > if (error != -ENOSPC) { > - error2 = xfs_trans_commit(args->trans); > - args->trans = NULL; > - return error ? error : error2; > + return error; > } > /* > * It won't fit in the shortform, transform to a leaf block. GROT: > @@ -258,18 +262,10 @@ xfs_attr_set_shortform( > /* > * Prevent the leaf buffer from being unlocked so that a concurrent AIL > * push cannot grab the half-baked leaf buffer and run into problems > - * with the write verifier. Once we're done rolling the transaction we > - * can release the hold and add the attr to the leaf. > + * with the write verifier. > */ > xfs_trans_bhold(args->trans, *leaf_bp); > - error = xfs_defer_finish(&args->trans); > - xfs_trans_bhold_release(args->trans, *leaf_bp); > - if (error) { > - xfs_trans_brelse(args->trans, *leaf_bp); > - return error; > - } > - > - return 0; > + return -EAGAIN; > } > > /* > @@ -279,9 +275,83 @@ int > xfs_attr_set_args( > struct xfs_da_args *args) > { > - struct xfs_inode *dp = args->dp; > - struct xfs_buf *leaf_bp = NULL; > - int error = 0; > + struct xfs_buf *leaf_bp = NULL; > + int error = 0; > + struct xfs_delattr_context dac; > + > + xfs_delattr_context_init(&dac, args); > + > + do { > + error = xfs_attr_set_iter(&dac, &leaf_bp); > + if (error != -EAGAIN) > + break; > + > + if (dac.flags & XFS_DAC_DEFER_FINISH) { > + dac.flags &= ~XFS_DAC_DEFER_FINISH; > + error = xfs_defer_finish(&args->trans); > + if (error) > + break; > + } > + > + error = xfs_trans_roll_inode(&args->trans, args->dp); > + if (error) > + break; > + > + if (leaf_bp) { > + xfs_trans_bjoin(args->trans, leaf_bp); > + xfs_trans_bhold(args->trans, leaf_bp); > + } > + > + } while (true); > + > + return error; > +} > + > +/* > + * Set the attribute specified in @args. > + * This routine is meant to function as a 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. > + */ > +int > +xfs_attr_set_iter( > + struct xfs_delattr_context *dac, > + struct xfs_buf **leaf_bp) > +{ > + struct xfs_da_args *args = dac->da_args; > + struct xfs_inode *dp = args->dp; > + int error = 0; > + int sf_size; > + > + /* State machine switch */ > + switch (dac->dela_state) { > + case XFS_DAS_ADD_LEAF: > + goto das_add_leaf; > + case XFS_DAS_ALLOC_LEAF: > + case XFS_DAS_FLIP_LFLAG: > + case XFS_DAS_FOUND_LBLK: > + goto das_leaf; > + case XFS_DAS_FOUND_NBLK: > + case XFS_DAS_FLIP_NFLAG: > + case XFS_DAS_ALLOC_NODE: > + case XFS_DAS_LEAF_TO_NODE: > + goto das_node; > + default: > + break; > + } > + > + /* > + * New inodes may not have an attribute fork yet. So set the attribute > + * fork appropriately > + */ > + if (XFS_IFORK_Q((args->dp)) == 0) { > + sf_size = sizeof(struct xfs_attr_sf_hdr) + > + XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen); > + xfs_bmap_set_attrforkoff(args->dp, sf_size, NULL); > + args->dp->i_afp = kmem_zone_zalloc(xfs_ifork_zone, 0); > + args->dp->i_afp->if_flags = XFS_IFEXTENTS; > + } > Is this hunk moved from somewhere? If so, we should probably handle that in a separate patch. I think we really want these last couple of patches to introduce the state/markers and not much else. > /* > * If the attribute list is already in leaf format, jump straight to > @@ -292,40 +362,53 @@ xfs_attr_set_args( > if (xfs_attr_is_shortform(dp)) { > > /* > - * If the attr was successfully set in shortform, the > - * transaction is committed and set to NULL. Otherwise, is it > - * converted from shortform to leaf, and the transaction is > - * retained. > + * If the attr was successfully set in shortform, no need to > + * continue. Otherwise, is it converted from shortform to leaf > + * and -EAGAIN is returned. > */ > - error = xfs_attr_set_shortform(args, &leaf_bp); > - if (error || !args->trans) > - return error; > + error = xfs_attr_set_shortform(args, leaf_bp); > + if (error == -EAGAIN) { > + dac->flags |= XFS_DAC_DEFER_FINISH; > + dac->dela_state = XFS_DAS_ADD_LEAF; > + } > + return error; Similar to the previous patch, I wonder if we need the explicit states that are otherwise handled by existing inode state. For example, if the above returns -EAGAIN, xfs_attr_is_shortform() is no longer true on reentry, right? If that's the case for the other conversions, it seems like we might only need one state (XFS_DAS_FOUND_LBLK) for this function. BTW, that general approach might be more clear if we lifted the format conversions into this level from down in the format specific add handlers. The goal would be to make the high level flow look something like: if (shortform) { error = sf_try_add(); if (error == -ENOSPC) { shortform_to_leaf(...); ... return -EAGAIN; } } else if (xfs_bmap_one_block(...)) { error = xfs_attr_leaf_try_add(args, *leaf_bp); if (error == -ENOSPC) { leaf_to_node(...); return -EAGAIN; } ... state stuff for leaf add ... } else { error = xfs_attr_node_addname(dac); } Hm? Of course, something like that should be incorporated via independent refactoring patches. > } > > - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > - error = xfs_attr_leaf_addname(args); > - if (error != -ENOSPC) > - return error; > +das_add_leaf: > > - /* > - * Commit that transaction so that the node_addname() > - * call can manage its own transactions. > - */ > - error = xfs_defer_finish(&args->trans); > - if (error) > - return error; > + /* > + * After a shortform to leaf conversion, we need to hold the leaf and > + * cylce out the transaction. When we get back, we need to release > + * the leaf. > + */ > + if (*leaf_bp != NULL) { > + xfs_trans_brelse(args->trans, *leaf_bp); > + *leaf_bp = NULL; > + } > > - /* > - * Commit the current trans (including the inode) and > - * start a new one. > - */ > - error = xfs_trans_roll_inode(&args->trans, dp); > - if (error) > + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > + error = xfs_attr_leaf_try_add(args, *leaf_bp); > + switch (error) { > + case -ENOSPC: > + dac->flags |= XFS_DAC_DEFER_FINISH; > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > + return -EAGAIN; > + case 0: > + dac->dela_state = XFS_DAS_FOUND_LBLK; > + return -EAGAIN; > + default: > return error; > - > + } > +das_leaf: > + error = xfs_attr_leaf_addname(dac); > + if (error == -ENOSPC) { > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > + return -EAGAIN; > + } > + return error; > } > - > - error = xfs_attr_node_addname(args); > +das_node: > + error = xfs_attr_node_addname(dac); > return error; > } > > @@ -716,28 +799,32 @@ xfs_attr_leaf_try_add( > * > * This leaf block cannot have a "remote" value, we only call this routine > * if bmap_one_block() says there is only one block (ie: no remote blks). > + * > + * This routine is meant to function as a 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_leaf_addname( > - struct xfs_da_args *args) > + struct xfs_delattr_context *dac) > { > - int error, forkoff; > - struct xfs_buf *bp = NULL; > - struct xfs_inode *dp = args->dp; > - > - trace_xfs_attr_leaf_addname(args); > - > - error = xfs_attr_leaf_try_add(args, bp); > - if (error) > - return error; > + struct xfs_da_args *args = dac->da_args; > + struct xfs_buf *bp = NULL; > + int error, forkoff; > + struct xfs_inode *dp = args->dp; > > - /* > - * Commit the transaction that added the attr name so that > - * later routines can manage their own transactions. > - */ > - error = xfs_trans_roll_inode(&args->trans, dp); > - if (error) > - return error; > + /* State machine switch */ > + switch (dac->dela_state) { > + case XFS_DAS_FLIP_LFLAG: > + goto das_flip_flag; > + case XFS_DAS_ALLOC_LEAF: > + goto das_alloc_leaf; > + case XFS_DAS_RM_LBLK: > + goto das_rm_lblk; > + default: > + break; > + } > > /* > * If there was an out-of-line value, allocate the blocks we > @@ -746,7 +833,28 @@ xfs_attr_leaf_addname( > * maximum size of a transaction and/or hit a deadlock. > */ > if (args->rmtblkno > 0) { > - error = xfs_attr_rmtval_set(args); > + > + /* Open coded xfs_attr_rmtval_set without trans handling */ > + error = xfs_attr_rmtval_set_init(dac); > + if (error) > + return error; > + > + /* > + * Roll through the "value", allocating blocks on disk as > + * required. > + */ > +das_alloc_leaf: If we filter out the setup above, it seems like this state could be reduced to check for ->blkcnt > 0. > + while (dac->blkcnt > 0) { > + error = xfs_attr_rmtval_set_blk(dac); > + if (error) > + return error; > + > + dac->flags |= XFS_DAC_DEFER_FINISH; > + dac->dela_state = XFS_DAS_ALLOC_LEAF; > + return -EAGAIN; > + } > + > + error = xfs_attr_rmtval_set_value(args); > if (error) > return error; > } > @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( > error = xfs_attr3_leaf_flipflags(args); > if (error) > return error; > - /* > - * Commit the flag value change and start the next trans in > - * series. > - */ > - error = xfs_trans_roll_inode(&args->trans, args->dp); > - if (error) > - return error; > - > + dac->dela_state = XFS_DAS_FLIP_LFLAG; > + return -EAGAIN; > +das_flip_flag: > /* > * Dismantle the "old" attribute/value pair by removing > * a "remote" value (if it exists). > */ > xfs_attr_restore_rmt_blk(args); > > + xfs_attr_rmtval_invalidate(args); > +das_rm_lblk: > if (args->rmtblkno) { > - error = xfs_attr_rmtval_remove(args); > + error = __xfs_attr_rmtval_remove(args); > + > + if (error == -EAGAIN) { > + dac->dela_state = XFS_DAS_RM_LBLK; > + return -EAGAIN; > + } > + This whole function looks like it could use more refactoring to split out the rename case. > if (error) > return error; > } > @@ -799,15 +910,11 @@ xfs_attr_leaf_addname( > /* > * If the result is small enough, shrink it all into the inode. > */ > - if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) { > + forkoff = xfs_attr_shortform_allfit(bp, dp); > + if (forkoff) > error = xfs_attr3_leaf_to_shortform(bp, args, forkoff); > - /* bp is gone due to xfs_da_shrink_inode */ > - if (error) > - return error; > - error = xfs_defer_finish(&args->trans); > - if (error) > - return error; > - } > + > + dac->flags |= XFS_DAC_DEFER_FINISH; > > } else if (args->rmtblkno > 0) { > /* > @@ -967,16 +1074,23 @@ xfs_attr_node_hasname( > * > * "Remote" attribute values confuse the issue and atomic rename operations > * add a whole extra layer of confusion on top of that. > + * > + * This routine is meant to function as a 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_addname( > - struct xfs_da_args *args) > + struct xfs_delattr_context *dac) > { > - struct xfs_da_state *state; > - struct xfs_da_state_blk *blk; > - struct xfs_inode *dp; > - struct xfs_mount *mp; > - int retval, error; > + struct xfs_da_args *args = dac->da_args; > + struct xfs_da_state *state = NULL; > + struct xfs_da_state_blk *blk; > + struct xfs_inode *dp; > + struct xfs_mount *mp; > + int retval = 0; > + int error = 0; > > trace_xfs_attr_node_addname(args); > > @@ -985,7 +1099,21 @@ xfs_attr_node_addname( > */ > dp = args->dp; > mp = dp->i_mount; > -restart: > + > + /* State machine switch */ > + switch (dac->dela_state) { > + case XFS_DAS_FLIP_NFLAG: > + goto das_flip_flag; > + case XFS_DAS_FOUND_NBLK: > + goto das_found_nblk; > + case XFS_DAS_ALLOC_NODE: > + goto das_alloc_node; > + case XFS_DAS_RM_NBLK: > + goto das_rm_nblk; > + default: > + break; > + } > + > /* > * Search to see if name already exists, and get back a pointer > * to where it should go. > @@ -1031,19 +1159,13 @@ xfs_attr_node_addname( > error = xfs_attr3_leaf_to_node(args); > if (error) > goto out; > - error = xfs_defer_finish(&args->trans); > - if (error) > - goto out; > > /* > - * Commit the node conversion and start the next > - * trans in the chain. > + * Restart routine from the top. No need to set the > + * state > */ > - error = xfs_trans_roll_inode(&args->trans, dp); > - if (error) > - goto out; > - > - goto restart; > + dac->flags |= XFS_DAC_DEFER_FINISH; > + return -EAGAIN; > } > > /* > @@ -1055,9 +1177,7 @@ xfs_attr_node_addname( > error = xfs_da3_split(state); > if (error) > goto out; > - error = xfs_defer_finish(&args->trans); > - if (error) > - goto out; > + dac->flags |= XFS_DAC_DEFER_FINISH; > } else { > /* > * Addition succeeded, update Btree hashvals. > @@ -1072,13 +1192,9 @@ xfs_attr_node_addname( > xfs_da_state_free(state); > state = NULL; > > - /* > - * Commit the leaf addition or btree split and start the next > - * trans in the chain. > - */ > - error = xfs_trans_roll_inode(&args->trans, dp); > - if (error) > - goto out; > + dac->dela_state = XFS_DAS_FOUND_NBLK; > + return -EAGAIN; > +das_found_nblk: Same deal here. Any time we have this return -EAGAIN followed by a label pattern I think we're going to want to think about refactoring things more first to avoid dumping it in the middle of some unnecessarily large function. Brian > > /* > * If there was an out-of-line value, allocate the blocks we > @@ -1087,7 +1203,27 @@ xfs_attr_node_addname( > * maximum size of a transaction and/or hit a deadlock. > */ > if (args->rmtblkno > 0) { > - error = xfs_attr_rmtval_set(args); > + /* Open coded xfs_attr_rmtval_set without trans handling */ > + error = xfs_attr_rmtval_set_init(dac); > + if (error) > + return error; > + > + /* > + * Roll through the "value", allocating blocks on disk as > + * required. > + */ > +das_alloc_node: > + while (dac->blkcnt > 0) { > + error = xfs_attr_rmtval_set_blk(dac); > + if (error) > + return error; > + > + dac->flags |= XFS_DAC_DEFER_FINISH; > + dac->dela_state = XFS_DAS_ALLOC_NODE; > + return -EAGAIN; > + } > + > + error = xfs_attr_rmtval_set_value(args); > if (error) > return error; > } > @@ -1110,18 +1246,26 @@ xfs_attr_node_addname( > * Commit the flag value change and start the next trans in > * series > */ > - error = xfs_trans_roll_inode(&args->trans, args->dp); > - if (error) > - goto out; > - > + dac->dela_state = XFS_DAS_FLIP_NFLAG; > + return -EAGAIN; > +das_flip_flag: > /* > * Dismantle the "old" attribute/value pair by removing > * a "remote" value (if it exists). > */ > xfs_attr_restore_rmt_blk(args); > > + xfs_attr_rmtval_invalidate(args); > + > +das_rm_nblk: > if (args->rmtblkno) { > - error = xfs_attr_rmtval_remove(args); > + error = __xfs_attr_rmtval_remove(args); > + > + if (error == -EAGAIN) { > + dac->dela_state = XFS_DAS_RM_NBLK; > + return -EAGAIN; > + } > + > if (error) > return error; > } > @@ -1139,7 +1283,6 @@ xfs_attr_node_addname( > error = xfs_da3_node_lookup_int(state, &retval); > if (error) > goto out; > - > /* > * Remove the name and update the hashvals in the tree. > */ > @@ -1147,7 +1290,6 @@ xfs_attr_node_addname( > ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC); > error = xfs_attr3_leaf_remove(blk->bp, args); > xfs_da3_fixhashpath(state, &state->path); > - > /* > * Check to see if the tree needs to be collapsed. > */ > @@ -1155,11 +1297,9 @@ xfs_attr_node_addname( > error = xfs_da3_join(state); > if (error) > goto out; > - error = xfs_defer_finish(&args->trans); > - if (error) > - goto out; > - } > > + dac->flags |= XFS_DAC_DEFER_FINISH; > + } > } else if (args->rmtblkno > 0) { > /* > * Added a "remote" value, just clear the incomplete flag. > diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h > index 0e8ae1a..67af9d1 100644 > --- a/fs/xfs/libxfs/xfs_attr.h > +++ b/fs/xfs/libxfs/xfs_attr.h > @@ -93,6 +93,16 @@ enum xfs_delattr_state { > /* Zero is uninitalized */ > XFS_DAS_RM_SHRINK = 1, /* We are shrinking the tree */ > XFS_DAS_RMTVAL_REMOVE, /* We are removing remote value blocks */ > + XFS_DAS_ADD_LEAF, /* We are adding a leaf attr */ > + XFS_DAS_FOUND_LBLK, /* We found leaf blk for attr */ > + XFS_DAS_LEAF_TO_NODE, /* Converted leaf to node */ > + XFS_DAS_FOUND_NBLK, /* We found node blk for attr */ > + XFS_DAS_ALLOC_LEAF, /* We are allocating leaf blocks */ > + XFS_DAS_FLIP_LFLAG, /* Flipped leaf INCOMPLETE attr flag */ > + XFS_DAS_RM_LBLK, /* A rename is removing leaf blocks */ > + XFS_DAS_ALLOC_NODE, /* We are allocating node blocks */ > + XFS_DAS_FLIP_NFLAG, /* Flipped node INCOMPLETE attr flag */ > + XFS_DAS_RM_NBLK, /* A rename is removing node blocks */ > }; > > /* > @@ -105,8 +115,13 @@ enum xfs_delattr_state { > */ > struct xfs_delattr_context { > struct xfs_da_args *da_args; > + struct xfs_bmbt_irec map; > + struct xfs_buf *leaf_bp; > + xfs_fileoff_t lfileoff; > struct xfs_da_state *da_state; > struct xfs_da_state_blk *blk; > + xfs_dablk_t lblkno; > + int blkcnt; > unsigned int flags; > enum xfs_delattr_state dela_state; > }; > @@ -126,6 +141,7 @@ int xfs_attr_get_ilocked(struct xfs_da_args *args); > int xfs_attr_get(struct xfs_da_args *args); > int xfs_attr_set(struct xfs_da_args *args); > int xfs_attr_set_args(struct xfs_da_args *args); > +int xfs_attr_set_iter(struct xfs_delattr_context *dac, struct xfs_buf **leaf_bp); > 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); > diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c > index f55402b..4d15f45 100644 > --- a/fs/xfs/libxfs/xfs_attr_leaf.c > +++ b/fs/xfs/libxfs/xfs_attr_leaf.c > @@ -19,6 +19,7 @@ > #include "xfs_bmap_btree.h" > #include "xfs_bmap.h" > #include "xfs_attr_sf.h" > +#include "xfs_attr.h" > #include "xfs_attr_remote.h" > #include "xfs_attr.h" > #include "xfs_attr_leaf.h" > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c > index fd4be9d..9607fd2 100644 > --- a/fs/xfs/libxfs/xfs_attr_remote.c > +++ b/fs/xfs/libxfs/xfs_attr_remote.c > @@ -443,7 +443,7 @@ xfs_attr_rmtval_get( > * Find a "hole" in the attribute address space large enough for us to drop the > * new attribute's value into > */ > -STATIC int > +int > xfs_attr_rmt_find_hole( > struct xfs_da_args *args) > { > @@ -470,7 +470,7 @@ xfs_attr_rmt_find_hole( > return 0; > } > > -STATIC int > +int > xfs_attr_rmtval_set_value( > struct xfs_da_args *args) > { > @@ -630,6 +630,71 @@ xfs_attr_rmtval_set( > } > > /* > + * Find a hole for the attr and store it in the delayed attr context. This > + * initializes the context to roll through allocating an attr extent for a > + * delayed attr operation > + */ > +int > +xfs_attr_rmtval_set_init( > + struct xfs_delattr_context *dac) > +{ > + struct xfs_da_args *args = dac->da_args; > + struct xfs_bmbt_irec *map = &dac->map; > + int error; > + > + dac->lblkno = 0; > + dac->lfileoff = 0; > + dac->blkcnt = 0; > + args->rmtblkcnt = 0; > + args->rmtblkno = 0; > + memset(map, 0, sizeof(struct xfs_bmbt_irec)); > + > + error = xfs_attr_rmt_find_hole(args); > + if (error) > + return error; > + > + dac->blkcnt = args->rmtblkcnt; > + dac->lblkno = args->rmtblkno; > + > + return error; > +} > + > +/* > + * Write one block of the value associated with an attribute into the > + * out-of-line buffer that we have defined for it. This is similar to a subset > + * of xfs_attr_rmtval_set, but records the current block to the delayed attr > + * context, and leaves transaction handling to the caller. > + */ > +int > +xfs_attr_rmtval_set_blk( > + struct xfs_delattr_context *dac) > +{ > + struct xfs_da_args *args = dac->da_args; > + struct xfs_inode *dp = args->dp; > + struct xfs_bmbt_irec *map = &dac->map; > + int nmap; > + int error; > + > + nmap = 1; > + error = xfs_bmapi_write(args->trans, dp, > + (xfs_fileoff_t)dac->lblkno, > + dac->blkcnt, XFS_BMAPI_ATTRFORK, > + args->total, map, &nmap); > + if (error) > + return error; > + > + ASSERT(nmap == 1); > + ASSERT((map->br_startblock != DELAYSTARTBLOCK) && > + (map->br_startblock != HOLESTARTBLOCK)); > + > + /* roll attribute extent map forwards */ > + dac->lblkno += map->br_blockcount; > + dac->blkcnt -= map->br_blockcount; > + > + return 0; > +} > + > +/* > * Remove the value associated with an attribute by deleting the > * out-of-line buffer that it is stored on. > */ > @@ -671,48 +736,6 @@ xfs_attr_rmtval_invalidate( > } > > /* > - * Remove the value associated with an attribute by deleting the > - * out-of-line buffer that it is stored on. > - */ > -int > -xfs_attr_rmtval_remove( > - struct xfs_da_args *args) > -{ > - xfs_dablk_t lblkno; > - int blkcnt; > - int error = 0; > - int done = 0; > - > - trace_xfs_attr_rmtval_remove(args); > - > - error = xfs_attr_rmtval_invalidate(args); > - if (error) > - return error; > - /* > - * Keep de-allocating extents until the remote-value region is gone. > - */ > - lblkno = args->rmtblkno; > - blkcnt = args->rmtblkcnt; > - while (!done) { > - error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt, > - XFS_BMAPI_ATTRFORK, 1, &done); > - if (error) > - return error; > - error = xfs_defer_finish(&args->trans); > - if (error) > - return error; > - > - /* > - * Close out trans and start the next one in the chain. > - */ > - error = xfs_trans_roll_inode(&args->trans, args->dp); > - if (error) > - return error; > - } > - return 0; > -} > - > -/* > * Remove the value associated with an attribute by deleting the out-of-line > * buffer that it is stored on. Returns EAGAIN for the caller to refresh the > * transaction and recall the function > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h > index ee3337b..482dff9 100644 > --- a/fs/xfs/libxfs/xfs_attr_remote.h > +++ b/fs/xfs/libxfs/xfs_attr_remote.h > @@ -15,4 +15,8 @@ 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_rmt_find_hole(struct xfs_da_args *args); > +int xfs_attr_rmtval_set_value(struct xfs_da_args *args); > +int xfs_attr_rmtval_set_blk(struct xfs_delattr_context *dac); > +int xfs_attr_rmtval_set_init(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 c42f90e..3e8cec5 100644 > --- a/fs/xfs/xfs_attr_inactive.c > +++ b/fs/xfs/xfs_attr_inactive.c > @@ -15,6 +15,7 @@ > #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" > diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h > index a4323a6..26dc8bf 100644 > --- a/fs/xfs/xfs_trace.h > +++ b/fs/xfs/xfs_trace.h > @@ -1784,7 +1784,6 @@ DEFINE_ATTR_EVENT(xfs_attr_refillstate); > > DEFINE_ATTR_EVENT(xfs_attr_rmtval_get); > DEFINE_ATTR_EVENT(xfs_attr_rmtval_set); > -DEFINE_ATTR_EVENT(xfs_attr_rmtval_remove); > > #define DEFINE_DA_EVENT(name) \ > DEFINE_EVENT(xfs_da_class, name, \ > -- > 2.7.4 >
On 4/13/20 6:40 AM, Brian Foster wrote: > On Fri, Apr 03, 2020 at 03:12:28PM -0700, Allison Collins wrote: >> This patch modifies the attr set 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_set_args has become xfs_attr_set_iter, which uses a >> state machine like switch to keep track of where it was when EAGAIN was >> returned. >> >> Two new helper functions have been added: xfs_attr_rmtval_set_init and >> xfs_attr_rmtval_set_blk. They provide a subset of logic similar to >> xfs_attr_rmtval_set, but they store the current block in the delay attr >> context to allow the caller to roll the transaction between allocations. >> This helps to simplify and consolidate code used by >> xfs_attr_leaf_addname and xfs_attr_node_addname. xfs_attr_set_args has >> now become a simple loop to refresh the transaction until the operation >> is completed. Lastly, xfs_attr_rmtval_remove is no longer used, and is >> removed. >> >> Below is a state machine diagram for attr set 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_set_iter() >> │ >> v >> need to upgrade >> from sf to leaf? ──n─┐ >> │ │ >> y │ >> │ │ >> V │ >> XFS_DAS_ADD_LEAF │ >> │ │ >> v │ >> ┌──────n── fork has <──────┘ >> │ only 1 blk? >> │ │ >> │ y >> │ │ >> │ v >> │ xfs_attr_leaf_try_add() >> │ │ >> │ v >> │ had enough >> ├──────n── space? >> │ │ >> │ y >> │ │ >> │ v >> │ XFS_DAS_FOUND_LBLK ──┐ >> │ │ >> │ XFS_DAS_FLIP_LFLAG ──┤ >> │ (subroutine state) │ >> │ │ >> │ XFS_DAS_ALLOC_LEAF ──┤ >> │ (subroutine state) │ >> │ └─>xfs_attr_leaf_addname() >> │ │ >> │ v >> │ ┌─────n── need to >> │ │ alloc blks? >> │ │ │ >> │ │ y >> │ │ │ >> │ │ v >> │ │ ┌─>XFS_DAS_ALLOC_LEAF >> │ │ │ │ >> │ │ │ v >> │ │ └──y── need to alloc >> │ │ more blocks? >> │ │ │ >> │ │ n >> │ │ │ >> │ │ v >> │ │ was this >> │ └────────> a rename? ──n─┐ >> │ │ │ >> │ y │ >> │ │ │ >> │ v │ >> │ flip incomplete │ >> │ flag │ >> │ │ │ >> │ v │ >> │ XFS_DAS_FLIP_LFLAG │ >> │ │ │ >> │ v │ >> │ remove │ >> │ XFS_DAS_RM_LBLK ─> old name │ >> │ ^ │ │ >> │ │ v │ >> │ └──────y── more to │ >> │ remove │ >> │ │ │ >> │ n │ >> │ │ │ >> │ v │ >> │ done <──────┘ >> └────> XFS_DAS_LEAF_TO_NODE ─┐ >> │ >> XFS_DAS_FOUND_NBLK ──┤ >> (subroutine state) │ >> │ >> XFS_DAS_ALLOC_NODE ──┤ >> (subroutine state) │ >> │ >> XFS_DAS_FLIP_NFLAG ──┤ >> (subroutine state) │ >> │ >> └─>xfs_attr_node_addname() >> │ >> v >> find space to store >> attr. Split if needed >> │ >> v >> XFS_DAS_FOUND_NBLK >> │ >> v >> ┌─────n── need to >> │ alloc blks? >> │ │ >> │ y >> │ │ >> │ v >> │ ┌─>XFS_DAS_ALLOC_NODE >> │ │ │ >> │ │ v >> │ └──y── need to alloc >> │ more blocks? >> │ │ >> │ n >> │ │ >> │ v >> │ was this >> └────────> a rename? ──n─┐ >> │ │ >> y │ >> │ │ >> v │ >> flip incomplete │ >> flag │ >> │ │ >> v │ >> XFS_DAS_FLIP_NFLAG │ >> │ │ >> v │ >> remove │ >> XFS_DAS_RM_NBLK ─> old name │ >> ^ │ │ >> │ v │ >> └──────y── more to │ >> remove │ >> │ │ >> n │ >> │ │ >> v │ >> done <──────┘ >> >> Signed-off-by: Allison Collins <allison.henderson@oracle.com> >> --- > > Only a cursory pass given the previous feedback... > >> fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- >> fs/xfs/libxfs/xfs_attr.h | 16 ++ >> fs/xfs/libxfs/xfs_attr_leaf.c | 1 + >> fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- >> fs/xfs/libxfs/xfs_attr_remote.h | 4 + >> fs/xfs/xfs_attr_inactive.c | 1 + >> fs/xfs/xfs_trace.h | 1 - >> 7 files changed, 351 insertions(+), 167 deletions(-) >> >> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c >> index f700976..c160b7a 100644 >> --- a/fs/xfs/libxfs/xfs_attr.c >> +++ b/fs/xfs/libxfs/xfs_attr.c >> @@ -44,7 +44,7 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args); >> * Internal routines when attribute list is one block. >> */ >> STATIC int xfs_attr_leaf_get(xfs_da_args_t *args); >> -STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args); >> +STATIC int xfs_attr_leaf_addname(struct xfs_delattr_context *dac); >> STATIC int xfs_attr_leaf_removename(struct xfs_delattr_context *dac); >> STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); >> >> @@ -52,12 +52,13 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); >> * Internal routines when attribute list is more than one block. >> */ >> 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_addname(struct xfs_delattr_context *dac); >> 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); >> STATIC int xfs_attr_refillstate(xfs_da_state_t *state); >> +STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp); >> >> STATIC void >> xfs_delattr_context_init( >> @@ -227,8 +228,11 @@ xfs_attr_is_shortform( >> >> /* >> * Attempts to set an attr in shortform, or converts the tree to leaf form if >> - * there is not enough room. If the attr is set, the transaction is committed >> - * and set to NULL. >> + * there is not enough room. This function is meant to operate as a helper >> + * routine to the delayed attribute functions. It returns -EAGAIN to indicate >> + * that the calling function should roll the transaction, and then proceed to >> + * add the attr in leaf form. This subroutine does not expect to be recalled >> + * again like the other delayed attr routines do. >> */ >> STATIC int >> xfs_attr_set_shortform( >> @@ -236,16 +240,16 @@ xfs_attr_set_shortform( >> struct xfs_buf **leaf_bp) >> { >> struct xfs_inode *dp = args->dp; >> - int error, error2 = 0; >> + int error = 0; >> >> /* >> * Try to add the attr to the attribute list in the inode. >> */ >> error = xfs_attr_try_sf_addname(dp, args); >> + >> + /* Should only be 0, -EEXIST or ENOSPC */ >> if (error != -ENOSPC) { >> - error2 = xfs_trans_commit(args->trans); >> - args->trans = NULL; >> - return error ? error : error2; >> + return error; >> } >> /* >> * It won't fit in the shortform, transform to a leaf block. GROT: >> @@ -258,18 +262,10 @@ xfs_attr_set_shortform( >> /* >> * Prevent the leaf buffer from being unlocked so that a concurrent AIL >> * push cannot grab the half-baked leaf buffer and run into problems >> - * with the write verifier. Once we're done rolling the transaction we >> - * can release the hold and add the attr to the leaf. >> + * with the write verifier. >> */ >> xfs_trans_bhold(args->trans, *leaf_bp); >> - error = xfs_defer_finish(&args->trans); >> - xfs_trans_bhold_release(args->trans, *leaf_bp); >> - if (error) { >> - xfs_trans_brelse(args->trans, *leaf_bp); >> - return error; >> - } >> - >> - return 0; >> + return -EAGAIN; >> } >> >> /* >> @@ -279,9 +275,83 @@ int >> xfs_attr_set_args( >> struct xfs_da_args *args) >> { >> - struct xfs_inode *dp = args->dp; >> - struct xfs_buf *leaf_bp = NULL; >> - int error = 0; >> + struct xfs_buf *leaf_bp = NULL; >> + int error = 0; >> + struct xfs_delattr_context dac; >> + >> + xfs_delattr_context_init(&dac, args); >> + >> + do { >> + error = xfs_attr_set_iter(&dac, &leaf_bp); >> + if (error != -EAGAIN) >> + break; >> + >> + if (dac.flags & XFS_DAC_DEFER_FINISH) { >> + dac.flags &= ~XFS_DAC_DEFER_FINISH; >> + error = xfs_defer_finish(&args->trans); >> + if (error) >> + break; >> + } >> + >> + error = xfs_trans_roll_inode(&args->trans, args->dp); >> + if (error) >> + break; >> + >> + if (leaf_bp) { >> + xfs_trans_bjoin(args->trans, leaf_bp); >> + xfs_trans_bhold(args->trans, leaf_bp); >> + } >> + >> + } while (true); >> + >> + return error; >> +} >> + >> +/* >> + * Set the attribute specified in @args. >> + * This routine is meant to function as a 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. >> + */ >> +int >> +xfs_attr_set_iter( >> + struct xfs_delattr_context *dac, >> + struct xfs_buf **leaf_bp) >> +{ >> + struct xfs_da_args *args = dac->da_args; >> + struct xfs_inode *dp = args->dp; >> + int error = 0; >> + int sf_size; >> + >> + /* State machine switch */ >> + switch (dac->dela_state) { >> + case XFS_DAS_ADD_LEAF: >> + goto das_add_leaf; >> + case XFS_DAS_ALLOC_LEAF: >> + case XFS_DAS_FLIP_LFLAG: >> + case XFS_DAS_FOUND_LBLK: >> + goto das_leaf; >> + case XFS_DAS_FOUND_NBLK: >> + case XFS_DAS_FLIP_NFLAG: >> + case XFS_DAS_ALLOC_NODE: >> + case XFS_DAS_LEAF_TO_NODE: >> + goto das_node; >> + default: >> + break; >> + } >> + >> + /* >> + * New inodes may not have an attribute fork yet. So set the attribute >> + * fork appropriately >> + */ >> + if (XFS_IFORK_Q((args->dp)) == 0) { >> + sf_size = sizeof(struct xfs_attr_sf_hdr) + >> + XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen); >> + xfs_bmap_set_attrforkoff(args->dp, sf_size, NULL); >> + args->dp->i_afp = kmem_zone_zalloc(xfs_ifork_zone, 0); >> + args->dp->i_afp->if_flags = XFS_IFEXTENTS; >> + } >> > > Is this hunk moved from somewhere? If so, we should probably handle that > in a separate patch. I think we really want these last couple of patches > to introduce the state/markers and not much else. Oh, this hunk is new, not moved, I believe it's been here since I picked up the series quite a while ago. It actually has more to do with parent pointers than delayed atts. Basically when we try to add the parent pointer during a create, the inode isnt fully constructed yet, so we have add the fork here. I will put this in a separate patch and move it further up the set. > >> /* >> * If the attribute list is already in leaf format, jump straight to >> @@ -292,40 +362,53 @@ xfs_attr_set_args( >> if (xfs_attr_is_shortform(dp)) { >> >> /* >> - * If the attr was successfully set in shortform, the >> - * transaction is committed and set to NULL. Otherwise, is it >> - * converted from shortform to leaf, and the transaction is >> - * retained. >> + * If the attr was successfully set in shortform, no need to >> + * continue. Otherwise, is it converted from shortform to leaf >> + * and -EAGAIN is returned. >> */ >> - error = xfs_attr_set_shortform(args, &leaf_bp); >> - if (error || !args->trans) >> - return error; >> + error = xfs_attr_set_shortform(args, leaf_bp); >> + if (error == -EAGAIN) { >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> + dac->dela_state = XFS_DAS_ADD_LEAF; >> + } >> + return error; > > Similar to the previous patch, I wonder if we need the explicit states > that are otherwise handled by existing inode state. For example, if the > above returns -EAGAIN, xfs_attr_is_shortform() is no longer true on > reentry, right? If that's the case for the other conversions, it seems > like we might only need one state (XFS_DAS_FOUND_LBLK) for this > function. Ok, it looks like I can get away with out XFS_DAS_ADD_LEAF and XFS_DAS_LEAF_TO_NODE. This is actually a little similar to how it was done when I was trying to use flags instead of the state enums a long time ago. At the time, I got the impression people were concerned about the complexity and maintainability. So we went back to the enums because the explicite jumping made it clear where the re-entry was to resume, and reduced the risk that a person may mistakenly introduce a change that disrupts the reentry flow. Though perhaps the current arrangement of refactoring has made that less concerning? > > BTW, that general approach might be more clear if we lifted the format > conversions into this level from down in the format specific add > handlers. The goal would be to make the high level flow look something > like: > > if (shortform) { > error = sf_try_add(); > if (error == -ENOSPC) { > shortform_to_leaf(...); > ... > return -EAGAIN; > } Well, actually this piece was hoisted out into the helper function in patch 13. So pulling this up is sort of like letting go of patch 13. :-) The history is: initially I had tried to reduce indentation here in v6 by taking advantage of the _add_leaf label. Because I think in v5 we were trying to unnest where some of the jump points were. So in v6 I had: "if(!shortform) goto leaf;" Which un-nests this shortform code. Then in the v7 review I think Dave suggested I should add the helper and invert the check. So now we have "if(shortform()){ call_helper(); handle -EAGAIN; }" So pulling up is a bit of a circle now. But still functional if people prefer it as it was. Just need to keep track of the history so we avoid going around again. :-) > } else if (xfs_bmap_one_block(...)) { Mmm.... cant jump into the leaf logic right away. We need to handle releasing leaf_bp which looks like it got lost in the pseudo code? IIRC there is a race condition with the AIL that is resolved by holding the leaf buffer across the transaction roll. So we need to check for that and release it upon reentry. > error = xfs_attr_leaf_try_add(args, *leaf_bp); > if (error == -ENOSPC) { > leaf_to_node(...); > return -EAGAIN; > } Ok, so lift ENOSPC handler from xfs_attr_leaf_try_add > > ... state stuff for leaf add ... > } else { > error = xfs_attr_node_addname(dac); > } > > Hm? Of course, something like that should be incorporated via > independent refactoring patches. I think what this becomes right now is: Drop patch 13 Add new patch xfs: Lift ENOSPC handler from xfs_attr_leaf_try_add ? > >> } >> >> - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { >> - error = xfs_attr_leaf_addname(args); >> - if (error != -ENOSPC) >> - return error; >> +das_add_leaf: >> >> - /* >> - * Commit that transaction so that the node_addname() >> - * call can manage its own transactions. >> - */ >> - error = xfs_defer_finish(&args->trans); >> - if (error) >> - return error; >> + /* >> + * After a shortform to leaf conversion, we need to hold the leaf and >> + * cylce out the transaction. When we get back, we need to release >> + * the leaf. >> + */ >> + if (*leaf_bp != NULL) { >> + xfs_trans_brelse(args->trans, *leaf_bp); >> + *leaf_bp = NULL; >> + } >> >> - /* >> - * Commit the current trans (including the inode) and >> - * start a new one. >> - */ >> - error = xfs_trans_roll_inode(&args->trans, dp); >> - if (error) >> + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { >> + error = xfs_attr_leaf_try_add(args, *leaf_bp); >> + switch (error) { >> + case -ENOSPC: >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> + dac->dela_state = XFS_DAS_LEAF_TO_NODE; >> + return -EAGAIN; >> + case 0: >> + dac->dela_state = XFS_DAS_FOUND_LBLK; >> + return -EAGAIN; >> + default: >> return error; >> - >> + } >> +das_leaf: >> + error = xfs_attr_leaf_addname(dac); >> + if (error == -ENOSPC) { >> + dac->dela_state = XFS_DAS_LEAF_TO_NODE; >> + return -EAGAIN; >> + } >> + return error; >> } >> - >> - error = xfs_attr_node_addname(args); >> +das_node: >> + error = xfs_attr_node_addname(dac); >> return error; >> } >> >> @@ -716,28 +799,32 @@ xfs_attr_leaf_try_add( >> * >> * This leaf block cannot have a "remote" value, we only call this routine >> * if bmap_one_block() says there is only one block (ie: no remote blks). >> + * >> + * This routine is meant to function as a 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_leaf_addname( >> - struct xfs_da_args *args) >> + struct xfs_delattr_context *dac) >> { >> - int error, forkoff; >> - struct xfs_buf *bp = NULL; >> - struct xfs_inode *dp = args->dp; >> - >> - trace_xfs_attr_leaf_addname(args); >> - >> - error = xfs_attr_leaf_try_add(args, bp); >> - if (error) >> - return error; >> + struct xfs_da_args *args = dac->da_args; >> + struct xfs_buf *bp = NULL; >> + int error, forkoff; >> + struct xfs_inode *dp = args->dp; >> >> - /* >> - * Commit the transaction that added the attr name so that >> - * later routines can manage their own transactions. >> - */ >> - error = xfs_trans_roll_inode(&args->trans, dp); >> - if (error) >> - return error; >> + /* State machine switch */ >> + switch (dac->dela_state) { >> + case XFS_DAS_FLIP_LFLAG: >> + goto das_flip_flag; >> + case XFS_DAS_ALLOC_LEAF: >> + goto das_alloc_leaf; >> + case XFS_DAS_RM_LBLK: >> + goto das_rm_lblk; >> + default: >> + break; >> + } >> >> /* >> * If there was an out-of-line value, allocate the blocks we >> @@ -746,7 +833,28 @@ xfs_attr_leaf_addname( >> * maximum size of a transaction and/or hit a deadlock. >> */ >> if (args->rmtblkno > 0) { >> - error = xfs_attr_rmtval_set(args); >> + >> + /* Open coded xfs_attr_rmtval_set without trans handling */ >> + error = xfs_attr_rmtval_set_init(dac); >> + if (error) >> + return error; >> + >> + /* >> + * Roll through the "value", allocating blocks on disk as >> + * required. >> + */ >> +das_alloc_leaf: > > If we filter out the setup above, it seems like this state could be > reduced to check for ->blkcnt > 0. By filter out the set up, you mean to introduce the setup flag like you mentioned earlier? For example: if(flags & setup == 0){ setup(); flags |= setup; } To be clear, if we did that, we're talking about adding an "int setup_flags" to the dac. And then defineing a XFS_DAC_SETUP_* scheme. Like a XFS_DAC_SETUP_ADD_NAME, and an XFS_DAC_SETUP_NODE_REMVE_NAME and so on. Because we cant have multiple functions sharing the same set up flag if they ever call each others. Is that what you mean to imply? Otherwise I may not be clear on what you mean by filtering out the set up. > >> + while (dac->blkcnt > 0) { >> + error = xfs_attr_rmtval_set_blk(dac); >> + if (error) >> + return error; >> + >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> + dac->dela_state = XFS_DAS_ALLOC_LEAF; >> + return -EAGAIN; >> + } >> + >> + error = xfs_attr_rmtval_set_value(args); >> if (error) >> return error; >> } >> @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( >> error = xfs_attr3_leaf_flipflags(args); >> if (error) >> return error; >> - /* >> - * Commit the flag value change and start the next trans in >> - * series. >> - */ >> - error = xfs_trans_roll_inode(&args->trans, args->dp); >> - if (error) >> - return error; >> - >> + dac->dela_state = XFS_DAS_FLIP_LFLAG; >> + return -EAGAIN; >> +das_flip_flag: >> /* >> * Dismantle the "old" attribute/value pair by removing >> * a "remote" value (if it exists). >> */ >> xfs_attr_restore_rmt_blk(args); >> >> + xfs_attr_rmtval_invalidate(args); >> +das_rm_lblk: >> if (args->rmtblkno) { >> - error = xfs_attr_rmtval_remove(args); >> + error = __xfs_attr_rmtval_remove(args); >> + >> + if (error == -EAGAIN) { >> + dac->dela_state = XFS_DAS_RM_LBLK; >> + return -EAGAIN; >> + } >> + > > This whole function looks like it could use more refactoring to split > out the rename case. Hmm, how about we take every thing inside the "if (args->op_flags & XFS_DA_OP_RENAME) {" and put it in a xfs_attr_leaf_rename() helper? Then pull the helper into the calling function? > >> if (error) >> return error; >> } >> @@ -799,15 +910,11 @@ xfs_attr_leaf_addname( >> /* >> * If the result is small enough, shrink it all into the inode. >> */ >> - if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) { >> + forkoff = xfs_attr_shortform_allfit(bp, dp); >> + if (forkoff) >> error = xfs_attr3_leaf_to_shortform(bp, args, forkoff); >> - /* bp is gone due to xfs_da_shrink_inode */ >> - if (error) >> - return error; >> - error = xfs_defer_finish(&args->trans); >> - if (error) >> - return error; >> - } >> + >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> >> } else if (args->rmtblkno > 0) { >> /* >> @@ -967,16 +1074,23 @@ xfs_attr_node_hasname( >> * >> * "Remote" attribute values confuse the issue and atomic rename operations >> * add a whole extra layer of confusion on top of that. >> + * >> + * This routine is meant to function as a 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_addname( >> - struct xfs_da_args *args) >> + struct xfs_delattr_context *dac) >> { >> - struct xfs_da_state *state; >> - struct xfs_da_state_blk *blk; >> - struct xfs_inode *dp; >> - struct xfs_mount *mp; >> - int retval, error; >> + struct xfs_da_args *args = dac->da_args; >> + struct xfs_da_state *state = NULL; >> + struct xfs_da_state_blk *blk; >> + struct xfs_inode *dp; >> + struct xfs_mount *mp; >> + int retval = 0; >> + int error = 0; >> >> trace_xfs_attr_node_addname(args); >> >> @@ -985,7 +1099,21 @@ xfs_attr_node_addname( >> */ >> dp = args->dp; >> mp = dp->i_mount; >> -restart: >> + >> + /* State machine switch */ >> + switch (dac->dela_state) { >> + case XFS_DAS_FLIP_NFLAG: >> + goto das_flip_flag; >> + case XFS_DAS_FOUND_NBLK: >> + goto das_found_nblk; >> + case XFS_DAS_ALLOC_NODE: >> + goto das_alloc_node; >> + case XFS_DAS_RM_NBLK: >> + goto das_rm_nblk; >> + default: >> + break; >> + } >> + >> /* >> * Search to see if name already exists, and get back a pointer >> * to where it should go. >> @@ -1031,19 +1159,13 @@ xfs_attr_node_addname( >> error = xfs_attr3_leaf_to_node(args); >> if (error) >> goto out; >> - error = xfs_defer_finish(&args->trans); >> - if (error) >> - goto out; >> >> /* >> - * Commit the node conversion and start the next >> - * trans in the chain. >> + * Restart routine from the top. No need to set the >> + * state >> */ >> - error = xfs_trans_roll_inode(&args->trans, dp); >> - if (error) >> - goto out; >> - >> - goto restart; >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> + return -EAGAIN; >> } >> >> /* >> @@ -1055,9 +1177,7 @@ xfs_attr_node_addname( >> error = xfs_da3_split(state); >> if (error) >> goto out; >> - error = xfs_defer_finish(&args->trans); >> - if (error) >> - goto out; >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> } else { >> /* >> * Addition succeeded, update Btree hashvals. >> @@ -1072,13 +1192,9 @@ xfs_attr_node_addname( >> xfs_da_state_free(state); >> state = NULL; >> >> - /* >> - * Commit the leaf addition or btree split and start the next >> - * trans in the chain. >> - */ >> - error = xfs_trans_roll_inode(&args->trans, dp); >> - if (error) >> - goto out; >> + dac->dela_state = XFS_DAS_FOUND_NBLK; >> + return -EAGAIN; >> +das_found_nblk: > > Same deal here. Any time we have this return -EAGAIN followed by a label > pattern I think we're going to want to think about refactoring things > more first to avoid dumping it in the middle of some unnecessarily large > function. Ok, similar pattern here too then? Everything in "if (args->op_flags & XFS_DA_OP_RENAME) {" goes in a new xfs_attr_node_rename() helper? Then hoist upwards? Thanks for the reviewing!! Allison > > Brian > >> >> /* >> * If there was an out-of-line value, allocate the blocks we >> @@ -1087,7 +1203,27 @@ xfs_attr_node_addname( >> * maximum size of a transaction and/or hit a deadlock. >> */ >> if (args->rmtblkno > 0) { >> - error = xfs_attr_rmtval_set(args); >> + /* Open coded xfs_attr_rmtval_set without trans handling */ >> + error = xfs_attr_rmtval_set_init(dac); >> + if (error) >> + return error; >> + >> + /* >> + * Roll through the "value", allocating blocks on disk as >> + * required. >> + */ >> +das_alloc_node: >> + while (dac->blkcnt > 0) { >> + error = xfs_attr_rmtval_set_blk(dac); >> + if (error) >> + return error; >> + >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> + dac->dela_state = XFS_DAS_ALLOC_NODE; >> + return -EAGAIN; >> + } >> + >> + error = xfs_attr_rmtval_set_value(args); >> if (error) >> return error; >> } >> @@ -1110,18 +1246,26 @@ xfs_attr_node_addname( >> * Commit the flag value change and start the next trans in >> * series >> */ >> - error = xfs_trans_roll_inode(&args->trans, args->dp); >> - if (error) >> - goto out; >> - >> + dac->dela_state = XFS_DAS_FLIP_NFLAG; >> + return -EAGAIN; >> +das_flip_flag: >> /* >> * Dismantle the "old" attribute/value pair by removing >> * a "remote" value (if it exists). >> */ >> xfs_attr_restore_rmt_blk(args); >> >> + xfs_attr_rmtval_invalidate(args); >> + >> +das_rm_nblk: >> if (args->rmtblkno) { >> - error = xfs_attr_rmtval_remove(args); >> + error = __xfs_attr_rmtval_remove(args); >> + >> + if (error == -EAGAIN) { >> + dac->dela_state = XFS_DAS_RM_NBLK; >> + return -EAGAIN; >> + } >> + >> if (error) >> return error; >> } >> @@ -1139,7 +1283,6 @@ xfs_attr_node_addname( >> error = xfs_da3_node_lookup_int(state, &retval); >> if (error) >> goto out; >> - >> /* >> * Remove the name and update the hashvals in the tree. >> */ >> @@ -1147,7 +1290,6 @@ xfs_attr_node_addname( >> ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC); >> error = xfs_attr3_leaf_remove(blk->bp, args); >> xfs_da3_fixhashpath(state, &state->path); >> - >> /* >> * Check to see if the tree needs to be collapsed. >> */ >> @@ -1155,11 +1297,9 @@ xfs_attr_node_addname( >> error = xfs_da3_join(state); >> if (error) >> goto out; >> - error = xfs_defer_finish(&args->trans); >> - if (error) >> - goto out; >> - } >> >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> + } >> } else if (args->rmtblkno > 0) { >> /* >> * Added a "remote" value, just clear the incomplete flag. >> diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h >> index 0e8ae1a..67af9d1 100644 >> --- a/fs/xfs/libxfs/xfs_attr.h >> +++ b/fs/xfs/libxfs/xfs_attr.h >> @@ -93,6 +93,16 @@ enum xfs_delattr_state { >> /* Zero is uninitalized */ >> XFS_DAS_RM_SHRINK = 1, /* We are shrinking the tree */ >> XFS_DAS_RMTVAL_REMOVE, /* We are removing remote value blocks */ >> + XFS_DAS_ADD_LEAF, /* We are adding a leaf attr */ >> + XFS_DAS_FOUND_LBLK, /* We found leaf blk for attr */ >> + XFS_DAS_LEAF_TO_NODE, /* Converted leaf to node */ >> + XFS_DAS_FOUND_NBLK, /* We found node blk for attr */ >> + XFS_DAS_ALLOC_LEAF, /* We are allocating leaf blocks */ >> + XFS_DAS_FLIP_LFLAG, /* Flipped leaf INCOMPLETE attr flag */ >> + XFS_DAS_RM_LBLK, /* A rename is removing leaf blocks */ >> + XFS_DAS_ALLOC_NODE, /* We are allocating node blocks */ >> + XFS_DAS_FLIP_NFLAG, /* Flipped node INCOMPLETE attr flag */ >> + XFS_DAS_RM_NBLK, /* A rename is removing node blocks */ >> }; >> >> /* >> @@ -105,8 +115,13 @@ enum xfs_delattr_state { >> */ >> struct xfs_delattr_context { >> struct xfs_da_args *da_args; >> + struct xfs_bmbt_irec map; >> + struct xfs_buf *leaf_bp; >> + xfs_fileoff_t lfileoff; >> struct xfs_da_state *da_state; >> struct xfs_da_state_blk *blk; >> + xfs_dablk_t lblkno; >> + int blkcnt; >> unsigned int flags; >> enum xfs_delattr_state dela_state; >> }; >> @@ -126,6 +141,7 @@ int xfs_attr_get_ilocked(struct xfs_da_args *args); >> int xfs_attr_get(struct xfs_da_args *args); >> int xfs_attr_set(struct xfs_da_args *args); >> int xfs_attr_set_args(struct xfs_da_args *args); >> +int xfs_attr_set_iter(struct xfs_delattr_context *dac, struct xfs_buf **leaf_bp); >> 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); >> diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c >> index f55402b..4d15f45 100644 >> --- a/fs/xfs/libxfs/xfs_attr_leaf.c >> +++ b/fs/xfs/libxfs/xfs_attr_leaf.c >> @@ -19,6 +19,7 @@ >> #include "xfs_bmap_btree.h" >> #include "xfs_bmap.h" >> #include "xfs_attr_sf.h" >> +#include "xfs_attr.h" >> #include "xfs_attr_remote.h" >> #include "xfs_attr.h" >> #include "xfs_attr_leaf.h" >> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c >> index fd4be9d..9607fd2 100644 >> --- a/fs/xfs/libxfs/xfs_attr_remote.c >> +++ b/fs/xfs/libxfs/xfs_attr_remote.c >> @@ -443,7 +443,7 @@ xfs_attr_rmtval_get( >> * Find a "hole" in the attribute address space large enough for us to drop the >> * new attribute's value into >> */ >> -STATIC int >> +int >> xfs_attr_rmt_find_hole( >> struct xfs_da_args *args) >> { >> @@ -470,7 +470,7 @@ xfs_attr_rmt_find_hole( >> return 0; >> } >> >> -STATIC int >> +int >> xfs_attr_rmtval_set_value( >> struct xfs_da_args *args) >> { >> @@ -630,6 +630,71 @@ xfs_attr_rmtval_set( >> } >> >> /* >> + * Find a hole for the attr and store it in the delayed attr context. This >> + * initializes the context to roll through allocating an attr extent for a >> + * delayed attr operation >> + */ >> +int >> +xfs_attr_rmtval_set_init( >> + struct xfs_delattr_context *dac) >> +{ >> + struct xfs_da_args *args = dac->da_args; >> + struct xfs_bmbt_irec *map = &dac->map; >> + int error; >> + >> + dac->lblkno = 0; >> + dac->lfileoff = 0; >> + dac->blkcnt = 0; >> + args->rmtblkcnt = 0; >> + args->rmtblkno = 0; >> + memset(map, 0, sizeof(struct xfs_bmbt_irec)); >> + >> + error = xfs_attr_rmt_find_hole(args); >> + if (error) >> + return error; >> + >> + dac->blkcnt = args->rmtblkcnt; >> + dac->lblkno = args->rmtblkno; >> + >> + return error; >> +} >> + >> +/* >> + * Write one block of the value associated with an attribute into the >> + * out-of-line buffer that we have defined for it. This is similar to a subset >> + * of xfs_attr_rmtval_set, but records the current block to the delayed attr >> + * context, and leaves transaction handling to the caller. >> + */ >> +int >> +xfs_attr_rmtval_set_blk( >> + struct xfs_delattr_context *dac) >> +{ >> + struct xfs_da_args *args = dac->da_args; >> + struct xfs_inode *dp = args->dp; >> + struct xfs_bmbt_irec *map = &dac->map; >> + int nmap; >> + int error; >> + >> + nmap = 1; >> + error = xfs_bmapi_write(args->trans, dp, >> + (xfs_fileoff_t)dac->lblkno, >> + dac->blkcnt, XFS_BMAPI_ATTRFORK, >> + args->total, map, &nmap); >> + if (error) >> + return error; >> + >> + ASSERT(nmap == 1); >> + ASSERT((map->br_startblock != DELAYSTARTBLOCK) && >> + (map->br_startblock != HOLESTARTBLOCK)); >> + >> + /* roll attribute extent map forwards */ >> + dac->lblkno += map->br_blockcount; >> + dac->blkcnt -= map->br_blockcount; >> + >> + return 0; >> +} >> + >> +/* >> * Remove the value associated with an attribute by deleting the >> * out-of-line buffer that it is stored on. >> */ >> @@ -671,48 +736,6 @@ xfs_attr_rmtval_invalidate( >> } >> >> /* >> - * Remove the value associated with an attribute by deleting the >> - * out-of-line buffer that it is stored on. >> - */ >> -int >> -xfs_attr_rmtval_remove( >> - struct xfs_da_args *args) >> -{ >> - xfs_dablk_t lblkno; >> - int blkcnt; >> - int error = 0; >> - int done = 0; >> - >> - trace_xfs_attr_rmtval_remove(args); >> - >> - error = xfs_attr_rmtval_invalidate(args); >> - if (error) >> - return error; >> - /* >> - * Keep de-allocating extents until the remote-value region is gone. >> - */ >> - lblkno = args->rmtblkno; >> - blkcnt = args->rmtblkcnt; >> - while (!done) { >> - error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt, >> - XFS_BMAPI_ATTRFORK, 1, &done); >> - if (error) >> - return error; >> - error = xfs_defer_finish(&args->trans); >> - if (error) >> - return error; >> - >> - /* >> - * Close out trans and start the next one in the chain. >> - */ >> - error = xfs_trans_roll_inode(&args->trans, args->dp); >> - if (error) >> - return error; >> - } >> - return 0; >> -} >> - >> -/* >> * Remove the value associated with an attribute by deleting the out-of-line >> * buffer that it is stored on. Returns EAGAIN for the caller to refresh the >> * transaction and recall the function >> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h >> index ee3337b..482dff9 100644 >> --- a/fs/xfs/libxfs/xfs_attr_remote.h >> +++ b/fs/xfs/libxfs/xfs_attr_remote.h >> @@ -15,4 +15,8 @@ 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_rmt_find_hole(struct xfs_da_args *args); >> +int xfs_attr_rmtval_set_value(struct xfs_da_args *args); >> +int xfs_attr_rmtval_set_blk(struct xfs_delattr_context *dac); >> +int xfs_attr_rmtval_set_init(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 c42f90e..3e8cec5 100644 >> --- a/fs/xfs/xfs_attr_inactive.c >> +++ b/fs/xfs/xfs_attr_inactive.c >> @@ -15,6 +15,7 @@ >> #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" >> diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h >> index a4323a6..26dc8bf 100644 >> --- a/fs/xfs/xfs_trace.h >> +++ b/fs/xfs/xfs_trace.h >> @@ -1784,7 +1784,6 @@ DEFINE_ATTR_EVENT(xfs_attr_refillstate); >> >> DEFINE_ATTR_EVENT(xfs_attr_rmtval_get); >> DEFINE_ATTR_EVENT(xfs_attr_rmtval_set); >> -DEFINE_ATTR_EVENT(xfs_attr_rmtval_remove); >> >> #define DEFINE_DA_EVENT(name) \ >> DEFINE_EVENT(xfs_da_class, name, \ >> -- >> 2.7.4 >> >
On Wed, Apr 15, 2020 at 03:08:11PM -0700, Allison Collins wrote: > > > On 4/13/20 6:40 AM, Brian Foster wrote: > > On Fri, Apr 03, 2020 at 03:12:28PM -0700, Allison Collins wrote: > > > This patch modifies the attr set 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_set_args has become xfs_attr_set_iter, which uses a > > > state machine like switch to keep track of where it was when EAGAIN was > > > returned. > > > > > > Two new helper functions have been added: xfs_attr_rmtval_set_init and > > > xfs_attr_rmtval_set_blk. They provide a subset of logic similar to > > > xfs_attr_rmtval_set, but they store the current block in the delay attr > > > context to allow the caller to roll the transaction between allocations. > > > This helps to simplify and consolidate code used by > > > xfs_attr_leaf_addname and xfs_attr_node_addname. xfs_attr_set_args has > > > now become a simple loop to refresh the transaction until the operation > > > is completed. Lastly, xfs_attr_rmtval_remove is no longer used, and is > > > removed. > > > > > > Below is a state machine diagram for attr set 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_set_iter() > > > │ > > > v > > > need to upgrade > > > from sf to leaf? ──n─┐ > > > │ │ > > > y │ > > > │ │ > > > V │ > > > XFS_DAS_ADD_LEAF │ > > > │ │ > > > v │ > > > ┌──────n── fork has <──────┘ > > > │ only 1 blk? > > > │ │ > > > │ y > > > │ │ > > > │ v > > > │ xfs_attr_leaf_try_add() > > > │ │ > > > │ v > > > │ had enough > > > ├──────n── space? > > > │ │ > > > │ y > > > │ │ > > > │ v > > > │ XFS_DAS_FOUND_LBLK ──┐ > > > │ │ > > > │ XFS_DAS_FLIP_LFLAG ──┤ > > > │ (subroutine state) │ > > > │ │ > > > │ XFS_DAS_ALLOC_LEAF ──┤ > > > │ (subroutine state) │ > > > │ └─>xfs_attr_leaf_addname() > > > │ │ > > > │ v > > > │ ┌─────n── need to > > > │ │ alloc blks? > > > │ │ │ > > > │ │ y > > > │ │ │ > > > │ │ v > > > │ │ ┌─>XFS_DAS_ALLOC_LEAF > > > │ │ │ │ > > > │ │ │ v > > > │ │ └──y── need to alloc > > > │ │ more blocks? > > > │ │ │ > > > │ │ n > > > │ │ │ > > > │ │ v > > > │ │ was this > > > │ └────────> a rename? ──n─┐ > > > │ │ │ > > > │ y │ > > > │ │ │ > > > │ v │ > > > │ flip incomplete │ > > > │ flag │ > > > │ │ │ > > > │ v │ > > > │ XFS_DAS_FLIP_LFLAG │ > > > │ │ │ > > > │ v │ > > > │ remove │ > > > │ XFS_DAS_RM_LBLK ─> old name │ > > > │ ^ │ │ > > > │ │ v │ > > > │ └──────y── more to │ > > > │ remove │ > > > │ │ │ > > > │ n │ > > > │ │ │ > > > │ v │ > > > │ done <──────┘ > > > └────> XFS_DAS_LEAF_TO_NODE ─┐ > > > │ > > > XFS_DAS_FOUND_NBLK ──┤ > > > (subroutine state) │ > > > │ > > > XFS_DAS_ALLOC_NODE ──┤ > > > (subroutine state) │ > > > │ > > > XFS_DAS_FLIP_NFLAG ──┤ > > > (subroutine state) │ > > > │ > > > └─>xfs_attr_node_addname() > > > │ > > > v > > > find space to store > > > attr. Split if needed > > > │ > > > v > > > XFS_DAS_FOUND_NBLK > > > │ > > > v > > > ┌─────n── need to > > > │ alloc blks? > > > │ │ > > > │ y > > > │ │ > > > │ v > > > │ ┌─>XFS_DAS_ALLOC_NODE > > > │ │ │ > > > │ │ v > > > │ └──y── need to alloc > > > │ more blocks? > > > │ │ > > > │ n > > > │ │ > > > │ v > > > │ was this > > > └────────> a rename? ──n─┐ > > > │ │ > > > y │ > > > │ │ > > > v │ > > > flip incomplete │ > > > flag │ > > > │ │ > > > v │ > > > XFS_DAS_FLIP_NFLAG │ > > > │ │ > > > v │ > > > remove │ > > > XFS_DAS_RM_NBLK ─> old name │ > > > ^ │ │ > > > │ v │ > > > └──────y── more to │ > > > remove │ > > > │ │ > > > n │ > > > │ │ > > > v │ > > > done <──────┘ > > > > > > Signed-off-by: Allison Collins <allison.henderson@oracle.com> > > > --- > > > > Only a cursory pass given the previous feedback... > > > > > fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- > > > fs/xfs/libxfs/xfs_attr.h | 16 ++ > > > fs/xfs/libxfs/xfs_attr_leaf.c | 1 + > > > fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- > > > fs/xfs/libxfs/xfs_attr_remote.h | 4 + > > > fs/xfs/xfs_attr_inactive.c | 1 + > > > fs/xfs/xfs_trace.h | 1 - > > > 7 files changed, 351 insertions(+), 167 deletions(-) > > > > > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > > > index f700976..c160b7a 100644 > > > --- a/fs/xfs/libxfs/xfs_attr.c > > > +++ b/fs/xfs/libxfs/xfs_attr.c > > > @@ -44,7 +44,7 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args); > > > * Internal routines when attribute list is one block. > > > */ > > > STATIC int xfs_attr_leaf_get(xfs_da_args_t *args); > > > -STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args); > > > +STATIC int xfs_attr_leaf_addname(struct xfs_delattr_context *dac); > > > STATIC int xfs_attr_leaf_removename(struct xfs_delattr_context *dac); > > > STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > > > @@ -52,12 +52,13 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > > > * Internal routines when attribute list is more than one block. > > > */ > > > 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_addname(struct xfs_delattr_context *dac); > > > 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); > > > STATIC int xfs_attr_refillstate(xfs_da_state_t *state); > > > +STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp); > > > STATIC void > > > xfs_delattr_context_init( > > > @@ -227,8 +228,11 @@ xfs_attr_is_shortform( > > > /* > > > * Attempts to set an attr in shortform, or converts the tree to leaf form if > > > - * there is not enough room. If the attr is set, the transaction is committed > > > - * and set to NULL. > > > + * there is not enough room. This function is meant to operate as a helper > > > + * routine to the delayed attribute functions. It returns -EAGAIN to indicate > > > + * that the calling function should roll the transaction, and then proceed to > > > + * add the attr in leaf form. This subroutine does not expect to be recalled > > > + * again like the other delayed attr routines do. > > > */ > > > STATIC int > > > xfs_attr_set_shortform( > > > @@ -236,16 +240,16 @@ xfs_attr_set_shortform( > > > struct xfs_buf **leaf_bp) > > > { > > > struct xfs_inode *dp = args->dp; > > > - int error, error2 = 0; > > > + int error = 0; > > > /* > > > * Try to add the attr to the attribute list in the inode. > > > */ > > > error = xfs_attr_try_sf_addname(dp, args); > > > + > > > + /* Should only be 0, -EEXIST or ENOSPC */ > > > if (error != -ENOSPC) { > > > - error2 = xfs_trans_commit(args->trans); > > > - args->trans = NULL; > > > - return error ? error : error2; > > > + return error; > > > } > > > /* > > > * It won't fit in the shortform, transform to a leaf block. GROT: > > > @@ -258,18 +262,10 @@ xfs_attr_set_shortform( > > > /* > > > * Prevent the leaf buffer from being unlocked so that a concurrent AIL > > > * push cannot grab the half-baked leaf buffer and run into problems > > > - * with the write verifier. Once we're done rolling the transaction we > > > - * can release the hold and add the attr to the leaf. > > > + * with the write verifier. > > > */ > > > xfs_trans_bhold(args->trans, *leaf_bp); > > > - error = xfs_defer_finish(&args->trans); > > > - xfs_trans_bhold_release(args->trans, *leaf_bp); > > > - if (error) { > > > - xfs_trans_brelse(args->trans, *leaf_bp); > > > - return error; > > > - } > > > - > > > - return 0; > > > + return -EAGAIN; > > > } > > > /* > > > @@ -279,9 +275,83 @@ int > > > xfs_attr_set_args( > > > struct xfs_da_args *args) > > > { > > > - struct xfs_inode *dp = args->dp; > > > - struct xfs_buf *leaf_bp = NULL; > > > - int error = 0; > > > + struct xfs_buf *leaf_bp = NULL; > > > + int error = 0; > > > + struct xfs_delattr_context dac; > > > + > > > + xfs_delattr_context_init(&dac, args); > > > + > > > + do { > > > + error = xfs_attr_set_iter(&dac, &leaf_bp); > > > + if (error != -EAGAIN) > > > + break; > > > + > > > + if (dac.flags & XFS_DAC_DEFER_FINISH) { > > > + dac.flags &= ~XFS_DAC_DEFER_FINISH; > > > + error = xfs_defer_finish(&args->trans); > > > + if (error) > > > + break; > > > + } > > > + > > > + error = xfs_trans_roll_inode(&args->trans, args->dp); > > > + if (error) > > > + break; > > > + > > > + if (leaf_bp) { > > > + xfs_trans_bjoin(args->trans, leaf_bp); > > > + xfs_trans_bhold(args->trans, leaf_bp); > > > + } > > > + > > > + } while (true); > > > + > > > + return error; > > > +} > > > + > > > +/* > > > + * Set the attribute specified in @args. > > > + * This routine is meant to function as a 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. > > > + */ > > > +int > > > +xfs_attr_set_iter( > > > + struct xfs_delattr_context *dac, > > > + struct xfs_buf **leaf_bp) > > > +{ > > > + struct xfs_da_args *args = dac->da_args; > > > + struct xfs_inode *dp = args->dp; > > > + int error = 0; > > > + int sf_size; > > > + > > > + /* State machine switch */ > > > + switch (dac->dela_state) { > > > + case XFS_DAS_ADD_LEAF: > > > + goto das_add_leaf; > > > + case XFS_DAS_ALLOC_LEAF: > > > + case XFS_DAS_FLIP_LFLAG: > > > + case XFS_DAS_FOUND_LBLK: > > > + goto das_leaf; > > > + case XFS_DAS_FOUND_NBLK: > > > + case XFS_DAS_FLIP_NFLAG: > > > + case XFS_DAS_ALLOC_NODE: > > > + case XFS_DAS_LEAF_TO_NODE: > > > + goto das_node; > > > + default: > > > + break; > > > + } > > > + > > > + /* > > > + * New inodes may not have an attribute fork yet. So set the attribute > > > + * fork appropriately > > > + */ > > > + if (XFS_IFORK_Q((args->dp)) == 0) { > > > + sf_size = sizeof(struct xfs_attr_sf_hdr) + > > > + XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen); > > > + xfs_bmap_set_attrforkoff(args->dp, sf_size, NULL); > > > + args->dp->i_afp = kmem_zone_zalloc(xfs_ifork_zone, 0); > > > + args->dp->i_afp->if_flags = XFS_IFEXTENTS; > > > + } > > > > Is this hunk moved from somewhere? If so, we should probably handle that > > in a separate patch. I think we really want these last couple of patches > > to introduce the state/markers and not much else. > Oh, this hunk is new, not moved, I believe it's been here since I picked up > the series quite a while ago. It actually has more to do with parent > pointers than delayed atts. Basically when we try to add the parent pointer > during a create, the inode isnt fully constructed yet, so we have add the > fork here. I will put this in a separate patch and move it further up the > set. > Ok. > > > > > /* > > > * If the attribute list is already in leaf format, jump straight to > > > @@ -292,40 +362,53 @@ xfs_attr_set_args( > > > if (xfs_attr_is_shortform(dp)) { > > > /* > > > - * If the attr was successfully set in shortform, the > > > - * transaction is committed and set to NULL. Otherwise, is it > > > - * converted from shortform to leaf, and the transaction is > > > - * retained. > > > + * If the attr was successfully set in shortform, no need to > > > + * continue. Otherwise, is it converted from shortform to leaf > > > + * and -EAGAIN is returned. > > > */ > > > - error = xfs_attr_set_shortform(args, &leaf_bp); > > > - if (error || !args->trans) > > > - return error; > > > + error = xfs_attr_set_shortform(args, leaf_bp); > > > + if (error == -EAGAIN) { > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > + dac->dela_state = XFS_DAS_ADD_LEAF; > > > + } > > > + return error; > > > > Similar to the previous patch, I wonder if we need the explicit states > > that are otherwise handled by existing inode state. For example, if the > > above returns -EAGAIN, xfs_attr_is_shortform() is no longer true on > > reentry, right? If that's the case for the other conversions, it seems > > like we might only need one state (XFS_DAS_FOUND_LBLK) for this > > function. > Ok, it looks like I can get away with out XFS_DAS_ADD_LEAF and > XFS_DAS_LEAF_TO_NODE. > > This is actually a little similar to how it was done when I was trying to > use flags instead of the state enums a long time ago. At the time, I got > the impression people were concerned about the complexity and > maintainability. So we went back to the enums because the explicite jumping > made it clear where the re-entry was to resume, and reduced the risk that a > person may mistakenly introduce a change that disrupts the reentry flow. > Though perhaps the current arrangement of refactoring has made that less > concerning? > Perhaps. I think it just depends on context. It might not make sense to try and implement such a reentry model down in the node code where we have N different non-deterministic states to deal with; as opposed to here where we're looking at one of a handful of high level formats and the behavior is very predictable: try to add to the current format, else convert to the next. > > > > BTW, that general approach might be more clear if we lifted the format > > conversions into this level from down in the format specific add > > handlers. The goal would be to make the high level flow look something > > like: > > > > if (shortform) { > > error = sf_try_add(); > > if (error == -ENOSPC) { > > shortform_to_leaf(...); > > ... > > return -EAGAIN; > > } > Well, actually this piece was hoisted out into the helper function in patch > 13. So pulling this up is sort of like letting go of patch 13. :-) > > The history is: initially I had tried to reduce indentation here in v6 by > taking advantage of the _add_leaf label. Because I think in v5 we were > trying to unnest where some of the jump points were. So in v6 I had: > "if(!shortform) goto leaf;" Which un-nests this shortform code. > > Then in the v7 review I think Dave suggested I should add the helper and > invert the check. So now we have "if(shortform()){ call_helper(); handle > -EAGAIN; }" > > So pulling up is a bit of a circle now. But still functional if people > prefer it as it was. Just need to keep track of the history so we avoid > going around again. :-) > I don't think you need to drop that patch necessarily, but perhaps keep the conversion part to a separate helper and then introduce a similar patch for other formats..? In any event, that was just a followup thought so it might be reasonable to defer it to the next version and for now just see what it looks like to rely on the format states. > > > } else if (xfs_bmap_one_block(...)) { > Mmm.... cant jump into the leaf logic right away. We need to handle > releasing leaf_bp which looks like it got lost in the pseudo code? IIRC > there is a race condition with the AIL that is resolved by holding the leaf > buffer across the transaction roll. So we need to check for that and > release it upon reentry. > That sounds more like an implementation detail. leaf_bp is set when we convert from shortform, so isn't this the next state either way? > > > error = xfs_attr_leaf_try_add(args, *leaf_bp); > > if (error == -ENOSPC) { > > leaf_to_node(...); > > return -EAGAIN; > > } > Ok, so lift ENOSPC handler from xfs_attr_leaf_try_add > > > > > ... state stuff for leaf add ... > > } else { > > error = xfs_attr_node_addname(dac); > > } > > > > Hm? Of course, something like that should be incorporated via > > independent refactoring patches. > I think what this becomes right now is: > > Drop patch 13 > Add new patch xfs: Lift ENOSPC handler from xfs_attr_leaf_try_add > > ? > Re: above, I don't think the patch needs to necessarily go away. Feel free to table this for now, just something to think about... > > > > > } > > > - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > > > - error = xfs_attr_leaf_addname(args); > > > - if (error != -ENOSPC) > > > - return error; > > > +das_add_leaf: > > > - /* > > > - * Commit that transaction so that the node_addname() > > > - * call can manage its own transactions. > > > - */ > > > - error = xfs_defer_finish(&args->trans); > > > - if (error) > > > - return error; > > > + /* > > > + * After a shortform to leaf conversion, we need to hold the leaf and > > > + * cylce out the transaction. When we get back, we need to release > > > + * the leaf. > > > + */ > > > + if (*leaf_bp != NULL) { > > > + xfs_trans_brelse(args->trans, *leaf_bp); > > > + *leaf_bp = NULL; > > > + } > > > - /* > > > - * Commit the current trans (including the inode) and > > > - * start a new one. > > > - */ > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > - if (error) > > > + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > > > + error = xfs_attr_leaf_try_add(args, *leaf_bp); > > > + switch (error) { > > > + case -ENOSPC: > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > > > + return -EAGAIN; > > > + case 0: > > > + dac->dela_state = XFS_DAS_FOUND_LBLK; > > > + return -EAGAIN; > > > + default: > > > return error; > > > - > > > + } > > > +das_leaf: > > > + error = xfs_attr_leaf_addname(dac); > > > + if (error == -ENOSPC) { > > > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > > > + return -EAGAIN; > > > + } > > > + return error; > > > } > > > - > > > - error = xfs_attr_node_addname(args); > > > +das_node: > > > + error = xfs_attr_node_addname(dac); > > > return error; > > > } > > > @@ -716,28 +799,32 @@ xfs_attr_leaf_try_add( > > > * > > > * This leaf block cannot have a "remote" value, we only call this routine > > > * if bmap_one_block() says there is only one block (ie: no remote blks). > > > + * > > > + * This routine is meant to function as a 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_leaf_addname( > > > - struct xfs_da_args *args) > > > + struct xfs_delattr_context *dac) > > > { > > > - int error, forkoff; > > > - struct xfs_buf *bp = NULL; > > > - struct xfs_inode *dp = args->dp; > > > - > > > - trace_xfs_attr_leaf_addname(args); > > > - > > > - error = xfs_attr_leaf_try_add(args, bp); > > > - if (error) > > > - return error; > > > + struct xfs_da_args *args = dac->da_args; > > > + struct xfs_buf *bp = NULL; > > > + int error, forkoff; > > > + struct xfs_inode *dp = args->dp; > > > - /* > > > - * Commit the transaction that added the attr name so that > > > - * later routines can manage their own transactions. > > > - */ > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > - if (error) > > > - return error; > > > + /* State machine switch */ > > > + switch (dac->dela_state) { > > > + case XFS_DAS_FLIP_LFLAG: > > > + goto das_flip_flag; > > > + case XFS_DAS_ALLOC_LEAF: > > > + goto das_alloc_leaf; > > > + case XFS_DAS_RM_LBLK: > > > + goto das_rm_lblk; > > > + default: > > > + break; > > > + } > > > /* > > > * If there was an out-of-line value, allocate the blocks we > > > @@ -746,7 +833,28 @@ xfs_attr_leaf_addname( > > > * maximum size of a transaction and/or hit a deadlock. > > > */ > > > if (args->rmtblkno > 0) { > > > - error = xfs_attr_rmtval_set(args); > > > + > > > + /* Open coded xfs_attr_rmtval_set without trans handling */ > > > + error = xfs_attr_rmtval_set_init(dac); > > > + if (error) > > > + return error; > > > + > > > + /* > > > + * Roll through the "value", allocating blocks on disk as > > > + * required. > > > + */ > > > +das_alloc_leaf: > > > > If we filter out the setup above, it seems like this state could be > > reduced to check for ->blkcnt > 0. > By filter out the set up, you mean to introduce the setup flag like you > mentioned earlier? For example: > if(flags & setup == 0){ > setup(); > flags |= setup; > } > > To be clear, if we did that, we're talking about adding an "int setup_flags" > to the dac. And then defineing a XFS_DAC_SETUP_* scheme. Like a > XFS_DAC_SETUP_ADD_NAME, and an XFS_DAC_SETUP_NODE_REMVE_NAME and so on. > Because we cant have multiple functions sharing the same set up flag if they > ever call each others. > > Is that what you mean to imply? Otherwise I may not be clear on what you > mean by filtering out the set up. > Well we already have a flag scheme for the defer thing, right? Is there any reason we couldn't define a similar DAC_DEFER_INIT and let each operation (add/remove) use it as appropriate? Note again that I'm just trying to think about potential simplifications. If there are isolated functions that can be made clearly/simply idempotent (like init/setup helpers tend to be), then doing so might be helpful. I.e., if xfs_attr_rmtval_set_init() could do something like: { if (dax->flags & DAC_RMTVAL_SET_INIT) return; dax->flags |= DAC_RMTVAL_SET_INIT; ... } ... and that removes an execution state, then that might be a win. If it requires more complexity than that, then perhaps the suggestion is not worthwile. ;P > > > > > + while (dac->blkcnt > 0) { > > > + error = xfs_attr_rmtval_set_blk(dac); > > > + if (error) > > > + return error; > > > + > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > + dac->dela_state = XFS_DAS_ALLOC_LEAF; > > > + return -EAGAIN; > > > + } > > > + > > > + error = xfs_attr_rmtval_set_value(args); > > > if (error) > > > return error; > > > } > > > @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( > > > error = xfs_attr3_leaf_flipflags(args); > > > if (error) > > > return error; > > > - /* > > > - * Commit the flag value change and start the next trans in > > > - * series. > > > - */ > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > - if (error) > > > - return error; > > > - > > > + dac->dela_state = XFS_DAS_FLIP_LFLAG; > > > + return -EAGAIN; > > > +das_flip_flag: > > > /* > > > * Dismantle the "old" attribute/value pair by removing > > > * a "remote" value (if it exists). > > > */ > > > xfs_attr_restore_rmt_blk(args); > > > + xfs_attr_rmtval_invalidate(args); > > > +das_rm_lblk: > > > if (args->rmtblkno) { > > > - error = xfs_attr_rmtval_remove(args); > > > + error = __xfs_attr_rmtval_remove(args); > > > + > > > + if (error == -EAGAIN) { > > > + dac->dela_state = XFS_DAS_RM_LBLK; > > > + return -EAGAIN; > > > + } > > > + > > > > This whole function looks like it could use more refactoring to split > > out the rename case. > Hmm, how about we take every thing inside the "if (args->op_flags & > XFS_DA_OP_RENAME) {" and put it in a xfs_attr_leaf_rename() helper? Then > pull the helper into the calling function? > That would help. I'm not sure what the optimal breakdown is off the top of my head tbh. This function is kind of a logical beast between dealing with remote blocks and rename and combinations of the two, so I'd have to stare at that one some more. In general the goal should be to try and avoid jumping in the middle of logic branches as much as possible so there's a nice and visible distinction between the state code and the functional code. Brian > > > > > if (error) > > > return error; > > > } > > > @@ -799,15 +910,11 @@ xfs_attr_leaf_addname( > > > /* > > > * If the result is small enough, shrink it all into the inode. > > > */ > > > - if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) { > > > + forkoff = xfs_attr_shortform_allfit(bp, dp); > > > + if (forkoff) > > > error = xfs_attr3_leaf_to_shortform(bp, args, forkoff); > > > - /* bp is gone due to xfs_da_shrink_inode */ > > > - if (error) > > > - return error; > > > - error = xfs_defer_finish(&args->trans); > > > - if (error) > > > - return error; > > > - } > > > + > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > } else if (args->rmtblkno > 0) { > > > /* > > > @@ -967,16 +1074,23 @@ xfs_attr_node_hasname( > > > * > > > * "Remote" attribute values confuse the issue and atomic rename operations > > > * add a whole extra layer of confusion on top of that. > > > + * > > > + * This routine is meant to function as a 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_addname( > > > - struct xfs_da_args *args) > > > + struct xfs_delattr_context *dac) > > > { > > > - struct xfs_da_state *state; > > > - struct xfs_da_state_blk *blk; > > > - struct xfs_inode *dp; > > > - struct xfs_mount *mp; > > > - int retval, error; > > > + struct xfs_da_args *args = dac->da_args; > > > + struct xfs_da_state *state = NULL; > > > + struct xfs_da_state_blk *blk; > > > + struct xfs_inode *dp; > > > + struct xfs_mount *mp; > > > + int retval = 0; > > > + int error = 0; > > > trace_xfs_attr_node_addname(args); > > > @@ -985,7 +1099,21 @@ xfs_attr_node_addname( > > > */ > > > dp = args->dp; > > > mp = dp->i_mount; > > > -restart: > > > + > > > + /* State machine switch */ > > > + switch (dac->dela_state) { > > > + case XFS_DAS_FLIP_NFLAG: > > > + goto das_flip_flag; > > > + case XFS_DAS_FOUND_NBLK: > > > + goto das_found_nblk; > > > + case XFS_DAS_ALLOC_NODE: > > > + goto das_alloc_node; > > > + case XFS_DAS_RM_NBLK: > > > + goto das_rm_nblk; > > > + default: > > > + break; > > > + } > > > + > > > /* > > > * Search to see if name already exists, and get back a pointer > > > * to where it should go. > > > @@ -1031,19 +1159,13 @@ xfs_attr_node_addname( > > > error = xfs_attr3_leaf_to_node(args); > > > if (error) > > > goto out; > > > - error = xfs_defer_finish(&args->trans); > > > - if (error) > > > - goto out; > > > /* > > > - * Commit the node conversion and start the next > > > - * trans in the chain. > > > + * Restart routine from the top. No need to set the > > > + * state > > > */ > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > - if (error) > > > - goto out; > > > - > > > - goto restart; > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > + return -EAGAIN; > > > } > > > /* > > > @@ -1055,9 +1177,7 @@ xfs_attr_node_addname( > > > error = xfs_da3_split(state); > > > if (error) > > > goto out; > > > - error = xfs_defer_finish(&args->trans); > > > - if (error) > > > - goto out; > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > } else { > > > /* > > > * Addition succeeded, update Btree hashvals. > > > @@ -1072,13 +1192,9 @@ xfs_attr_node_addname( > > > xfs_da_state_free(state); > > > state = NULL; > > > - /* > > > - * Commit the leaf addition or btree split and start the next > > > - * trans in the chain. > > > - */ > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > - if (error) > > > - goto out; > > > + dac->dela_state = XFS_DAS_FOUND_NBLK; > > > + return -EAGAIN; > > > +das_found_nblk: > > > > Same deal here. Any time we have this return -EAGAIN followed by a label > > pattern I think we're going to want to think about refactoring things > > more first to avoid dumping it in the middle of some unnecessarily large > > function. > Ok, similar pattern here too then? Everything in "if (args->op_flags & > XFS_DA_OP_RENAME) {" goes in a new xfs_attr_node_rename() helper? Then > hoist upwards? > > Thanks for the reviewing!! > Allison > > > > > Brian > > > > > /* > > > * If there was an out-of-line value, allocate the blocks we > > > @@ -1087,7 +1203,27 @@ xfs_attr_node_addname( > > > * maximum size of a transaction and/or hit a deadlock. > > > */ > > > if (args->rmtblkno > 0) { > > > - error = xfs_attr_rmtval_set(args); > > > + /* Open coded xfs_attr_rmtval_set without trans handling */ > > > + error = xfs_attr_rmtval_set_init(dac); > > > + if (error) > > > + return error; > > > + > > > + /* > > > + * Roll through the "value", allocating blocks on disk as > > > + * required. > > > + */ > > > +das_alloc_node: > > > + while (dac->blkcnt > 0) { > > > + error = xfs_attr_rmtval_set_blk(dac); > > > + if (error) > > > + return error; > > > + > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > + dac->dela_state = XFS_DAS_ALLOC_NODE; > > > + return -EAGAIN; > > > + } > > > + > > > + error = xfs_attr_rmtval_set_value(args); > > > if (error) > > > return error; > > > } > > > @@ -1110,18 +1246,26 @@ xfs_attr_node_addname( > > > * Commit the flag value change and start the next trans in > > > * series > > > */ > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > - if (error) > > > - goto out; > > > - > > > + dac->dela_state = XFS_DAS_FLIP_NFLAG; > > > + return -EAGAIN; > > > +das_flip_flag: > > > /* > > > * Dismantle the "old" attribute/value pair by removing > > > * a "remote" value (if it exists). > > > */ > > > xfs_attr_restore_rmt_blk(args); > > > + xfs_attr_rmtval_invalidate(args); > > > + > > > +das_rm_nblk: > > > if (args->rmtblkno) { > > > - error = xfs_attr_rmtval_remove(args); > > > + error = __xfs_attr_rmtval_remove(args); > > > + > > > + if (error == -EAGAIN) { > > > + dac->dela_state = XFS_DAS_RM_NBLK; > > > + return -EAGAIN; > > > + } > > > + > > > if (error) > > > return error; > > > } > > > @@ -1139,7 +1283,6 @@ xfs_attr_node_addname( > > > error = xfs_da3_node_lookup_int(state, &retval); > > > if (error) > > > goto out; > > > - > > > /* > > > * Remove the name and update the hashvals in the tree. > > > */ > > > @@ -1147,7 +1290,6 @@ xfs_attr_node_addname( > > > ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC); > > > error = xfs_attr3_leaf_remove(blk->bp, args); > > > xfs_da3_fixhashpath(state, &state->path); > > > - > > > /* > > > * Check to see if the tree needs to be collapsed. > > > */ > > > @@ -1155,11 +1297,9 @@ xfs_attr_node_addname( > > > error = xfs_da3_join(state); > > > if (error) > > > goto out; > > > - error = xfs_defer_finish(&args->trans); > > > - if (error) > > > - goto out; > > > - } > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > + } > > > } else if (args->rmtblkno > 0) { > > > /* > > > * Added a "remote" value, just clear the incomplete flag. > > > diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h > > > index 0e8ae1a..67af9d1 100644 > > > --- a/fs/xfs/libxfs/xfs_attr.h > > > +++ b/fs/xfs/libxfs/xfs_attr.h > > > @@ -93,6 +93,16 @@ enum xfs_delattr_state { > > > /* Zero is uninitalized */ > > > XFS_DAS_RM_SHRINK = 1, /* We are shrinking the tree */ > > > XFS_DAS_RMTVAL_REMOVE, /* We are removing remote value blocks */ > > > + XFS_DAS_ADD_LEAF, /* We are adding a leaf attr */ > > > + XFS_DAS_FOUND_LBLK, /* We found leaf blk for attr */ > > > + XFS_DAS_LEAF_TO_NODE, /* Converted leaf to node */ > > > + XFS_DAS_FOUND_NBLK, /* We found node blk for attr */ > > > + XFS_DAS_ALLOC_LEAF, /* We are allocating leaf blocks */ > > > + XFS_DAS_FLIP_LFLAG, /* Flipped leaf INCOMPLETE attr flag */ > > > + XFS_DAS_RM_LBLK, /* A rename is removing leaf blocks */ > > > + XFS_DAS_ALLOC_NODE, /* We are allocating node blocks */ > > > + XFS_DAS_FLIP_NFLAG, /* Flipped node INCOMPLETE attr flag */ > > > + XFS_DAS_RM_NBLK, /* A rename is removing node blocks */ > > > }; > > > /* > > > @@ -105,8 +115,13 @@ enum xfs_delattr_state { > > > */ > > > struct xfs_delattr_context { > > > struct xfs_da_args *da_args; > > > + struct xfs_bmbt_irec map; > > > + struct xfs_buf *leaf_bp; > > > + xfs_fileoff_t lfileoff; > > > struct xfs_da_state *da_state; > > > struct xfs_da_state_blk *blk; > > > + xfs_dablk_t lblkno; > > > + int blkcnt; > > > unsigned int flags; > > > enum xfs_delattr_state dela_state; > > > }; > > > @@ -126,6 +141,7 @@ int xfs_attr_get_ilocked(struct xfs_da_args *args); > > > int xfs_attr_get(struct xfs_da_args *args); > > > int xfs_attr_set(struct xfs_da_args *args); > > > int xfs_attr_set_args(struct xfs_da_args *args); > > > +int xfs_attr_set_iter(struct xfs_delattr_context *dac, struct xfs_buf **leaf_bp); > > > 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); > > > diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c > > > index f55402b..4d15f45 100644 > > > --- a/fs/xfs/libxfs/xfs_attr_leaf.c > > > +++ b/fs/xfs/libxfs/xfs_attr_leaf.c > > > @@ -19,6 +19,7 @@ > > > #include "xfs_bmap_btree.h" > > > #include "xfs_bmap.h" > > > #include "xfs_attr_sf.h" > > > +#include "xfs_attr.h" > > > #include "xfs_attr_remote.h" > > > #include "xfs_attr.h" > > > #include "xfs_attr_leaf.h" > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c > > > index fd4be9d..9607fd2 100644 > > > --- a/fs/xfs/libxfs/xfs_attr_remote.c > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.c > > > @@ -443,7 +443,7 @@ xfs_attr_rmtval_get( > > > * Find a "hole" in the attribute address space large enough for us to drop the > > > * new attribute's value into > > > */ > > > -STATIC int > > > +int > > > xfs_attr_rmt_find_hole( > > > struct xfs_da_args *args) > > > { > > > @@ -470,7 +470,7 @@ xfs_attr_rmt_find_hole( > > > return 0; > > > } > > > -STATIC int > > > +int > > > xfs_attr_rmtval_set_value( > > > struct xfs_da_args *args) > > > { > > > @@ -630,6 +630,71 @@ xfs_attr_rmtval_set( > > > } > > > /* > > > + * Find a hole for the attr and store it in the delayed attr context. This > > > + * initializes the context to roll through allocating an attr extent for a > > > + * delayed attr operation > > > + */ > > > +int > > > +xfs_attr_rmtval_set_init( > > > + struct xfs_delattr_context *dac) > > > +{ > > > + struct xfs_da_args *args = dac->da_args; > > > + struct xfs_bmbt_irec *map = &dac->map; > > > + int error; > > > + > > > + dac->lblkno = 0; > > > + dac->lfileoff = 0; > > > + dac->blkcnt = 0; > > > + args->rmtblkcnt = 0; > > > + args->rmtblkno = 0; > > > + memset(map, 0, sizeof(struct xfs_bmbt_irec)); > > > + > > > + error = xfs_attr_rmt_find_hole(args); > > > + if (error) > > > + return error; > > > + > > > + dac->blkcnt = args->rmtblkcnt; > > > + dac->lblkno = args->rmtblkno; > > > + > > > + return error; > > > +} > > > + > > > +/* > > > + * Write one block of the value associated with an attribute into the > > > + * out-of-line buffer that we have defined for it. This is similar to a subset > > > + * of xfs_attr_rmtval_set, but records the current block to the delayed attr > > > + * context, and leaves transaction handling to the caller. > > > + */ > > > +int > > > +xfs_attr_rmtval_set_blk( > > > + struct xfs_delattr_context *dac) > > > +{ > > > + struct xfs_da_args *args = dac->da_args; > > > + struct xfs_inode *dp = args->dp; > > > + struct xfs_bmbt_irec *map = &dac->map; > > > + int nmap; > > > + int error; > > > + > > > + nmap = 1; > > > + error = xfs_bmapi_write(args->trans, dp, > > > + (xfs_fileoff_t)dac->lblkno, > > > + dac->blkcnt, XFS_BMAPI_ATTRFORK, > > > + args->total, map, &nmap); > > > + if (error) > > > + return error; > > > + > > > + ASSERT(nmap == 1); > > > + ASSERT((map->br_startblock != DELAYSTARTBLOCK) && > > > + (map->br_startblock != HOLESTARTBLOCK)); > > > + > > > + /* roll attribute extent map forwards */ > > > + dac->lblkno += map->br_blockcount; > > > + dac->blkcnt -= map->br_blockcount; > > > + > > > + return 0; > > > +} > > > + > > > +/* > > > * Remove the value associated with an attribute by deleting the > > > * out-of-line buffer that it is stored on. > > > */ > > > @@ -671,48 +736,6 @@ xfs_attr_rmtval_invalidate( > > > } > > > /* > > > - * Remove the value associated with an attribute by deleting the > > > - * out-of-line buffer that it is stored on. > > > - */ > > > -int > > > -xfs_attr_rmtval_remove( > > > - struct xfs_da_args *args) > > > -{ > > > - xfs_dablk_t lblkno; > > > - int blkcnt; > > > - int error = 0; > > > - int done = 0; > > > - > > > - trace_xfs_attr_rmtval_remove(args); > > > - > > > - error = xfs_attr_rmtval_invalidate(args); > > > - if (error) > > > - return error; > > > - /* > > > - * Keep de-allocating extents until the remote-value region is gone. > > > - */ > > > - lblkno = args->rmtblkno; > > > - blkcnt = args->rmtblkcnt; > > > - while (!done) { > > > - error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt, > > > - XFS_BMAPI_ATTRFORK, 1, &done); > > > - if (error) > > > - return error; > > > - error = xfs_defer_finish(&args->trans); > > > - if (error) > > > - return error; > > > - > > > - /* > > > - * Close out trans and start the next one in the chain. > > > - */ > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > - if (error) > > > - return error; > > > - } > > > - return 0; > > > -} > > > - > > > -/* > > > * Remove the value associated with an attribute by deleting the out-of-line > > > * buffer that it is stored on. Returns EAGAIN for the caller to refresh the > > > * transaction and recall the function > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h > > > index ee3337b..482dff9 100644 > > > --- a/fs/xfs/libxfs/xfs_attr_remote.h > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.h > > > @@ -15,4 +15,8 @@ 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_rmt_find_hole(struct xfs_da_args *args); > > > +int xfs_attr_rmtval_set_value(struct xfs_da_args *args); > > > +int xfs_attr_rmtval_set_blk(struct xfs_delattr_context *dac); > > > +int xfs_attr_rmtval_set_init(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 c42f90e..3e8cec5 100644 > > > --- a/fs/xfs/xfs_attr_inactive.c > > > +++ b/fs/xfs/xfs_attr_inactive.c > > > @@ -15,6 +15,7 @@ > > > #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" > > > diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h > > > index a4323a6..26dc8bf 100644 > > > --- a/fs/xfs/xfs_trace.h > > > +++ b/fs/xfs/xfs_trace.h > > > @@ -1784,7 +1784,6 @@ DEFINE_ATTR_EVENT(xfs_attr_refillstate); > > > DEFINE_ATTR_EVENT(xfs_attr_rmtval_get); > > > DEFINE_ATTR_EVENT(xfs_attr_rmtval_set); > > > -DEFINE_ATTR_EVENT(xfs_attr_rmtval_remove); > > > #define DEFINE_DA_EVENT(name) \ > > > DEFINE_EVENT(xfs_da_class, name, \ > > > -- > > > 2.7.4 > > > > > >
On 4/16/20 4:01 AM, Brian Foster wrote: > On Wed, Apr 15, 2020 at 03:08:11PM -0700, Allison Collins wrote: >> >> >> On 4/13/20 6:40 AM, Brian Foster wrote: >>> On Fri, Apr 03, 2020 at 03:12:28PM -0700, Allison Collins wrote: >>>> This patch modifies the attr set 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_set_args has become xfs_attr_set_iter, which uses a >>>> state machine like switch to keep track of where it was when EAGAIN was >>>> returned. >>>> >>>> Two new helper functions have been added: xfs_attr_rmtval_set_init and >>>> xfs_attr_rmtval_set_blk. They provide a subset of logic similar to >>>> xfs_attr_rmtval_set, but they store the current block in the delay attr >>>> context to allow the caller to roll the transaction between allocations. >>>> This helps to simplify and consolidate code used by >>>> xfs_attr_leaf_addname and xfs_attr_node_addname. xfs_attr_set_args has >>>> now become a simple loop to refresh the transaction until the operation >>>> is completed. Lastly, xfs_attr_rmtval_remove is no longer used, and is >>>> removed. >>>> >>>> Below is a state machine diagram for attr set 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_set_iter() >>>> │ >>>> v >>>> need to upgrade >>>> from sf to leaf? ──n─┐ >>>> │ │ >>>> y │ >>>> │ │ >>>> V │ >>>> XFS_DAS_ADD_LEAF │ >>>> │ │ >>>> v │ >>>> ┌──────n── fork has <──────┘ >>>> │ only 1 blk? >>>> │ │ >>>> │ y >>>> │ │ >>>> │ v >>>> │ xfs_attr_leaf_try_add() >>>> │ │ >>>> │ v >>>> │ had enough >>>> ├──────n── space? >>>> │ │ >>>> │ y >>>> │ │ >>>> │ v >>>> │ XFS_DAS_FOUND_LBLK ──┐ >>>> │ │ >>>> │ XFS_DAS_FLIP_LFLAG ──┤ >>>> │ (subroutine state) │ >>>> │ │ >>>> │ XFS_DAS_ALLOC_LEAF ──┤ >>>> │ (subroutine state) │ >>>> │ └─>xfs_attr_leaf_addname() >>>> │ │ >>>> │ v >>>> │ ┌─────n── need to >>>> │ │ alloc blks? >>>> │ │ │ >>>> │ │ y >>>> │ │ │ >>>> │ │ v >>>> │ │ ┌─>XFS_DAS_ALLOC_LEAF >>>> │ │ │ │ >>>> │ │ │ v >>>> │ │ └──y── need to alloc >>>> │ │ more blocks? >>>> │ │ │ >>>> │ │ n >>>> │ │ │ >>>> │ │ v >>>> │ │ was this >>>> │ └────────> a rename? ──n─┐ >>>> │ │ │ >>>> │ y │ >>>> │ │ │ >>>> │ v │ >>>> │ flip incomplete │ >>>> │ flag │ >>>> │ │ │ >>>> │ v │ >>>> │ XFS_DAS_FLIP_LFLAG │ >>>> │ │ │ >>>> │ v │ >>>> │ remove │ >>>> │ XFS_DAS_RM_LBLK ─> old name │ >>>> │ ^ │ │ >>>> │ │ v │ >>>> │ └──────y── more to │ >>>> │ remove │ >>>> │ │ │ >>>> │ n │ >>>> │ │ │ >>>> │ v │ >>>> │ done <──────┘ >>>> └────> XFS_DAS_LEAF_TO_NODE ─┐ >>>> │ >>>> XFS_DAS_FOUND_NBLK ──┤ >>>> (subroutine state) │ >>>> │ >>>> XFS_DAS_ALLOC_NODE ──┤ >>>> (subroutine state) │ >>>> │ >>>> XFS_DAS_FLIP_NFLAG ──┤ >>>> (subroutine state) │ >>>> │ >>>> └─>xfs_attr_node_addname() >>>> │ >>>> v >>>> find space to store >>>> attr. Split if needed >>>> │ >>>> v >>>> XFS_DAS_FOUND_NBLK >>>> │ >>>> v >>>> ┌─────n── need to >>>> │ alloc blks? >>>> │ │ >>>> │ y >>>> │ │ >>>> │ v >>>> │ ┌─>XFS_DAS_ALLOC_NODE >>>> │ │ │ >>>> │ │ v >>>> │ └──y── need to alloc >>>> │ more blocks? >>>> │ │ >>>> │ n >>>> │ │ >>>> │ v >>>> │ was this >>>> └────────> a rename? ──n─┐ >>>> │ │ >>>> y │ >>>> │ │ >>>> v │ >>>> flip incomplete │ >>>> flag │ >>>> │ │ >>>> v │ >>>> XFS_DAS_FLIP_NFLAG │ >>>> │ │ >>>> v │ >>>> remove │ >>>> XFS_DAS_RM_NBLK ─> old name │ >>>> ^ │ │ >>>> │ v │ >>>> └──────y── more to │ >>>> remove │ >>>> │ │ >>>> n │ >>>> │ │ >>>> v │ >>>> done <──────┘ >>>> >>>> Signed-off-by: Allison Collins <allison.henderson@oracle.com> >>>> --- >>> >>> Only a cursory pass given the previous feedback... >>> >>>> fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- >>>> fs/xfs/libxfs/xfs_attr.h | 16 ++ >>>> fs/xfs/libxfs/xfs_attr_leaf.c | 1 + >>>> fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- >>>> fs/xfs/libxfs/xfs_attr_remote.h | 4 + >>>> fs/xfs/xfs_attr_inactive.c | 1 + >>>> fs/xfs/xfs_trace.h | 1 - >>>> 7 files changed, 351 insertions(+), 167 deletions(-) >>>> >>>> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c >>>> index f700976..c160b7a 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr.c >>>> +++ b/fs/xfs/libxfs/xfs_attr.c >>>> @@ -44,7 +44,7 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args); >>>> * Internal routines when attribute list is one block. >>>> */ >>>> STATIC int xfs_attr_leaf_get(xfs_da_args_t *args); >>>> -STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args); >>>> +STATIC int xfs_attr_leaf_addname(struct xfs_delattr_context *dac); >>>> STATIC int xfs_attr_leaf_removename(struct xfs_delattr_context *dac); >>>> STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); >>>> @@ -52,12 +52,13 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); >>>> * Internal routines when attribute list is more than one block. >>>> */ >>>> 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_addname(struct xfs_delattr_context *dac); >>>> 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); >>>> STATIC int xfs_attr_refillstate(xfs_da_state_t *state); >>>> +STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp); >>>> STATIC void >>>> xfs_delattr_context_init( >>>> @@ -227,8 +228,11 @@ xfs_attr_is_shortform( >>>> /* >>>> * Attempts to set an attr in shortform, or converts the tree to leaf form if >>>> - * there is not enough room. If the attr is set, the transaction is committed >>>> - * and set to NULL. >>>> + * there is not enough room. This function is meant to operate as a helper >>>> + * routine to the delayed attribute functions. It returns -EAGAIN to indicate >>>> + * that the calling function should roll the transaction, and then proceed to >>>> + * add the attr in leaf form. This subroutine does not expect to be recalled >>>> + * again like the other delayed attr routines do. >>>> */ >>>> STATIC int >>>> xfs_attr_set_shortform( >>>> @@ -236,16 +240,16 @@ xfs_attr_set_shortform( >>>> struct xfs_buf **leaf_bp) >>>> { >>>> struct xfs_inode *dp = args->dp; >>>> - int error, error2 = 0; >>>> + int error = 0; >>>> /* >>>> * Try to add the attr to the attribute list in the inode. >>>> */ >>>> error = xfs_attr_try_sf_addname(dp, args); >>>> + >>>> + /* Should only be 0, -EEXIST or ENOSPC */ >>>> if (error != -ENOSPC) { >>>> - error2 = xfs_trans_commit(args->trans); >>>> - args->trans = NULL; >>>> - return error ? error : error2; >>>> + return error; >>>> } >>>> /* >>>> * It won't fit in the shortform, transform to a leaf block. GROT: >>>> @@ -258,18 +262,10 @@ xfs_attr_set_shortform( >>>> /* >>>> * Prevent the leaf buffer from being unlocked so that a concurrent AIL >>>> * push cannot grab the half-baked leaf buffer and run into problems >>>> - * with the write verifier. Once we're done rolling the transaction we >>>> - * can release the hold and add the attr to the leaf. >>>> + * with the write verifier. >>>> */ >>>> xfs_trans_bhold(args->trans, *leaf_bp); >>>> - error = xfs_defer_finish(&args->trans); >>>> - xfs_trans_bhold_release(args->trans, *leaf_bp); >>>> - if (error) { >>>> - xfs_trans_brelse(args->trans, *leaf_bp); >>>> - return error; >>>> - } >>>> - >>>> - return 0; >>>> + return -EAGAIN; >>>> } >>>> /* >>>> @@ -279,9 +275,83 @@ int >>>> xfs_attr_set_args( >>>> struct xfs_da_args *args) >>>> { >>>> - struct xfs_inode *dp = args->dp; >>>> - struct xfs_buf *leaf_bp = NULL; >>>> - int error = 0; >>>> + struct xfs_buf *leaf_bp = NULL; >>>> + int error = 0; >>>> + struct xfs_delattr_context dac; >>>> + >>>> + xfs_delattr_context_init(&dac, args); >>>> + >>>> + do { >>>> + error = xfs_attr_set_iter(&dac, &leaf_bp); >>>> + if (error != -EAGAIN) >>>> + break; >>>> + >>>> + if (dac.flags & XFS_DAC_DEFER_FINISH) { >>>> + dac.flags &= ~XFS_DAC_DEFER_FINISH; >>>> + error = xfs_defer_finish(&args->trans); >>>> + if (error) >>>> + break; >>>> + } >>>> + >>>> + error = xfs_trans_roll_inode(&args->trans, args->dp); >>>> + if (error) >>>> + break; >>>> + >>>> + if (leaf_bp) { >>>> + xfs_trans_bjoin(args->trans, leaf_bp); >>>> + xfs_trans_bhold(args->trans, leaf_bp); >>>> + } >>>> + >>>> + } while (true); >>>> + >>>> + return error; >>>> +} >>>> + >>>> +/* >>>> + * Set the attribute specified in @args. >>>> + * This routine is meant to function as a 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. >>>> + */ >>>> +int >>>> +xfs_attr_set_iter( >>>> + struct xfs_delattr_context *dac, >>>> + struct xfs_buf **leaf_bp) >>>> +{ >>>> + struct xfs_da_args *args = dac->da_args; >>>> + struct xfs_inode *dp = args->dp; >>>> + int error = 0; >>>> + int sf_size; >>>> + >>>> + /* State machine switch */ >>>> + switch (dac->dela_state) { >>>> + case XFS_DAS_ADD_LEAF: >>>> + goto das_add_leaf; >>>> + case XFS_DAS_ALLOC_LEAF: >>>> + case XFS_DAS_FLIP_LFLAG: >>>> + case XFS_DAS_FOUND_LBLK: >>>> + goto das_leaf; >>>> + case XFS_DAS_FOUND_NBLK: >>>> + case XFS_DAS_FLIP_NFLAG: >>>> + case XFS_DAS_ALLOC_NODE: >>>> + case XFS_DAS_LEAF_TO_NODE: >>>> + goto das_node; >>>> + default: >>>> + break; >>>> + } >>>> + >>>> + /* >>>> + * New inodes may not have an attribute fork yet. So set the attribute >>>> + * fork appropriately >>>> + */ >>>> + if (XFS_IFORK_Q((args->dp)) == 0) { >>>> + sf_size = sizeof(struct xfs_attr_sf_hdr) + >>>> + XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen); >>>> + xfs_bmap_set_attrforkoff(args->dp, sf_size, NULL); >>>> + args->dp->i_afp = kmem_zone_zalloc(xfs_ifork_zone, 0); >>>> + args->dp->i_afp->if_flags = XFS_IFEXTENTS; >>>> + } >>> >>> Is this hunk moved from somewhere? If so, we should probably handle that >>> in a separate patch. I think we really want these last couple of patches >>> to introduce the state/markers and not much else. >> Oh, this hunk is new, not moved, I believe it's been here since I picked up >> the series quite a while ago. It actually has more to do with parent >> pointers than delayed atts. Basically when we try to add the parent pointer >> during a create, the inode isnt fully constructed yet, so we have add the >> fork here. I will put this in a separate patch and move it further up the >> set. >> > > Ok. > >>> >>>> /* >>>> * If the attribute list is already in leaf format, jump straight to >>>> @@ -292,40 +362,53 @@ xfs_attr_set_args( >>>> if (xfs_attr_is_shortform(dp)) { >>>> /* >>>> - * If the attr was successfully set in shortform, the >>>> - * transaction is committed and set to NULL. Otherwise, is it >>>> - * converted from shortform to leaf, and the transaction is >>>> - * retained. >>>> + * If the attr was successfully set in shortform, no need to >>>> + * continue. Otherwise, is it converted from shortform to leaf >>>> + * and -EAGAIN is returned. >>>> */ >>>> - error = xfs_attr_set_shortform(args, &leaf_bp); >>>> - if (error || !args->trans) >>>> - return error; >>>> + error = xfs_attr_set_shortform(args, leaf_bp); >>>> + if (error == -EAGAIN) { >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> + dac->dela_state = XFS_DAS_ADD_LEAF; >>>> + } >>>> + return error; >>> >>> Similar to the previous patch, I wonder if we need the explicit states >>> that are otherwise handled by existing inode state. For example, if the >>> above returns -EAGAIN, xfs_attr_is_shortform() is no longer true on >>> reentry, right? If that's the case for the other conversions, it seems >>> like we might only need one state (XFS_DAS_FOUND_LBLK) for this >>> function. >> Ok, it looks like I can get away with out XFS_DAS_ADD_LEAF and >> XFS_DAS_LEAF_TO_NODE. >> >> This is actually a little similar to how it was done when I was trying to >> use flags instead of the state enums a long time ago. At the time, I got >> the impression people were concerned about the complexity and >> maintainability. So we went back to the enums because the explicite jumping >> made it clear where the re-entry was to resume, and reduced the risk that a >> person may mistakenly introduce a change that disrupts the reentry flow. >> Though perhaps the current arrangement of refactoring has made that less >> concerning? >> > > Perhaps. I think it just depends on context. It might not make sense to > try and implement such a reentry model down in the node code where we > have N different non-deterministic states to deal with; as opposed to > here where we're looking at one of a handful of high level formats and > the behavior is very predictable: try to add to the current format, else > convert to the next. Alrighty then, that sounds reasonable. I try my best to recall history of the decisions just as a sort of guide of where we've been and where to go next. > >>> >>> BTW, that general approach might be more clear if we lifted the format >>> conversions into this level from down in the format specific add >>> handlers. The goal would be to make the high level flow look something >>> like: >>> >>> if (shortform) { >>> error = sf_try_add(); >>> if (error == -ENOSPC) { >>> shortform_to_leaf(...); >>> ... >>> return -EAGAIN; >>> } >> Well, actually this piece was hoisted out into the helper function in patch >> 13. So pulling this up is sort of like letting go of patch 13. :-) >> >> The history is: initially I had tried to reduce indentation here in v6 by >> taking advantage of the _add_leaf label. Because I think in v5 we were >> trying to unnest where some of the jump points were. So in v6 I had: >> "if(!shortform) goto leaf;" Which un-nests this shortform code. >> >> Then in the v7 review I think Dave suggested I should add the helper and >> invert the check. So now we have "if(shortform()){ call_helper(); handle >> -EAGAIN; }" >> >> So pulling up is a bit of a circle now. But still functional if people >> prefer it as it was. Just need to keep track of the history so we avoid >> going around again. :-) >> > > I don't think you need to drop that patch necessarily, but perhaps keep > the conversion part to a separate helper and then introduce a similar > patch for other formats..? In any event, that was just a followup > thought so it might be reasonable to defer it to the next version and > for now just see what it looks like to rely on the format states. > >> >>> } else if (xfs_bmap_one_block(...)) { >> Mmm.... cant jump into the leaf logic right away. We need to handle >> releasing leaf_bp which looks like it got lost in the pseudo code? IIRC >> there is a race condition with the AIL that is resolved by holding the leaf >> buffer across the transaction roll. So we need to check for that and >> release it upon reentry. >> > > That sounds more like an implementation detail. leaf_bp is set when we > convert from shortform, so isn't this the next state either way? Sort of, I think when we convert out of shortform, and we dont fit in a block, we skip over the leaf logic. But the leaf_bp release should happen either way, so it cant belong to either state. It is an implementation detail, it just sort of breaks up the if/else logic in this example. So I just meant to point it out so that there isnt confusion as to why the next patch may not look quite like the pseudo code. Or unless I'm overlooking something? > >> >>> error = xfs_attr_leaf_try_add(args, *leaf_bp); >>> if (error == -ENOSPC) { >>> leaf_to_node(...); >>> return -EAGAIN; >>> } >> Ok, so lift ENOSPC handler from xfs_attr_leaf_try_add >> >>> >>> ... state stuff for leaf add ... >>> } else { >>> error = xfs_attr_node_addname(dac); >>> } >>> >>> Hm? Of course, something like that should be incorporated via >>> independent refactoring patches. >> I think what this becomes right now is: >> >> Drop patch 13 >> Add new patch xfs: Lift ENOSPC handler from xfs_attr_leaf_try_add >> >> ? >> > > Re: above, I don't think the patch needs to necessarily go away. Feel > free to table this for now, just something to think about... Ok, I may fiddle around with some other arrangements or table it if I dont see anything particuarly elegant > >>> >>>> } >>>> - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { >>>> - error = xfs_attr_leaf_addname(args); >>>> - if (error != -ENOSPC) >>>> - return error; >>>> +das_add_leaf: >>>> - /* >>>> - * Commit that transaction so that the node_addname() >>>> - * call can manage its own transactions. >>>> - */ >>>> - error = xfs_defer_finish(&args->trans); >>>> - if (error) >>>> - return error; >>>> + /* >>>> + * After a shortform to leaf conversion, we need to hold the leaf and >>>> + * cylce out the transaction. When we get back, we need to release >>>> + * the leaf. >>>> + */ >>>> + if (*leaf_bp != NULL) { >>>> + xfs_trans_brelse(args->trans, *leaf_bp); >>>> + *leaf_bp = NULL; >>>> + } >>>> - /* >>>> - * Commit the current trans (including the inode) and >>>> - * start a new one. >>>> - */ >>>> - error = xfs_trans_roll_inode(&args->trans, dp); >>>> - if (error) >>>> + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { >>>> + error = xfs_attr_leaf_try_add(args, *leaf_bp); >>>> + switch (error) { >>>> + case -ENOSPC: >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> + dac->dela_state = XFS_DAS_LEAF_TO_NODE; >>>> + return -EAGAIN; >>>> + case 0: >>>> + dac->dela_state = XFS_DAS_FOUND_LBLK; >>>> + return -EAGAIN; >>>> + default: >>>> return error; >>>> - >>>> + } >>>> +das_leaf: >>>> + error = xfs_attr_leaf_addname(dac); >>>> + if (error == -ENOSPC) { >>>> + dac->dela_state = XFS_DAS_LEAF_TO_NODE; >>>> + return -EAGAIN; >>>> + } >>>> + return error; >>>> } >>>> - >>>> - error = xfs_attr_node_addname(args); >>>> +das_node: >>>> + error = xfs_attr_node_addname(dac); >>>> return error; >>>> } >>>> @@ -716,28 +799,32 @@ xfs_attr_leaf_try_add( >>>> * >>>> * This leaf block cannot have a "remote" value, we only call this routine >>>> * if bmap_one_block() says there is only one block (ie: no remote blks). >>>> + * >>>> + * This routine is meant to function as a 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_leaf_addname( >>>> - struct xfs_da_args *args) >>>> + struct xfs_delattr_context *dac) >>>> { >>>> - int error, forkoff; >>>> - struct xfs_buf *bp = NULL; >>>> - struct xfs_inode *dp = args->dp; >>>> - >>>> - trace_xfs_attr_leaf_addname(args); >>>> - >>>> - error = xfs_attr_leaf_try_add(args, bp); >>>> - if (error) >>>> - return error; >>>> + struct xfs_da_args *args = dac->da_args; >>>> + struct xfs_buf *bp = NULL; >>>> + int error, forkoff; >>>> + struct xfs_inode *dp = args->dp; >>>> - /* >>>> - * Commit the transaction that added the attr name so that >>>> - * later routines can manage their own transactions. >>>> - */ >>>> - error = xfs_trans_roll_inode(&args->trans, dp); >>>> - if (error) >>>> - return error; >>>> + /* State machine switch */ >>>> + switch (dac->dela_state) { >>>> + case XFS_DAS_FLIP_LFLAG: >>>> + goto das_flip_flag; >>>> + case XFS_DAS_ALLOC_LEAF: >>>> + goto das_alloc_leaf; >>>> + case XFS_DAS_RM_LBLK: >>>> + goto das_rm_lblk; >>>> + default: >>>> + break; >>>> + } >>>> /* >>>> * If there was an out-of-line value, allocate the blocks we >>>> @@ -746,7 +833,28 @@ xfs_attr_leaf_addname( >>>> * maximum size of a transaction and/or hit a deadlock. >>>> */ >>>> if (args->rmtblkno > 0) { >>>> - error = xfs_attr_rmtval_set(args); >>>> + >>>> + /* Open coded xfs_attr_rmtval_set without trans handling */ >>>> + error = xfs_attr_rmtval_set_init(dac); >>>> + if (error) >>>> + return error; >>>> + >>>> + /* >>>> + * Roll through the "value", allocating blocks on disk as >>>> + * required. >>>> + */ >>>> +das_alloc_leaf: >>> >>> If we filter out the setup above, it seems like this state could be >>> reduced to check for ->blkcnt > 0. >> By filter out the set up, you mean to introduce the setup flag like you >> mentioned earlier? For example: >> if(flags & setup == 0){ >> setup(); >> flags |= setup; >> } >> >> To be clear, if we did that, we're talking about adding an "int setup_flags" >> to the dac. And then defineing a XFS_DAC_SETUP_* scheme. Like a >> XFS_DAC_SETUP_ADD_NAME, and an XFS_DAC_SETUP_NODE_REMVE_NAME and so on. >> Because we cant have multiple functions sharing the same set up flag if they >> ever call each others. >> >> Is that what you mean to imply? Otherwise I may not be clear on what you >> mean by filtering out the set up. >> > > Well we already have a flag scheme for the defer thing, right? Is there > any reason we couldn't define a similar DAC_DEFER_INIT and let each > operation (add/remove) use it as appropriate? You're right, I momentarily forgot I had added the flag memmber to hoist the defer finish out. Yes, I think using it here would be appropriate. :-) Note again that I'm just > trying to think about potential simplifications. If there are isolated > functions that can be made clearly/simply idempotent (like init/setup > helpers tend to be), then doing so might be helpful. I.e., if > xfs_attr_rmtval_set_init() could do something like: > > { > if (dax->flags & DAC_RMTVAL_SET_INIT) > return; > > dax->flags |= DAC_RMTVAL_SET_INIT; > ... > } This seems reasonable, sorry for the earlier confusion. On a side note though, a bit ago Dave had suggested we select a different acronym because "dac" is too easy to mix up with other existing schemes like dax ;-) I'm not super concerned about it right now because it's purely an asthetic thing, but eventuallly we probably should come up with something. :-) > > ... and that removes an execution state, then that might be a win. If it > requires more complexity than that, then perhaps the suggestion is not > worthwile. ;P I dont think it will be too complicated, and may be of use in the next phase of the series. I will move forward with an init flag scheme in v9 > >>> >>>> + while (dac->blkcnt > 0) { >>>> + error = xfs_attr_rmtval_set_blk(dac); >>>> + if (error) >>>> + return error; >>>> + >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> + dac->dela_state = XFS_DAS_ALLOC_LEAF; >>>> + return -EAGAIN; >>>> + } >>>> + >>>> + error = xfs_attr_rmtval_set_value(args); >>>> if (error) >>>> return error; >>>> } >>>> @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( >>>> error = xfs_attr3_leaf_flipflags(args); >>>> if (error) >>>> return error; >>>> - /* >>>> - * Commit the flag value change and start the next trans in >>>> - * series. >>>> - */ >>>> - error = xfs_trans_roll_inode(&args->trans, args->dp); >>>> - if (error) >>>> - return error; >>>> - >>>> + dac->dela_state = XFS_DAS_FLIP_LFLAG; >>>> + return -EAGAIN; >>>> +das_flip_flag: >>>> /* >>>> * Dismantle the "old" attribute/value pair by removing >>>> * a "remote" value (if it exists). >>>> */ >>>> xfs_attr_restore_rmt_blk(args); >>>> + xfs_attr_rmtval_invalidate(args); >>>> +das_rm_lblk: >>>> if (args->rmtblkno) { >>>> - error = xfs_attr_rmtval_remove(args); >>>> + error = __xfs_attr_rmtval_remove(args); >>>> + >>>> + if (error == -EAGAIN) { >>>> + dac->dela_state = XFS_DAS_RM_LBLK; >>>> + return -EAGAIN; >>>> + } >>>> + >>> >>> This whole function looks like it could use more refactoring to split >>> out the rename case. >> Hmm, how about we take every thing inside the "if (args->op_flags & >> XFS_DA_OP_RENAME) {" and put it in a xfs_attr_leaf_rename() helper? Then >> pull the helper into the calling function? >> > > That would help. I'm not sure what the optimal breakdown is off the top > of my head tbh. This function is kind of a logical beast between dealing > with remote blocks and rename and combinations of the two, so I'd have > to stare at that one some more. No worries, I think some versions just have to become exploratory efforts for that reason. In general the goal should be to try and > avoid jumping in the middle of logic branches as much as possible so > there's a nice and visible distinction between the state code and the > functional code. > > Brian Ok, I think the *_rename() helpers will help with the unnesting here. I'll put it together and see what folks think. Thank you thank you thank you, for all your time and pateince with this, I know its super complicated and I really appreciate all the reviews! Allison > >>> >>>> if (error) >>>> return error; >>>> } >>>> @@ -799,15 +910,11 @@ xfs_attr_leaf_addname( >>>> /* >>>> * If the result is small enough, shrink it all into the inode. >>>> */ >>>> - if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) { >>>> + forkoff = xfs_attr_shortform_allfit(bp, dp); >>>> + if (forkoff) >>>> error = xfs_attr3_leaf_to_shortform(bp, args, forkoff); >>>> - /* bp is gone due to xfs_da_shrink_inode */ >>>> - if (error) >>>> - return error; >>>> - error = xfs_defer_finish(&args->trans); >>>> - if (error) >>>> - return error; >>>> - } >>>> + >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> } else if (args->rmtblkno > 0) { >>>> /* >>>> @@ -967,16 +1074,23 @@ xfs_attr_node_hasname( >>>> * >>>> * "Remote" attribute values confuse the issue and atomic rename operations >>>> * add a whole extra layer of confusion on top of that. >>>> + * >>>> + * This routine is meant to function as a 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_addname( >>>> - struct xfs_da_args *args) >>>> + struct xfs_delattr_context *dac) >>>> { >>>> - struct xfs_da_state *state; >>>> - struct xfs_da_state_blk *blk; >>>> - struct xfs_inode *dp; >>>> - struct xfs_mount *mp; >>>> - int retval, error; >>>> + struct xfs_da_args *args = dac->da_args; >>>> + struct xfs_da_state *state = NULL; >>>> + struct xfs_da_state_blk *blk; >>>> + struct xfs_inode *dp; >>>> + struct xfs_mount *mp; >>>> + int retval = 0; >>>> + int error = 0; >>>> trace_xfs_attr_node_addname(args); >>>> @@ -985,7 +1099,21 @@ xfs_attr_node_addname( >>>> */ >>>> dp = args->dp; >>>> mp = dp->i_mount; >>>> -restart: >>>> + >>>> + /* State machine switch */ >>>> + switch (dac->dela_state) { >>>> + case XFS_DAS_FLIP_NFLAG: >>>> + goto das_flip_flag; >>>> + case XFS_DAS_FOUND_NBLK: >>>> + goto das_found_nblk; >>>> + case XFS_DAS_ALLOC_NODE: >>>> + goto das_alloc_node; >>>> + case XFS_DAS_RM_NBLK: >>>> + goto das_rm_nblk; >>>> + default: >>>> + break; >>>> + } >>>> + >>>> /* >>>> * Search to see if name already exists, and get back a pointer >>>> * to where it should go. >>>> @@ -1031,19 +1159,13 @@ xfs_attr_node_addname( >>>> error = xfs_attr3_leaf_to_node(args); >>>> if (error) >>>> goto out; >>>> - error = xfs_defer_finish(&args->trans); >>>> - if (error) >>>> - goto out; >>>> /* >>>> - * Commit the node conversion and start the next >>>> - * trans in the chain. >>>> + * Restart routine from the top. No need to set the >>>> + * state >>>> */ >>>> - error = xfs_trans_roll_inode(&args->trans, dp); >>>> - if (error) >>>> - goto out; >>>> - >>>> - goto restart; >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> + return -EAGAIN; >>>> } >>>> /* >>>> @@ -1055,9 +1177,7 @@ xfs_attr_node_addname( >>>> error = xfs_da3_split(state); >>>> if (error) >>>> goto out; >>>> - error = xfs_defer_finish(&args->trans); >>>> - if (error) >>>> - goto out; >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> } else { >>>> /* >>>> * Addition succeeded, update Btree hashvals. >>>> @@ -1072,13 +1192,9 @@ xfs_attr_node_addname( >>>> xfs_da_state_free(state); >>>> state = NULL; >>>> - /* >>>> - * Commit the leaf addition or btree split and start the next >>>> - * trans in the chain. >>>> - */ >>>> - error = xfs_trans_roll_inode(&args->trans, dp); >>>> - if (error) >>>> - goto out; >>>> + dac->dela_state = XFS_DAS_FOUND_NBLK; >>>> + return -EAGAIN; >>>> +das_found_nblk: >>> >>> Same deal here. Any time we have this return -EAGAIN followed by a label >>> pattern I think we're going to want to think about refactoring things >>> more first to avoid dumping it in the middle of some unnecessarily large >>> function. >> Ok, similar pattern here too then? Everything in "if (args->op_flags & >> XFS_DA_OP_RENAME) {" goes in a new xfs_attr_node_rename() helper? Then >> hoist upwards? >> >> Thanks for the reviewing!! >> Allison >> >>> >>> Brian >>> >>>> /* >>>> * If there was an out-of-line value, allocate the blocks we >>>> @@ -1087,7 +1203,27 @@ xfs_attr_node_addname( >>>> * maximum size of a transaction and/or hit a deadlock. >>>> */ >>>> if (args->rmtblkno > 0) { >>>> - error = xfs_attr_rmtval_set(args); >>>> + /* Open coded xfs_attr_rmtval_set without trans handling */ >>>> + error = xfs_attr_rmtval_set_init(dac); >>>> + if (error) >>>> + return error; >>>> + >>>> + /* >>>> + * Roll through the "value", allocating blocks on disk as >>>> + * required. >>>> + */ >>>> +das_alloc_node: >>>> + while (dac->blkcnt > 0) { >>>> + error = xfs_attr_rmtval_set_blk(dac); >>>> + if (error) >>>> + return error; >>>> + >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> + dac->dela_state = XFS_DAS_ALLOC_NODE; >>>> + return -EAGAIN; >>>> + } >>>> + >>>> + error = xfs_attr_rmtval_set_value(args); >>>> if (error) >>>> return error; >>>> } >>>> @@ -1110,18 +1246,26 @@ xfs_attr_node_addname( >>>> * Commit the flag value change and start the next trans in >>>> * series >>>> */ >>>> - error = xfs_trans_roll_inode(&args->trans, args->dp); >>>> - if (error) >>>> - goto out; >>>> - >>>> + dac->dela_state = XFS_DAS_FLIP_NFLAG; >>>> + return -EAGAIN; >>>> +das_flip_flag: >>>> /* >>>> * Dismantle the "old" attribute/value pair by removing >>>> * a "remote" value (if it exists). >>>> */ >>>> xfs_attr_restore_rmt_blk(args); >>>> + xfs_attr_rmtval_invalidate(args); >>>> + >>>> +das_rm_nblk: >>>> if (args->rmtblkno) { >>>> - error = xfs_attr_rmtval_remove(args); >>>> + error = __xfs_attr_rmtval_remove(args); >>>> + >>>> + if (error == -EAGAIN) { >>>> + dac->dela_state = XFS_DAS_RM_NBLK; >>>> + return -EAGAIN; >>>> + } >>>> + >>>> if (error) >>>> return error; >>>> } >>>> @@ -1139,7 +1283,6 @@ xfs_attr_node_addname( >>>> error = xfs_da3_node_lookup_int(state, &retval); >>>> if (error) >>>> goto out; >>>> - >>>> /* >>>> * Remove the name and update the hashvals in the tree. >>>> */ >>>> @@ -1147,7 +1290,6 @@ xfs_attr_node_addname( >>>> ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC); >>>> error = xfs_attr3_leaf_remove(blk->bp, args); >>>> xfs_da3_fixhashpath(state, &state->path); >>>> - >>>> /* >>>> * Check to see if the tree needs to be collapsed. >>>> */ >>>> @@ -1155,11 +1297,9 @@ xfs_attr_node_addname( >>>> error = xfs_da3_join(state); >>>> if (error) >>>> goto out; >>>> - error = xfs_defer_finish(&args->trans); >>>> - if (error) >>>> - goto out; >>>> - } >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> + } >>>> } else if (args->rmtblkno > 0) { >>>> /* >>>> * Added a "remote" value, just clear the incomplete flag. >>>> diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h >>>> index 0e8ae1a..67af9d1 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr.h >>>> +++ b/fs/xfs/libxfs/xfs_attr.h >>>> @@ -93,6 +93,16 @@ enum xfs_delattr_state { >>>> /* Zero is uninitalized */ >>>> XFS_DAS_RM_SHRINK = 1, /* We are shrinking the tree */ >>>> XFS_DAS_RMTVAL_REMOVE, /* We are removing remote value blocks */ >>>> + XFS_DAS_ADD_LEAF, /* We are adding a leaf attr */ >>>> + XFS_DAS_FOUND_LBLK, /* We found leaf blk for attr */ >>>> + XFS_DAS_LEAF_TO_NODE, /* Converted leaf to node */ >>>> + XFS_DAS_FOUND_NBLK, /* We found node blk for attr */ >>>> + XFS_DAS_ALLOC_LEAF, /* We are allocating leaf blocks */ >>>> + XFS_DAS_FLIP_LFLAG, /* Flipped leaf INCOMPLETE attr flag */ >>>> + XFS_DAS_RM_LBLK, /* A rename is removing leaf blocks */ >>>> + XFS_DAS_ALLOC_NODE, /* We are allocating node blocks */ >>>> + XFS_DAS_FLIP_NFLAG, /* Flipped node INCOMPLETE attr flag */ >>>> + XFS_DAS_RM_NBLK, /* A rename is removing node blocks */ >>>> }; >>>> /* >>>> @@ -105,8 +115,13 @@ enum xfs_delattr_state { >>>> */ >>>> struct xfs_delattr_context { >>>> struct xfs_da_args *da_args; >>>> + struct xfs_bmbt_irec map; >>>> + struct xfs_buf *leaf_bp; >>>> + xfs_fileoff_t lfileoff; >>>> struct xfs_da_state *da_state; >>>> struct xfs_da_state_blk *blk; >>>> + xfs_dablk_t lblkno; >>>> + int blkcnt; >>>> unsigned int flags; >>>> enum xfs_delattr_state dela_state; >>>> }; >>>> @@ -126,6 +141,7 @@ int xfs_attr_get_ilocked(struct xfs_da_args *args); >>>> int xfs_attr_get(struct xfs_da_args *args); >>>> int xfs_attr_set(struct xfs_da_args *args); >>>> int xfs_attr_set_args(struct xfs_da_args *args); >>>> +int xfs_attr_set_iter(struct xfs_delattr_context *dac, struct xfs_buf **leaf_bp); >>>> 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); >>>> diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c >>>> index f55402b..4d15f45 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr_leaf.c >>>> +++ b/fs/xfs/libxfs/xfs_attr_leaf.c >>>> @@ -19,6 +19,7 @@ >>>> #include "xfs_bmap_btree.h" >>>> #include "xfs_bmap.h" >>>> #include "xfs_attr_sf.h" >>>> +#include "xfs_attr.h" >>>> #include "xfs_attr_remote.h" >>>> #include "xfs_attr.h" >>>> #include "xfs_attr_leaf.h" >>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c >>>> index fd4be9d..9607fd2 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr_remote.c >>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.c >>>> @@ -443,7 +443,7 @@ xfs_attr_rmtval_get( >>>> * Find a "hole" in the attribute address space large enough for us to drop the >>>> * new attribute's value into >>>> */ >>>> -STATIC int >>>> +int >>>> xfs_attr_rmt_find_hole( >>>> struct xfs_da_args *args) >>>> { >>>> @@ -470,7 +470,7 @@ xfs_attr_rmt_find_hole( >>>> return 0; >>>> } >>>> -STATIC int >>>> +int >>>> xfs_attr_rmtval_set_value( >>>> struct xfs_da_args *args) >>>> { >>>> @@ -630,6 +630,71 @@ xfs_attr_rmtval_set( >>>> } >>>> /* >>>> + * Find a hole for the attr and store it in the delayed attr context. This >>>> + * initializes the context to roll through allocating an attr extent for a >>>> + * delayed attr operation >>>> + */ >>>> +int >>>> +xfs_attr_rmtval_set_init( >>>> + struct xfs_delattr_context *dac) >>>> +{ >>>> + struct xfs_da_args *args = dac->da_args; >>>> + struct xfs_bmbt_irec *map = &dac->map; >>>> + int error; >>>> + >>>> + dac->lblkno = 0; >>>> + dac->lfileoff = 0; >>>> + dac->blkcnt = 0; >>>> + args->rmtblkcnt = 0; >>>> + args->rmtblkno = 0; >>>> + memset(map, 0, sizeof(struct xfs_bmbt_irec)); >>>> + >>>> + error = xfs_attr_rmt_find_hole(args); >>>> + if (error) >>>> + return error; >>>> + >>>> + dac->blkcnt = args->rmtblkcnt; >>>> + dac->lblkno = args->rmtblkno; >>>> + >>>> + return error; >>>> +} >>>> + >>>> +/* >>>> + * Write one block of the value associated with an attribute into the >>>> + * out-of-line buffer that we have defined for it. This is similar to a subset >>>> + * of xfs_attr_rmtval_set, but records the current block to the delayed attr >>>> + * context, and leaves transaction handling to the caller. >>>> + */ >>>> +int >>>> +xfs_attr_rmtval_set_blk( >>>> + struct xfs_delattr_context *dac) >>>> +{ >>>> + struct xfs_da_args *args = dac->da_args; >>>> + struct xfs_inode *dp = args->dp; >>>> + struct xfs_bmbt_irec *map = &dac->map; >>>> + int nmap; >>>> + int error; >>>> + >>>> + nmap = 1; >>>> + error = xfs_bmapi_write(args->trans, dp, >>>> + (xfs_fileoff_t)dac->lblkno, >>>> + dac->blkcnt, XFS_BMAPI_ATTRFORK, >>>> + args->total, map, &nmap); >>>> + if (error) >>>> + return error; >>>> + >>>> + ASSERT(nmap == 1); >>>> + ASSERT((map->br_startblock != DELAYSTARTBLOCK) && >>>> + (map->br_startblock != HOLESTARTBLOCK)); >>>> + >>>> + /* roll attribute extent map forwards */ >>>> + dac->lblkno += map->br_blockcount; >>>> + dac->blkcnt -= map->br_blockcount; >>>> + >>>> + return 0; >>>> +} >>>> + >>>> +/* >>>> * Remove the value associated with an attribute by deleting the >>>> * out-of-line buffer that it is stored on. >>>> */ >>>> @@ -671,48 +736,6 @@ xfs_attr_rmtval_invalidate( >>>> } >>>> /* >>>> - * Remove the value associated with an attribute by deleting the >>>> - * out-of-line buffer that it is stored on. >>>> - */ >>>> -int >>>> -xfs_attr_rmtval_remove( >>>> - struct xfs_da_args *args) >>>> -{ >>>> - xfs_dablk_t lblkno; >>>> - int blkcnt; >>>> - int error = 0; >>>> - int done = 0; >>>> - >>>> - trace_xfs_attr_rmtval_remove(args); >>>> - >>>> - error = xfs_attr_rmtval_invalidate(args); >>>> - if (error) >>>> - return error; >>>> - /* >>>> - * Keep de-allocating extents until the remote-value region is gone. >>>> - */ >>>> - lblkno = args->rmtblkno; >>>> - blkcnt = args->rmtblkcnt; >>>> - while (!done) { >>>> - error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt, >>>> - XFS_BMAPI_ATTRFORK, 1, &done); >>>> - if (error) >>>> - return error; >>>> - error = xfs_defer_finish(&args->trans); >>>> - if (error) >>>> - return error; >>>> - >>>> - /* >>>> - * Close out trans and start the next one in the chain. >>>> - */ >>>> - error = xfs_trans_roll_inode(&args->trans, args->dp); >>>> - if (error) >>>> - return error; >>>> - } >>>> - return 0; >>>> -} >>>> - >>>> -/* >>>> * Remove the value associated with an attribute by deleting the out-of-line >>>> * buffer that it is stored on. Returns EAGAIN for the caller to refresh the >>>> * transaction and recall the function >>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h >>>> index ee3337b..482dff9 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr_remote.h >>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.h >>>> @@ -15,4 +15,8 @@ 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_rmt_find_hole(struct xfs_da_args *args); >>>> +int xfs_attr_rmtval_set_value(struct xfs_da_args *args); >>>> +int xfs_attr_rmtval_set_blk(struct xfs_delattr_context *dac); >>>> +int xfs_attr_rmtval_set_init(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 c42f90e..3e8cec5 100644 >>>> --- a/fs/xfs/xfs_attr_inactive.c >>>> +++ b/fs/xfs/xfs_attr_inactive.c >>>> @@ -15,6 +15,7 @@ >>>> #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" >>>> diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h >>>> index a4323a6..26dc8bf 100644 >>>> --- a/fs/xfs/xfs_trace.h >>>> +++ b/fs/xfs/xfs_trace.h >>>> @@ -1784,7 +1784,6 @@ DEFINE_ATTR_EVENT(xfs_attr_refillstate); >>>> DEFINE_ATTR_EVENT(xfs_attr_rmtval_get); >>>> DEFINE_ATTR_EVENT(xfs_attr_rmtval_set); >>>> -DEFINE_ATTR_EVENT(xfs_attr_rmtval_remove); >>>> #define DEFINE_DA_EVENT(name) \ >>>> DEFINE_EVENT(xfs_da_class, name, \ >>>> -- >>>> 2.7.4 >>>> >>> >> >
On Thu, Apr 16, 2020 at 03:54:50PM -0700, Allison Collins wrote: > > > On 4/16/20 4:01 AM, Brian Foster wrote: > > On Wed, Apr 15, 2020 at 03:08:11PM -0700, Allison Collins wrote: > > > > > > > > > On 4/13/20 6:40 AM, Brian Foster wrote: > > > > On Fri, Apr 03, 2020 at 03:12:28PM -0700, Allison Collins wrote: > > > > > This patch modifies the attr set 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_set_args has become xfs_attr_set_iter, which uses a > > > > > state machine like switch to keep track of where it was when EAGAIN was > > > > > returned. > > > > > > > > > > Two new helper functions have been added: xfs_attr_rmtval_set_init and > > > > > xfs_attr_rmtval_set_blk. They provide a subset of logic similar to > > > > > xfs_attr_rmtval_set, but they store the current block in the delay attr > > > > > context to allow the caller to roll the transaction between allocations. > > > > > This helps to simplify and consolidate code used by > > > > > xfs_attr_leaf_addname and xfs_attr_node_addname. xfs_attr_set_args has > > > > > now become a simple loop to refresh the transaction until the operation > > > > > is completed. Lastly, xfs_attr_rmtval_remove is no longer used, and is > > > > > removed. > > > > > > > > > > Below is a state machine diagram for attr set 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_set_iter() > > > > > │ > > > > > v > > > > > need to upgrade > > > > > from sf to leaf? ──n─┐ > > > > > │ │ > > > > > y │ > > > > > │ │ > > > > > V │ > > > > > XFS_DAS_ADD_LEAF │ > > > > > │ │ > > > > > v │ > > > > > ┌──────n── fork has <──────┘ > > > > > │ only 1 blk? > > > > > │ │ > > > > > │ y > > > > > │ │ > > > > > │ v > > > > > │ xfs_attr_leaf_try_add() > > > > > │ │ > > > > > │ v > > > > > │ had enough > > > > > ├──────n── space? > > > > > │ │ > > > > > │ y > > > > > │ │ > > > > > │ v > > > > > │ XFS_DAS_FOUND_LBLK ──┐ > > > > > │ │ > > > > > │ XFS_DAS_FLIP_LFLAG ──┤ > > > > > │ (subroutine state) │ > > > > > │ │ > > > > > │ XFS_DAS_ALLOC_LEAF ──┤ > > > > > │ (subroutine state) │ > > > > > │ └─>xfs_attr_leaf_addname() > > > > > │ │ > > > > > │ v > > > > > │ ┌─────n── need to > > > > > │ │ alloc blks? > > > > > │ │ │ > > > > > │ │ y > > > > > │ │ │ > > > > > │ │ v > > > > > │ │ ┌─>XFS_DAS_ALLOC_LEAF > > > > > │ │ │ │ > > > > > │ │ │ v > > > > > │ │ └──y── need to alloc > > > > > │ │ more blocks? > > > > > │ │ │ > > > > > │ │ n > > > > > │ │ │ > > > > > │ │ v > > > > > │ │ was this > > > > > │ └────────> a rename? ──n─┐ > > > > > │ │ │ > > > > > │ y │ > > > > > │ │ │ > > > > > │ v │ > > > > > │ flip incomplete │ > > > > > │ flag │ > > > > > │ │ │ > > > > > │ v │ > > > > > │ XFS_DAS_FLIP_LFLAG │ > > > > > │ │ │ > > > > > │ v │ > > > > > │ remove │ > > > > > │ XFS_DAS_RM_LBLK ─> old name │ > > > > > │ ^ │ │ > > > > > │ │ v │ > > > > > │ └──────y── more to │ > > > > > │ remove │ > > > > > │ │ │ > > > > > │ n │ > > > > > │ │ │ > > > > > │ v │ > > > > > │ done <──────┘ > > > > > └────> XFS_DAS_LEAF_TO_NODE ─┐ > > > > > │ > > > > > XFS_DAS_FOUND_NBLK ──┤ > > > > > (subroutine state) │ > > > > > │ > > > > > XFS_DAS_ALLOC_NODE ──┤ > > > > > (subroutine state) │ > > > > > │ > > > > > XFS_DAS_FLIP_NFLAG ──┤ > > > > > (subroutine state) │ > > > > > │ > > > > > └─>xfs_attr_node_addname() > > > > > │ > > > > > v > > > > > find space to store > > > > > attr. Split if needed > > > > > │ > > > > > v > > > > > XFS_DAS_FOUND_NBLK > > > > > │ > > > > > v > > > > > ┌─────n── need to > > > > > │ alloc blks? > > > > > │ │ > > > > > │ y > > > > > │ │ > > > > > │ v > > > > > │ ┌─>XFS_DAS_ALLOC_NODE > > > > > │ │ │ > > > > > │ │ v > > > > > │ └──y── need to alloc > > > > > │ more blocks? > > > > > │ │ > > > > > │ n > > > > > │ │ > > > > > │ v > > > > > │ was this > > > > > └────────> a rename? ──n─┐ > > > > > │ │ > > > > > y │ > > > > > │ │ > > > > > v │ > > > > > flip incomplete │ > > > > > flag │ > > > > > │ │ > > > > > v │ > > > > > XFS_DAS_FLIP_NFLAG │ > > > > > │ │ > > > > > v │ > > > > > remove │ > > > > > XFS_DAS_RM_NBLK ─> old name │ > > > > > ^ │ │ > > > > > │ v │ > > > > > └──────y── more to │ > > > > > remove │ > > > > > │ │ > > > > > n │ > > > > > │ │ > > > > > v │ > > > > > done <──────┘ > > > > > > > > > > Signed-off-by: Allison Collins <allison.henderson@oracle.com> > > > > > --- > > > > > > > > Only a cursory pass given the previous feedback... > > > > > > > > > fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- > > > > > fs/xfs/libxfs/xfs_attr.h | 16 ++ > > > > > fs/xfs/libxfs/xfs_attr_leaf.c | 1 + > > > > > fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- > > > > > fs/xfs/libxfs/xfs_attr_remote.h | 4 + > > > > > fs/xfs/xfs_attr_inactive.c | 1 + > > > > > fs/xfs/xfs_trace.h | 1 - > > > > > 7 files changed, 351 insertions(+), 167 deletions(-) > > > > > > > > > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > > > > > index f700976..c160b7a 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr.c > > > > > +++ b/fs/xfs/libxfs/xfs_attr.c > > > > > @@ -44,7 +44,7 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args); > > > > > * Internal routines when attribute list is one block. > > > > > */ > > > > > STATIC int xfs_attr_leaf_get(xfs_da_args_t *args); > > > > > -STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args); > > > > > +STATIC int xfs_attr_leaf_addname(struct xfs_delattr_context *dac); > > > > > STATIC int xfs_attr_leaf_removename(struct xfs_delattr_context *dac); > > > > > STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > > > > > @@ -52,12 +52,13 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > > > > > * Internal routines when attribute list is more than one block. > > > > > */ > > > > > 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_addname(struct xfs_delattr_context *dac); > > > > > 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); > > > > > STATIC int xfs_attr_refillstate(xfs_da_state_t *state); > > > > > +STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp); > > > > > STATIC void > > > > > xfs_delattr_context_init( > > > > > @@ -227,8 +228,11 @@ xfs_attr_is_shortform( > > > > > /* > > > > > * Attempts to set an attr in shortform, or converts the tree to leaf form if > > > > > - * there is not enough room. If the attr is set, the transaction is committed > > > > > - * and set to NULL. > > > > > + * there is not enough room. This function is meant to operate as a helper > > > > > + * routine to the delayed attribute functions. It returns -EAGAIN to indicate > > > > > + * that the calling function should roll the transaction, and then proceed to > > > > > + * add the attr in leaf form. This subroutine does not expect to be recalled > > > > > + * again like the other delayed attr routines do. > > > > > */ > > > > > STATIC int > > > > > xfs_attr_set_shortform( > > > > > @@ -236,16 +240,16 @@ xfs_attr_set_shortform( > > > > > struct xfs_buf **leaf_bp) > > > > > { > > > > > struct xfs_inode *dp = args->dp; > > > > > - int error, error2 = 0; > > > > > + int error = 0; > > > > > /* > > > > > * Try to add the attr to the attribute list in the inode. > > > > > */ > > > > > error = xfs_attr_try_sf_addname(dp, args); > > > > > + > > > > > + /* Should only be 0, -EEXIST or ENOSPC */ > > > > > if (error != -ENOSPC) { > > > > > - error2 = xfs_trans_commit(args->trans); > > > > > - args->trans = NULL; > > > > > - return error ? error : error2; > > > > > + return error; > > > > > } > > > > > /* > > > > > * It won't fit in the shortform, transform to a leaf block. GROT: > > > > > @@ -258,18 +262,10 @@ xfs_attr_set_shortform( > > > > > /* > > > > > * Prevent the leaf buffer from being unlocked so that a concurrent AIL > > > > > * push cannot grab the half-baked leaf buffer and run into problems > > > > > - * with the write verifier. Once we're done rolling the transaction we > > > > > - * can release the hold and add the attr to the leaf. > > > > > + * with the write verifier. > > > > > */ > > > > > xfs_trans_bhold(args->trans, *leaf_bp); > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - xfs_trans_bhold_release(args->trans, *leaf_bp); > > > > > - if (error) { > > > > > - xfs_trans_brelse(args->trans, *leaf_bp); > > > > > - return error; > > > > > - } > > > > > - > > > > > - return 0; > > > > > + return -EAGAIN; > > > > > } > > > > > /* > > > > > @@ -279,9 +275,83 @@ int > > > > > xfs_attr_set_args( > > > > > struct xfs_da_args *args) > > > > > { > > > > > - struct xfs_inode *dp = args->dp; > > > > > - struct xfs_buf *leaf_bp = NULL; > > > > > - int error = 0; > > > > > + struct xfs_buf *leaf_bp = NULL; > > > > > + int error = 0; > > > > > + struct xfs_delattr_context dac; > > > > > + > > > > > + xfs_delattr_context_init(&dac, args); > > > > > + > > > > > + do { > > > > > + error = xfs_attr_set_iter(&dac, &leaf_bp); > > > > > + if (error != -EAGAIN) > > > > > + break; > > > > > + > > > > > + if (dac.flags & XFS_DAC_DEFER_FINISH) { > > > > > + dac.flags &= ~XFS_DAC_DEFER_FINISH; > > > > > + error = xfs_defer_finish(&args->trans); > > > > > + if (error) > > > > > + break; > > > > > + } > > > > > + > > > > > + error = xfs_trans_roll_inode(&args->trans, args->dp); > > > > > + if (error) > > > > > + break; > > > > > + > > > > > + if (leaf_bp) { > > > > > + xfs_trans_bjoin(args->trans, leaf_bp); > > > > > + xfs_trans_bhold(args->trans, leaf_bp); > > > > > + } > > > > > + > > > > > + } while (true); > > > > > + > > > > > + return error; > > > > > +} > > > > > + > > > > > +/* > > > > > + * Set the attribute specified in @args. > > > > > + * This routine is meant to function as a 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. > > > > > + */ > > > > > +int > > > > > +xfs_attr_set_iter( > > > > > + struct xfs_delattr_context *dac, > > > > > + struct xfs_buf **leaf_bp) > > > > > +{ > > > > > + struct xfs_da_args *args = dac->da_args; > > > > > + struct xfs_inode *dp = args->dp; > > > > > + int error = 0; > > > > > + int sf_size; > > > > > + > > > > > + /* State machine switch */ > > > > > + switch (dac->dela_state) { > > > > > + case XFS_DAS_ADD_LEAF: > > > > > + goto das_add_leaf; > > > > > + case XFS_DAS_ALLOC_LEAF: > > > > > + case XFS_DAS_FLIP_LFLAG: > > > > > + case XFS_DAS_FOUND_LBLK: > > > > > + goto das_leaf; > > > > > + case XFS_DAS_FOUND_NBLK: > > > > > + case XFS_DAS_FLIP_NFLAG: > > > > > + case XFS_DAS_ALLOC_NODE: > > > > > + case XFS_DAS_LEAF_TO_NODE: > > > > > + goto das_node; > > > > > + default: > > > > > + break; > > > > > + } > > > > > + > > > > > + /* > > > > > + * New inodes may not have an attribute fork yet. So set the attribute > > > > > + * fork appropriately > > > > > + */ > > > > > + if (XFS_IFORK_Q((args->dp)) == 0) { > > > > > + sf_size = sizeof(struct xfs_attr_sf_hdr) + > > > > > + XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen); > > > > > + xfs_bmap_set_attrforkoff(args->dp, sf_size, NULL); > > > > > + args->dp->i_afp = kmem_zone_zalloc(xfs_ifork_zone, 0); > > > > > + args->dp->i_afp->if_flags = XFS_IFEXTENTS; > > > > > + } > > > > > > > > Is this hunk moved from somewhere? If so, we should probably handle that > > > > in a separate patch. I think we really want these last couple of patches > > > > to introduce the state/markers and not much else. > > > Oh, this hunk is new, not moved, I believe it's been here since I picked up > > > the series quite a while ago. It actually has more to do with parent > > > pointers than delayed atts. Basically when we try to add the parent pointer > > > during a create, the inode isnt fully constructed yet, so we have add the > > > fork here. I will put this in a separate patch and move it further up the > > > set. > > > > > > > Ok. > > > > > > > > > > > /* > > > > > * If the attribute list is already in leaf format, jump straight to > > > > > @@ -292,40 +362,53 @@ xfs_attr_set_args( > > > > > if (xfs_attr_is_shortform(dp)) { > > > > > /* > > > > > - * If the attr was successfully set in shortform, the > > > > > - * transaction is committed and set to NULL. Otherwise, is it > > > > > - * converted from shortform to leaf, and the transaction is > > > > > - * retained. > > > > > + * If the attr was successfully set in shortform, no need to > > > > > + * continue. Otherwise, is it converted from shortform to leaf > > > > > + * and -EAGAIN is returned. > > > > > */ > > > > > - error = xfs_attr_set_shortform(args, &leaf_bp); > > > > > - if (error || !args->trans) > > > > > - return error; > > > > > + error = xfs_attr_set_shortform(args, leaf_bp); > > > > > + if (error == -EAGAIN) { > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > + dac->dela_state = XFS_DAS_ADD_LEAF; > > > > > + } > > > > > + return error; > > > > > > > > Similar to the previous patch, I wonder if we need the explicit states > > > > that are otherwise handled by existing inode state. For example, if the > > > > above returns -EAGAIN, xfs_attr_is_shortform() is no longer true on > > > > reentry, right? If that's the case for the other conversions, it seems > > > > like we might only need one state (XFS_DAS_FOUND_LBLK) for this > > > > function. > > > Ok, it looks like I can get away with out XFS_DAS_ADD_LEAF and > > > XFS_DAS_LEAF_TO_NODE. > > > > > > This is actually a little similar to how it was done when I was trying to > > > use flags instead of the state enums a long time ago. At the time, I got > > > the impression people were concerned about the complexity and > > > maintainability. So we went back to the enums because the explicite jumping > > > made it clear where the re-entry was to resume, and reduced the risk that a > > > person may mistakenly introduce a change that disrupts the reentry flow. > > > Though perhaps the current arrangement of refactoring has made that less > > > concerning? > > > > > > > Perhaps. I think it just depends on context. It might not make sense to > > try and implement such a reentry model down in the node code where we > > have N different non-deterministic states to deal with; as opposed to > > here where we're looking at one of a handful of high level formats and > > the behavior is very predictable: try to add to the current format, else > > convert to the next. > Alrighty then, that sounds reasonable. I try my best to recall history of > the decisions just as a sort of guide of where we've been and where to go > next. > > > > > > > > > > > BTW, that general approach might be more clear if we lifted the format > > > > conversions into this level from down in the format specific add > > > > handlers. The goal would be to make the high level flow look something > > > > like: > > > > > > > > if (shortform) { > > > > error = sf_try_add(); > > > > if (error == -ENOSPC) { > > > > shortform_to_leaf(...); > > > > ... > > > > return -EAGAIN; > > > > } > > > Well, actually this piece was hoisted out into the helper function in patch > > > 13. So pulling this up is sort of like letting go of patch 13. :-) > > > > > > The history is: initially I had tried to reduce indentation here in v6 by > > > taking advantage of the _add_leaf label. Because I think in v5 we were > > > trying to unnest where some of the jump points were. So in v6 I had: > > > "if(!shortform) goto leaf;" Which un-nests this shortform code. > > > > > > Then in the v7 review I think Dave suggested I should add the helper and > > > invert the check. So now we have "if(shortform()){ call_helper(); handle > > > -EAGAIN; }" > > > > > > So pulling up is a bit of a circle now. But still functional if people > > > prefer it as it was. Just need to keep track of the history so we avoid > > > going around again. :-) > > > > > > > I don't think you need to drop that patch necessarily, but perhaps keep > > the conversion part to a separate helper and then introduce a similar > > patch for other formats..? In any event, that was just a followup > > thought so it might be reasonable to defer it to the next version and > > for now just see what it looks like to rely on the format states. > > > > > > > > > } else if (xfs_bmap_one_block(...)) { > > > Mmm.... cant jump into the leaf logic right away. We need to handle > > > releasing leaf_bp which looks like it got lost in the pseudo code? IIRC > > > there is a race condition with the AIL that is resolved by holding the leaf > > > buffer across the transaction roll. So we need to check for that and > > > release it upon reentry. > > > > > > > That sounds more like an implementation detail. leaf_bp is set when we > > convert from shortform, so isn't this the next state either way? > Sort of, I think when we convert out of shortform, and we dont fit in a > block, we skip over the leaf logic. But the leaf_bp release should happen > either way, so it cant belong to either state. > > It is an implementation detail, it just sort of breaks up the if/else logic > in this example. So I just meant to point it out so that there isnt > confusion as to why the next patch may not look quite like the pseudo code. > Or unless I'm overlooking something? > Ok. Perhaps if it's always true that leaf_bp is either NULL or needs to be released, we could encapsulate that somewhere like at the top of the function or maybe even in the caller. The latter might be more clear given the use case. IIRC the purpose of that little trick is simply to prevent writeback from grabbing hold of an empty leaf between transactions. It doesn't look like that leaf_bp reference is actually used for anything else even though the pointer is passed to leaf_try_add(), but I could be missing something. > > > > > > > > > error = xfs_attr_leaf_try_add(args, *leaf_bp); > > > > if (error == -ENOSPC) { > > > > leaf_to_node(...); > > > > return -EAGAIN; > > > > } > > > Ok, so lift ENOSPC handler from xfs_attr_leaf_try_add > > > > > > > > > > > ... state stuff for leaf add ... > > > > } else { > > > > error = xfs_attr_node_addname(dac); > > > > } > > > > > > > > Hm? Of course, something like that should be incorporated via > > > > independent refactoring patches. > > > I think what this becomes right now is: > > > > > > Drop patch 13 > > > Add new patch xfs: Lift ENOSPC handler from xfs_attr_leaf_try_add > > > > > > ? > > > > > > > Re: above, I don't think the patch needs to necessarily go away. Feel > > free to table this for now, just something to think about... > Ok, I may fiddle around with some other arrangements or table it if I dont > see anything particuarly elegant > > > > > > > > > > > > } > > > > > - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > > > > > - error = xfs_attr_leaf_addname(args); > > > > > - if (error != -ENOSPC) > > > > > - return error; > > > > > +das_add_leaf: > > > > > - /* > > > > > - * Commit that transaction so that the node_addname() > > > > > - * call can manage its own transactions. > > > > > - */ > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - if (error) > > > > > - return error; > > > > > + /* > > > > > + * After a shortform to leaf conversion, we need to hold the leaf and > > > > > + * cylce out the transaction. When we get back, we need to release > > > > > + * the leaf. > > > > > + */ > > > > > + if (*leaf_bp != NULL) { > > > > > + xfs_trans_brelse(args->trans, *leaf_bp); > > > > > + *leaf_bp = NULL; > > > > > + } > > > > > - /* > > > > > - * Commit the current trans (including the inode) and > > > > > - * start a new one. > > > > > - */ > > > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > > > - if (error) > > > > > + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > > > > > + error = xfs_attr_leaf_try_add(args, *leaf_bp); > > > > > + switch (error) { > > > > > + case -ENOSPC: > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > > > > > + return -EAGAIN; > > > > > + case 0: > > > > > + dac->dela_state = XFS_DAS_FOUND_LBLK; > > > > > + return -EAGAIN; > > > > > + default: > > > > > return error; > > > > > - > > > > > + } > > > > > +das_leaf: > > > > > + error = xfs_attr_leaf_addname(dac); > > > > > + if (error == -ENOSPC) { > > > > > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > > > > > + return -EAGAIN; > > > > > + } > > > > > + return error; > > > > > } > > > > > - > > > > > - error = xfs_attr_node_addname(args); > > > > > +das_node: > > > > > + error = xfs_attr_node_addname(dac); > > > > > return error; > > > > > } > > > > > @@ -716,28 +799,32 @@ xfs_attr_leaf_try_add( > > > > > * > > > > > * This leaf block cannot have a "remote" value, we only call this routine > > > > > * if bmap_one_block() says there is only one block (ie: no remote blks). > > > > > + * > > > > > + * This routine is meant to function as a 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_leaf_addname( > > > > > - struct xfs_da_args *args) > > > > > + struct xfs_delattr_context *dac) > > > > > { > > > > > - int error, forkoff; > > > > > - struct xfs_buf *bp = NULL; > > > > > - struct xfs_inode *dp = args->dp; > > > > > - > > > > > - trace_xfs_attr_leaf_addname(args); > > > > > - > > > > > - error = xfs_attr_leaf_try_add(args, bp); > > > > > - if (error) > > > > > - return error; > > > > > + struct xfs_da_args *args = dac->da_args; > > > > > + struct xfs_buf *bp = NULL; > > > > > + int error, forkoff; > > > > > + struct xfs_inode *dp = args->dp; > > > > > - /* > > > > > - * Commit the transaction that added the attr name so that > > > > > - * later routines can manage their own transactions. > > > > > - */ > > > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > > > - if (error) > > > > > - return error; > > > > > + /* State machine switch */ > > > > > + switch (dac->dela_state) { > > > > > + case XFS_DAS_FLIP_LFLAG: > > > > > + goto das_flip_flag; > > > > > + case XFS_DAS_ALLOC_LEAF: > > > > > + goto das_alloc_leaf; > > > > > + case XFS_DAS_RM_LBLK: > > > > > + goto das_rm_lblk; > > > > > + default: > > > > > + break; > > > > > + } > > > > > /* > > > > > * If there was an out-of-line value, allocate the blocks we > > > > > @@ -746,7 +833,28 @@ xfs_attr_leaf_addname( > > > > > * maximum size of a transaction and/or hit a deadlock. > > > > > */ > > > > > if (args->rmtblkno > 0) { > > > > > - error = xfs_attr_rmtval_set(args); > > > > > + > > > > > + /* Open coded xfs_attr_rmtval_set without trans handling */ > > > > > + error = xfs_attr_rmtval_set_init(dac); > > > > > + if (error) > > > > > + return error; > > > > > + > > > > > + /* > > > > > + * Roll through the "value", allocating blocks on disk as > > > > > + * required. > > > > > + */ > > > > > +das_alloc_leaf: > > > > > > > > If we filter out the setup above, it seems like this state could be > > > > reduced to check for ->blkcnt > 0. > > > By filter out the set up, you mean to introduce the setup flag like you > > > mentioned earlier? For example: > > > if(flags & setup == 0){ > > > setup(); > > > flags |= setup; > > > } > > > > > > To be clear, if we did that, we're talking about adding an "int setup_flags" > > > to the dac. And then defineing a XFS_DAC_SETUP_* scheme. Like a > > > XFS_DAC_SETUP_ADD_NAME, and an XFS_DAC_SETUP_NODE_REMVE_NAME and so on. > > > Because we cant have multiple functions sharing the same set up flag if they > > > ever call each others. > > > > > > Is that what you mean to imply? Otherwise I may not be clear on what you > > > mean by filtering out the set up. > > > > > > > Well we already have a flag scheme for the defer thing, right? Is there > > any reason we couldn't define a similar DAC_DEFER_INIT and let each > > operation (add/remove) use it as appropriate? > You're right, I momentarily forgot I had added the flag memmber to hoist the > defer finish out. Yes, I think using it here would be appropriate. :-) > > Note again that I'm just > > trying to think about potential simplifications. If there are isolated > > functions that can be made clearly/simply idempotent (like init/setup > > helpers tend to be), then doing so might be helpful. I.e., if > > xfs_attr_rmtval_set_init() could do something like: > > > > { > > if (dax->flags & DAC_RMTVAL_SET_INIT) > > return; > > > > dax->flags |= DAC_RMTVAL_SET_INIT; > > ... > > } > This seems reasonable, sorry for the earlier confusion. On a side note > though, a bit ago Dave had suggested we select a different acronym because > "dac" is too easy to mix up with other existing schemes like dax ;-) I'm > not super concerned about it right now because it's purely an asthetic > thing, but eventuallly we probably should come up with something. :-) > I agree. Besides the confusion with DAX, it's not very obvious what DAC is supposed to mean. Maybe something that reuses the XFS_ATTR_ prefix is better. Or if not, XATTR_STATE, XATTROP_ (with or without a leading XFS_) might also be options..? I don't have a strong preference atm. > > > > ... and that removes an execution state, then that might be a win. If it > > requires more complexity than that, then perhaps the suggestion is not > > worthwile. ;P > I dont think it will be too complicated, and may be of use in the next phase > of the series. I will move forward with an init flag scheme in v9 > > > > > > > > > > > > + while (dac->blkcnt > 0) { > > > > > + error = xfs_attr_rmtval_set_blk(dac); > > > > > + if (error) > > > > > + return error; > > > > > + > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > + dac->dela_state = XFS_DAS_ALLOC_LEAF; > > > > > + return -EAGAIN; > > > > > + } > > > > > + > > > > > + error = xfs_attr_rmtval_set_value(args); > > > > > if (error) > > > > > return error; > > > > > } > > > > > @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( > > > > > error = xfs_attr3_leaf_flipflags(args); > > > > > if (error) > > > > > return error; > > > > > - /* > > > > > - * Commit the flag value change and start the next trans in > > > > > - * series. > > > > > - */ > > > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > > > - if (error) > > > > > - return error; > > > > > - > > > > > + dac->dela_state = XFS_DAS_FLIP_LFLAG; > > > > > + return -EAGAIN; > > > > > +das_flip_flag: > > > > > /* > > > > > * Dismantle the "old" attribute/value pair by removing > > > > > * a "remote" value (if it exists). > > > > > */ > > > > > xfs_attr_restore_rmt_blk(args); > > > > > + xfs_attr_rmtval_invalidate(args); > > > > > +das_rm_lblk: > > > > > if (args->rmtblkno) { > > > > > - error = xfs_attr_rmtval_remove(args); > > > > > + error = __xfs_attr_rmtval_remove(args); > > > > > + > > > > > + if (error == -EAGAIN) { > > > > > + dac->dela_state = XFS_DAS_RM_LBLK; > > > > > + return -EAGAIN; > > > > > + } > > > > > + > > > > > > > > This whole function looks like it could use more refactoring to split > > > > out the rename case. > > > Hmm, how about we take every thing inside the "if (args->op_flags & > > > XFS_DA_OP_RENAME) {" and put it in a xfs_attr_leaf_rename() helper? Then > > > pull the helper into the calling function? > > > > > > > That would help. I'm not sure what the optimal breakdown is off the top > > of my head tbh. This function is kind of a logical beast between dealing > > with remote blocks and rename and combinations of the two, so I'd have > > to stare at that one some more. > No worries, I think some versions just have to become exploratory efforts > for that reason. > Unfortunately. :P > In general the goal should be to try and > > avoid jumping in the middle of logic branches as much as possible so > > there's a nice and visible distinction between the state code and the > > functional code. > > > > Brian > > Ok, I think the *_rename() helpers will help with the unnesting here. I'll > put it together and see what folks think. > > Thank you thank you thank you, for all your time and pateince with this, I > know its super complicated and I really appreciate all the reviews! > No problem and likewise. I know it can be painful at times to release 42 versions of a patchset. :) I'm hoping we can start to get some of the refactoring patches merged to shrink this back down to the core functional bits that are still under development. Brian > Allison > > > > > > > > > > > if (error) > > > > > return error; > > > > > } > > > > > @@ -799,15 +910,11 @@ xfs_attr_leaf_addname( > > > > > /* > > > > > * If the result is small enough, shrink it all into the inode. > > > > > */ > > > > > - if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) { > > > > > + forkoff = xfs_attr_shortform_allfit(bp, dp); > > > > > + if (forkoff) > > > > > error = xfs_attr3_leaf_to_shortform(bp, args, forkoff); > > > > > - /* bp is gone due to xfs_da_shrink_inode */ > > > > > - if (error) > > > > > - return error; > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - if (error) > > > > > - return error; > > > > > - } > > > > > + > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > } else if (args->rmtblkno > 0) { > > > > > /* > > > > > @@ -967,16 +1074,23 @@ xfs_attr_node_hasname( > > > > > * > > > > > * "Remote" attribute values confuse the issue and atomic rename operations > > > > > * add a whole extra layer of confusion on top of that. > > > > > + * > > > > > + * This routine is meant to function as a 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_addname( > > > > > - struct xfs_da_args *args) > > > > > + struct xfs_delattr_context *dac) > > > > > { > > > > > - struct xfs_da_state *state; > > > > > - struct xfs_da_state_blk *blk; > > > > > - struct xfs_inode *dp; > > > > > - struct xfs_mount *mp; > > > > > - int retval, error; > > > > > + struct xfs_da_args *args = dac->da_args; > > > > > + struct xfs_da_state *state = NULL; > > > > > + struct xfs_da_state_blk *blk; > > > > > + struct xfs_inode *dp; > > > > > + struct xfs_mount *mp; > > > > > + int retval = 0; > > > > > + int error = 0; > > > > > trace_xfs_attr_node_addname(args); > > > > > @@ -985,7 +1099,21 @@ xfs_attr_node_addname( > > > > > */ > > > > > dp = args->dp; > > > > > mp = dp->i_mount; > > > > > -restart: > > > > > + > > > > > + /* State machine switch */ > > > > > + switch (dac->dela_state) { > > > > > + case XFS_DAS_FLIP_NFLAG: > > > > > + goto das_flip_flag; > > > > > + case XFS_DAS_FOUND_NBLK: > > > > > + goto das_found_nblk; > > > > > + case XFS_DAS_ALLOC_NODE: > > > > > + goto das_alloc_node; > > > > > + case XFS_DAS_RM_NBLK: > > > > > + goto das_rm_nblk; > > > > > + default: > > > > > + break; > > > > > + } > > > > > + > > > > > /* > > > > > * Search to see if name already exists, and get back a pointer > > > > > * to where it should go. > > > > > @@ -1031,19 +1159,13 @@ xfs_attr_node_addname( > > > > > error = xfs_attr3_leaf_to_node(args); > > > > > if (error) > > > > > goto out; > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - if (error) > > > > > - goto out; > > > > > /* > > > > > - * Commit the node conversion and start the next > > > > > - * trans in the chain. > > > > > + * Restart routine from the top. No need to set the > > > > > + * state > > > > > */ > > > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > > > - if (error) > > > > > - goto out; > > > > > - > > > > > - goto restart; > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > + return -EAGAIN; > > > > > } > > > > > /* > > > > > @@ -1055,9 +1177,7 @@ xfs_attr_node_addname( > > > > > error = xfs_da3_split(state); > > > > > if (error) > > > > > goto out; > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - if (error) > > > > > - goto out; > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > } else { > > > > > /* > > > > > * Addition succeeded, update Btree hashvals. > > > > > @@ -1072,13 +1192,9 @@ xfs_attr_node_addname( > > > > > xfs_da_state_free(state); > > > > > state = NULL; > > > > > - /* > > > > > - * Commit the leaf addition or btree split and start the next > > > > > - * trans in the chain. > > > > > - */ > > > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > > > - if (error) > > > > > - goto out; > > > > > + dac->dela_state = XFS_DAS_FOUND_NBLK; > > > > > + return -EAGAIN; > > > > > +das_found_nblk: > > > > > > > > Same deal here. Any time we have this return -EAGAIN followed by a label > > > > pattern I think we're going to want to think about refactoring things > > > > more first to avoid dumping it in the middle of some unnecessarily large > > > > function. > > > Ok, similar pattern here too then? Everything in "if (args->op_flags & > > > XFS_DA_OP_RENAME) {" goes in a new xfs_attr_node_rename() helper? Then > > > hoist upwards? > > > > > > Thanks for the reviewing!! > > > Allison > > > > > > > > > > > Brian > > > > > > > > > /* > > > > > * If there was an out-of-line value, allocate the blocks we > > > > > @@ -1087,7 +1203,27 @@ xfs_attr_node_addname( > > > > > * maximum size of a transaction and/or hit a deadlock. > > > > > */ > > > > > if (args->rmtblkno > 0) { > > > > > - error = xfs_attr_rmtval_set(args); > > > > > + /* Open coded xfs_attr_rmtval_set without trans handling */ > > > > > + error = xfs_attr_rmtval_set_init(dac); > > > > > + if (error) > > > > > + return error; > > > > > + > > > > > + /* > > > > > + * Roll through the "value", allocating blocks on disk as > > > > > + * required. > > > > > + */ > > > > > +das_alloc_node: > > > > > + while (dac->blkcnt > 0) { > > > > > + error = xfs_attr_rmtval_set_blk(dac); > > > > > + if (error) > > > > > + return error; > > > > > + > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > + dac->dela_state = XFS_DAS_ALLOC_NODE; > > > > > + return -EAGAIN; > > > > > + } > > > > > + > > > > > + error = xfs_attr_rmtval_set_value(args); > > > > > if (error) > > > > > return error; > > > > > } > > > > > @@ -1110,18 +1246,26 @@ xfs_attr_node_addname( > > > > > * Commit the flag value change and start the next trans in > > > > > * series > > > > > */ > > > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > > > - if (error) > > > > > - goto out; > > > > > - > > > > > + dac->dela_state = XFS_DAS_FLIP_NFLAG; > > > > > + return -EAGAIN; > > > > > +das_flip_flag: > > > > > /* > > > > > * Dismantle the "old" attribute/value pair by removing > > > > > * a "remote" value (if it exists). > > > > > */ > > > > > xfs_attr_restore_rmt_blk(args); > > > > > + xfs_attr_rmtval_invalidate(args); > > > > > + > > > > > +das_rm_nblk: > > > > > if (args->rmtblkno) { > > > > > - error = xfs_attr_rmtval_remove(args); > > > > > + error = __xfs_attr_rmtval_remove(args); > > > > > + > > > > > + if (error == -EAGAIN) { > > > > > + dac->dela_state = XFS_DAS_RM_NBLK; > > > > > + return -EAGAIN; > > > > > + } > > > > > + > > > > > if (error) > > > > > return error; > > > > > } > > > > > @@ -1139,7 +1283,6 @@ xfs_attr_node_addname( > > > > > error = xfs_da3_node_lookup_int(state, &retval); > > > > > if (error) > > > > > goto out; > > > > > - > > > > > /* > > > > > * Remove the name and update the hashvals in the tree. > > > > > */ > > > > > @@ -1147,7 +1290,6 @@ xfs_attr_node_addname( > > > > > ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC); > > > > > error = xfs_attr3_leaf_remove(blk->bp, args); > > > > > xfs_da3_fixhashpath(state, &state->path); > > > > > - > > > > > /* > > > > > * Check to see if the tree needs to be collapsed. > > > > > */ > > > > > @@ -1155,11 +1297,9 @@ xfs_attr_node_addname( > > > > > error = xfs_da3_join(state); > > > > > if (error) > > > > > goto out; > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - if (error) > > > > > - goto out; > > > > > - } > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > + } > > > > > } else if (args->rmtblkno > 0) { > > > > > /* > > > > > * Added a "remote" value, just clear the incomplete flag. > > > > > diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h > > > > > index 0e8ae1a..67af9d1 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr.h > > > > > +++ b/fs/xfs/libxfs/xfs_attr.h > > > > > @@ -93,6 +93,16 @@ enum xfs_delattr_state { > > > > > /* Zero is uninitalized */ > > > > > XFS_DAS_RM_SHRINK = 1, /* We are shrinking the tree */ > > > > > XFS_DAS_RMTVAL_REMOVE, /* We are removing remote value blocks */ > > > > > + XFS_DAS_ADD_LEAF, /* We are adding a leaf attr */ > > > > > + XFS_DAS_FOUND_LBLK, /* We found leaf blk for attr */ > > > > > + XFS_DAS_LEAF_TO_NODE, /* Converted leaf to node */ > > > > > + XFS_DAS_FOUND_NBLK, /* We found node blk for attr */ > > > > > + XFS_DAS_ALLOC_LEAF, /* We are allocating leaf blocks */ > > > > > + XFS_DAS_FLIP_LFLAG, /* Flipped leaf INCOMPLETE attr flag */ > > > > > + XFS_DAS_RM_LBLK, /* A rename is removing leaf blocks */ > > > > > + XFS_DAS_ALLOC_NODE, /* We are allocating node blocks */ > > > > > + XFS_DAS_FLIP_NFLAG, /* Flipped node INCOMPLETE attr flag */ > > > > > + XFS_DAS_RM_NBLK, /* A rename is removing node blocks */ > > > > > }; > > > > > /* > > > > > @@ -105,8 +115,13 @@ enum xfs_delattr_state { > > > > > */ > > > > > struct xfs_delattr_context { > > > > > struct xfs_da_args *da_args; > > > > > + struct xfs_bmbt_irec map; > > > > > + struct xfs_buf *leaf_bp; > > > > > + xfs_fileoff_t lfileoff; > > > > > struct xfs_da_state *da_state; > > > > > struct xfs_da_state_blk *blk; > > > > > + xfs_dablk_t lblkno; > > > > > + int blkcnt; > > > > > unsigned int flags; > > > > > enum xfs_delattr_state dela_state; > > > > > }; > > > > > @@ -126,6 +141,7 @@ int xfs_attr_get_ilocked(struct xfs_da_args *args); > > > > > int xfs_attr_get(struct xfs_da_args *args); > > > > > int xfs_attr_set(struct xfs_da_args *args); > > > > > int xfs_attr_set_args(struct xfs_da_args *args); > > > > > +int xfs_attr_set_iter(struct xfs_delattr_context *dac, struct xfs_buf **leaf_bp); > > > > > 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); > > > > > diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c > > > > > index f55402b..4d15f45 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr_leaf.c > > > > > +++ b/fs/xfs/libxfs/xfs_attr_leaf.c > > > > > @@ -19,6 +19,7 @@ > > > > > #include "xfs_bmap_btree.h" > > > > > #include "xfs_bmap.h" > > > > > #include "xfs_attr_sf.h" > > > > > +#include "xfs_attr.h" > > > > > #include "xfs_attr_remote.h" > > > > > #include "xfs_attr.h" > > > > > #include "xfs_attr_leaf.h" > > > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c > > > > > index fd4be9d..9607fd2 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr_remote.c > > > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.c > > > > > @@ -443,7 +443,7 @@ xfs_attr_rmtval_get( > > > > > * Find a "hole" in the attribute address space large enough for us to drop the > > > > > * new attribute's value into > > > > > */ > > > > > -STATIC int > > > > > +int > > > > > xfs_attr_rmt_find_hole( > > > > > struct xfs_da_args *args) > > > > > { > > > > > @@ -470,7 +470,7 @@ xfs_attr_rmt_find_hole( > > > > > return 0; > > > > > } > > > > > -STATIC int > > > > > +int > > > > > xfs_attr_rmtval_set_value( > > > > > struct xfs_da_args *args) > > > > > { > > > > > @@ -630,6 +630,71 @@ xfs_attr_rmtval_set( > > > > > } > > > > > /* > > > > > + * Find a hole for the attr and store it in the delayed attr context. This > > > > > + * initializes the context to roll through allocating an attr extent for a > > > > > + * delayed attr operation > > > > > + */ > > > > > +int > > > > > +xfs_attr_rmtval_set_init( > > > > > + struct xfs_delattr_context *dac) > > > > > +{ > > > > > + struct xfs_da_args *args = dac->da_args; > > > > > + struct xfs_bmbt_irec *map = &dac->map; > > > > > + int error; > > > > > + > > > > > + dac->lblkno = 0; > > > > > + dac->lfileoff = 0; > > > > > + dac->blkcnt = 0; > > > > > + args->rmtblkcnt = 0; > > > > > + args->rmtblkno = 0; > > > > > + memset(map, 0, sizeof(struct xfs_bmbt_irec)); > > > > > + > > > > > + error = xfs_attr_rmt_find_hole(args); > > > > > + if (error) > > > > > + return error; > > > > > + > > > > > + dac->blkcnt = args->rmtblkcnt; > > > > > + dac->lblkno = args->rmtblkno; > > > > > + > > > > > + return error; > > > > > +} > > > > > + > > > > > +/* > > > > > + * Write one block of the value associated with an attribute into the > > > > > + * out-of-line buffer that we have defined for it. This is similar to a subset > > > > > + * of xfs_attr_rmtval_set, but records the current block to the delayed attr > > > > > + * context, and leaves transaction handling to the caller. > > > > > + */ > > > > > +int > > > > > +xfs_attr_rmtval_set_blk( > > > > > + struct xfs_delattr_context *dac) > > > > > +{ > > > > > + struct xfs_da_args *args = dac->da_args; > > > > > + struct xfs_inode *dp = args->dp; > > > > > + struct xfs_bmbt_irec *map = &dac->map; > > > > > + int nmap; > > > > > + int error; > > > > > + > > > > > + nmap = 1; > > > > > + error = xfs_bmapi_write(args->trans, dp, > > > > > + (xfs_fileoff_t)dac->lblkno, > > > > > + dac->blkcnt, XFS_BMAPI_ATTRFORK, > > > > > + args->total, map, &nmap); > > > > > + if (error) > > > > > + return error; > > > > > + > > > > > + ASSERT(nmap == 1); > > > > > + ASSERT((map->br_startblock != DELAYSTARTBLOCK) && > > > > > + (map->br_startblock != HOLESTARTBLOCK)); > > > > > + > > > > > + /* roll attribute extent map forwards */ > > > > > + dac->lblkno += map->br_blockcount; > > > > > + dac->blkcnt -= map->br_blockcount; > > > > > + > > > > > + return 0; > > > > > +} > > > > > + > > > > > +/* > > > > > * Remove the value associated with an attribute by deleting the > > > > > * out-of-line buffer that it is stored on. > > > > > */ > > > > > @@ -671,48 +736,6 @@ xfs_attr_rmtval_invalidate( > > > > > } > > > > > /* > > > > > - * Remove the value associated with an attribute by deleting the > > > > > - * out-of-line buffer that it is stored on. > > > > > - */ > > > > > -int > > > > > -xfs_attr_rmtval_remove( > > > > > - struct xfs_da_args *args) > > > > > -{ > > > > > - xfs_dablk_t lblkno; > > > > > - int blkcnt; > > > > > - int error = 0; > > > > > - int done = 0; > > > > > - > > > > > - trace_xfs_attr_rmtval_remove(args); > > > > > - > > > > > - error = xfs_attr_rmtval_invalidate(args); > > > > > - if (error) > > > > > - return error; > > > > > - /* > > > > > - * Keep de-allocating extents until the remote-value region is gone. > > > > > - */ > > > > > - lblkno = args->rmtblkno; > > > > > - blkcnt = args->rmtblkcnt; > > > > > - while (!done) { > > > > > - error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt, > > > > > - XFS_BMAPI_ATTRFORK, 1, &done); > > > > > - if (error) > > > > > - return error; > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - if (error) > > > > > - return error; > > > > > - > > > > > - /* > > > > > - * Close out trans and start the next one in the chain. > > > > > - */ > > > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > > > - if (error) > > > > > - return error; > > > > > - } > > > > > - return 0; > > > > > -} > > > > > - > > > > > -/* > > > > > * Remove the value associated with an attribute by deleting the out-of-line > > > > > * buffer that it is stored on. Returns EAGAIN for the caller to refresh the > > > > > * transaction and recall the function > > > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h > > > > > index ee3337b..482dff9 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr_remote.h > > > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.h > > > > > @@ -15,4 +15,8 @@ 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_rmt_find_hole(struct xfs_da_args *args); > > > > > +int xfs_attr_rmtval_set_value(struct xfs_da_args *args); > > > > > +int xfs_attr_rmtval_set_blk(struct xfs_delattr_context *dac); > > > > > +int xfs_attr_rmtval_set_init(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 c42f90e..3e8cec5 100644 > > > > > --- a/fs/xfs/xfs_attr_inactive.c > > > > > +++ b/fs/xfs/xfs_attr_inactive.c > > > > > @@ -15,6 +15,7 @@ > > > > > #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" > > > > > diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h > > > > > index a4323a6..26dc8bf 100644 > > > > > --- a/fs/xfs/xfs_trace.h > > > > > +++ b/fs/xfs/xfs_trace.h > > > > > @@ -1784,7 +1784,6 @@ DEFINE_ATTR_EVENT(xfs_attr_refillstate); > > > > > DEFINE_ATTR_EVENT(xfs_attr_rmtval_get); > > > > > DEFINE_ATTR_EVENT(xfs_attr_rmtval_set); > > > > > -DEFINE_ATTR_EVENT(xfs_attr_rmtval_remove); > > > > > #define DEFINE_DA_EVENT(name) \ > > > > > DEFINE_EVENT(xfs_da_class, name, \ > > > > > -- > > > > > 2.7.4 > > > > > > > > > > > > > > >
On Saturday, April 4, 2020 3:42 AM Allison Collins wrote: > This patch modifies the attr set 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_set_args has become xfs_attr_set_iter, which uses a > state machine like switch to keep track of where it was when EAGAIN was > returned. > > Two new helper functions have been added: xfs_attr_rmtval_set_init and > xfs_attr_rmtval_set_blk. They provide a subset of logic similar to > xfs_attr_rmtval_set, but they store the current block in the delay attr > context to allow the caller to roll the transaction between allocations. > This helps to simplify and consolidate code used by > xfs_attr_leaf_addname and xfs_attr_node_addname. xfs_attr_set_args has > now become a simple loop to refresh the transaction until the operation > is completed. Lastly, xfs_attr_rmtval_remove is no longer used, and is > removed. > > Below is a state machine diagram for attr set 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_set_iter() > │ > v > need to upgrade > from sf to leaf? ──n─┐ > │ │ > y │ > │ │ > V │ > XFS_DAS_ADD_LEAF │ > │ │ > v │ > ┌──────n── fork has <──────┘ > │ only 1 blk? > │ │ > │ y > │ │ > │ v > │ xfs_attr_leaf_try_add() > │ │ > │ v > │ had enough > ├──────n── space? > │ │ > │ y > │ │ > │ v > │ XFS_DAS_FOUND_LBLK ──┐ > │ │ > │ XFS_DAS_FLIP_LFLAG ──┤ > │ (subroutine state) │ > │ │ > │ XFS_DAS_ALLOC_LEAF ──┤ > │ (subroutine state) │ > │ └─>xfs_attr_leaf_addname() > │ │ > │ v > │ ┌─────n── need to > │ │ alloc blks? > │ │ │ > │ │ y > │ │ │ > │ │ v > │ │ ┌─>XFS_DAS_ALLOC_LEAF > │ │ │ │ > │ │ │ v > │ │ └──y── need to alloc > │ │ more blocks? > │ │ │ > │ │ n > │ │ │ > │ │ v > │ │ was this > │ └────────> a rename? ──n─┐ > │ │ │ > │ y │ > │ │ │ > │ v │ > │ flip incomplete │ > │ flag │ > │ │ │ > │ v │ > │ XFS_DAS_FLIP_LFLAG │ > │ │ │ > │ v │ > │ remove │ > │ XFS_DAS_RM_LBLK ─> old name │ > │ ^ │ │ > │ │ v │ > │ └──────y── more to │ > │ remove │ > │ │ │ > │ n │ > │ │ │ > │ v │ > │ done <──────┘ > └────> XFS_DAS_LEAF_TO_NODE ─┐ > │ > XFS_DAS_FOUND_NBLK ──┤ > (subroutine state) │ > │ > XFS_DAS_ALLOC_NODE ──┤ > (subroutine state) │ > │ > XFS_DAS_FLIP_NFLAG ──┤ > (subroutine state) │ > │ > └─>xfs_attr_node_addname() > │ > v > find space to store > attr. Split if needed > │ > v > XFS_DAS_FOUND_NBLK > │ > v > ┌─────n── need to > │ alloc blks? > │ │ > │ y > │ │ > │ v > │ ┌─>XFS_DAS_ALLOC_NODE > │ │ │ > │ │ v > │ └──y── need to alloc > │ more blocks? > │ │ > │ n > │ │ > │ v > │ was this > └────────> a rename? ──n─┐ > │ │ > y │ > │ │ > v │ > flip incomplete │ > flag │ > │ │ > v │ > XFS_DAS_FLIP_NFLAG │ > │ │ > v │ > remove │ > XFS_DAS_RM_NBLK ─> old name │ > ^ │ │ > │ v │ > └──────y── more to │ > remove │ > │ │ > n │ > │ │ > v │ > done <──────┘ > > Signed-off-by: Allison Collins <allison.henderson@oracle.com> > --- > fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- > fs/xfs/libxfs/xfs_attr.h | 16 ++ > fs/xfs/libxfs/xfs_attr_leaf.c | 1 + > fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- > fs/xfs/libxfs/xfs_attr_remote.h | 4 + > fs/xfs/xfs_attr_inactive.c | 1 + > fs/xfs/xfs_trace.h | 1 - > 7 files changed, 351 insertions(+), 167 deletions(-) > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > index f700976..c160b7a 100644 > --- a/fs/xfs/libxfs/xfs_attr.c > +++ b/fs/xfs/libxfs/xfs_attr.c > @@ -44,7 +44,7 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args); > * Internal routines when attribute list is one block. > */ > STATIC int xfs_attr_leaf_get(xfs_da_args_t *args); > -STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args); > +STATIC int xfs_attr_leaf_addname(struct xfs_delattr_context *dac); > STATIC int xfs_attr_leaf_removename(struct xfs_delattr_context *dac); > STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > > @@ -52,12 +52,13 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); > * Internal routines when attribute list is more than one block. > */ > 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_addname(struct xfs_delattr_context *dac); > 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); > STATIC int xfs_attr_refillstate(xfs_da_state_t *state); > +STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp); > > STATIC void > xfs_delattr_context_init( > @@ -227,8 +228,11 @@ xfs_attr_is_shortform( > > /* > * Attempts to set an attr in shortform, or converts the tree to leaf form if > - * there is not enough room. If the attr is set, the transaction is committed > - * and set to NULL. > + * there is not enough room. This function is meant to operate as a helper > + * routine to the delayed attribute functions. It returns -EAGAIN to indicate > + * that the calling function should roll the transaction, and then proceed to > + * add the attr in leaf form. This subroutine does not expect to be recalled > + * again like the other delayed attr routines do. > */ > STATIC int > xfs_attr_set_shortform( > @@ -236,16 +240,16 @@ xfs_attr_set_shortform( > struct xfs_buf **leaf_bp) > { > struct xfs_inode *dp = args->dp; > - int error, error2 = 0; > + int error = 0; > > /* > * Try to add the attr to the attribute list in the inode. > */ > error = xfs_attr_try_sf_addname(dp, args); > + > + /* Should only be 0, -EEXIST or ENOSPC */ > if (error != -ENOSPC) { > - error2 = xfs_trans_commit(args->trans); > - args->trans = NULL; > - return error ? error : error2; > + return error; > } > /* > * It won't fit in the shortform, transform to a leaf block. GROT: > @@ -258,18 +262,10 @@ xfs_attr_set_shortform( > /* > * Prevent the leaf buffer from being unlocked so that a concurrent AIL > * push cannot grab the half-baked leaf buffer and run into problems > - * with the write verifier. Once we're done rolling the transaction we > - * can release the hold and add the attr to the leaf. > + * with the write verifier. > */ > xfs_trans_bhold(args->trans, *leaf_bp); > - error = xfs_defer_finish(&args->trans); > - xfs_trans_bhold_release(args->trans, *leaf_bp); > - if (error) { > - xfs_trans_brelse(args->trans, *leaf_bp); > - return error; > - } > - > - return 0; > + return -EAGAIN; > } > > /* > @@ -279,9 +275,83 @@ int > xfs_attr_set_args( > struct xfs_da_args *args) > { > - struct xfs_inode *dp = args->dp; > - struct xfs_buf *leaf_bp = NULL; > - int error = 0; > + struct xfs_buf *leaf_bp = NULL; > + int error = 0; > + struct xfs_delattr_context dac; > + > + xfs_delattr_context_init(&dac, args); > + > + do { > + error = xfs_attr_set_iter(&dac, &leaf_bp); > + if (error != -EAGAIN) > + break; > + > + if (dac.flags & XFS_DAC_DEFER_FINISH) { > + dac.flags &= ~XFS_DAC_DEFER_FINISH; > + error = xfs_defer_finish(&args->trans); > + if (error) > + break; > + } > + > + error = xfs_trans_roll_inode(&args->trans, args->dp); > + if (error) > + break; > + > + if (leaf_bp) { > + xfs_trans_bjoin(args->trans, leaf_bp); > + xfs_trans_bhold(args->trans, leaf_bp); > + } > + > + } while (true); > + > + return error; > +} > + > +/* > + * Set the attribute specified in @args. > + * This routine is meant to function as a 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. > + */ > +int > +xfs_attr_set_iter( > + struct xfs_delattr_context *dac, > + struct xfs_buf **leaf_bp) > +{ > + struct xfs_da_args *args = dac->da_args; > + struct xfs_inode *dp = args->dp; > + int error = 0; > + int sf_size; > + > + /* State machine switch */ > + switch (dac->dela_state) { > + case XFS_DAS_ADD_LEAF: > + goto das_add_leaf; > + case XFS_DAS_ALLOC_LEAF: > + case XFS_DAS_FLIP_LFLAG: > + case XFS_DAS_FOUND_LBLK: > + goto das_leaf; > + case XFS_DAS_FOUND_NBLK: > + case XFS_DAS_FLIP_NFLAG: > + case XFS_DAS_ALLOC_NODE: > + case XFS_DAS_LEAF_TO_NODE: > + goto das_node; > + default: > + break; > + } > + > + /* > + * New inodes may not have an attribute fork yet. So set the attribute > + * fork appropriately > + */ > + if (XFS_IFORK_Q((args->dp)) == 0) { > + sf_size = sizeof(struct xfs_attr_sf_hdr) + > + XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen); > + xfs_bmap_set_attrforkoff(args->dp, sf_size, NULL); > + args->dp->i_afp = kmem_zone_zalloc(xfs_ifork_zone, 0); > + args->dp->i_afp->if_flags = XFS_IFEXTENTS; > + } > > /* > * If the attribute list is already in leaf format, jump straight to > @@ -292,40 +362,53 @@ xfs_attr_set_args( > if (xfs_attr_is_shortform(dp)) { > > /* > - * If the attr was successfully set in shortform, the > - * transaction is committed and set to NULL. Otherwise, is it > - * converted from shortform to leaf, and the transaction is > - * retained. > + * If the attr was successfully set in shortform, no need to > + * continue. Otherwise, is it converted from shortform to leaf > + * and -EAGAIN is returned. > */ > - error = xfs_attr_set_shortform(args, &leaf_bp); > - if (error || !args->trans) > - return error; > + error = xfs_attr_set_shortform(args, leaf_bp); > + if (error == -EAGAIN) { > + dac->flags |= XFS_DAC_DEFER_FINISH; > + dac->dela_state = XFS_DAS_ADD_LEAF; > + } > + return error; > } > > - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > - error = xfs_attr_leaf_addname(args); > - if (error != -ENOSPC) > - return error; > +das_add_leaf: > > - /* > - * Commit that transaction so that the node_addname() > - * call can manage its own transactions. > - */ > - error = xfs_defer_finish(&args->trans); > - if (error) > - return error; > + /* > + * After a shortform to leaf conversion, we need to hold the leaf and > + * cylce out the transaction. When we get back, we need to release > + * the leaf. > + */ > + if (*leaf_bp != NULL) { > + xfs_trans_brelse(args->trans, *leaf_bp); > + *leaf_bp = NULL; > + } > > - /* > - * Commit the current trans (including the inode) and > - * start a new one. > - */ > - error = xfs_trans_roll_inode(&args->trans, dp); > - if (error) > + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > + error = xfs_attr_leaf_try_add(args, *leaf_bp); > + switch (error) { > + case -ENOSPC: > + dac->flags |= XFS_DAC_DEFER_FINISH; > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > + return -EAGAIN; > + case 0: > + dac->dela_state = XFS_DAS_FOUND_LBLK; > + return -EAGAIN; > + default: > return error; > - > + } > +das_leaf: > + error = xfs_attr_leaf_addname(dac); > + if (error == -ENOSPC) { > + dac->dela_state = XFS_DAS_LEAF_TO_NODE; > + return -EAGAIN; > + } > + return error; > } > - > - error = xfs_attr_node_addname(args); > +das_node: > + error = xfs_attr_node_addname(dac); > return error; > } > > @@ -716,28 +799,32 @@ xfs_attr_leaf_try_add( > * > * This leaf block cannot have a "remote" value, we only call this routine > * if bmap_one_block() says there is only one block (ie: no remote blks). > + * > + * This routine is meant to function as a 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_leaf_addname( > - struct xfs_da_args *args) > + struct xfs_delattr_context *dac) > { > - int error, forkoff; > - struct xfs_buf *bp = NULL; > - struct xfs_inode *dp = args->dp; > - > - trace_xfs_attr_leaf_addname(args); > - > - error = xfs_attr_leaf_try_add(args, bp); > - if (error) > - return error; > + struct xfs_da_args *args = dac->da_args; > + struct xfs_buf *bp = NULL; > + int error, forkoff; > + struct xfs_inode *dp = args->dp; > > - /* > - * Commit the transaction that added the attr name so that > - * later routines can manage their own transactions. > - */ > - error = xfs_trans_roll_inode(&args->trans, dp); > - if (error) > - return error; > + /* State machine switch */ > + switch (dac->dela_state) { > + case XFS_DAS_FLIP_LFLAG: > + goto das_flip_flag; > + case XFS_DAS_ALLOC_LEAF: > + goto das_alloc_leaf; > + case XFS_DAS_RM_LBLK: > + goto das_rm_lblk; > + default: > + break; > + } > > /* > * If there was an out-of-line value, allocate the blocks we > @@ -746,7 +833,28 @@ xfs_attr_leaf_addname( > * maximum size of a transaction and/or hit a deadlock. > */ > if (args->rmtblkno > 0) { > - error = xfs_attr_rmtval_set(args); > + > + /* Open coded xfs_attr_rmtval_set without trans handling */ > + error = xfs_attr_rmtval_set_init(dac); > + if (error) > + return error; > + > + /* > + * Roll through the "value", allocating blocks on disk as > + * required. > + */ > +das_alloc_leaf: > + while (dac->blkcnt > 0) { > + error = xfs_attr_rmtval_set_blk(dac); > + if (error) > + return error; > + > + dac->flags |= XFS_DAC_DEFER_FINISH; > + dac->dela_state = XFS_DAS_ALLOC_LEAF; > + return -EAGAIN; > + } > + > + error = xfs_attr_rmtval_set_value(args); > if (error) > return error; > } > @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( > error = xfs_attr3_leaf_flipflags(args); > if (error) > return error; > - /* > - * Commit the flag value change and start the next trans in > - * series. > - */ > - error = xfs_trans_roll_inode(&args->trans, args->dp); > - if (error) > - return error; > - > + dac->dela_state = XFS_DAS_FLIP_LFLAG; > + return -EAGAIN; > +das_flip_flag: > /* > * Dismantle the "old" attribute/value pair by removing > * a "remote" value (if it exists). > */ > xfs_attr_restore_rmt_blk(args); > > + xfs_attr_rmtval_invalidate(args); > +das_rm_lblk: > if (args->rmtblkno) { > - error = xfs_attr_rmtval_remove(args); > + error = __xfs_attr_rmtval_remove(args); > + > + if (error == -EAGAIN) { > + dac->dela_state = XFS_DAS_RM_LBLK; Similar to what I had observed in the patch "Add delay ready attr remove routines", Shouldn't XFS_DAC_DEFER_FINISH be set in dac->flags? __xfs_attr_rmtval_remove() calls __xfs_bunmapi() which would have added items to the deferred list.
On Mon, Apr 20, 2020 at 05:15:08PM +0530, Chandan Rajendra wrote: > On Saturday, April 4, 2020 3:42 AM Allison Collins wrote: ... > > Signed-off-by: Allison Collins <allison.henderson@oracle.com> > > --- > > fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- > > fs/xfs/libxfs/xfs_attr.h | 16 ++ > > fs/xfs/libxfs/xfs_attr_leaf.c | 1 + > > fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- > > fs/xfs/libxfs/xfs_attr_remote.h | 4 + > > fs/xfs/xfs_attr_inactive.c | 1 + > > fs/xfs/xfs_trace.h | 1 - > > 7 files changed, 351 insertions(+), 167 deletions(-) > > > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > > index f700976..c160b7a 100644 > > --- a/fs/xfs/libxfs/xfs_attr.c > > +++ b/fs/xfs/libxfs/xfs_attr.c ... > > @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( > > error = xfs_attr3_leaf_flipflags(args); > > if (error) > > return error; > > - /* > > - * Commit the flag value change and start the next trans in > > - * series. > > - */ > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > - if (error) > > - return error; > > - > > + dac->dela_state = XFS_DAS_FLIP_LFLAG; > > + return -EAGAIN; > > +das_flip_flag: > > /* > > * Dismantle the "old" attribute/value pair by removing > > * a "remote" value (if it exists). > > */ > > xfs_attr_restore_rmt_blk(args); > > > > + xfs_attr_rmtval_invalidate(args); > > +das_rm_lblk: > > if (args->rmtblkno) { > > - error = xfs_attr_rmtval_remove(args); > > + error = __xfs_attr_rmtval_remove(args); > > + > > + if (error == -EAGAIN) { > > + dac->dela_state = XFS_DAS_RM_LBLK; > > Similar to what I had observed in the patch "Add delay ready attr remove > routines", > > Shouldn't XFS_DAC_DEFER_FINISH be set in dac->flags? > __xfs_attr_rmtval_remove() calls __xfs_bunmapi() which would > have added items to the deferred list. > Just note that transaction rolls don't currently finish deferred ops. So from the perspective of preserving current behavior it might make sense to set the flag here if there was an explicit xfs_defer_finish() that's been factored out, but not so if it was just a transaction roll. Brian > -- > chandan > > >
On 4/20/20 9:20 AM, Brian Foster wrote: > On Mon, Apr 20, 2020 at 05:15:08PM +0530, Chandan Rajendra wrote: >> On Saturday, April 4, 2020 3:42 AM Allison Collins wrote: > ... >>> Signed-off-by: Allison Collins <allison.henderson@oracle.com> >>> --- >>> fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- >>> fs/xfs/libxfs/xfs_attr.h | 16 ++ >>> fs/xfs/libxfs/xfs_attr_leaf.c | 1 + >>> fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- >>> fs/xfs/libxfs/xfs_attr_remote.h | 4 + >>> fs/xfs/xfs_attr_inactive.c | 1 + >>> fs/xfs/xfs_trace.h | 1 - >>> 7 files changed, 351 insertions(+), 167 deletions(-) >>> >>> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c >>> index f700976..c160b7a 100644 >>> --- a/fs/xfs/libxfs/xfs_attr.c >>> +++ b/fs/xfs/libxfs/xfs_attr.c > ... >>> @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( >>> error = xfs_attr3_leaf_flipflags(args); >>> if (error) >>> return error; >>> - /* >>> - * Commit the flag value change and start the next trans in >>> - * series. >>> - */ >>> - error = xfs_trans_roll_inode(&args->trans, args->dp); >>> - if (error) >>> - return error; >>> - >>> + dac->dela_state = XFS_DAS_FLIP_LFLAG; >>> + return -EAGAIN; >>> +das_flip_flag: >>> /* >>> * Dismantle the "old" attribute/value pair by removing >>> * a "remote" value (if it exists). >>> */ >>> xfs_attr_restore_rmt_blk(args); >>> >>> + xfs_attr_rmtval_invalidate(args); >>> +das_rm_lblk: >>> if (args->rmtblkno) { >>> - error = xfs_attr_rmtval_remove(args); >>> + error = __xfs_attr_rmtval_remove(args); >>> + >>> + if (error == -EAGAIN) { >>> + dac->dela_state = XFS_DAS_RM_LBLK; >> >> Similar to what I had observed in the patch "Add delay ready attr remove >> routines", >> >> Shouldn't XFS_DAC_DEFER_FINISH be set in dac->flags? >> __xfs_attr_rmtval_remove() calls __xfs_bunmapi() which would >> have added items to the deferred list. >> > > Just note that transaction rolls don't currently finish deferred ops. So > from the perspective of preserving current behavior it might make sense > to set the flag here if there was an explicit xfs_defer_finish() that's > been factored out, but not so if it was just a transaction roll. > > Brian Yep, I think Chandan is right, xfs_attr_rmtval_remove used to have a xfs_defer_finish and __xfs_attr_rmtval_remove does not. I think probably I'll keep the xfs_defer_finish in the __xfs_attr_rmtval_remove helper, and then when we get to this patch, turn it into the flag set. I believe that should be correct. Allison > >> -- >> chandan >> >> >> >
diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c index f700976..c160b7a 100644 --- a/fs/xfs/libxfs/xfs_attr.c +++ b/fs/xfs/libxfs/xfs_attr.c @@ -44,7 +44,7 @@ STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args); * Internal routines when attribute list is one block. */ STATIC int xfs_attr_leaf_get(xfs_da_args_t *args); -STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args); +STATIC int xfs_attr_leaf_addname(struct xfs_delattr_context *dac); STATIC int xfs_attr_leaf_removename(struct xfs_delattr_context *dac); STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); @@ -52,12 +52,13 @@ STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp); * Internal routines when attribute list is more than one block. */ 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_addname(struct xfs_delattr_context *dac); 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); STATIC int xfs_attr_refillstate(xfs_da_state_t *state); +STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp); STATIC void xfs_delattr_context_init( @@ -227,8 +228,11 @@ xfs_attr_is_shortform( /* * Attempts to set an attr in shortform, or converts the tree to leaf form if - * there is not enough room. If the attr is set, the transaction is committed - * and set to NULL. + * there is not enough room. This function is meant to operate as a helper + * routine to the delayed attribute functions. It returns -EAGAIN to indicate + * that the calling function should roll the transaction, and then proceed to + * add the attr in leaf form. This subroutine does not expect to be recalled + * again like the other delayed attr routines do. */ STATIC int xfs_attr_set_shortform( @@ -236,16 +240,16 @@ xfs_attr_set_shortform( struct xfs_buf **leaf_bp) { struct xfs_inode *dp = args->dp; - int error, error2 = 0; + int error = 0; /* * Try to add the attr to the attribute list in the inode. */ error = xfs_attr_try_sf_addname(dp, args); + + /* Should only be 0, -EEXIST or ENOSPC */ if (error != -ENOSPC) { - error2 = xfs_trans_commit(args->trans); - args->trans = NULL; - return error ? error : error2; + return error; } /* * It won't fit in the shortform, transform to a leaf block. GROT: @@ -258,18 +262,10 @@ xfs_attr_set_shortform( /* * Prevent the leaf buffer from being unlocked so that a concurrent AIL * push cannot grab the half-baked leaf buffer and run into problems - * with the write verifier. Once we're done rolling the transaction we - * can release the hold and add the attr to the leaf. + * with the write verifier. */ xfs_trans_bhold(args->trans, *leaf_bp); - error = xfs_defer_finish(&args->trans); - xfs_trans_bhold_release(args->trans, *leaf_bp); - if (error) { - xfs_trans_brelse(args->trans, *leaf_bp); - return error; - } - - return 0; + return -EAGAIN; } /* @@ -279,9 +275,83 @@ int xfs_attr_set_args( struct xfs_da_args *args) { - struct xfs_inode *dp = args->dp; - struct xfs_buf *leaf_bp = NULL; - int error = 0; + struct xfs_buf *leaf_bp = NULL; + int error = 0; + struct xfs_delattr_context dac; + + xfs_delattr_context_init(&dac, args); + + do { + error = xfs_attr_set_iter(&dac, &leaf_bp); + if (error != -EAGAIN) + break; + + if (dac.flags & XFS_DAC_DEFER_FINISH) { + dac.flags &= ~XFS_DAC_DEFER_FINISH; + error = xfs_defer_finish(&args->trans); + if (error) + break; + } + + error = xfs_trans_roll_inode(&args->trans, args->dp); + if (error) + break; + + if (leaf_bp) { + xfs_trans_bjoin(args->trans, leaf_bp); + xfs_trans_bhold(args->trans, leaf_bp); + } + + } while (true); + + return error; +} + +/* + * Set the attribute specified in @args. + * This routine is meant to function as a 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. + */ +int +xfs_attr_set_iter( + struct xfs_delattr_context *dac, + struct xfs_buf **leaf_bp) +{ + struct xfs_da_args *args = dac->da_args; + struct xfs_inode *dp = args->dp; + int error = 0; + int sf_size; + + /* State machine switch */ + switch (dac->dela_state) { + case XFS_DAS_ADD_LEAF: + goto das_add_leaf; + case XFS_DAS_ALLOC_LEAF: + case XFS_DAS_FLIP_LFLAG: + case XFS_DAS_FOUND_LBLK: + goto das_leaf; + case XFS_DAS_FOUND_NBLK: + case XFS_DAS_FLIP_NFLAG: + case XFS_DAS_ALLOC_NODE: + case XFS_DAS_LEAF_TO_NODE: + goto das_node; + default: + break; + } + + /* + * New inodes may not have an attribute fork yet. So set the attribute + * fork appropriately + */ + if (XFS_IFORK_Q((args->dp)) == 0) { + sf_size = sizeof(struct xfs_attr_sf_hdr) + + XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen); + xfs_bmap_set_attrforkoff(args->dp, sf_size, NULL); + args->dp->i_afp = kmem_zone_zalloc(xfs_ifork_zone, 0); + args->dp->i_afp->if_flags = XFS_IFEXTENTS; + } /* * If the attribute list is already in leaf format, jump straight to @@ -292,40 +362,53 @@ xfs_attr_set_args( if (xfs_attr_is_shortform(dp)) { /* - * If the attr was successfully set in shortform, the - * transaction is committed and set to NULL. Otherwise, is it - * converted from shortform to leaf, and the transaction is - * retained. + * If the attr was successfully set in shortform, no need to + * continue. Otherwise, is it converted from shortform to leaf + * and -EAGAIN is returned. */ - error = xfs_attr_set_shortform(args, &leaf_bp); - if (error || !args->trans) - return error; + error = xfs_attr_set_shortform(args, leaf_bp); + if (error == -EAGAIN) { + dac->flags |= XFS_DAC_DEFER_FINISH; + dac->dela_state = XFS_DAS_ADD_LEAF; + } + return error; } - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { - error = xfs_attr_leaf_addname(args); - if (error != -ENOSPC) - return error; +das_add_leaf: - /* - * Commit that transaction so that the node_addname() - * call can manage its own transactions. - */ - error = xfs_defer_finish(&args->trans); - if (error) - return error; + /* + * After a shortform to leaf conversion, we need to hold the leaf and + * cylce out the transaction. When we get back, we need to release + * the leaf. + */ + if (*leaf_bp != NULL) { + xfs_trans_brelse(args->trans, *leaf_bp); + *leaf_bp = NULL; + } - /* - * Commit the current trans (including the inode) and - * start a new one. - */ - error = xfs_trans_roll_inode(&args->trans, dp); - if (error) + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { + error = xfs_attr_leaf_try_add(args, *leaf_bp); + switch (error) { + case -ENOSPC: + dac->flags |= XFS_DAC_DEFER_FINISH; + dac->dela_state = XFS_DAS_LEAF_TO_NODE; + return -EAGAIN; + case 0: + dac->dela_state = XFS_DAS_FOUND_LBLK; + return -EAGAIN; + default: return error; - + } +das_leaf: + error = xfs_attr_leaf_addname(dac); + if (error == -ENOSPC) { + dac->dela_state = XFS_DAS_LEAF_TO_NODE; + return -EAGAIN; + } + return error; } - - error = xfs_attr_node_addname(args); +das_node: + error = xfs_attr_node_addname(dac); return error; } @@ -716,28 +799,32 @@ xfs_attr_leaf_try_add( * * This leaf block cannot have a "remote" value, we only call this routine * if bmap_one_block() says there is only one block (ie: no remote blks). + * + * This routine is meant to function as a 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_leaf_addname( - struct xfs_da_args *args) + struct xfs_delattr_context *dac) { - int error, forkoff; - struct xfs_buf *bp = NULL; - struct xfs_inode *dp = args->dp; - - trace_xfs_attr_leaf_addname(args); - - error = xfs_attr_leaf_try_add(args, bp); - if (error) - return error; + struct xfs_da_args *args = dac->da_args; + struct xfs_buf *bp = NULL; + int error, forkoff; + struct xfs_inode *dp = args->dp; - /* - * Commit the transaction that added the attr name so that - * later routines can manage their own transactions. - */ - error = xfs_trans_roll_inode(&args->trans, dp); - if (error) - return error; + /* State machine switch */ + switch (dac->dela_state) { + case XFS_DAS_FLIP_LFLAG: + goto das_flip_flag; + case XFS_DAS_ALLOC_LEAF: + goto das_alloc_leaf; + case XFS_DAS_RM_LBLK: + goto das_rm_lblk; + default: + break; + } /* * If there was an out-of-line value, allocate the blocks we @@ -746,7 +833,28 @@ xfs_attr_leaf_addname( * maximum size of a transaction and/or hit a deadlock. */ if (args->rmtblkno > 0) { - error = xfs_attr_rmtval_set(args); + + /* Open coded xfs_attr_rmtval_set without trans handling */ + error = xfs_attr_rmtval_set_init(dac); + if (error) + return error; + + /* + * Roll through the "value", allocating blocks on disk as + * required. + */ +das_alloc_leaf: + while (dac->blkcnt > 0) { + error = xfs_attr_rmtval_set_blk(dac); + if (error) + return error; + + dac->flags |= XFS_DAC_DEFER_FINISH; + dac->dela_state = XFS_DAS_ALLOC_LEAF; + return -EAGAIN; + } + + error = xfs_attr_rmtval_set_value(args); if (error) return error; } @@ -765,22 +873,25 @@ xfs_attr_leaf_addname( error = xfs_attr3_leaf_flipflags(args); if (error) return error; - /* - * Commit the flag value change and start the next trans in - * series. - */ - error = xfs_trans_roll_inode(&args->trans, args->dp); - if (error) - return error; - + dac->dela_state = XFS_DAS_FLIP_LFLAG; + return -EAGAIN; +das_flip_flag: /* * Dismantle the "old" attribute/value pair by removing * a "remote" value (if it exists). */ xfs_attr_restore_rmt_blk(args); + xfs_attr_rmtval_invalidate(args); +das_rm_lblk: if (args->rmtblkno) { - error = xfs_attr_rmtval_remove(args); + error = __xfs_attr_rmtval_remove(args); + + if (error == -EAGAIN) { + dac->dela_state = XFS_DAS_RM_LBLK; + return -EAGAIN; + } + if (error) return error; } @@ -799,15 +910,11 @@ xfs_attr_leaf_addname( /* * If the result is small enough, shrink it all into the inode. */ - if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) { + forkoff = xfs_attr_shortform_allfit(bp, dp); + if (forkoff) error = xfs_attr3_leaf_to_shortform(bp, args, forkoff); - /* bp is gone due to xfs_da_shrink_inode */ - if (error) - return error; - error = xfs_defer_finish(&args->trans); - if (error) - return error; - } + + dac->flags |= XFS_DAC_DEFER_FINISH; } else if (args->rmtblkno > 0) { /* @@ -967,16 +1074,23 @@ xfs_attr_node_hasname( * * "Remote" attribute values confuse the issue and atomic rename operations * add a whole extra layer of confusion on top of that. + * + * This routine is meant to function as a 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_addname( - struct xfs_da_args *args) + struct xfs_delattr_context *dac) { - struct xfs_da_state *state; - struct xfs_da_state_blk *blk; - struct xfs_inode *dp; - struct xfs_mount *mp; - int retval, error; + struct xfs_da_args *args = dac->da_args; + struct xfs_da_state *state = NULL; + struct xfs_da_state_blk *blk; + struct xfs_inode *dp; + struct xfs_mount *mp; + int retval = 0; + int error = 0; trace_xfs_attr_node_addname(args); @@ -985,7 +1099,21 @@ xfs_attr_node_addname( */ dp = args->dp; mp = dp->i_mount; -restart: + + /* State machine switch */ + switch (dac->dela_state) { + case XFS_DAS_FLIP_NFLAG: + goto das_flip_flag; + case XFS_DAS_FOUND_NBLK: + goto das_found_nblk; + case XFS_DAS_ALLOC_NODE: + goto das_alloc_node; + case XFS_DAS_RM_NBLK: + goto das_rm_nblk; + default: + break; + } + /* * Search to see if name already exists, and get back a pointer * to where it should go. @@ -1031,19 +1159,13 @@ xfs_attr_node_addname( error = xfs_attr3_leaf_to_node(args); if (error) goto out; - error = xfs_defer_finish(&args->trans); - if (error) - goto out; /* - * Commit the node conversion and start the next - * trans in the chain. + * Restart routine from the top. No need to set the + * state */ - error = xfs_trans_roll_inode(&args->trans, dp); - if (error) - goto out; - - goto restart; + dac->flags |= XFS_DAC_DEFER_FINISH; + return -EAGAIN; } /* @@ -1055,9 +1177,7 @@ xfs_attr_node_addname( error = xfs_da3_split(state); if (error) goto out; - error = xfs_defer_finish(&args->trans); - if (error) - goto out; + dac->flags |= XFS_DAC_DEFER_FINISH; } else { /* * Addition succeeded, update Btree hashvals. @@ -1072,13 +1192,9 @@ xfs_attr_node_addname( xfs_da_state_free(state); state = NULL; - /* - * Commit the leaf addition or btree split and start the next - * trans in the chain. - */ - error = xfs_trans_roll_inode(&args->trans, dp); - if (error) - goto out; + dac->dela_state = XFS_DAS_FOUND_NBLK; + return -EAGAIN; +das_found_nblk: /* * If there was an out-of-line value, allocate the blocks we @@ -1087,7 +1203,27 @@ xfs_attr_node_addname( * maximum size of a transaction and/or hit a deadlock. */ if (args->rmtblkno > 0) { - error = xfs_attr_rmtval_set(args); + /* Open coded xfs_attr_rmtval_set without trans handling */ + error = xfs_attr_rmtval_set_init(dac); + if (error) + return error; + + /* + * Roll through the "value", allocating blocks on disk as + * required. + */ +das_alloc_node: + while (dac->blkcnt > 0) { + error = xfs_attr_rmtval_set_blk(dac); + if (error) + return error; + + dac->flags |= XFS_DAC_DEFER_FINISH; + dac->dela_state = XFS_DAS_ALLOC_NODE; + return -EAGAIN; + } + + error = xfs_attr_rmtval_set_value(args); if (error) return error; } @@ -1110,18 +1246,26 @@ xfs_attr_node_addname( * Commit the flag value change and start the next trans in * series */ - error = xfs_trans_roll_inode(&args->trans, args->dp); - if (error) - goto out; - + dac->dela_state = XFS_DAS_FLIP_NFLAG; + return -EAGAIN; +das_flip_flag: /* * Dismantle the "old" attribute/value pair by removing * a "remote" value (if it exists). */ xfs_attr_restore_rmt_blk(args); + xfs_attr_rmtval_invalidate(args); + +das_rm_nblk: if (args->rmtblkno) { - error = xfs_attr_rmtval_remove(args); + error = __xfs_attr_rmtval_remove(args); + + if (error == -EAGAIN) { + dac->dela_state = XFS_DAS_RM_NBLK; + return -EAGAIN; + } + if (error) return error; } @@ -1139,7 +1283,6 @@ xfs_attr_node_addname( error = xfs_da3_node_lookup_int(state, &retval); if (error) goto out; - /* * Remove the name and update the hashvals in the tree. */ @@ -1147,7 +1290,6 @@ xfs_attr_node_addname( ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC); error = xfs_attr3_leaf_remove(blk->bp, args); xfs_da3_fixhashpath(state, &state->path); - /* * Check to see if the tree needs to be collapsed. */ @@ -1155,11 +1297,9 @@ xfs_attr_node_addname( error = xfs_da3_join(state); if (error) goto out; - error = xfs_defer_finish(&args->trans); - if (error) - goto out; - } + dac->flags |= XFS_DAC_DEFER_FINISH; + } } else if (args->rmtblkno > 0) { /* * Added a "remote" value, just clear the incomplete flag. diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h index 0e8ae1a..67af9d1 100644 --- a/fs/xfs/libxfs/xfs_attr.h +++ b/fs/xfs/libxfs/xfs_attr.h @@ -93,6 +93,16 @@ enum xfs_delattr_state { /* Zero is uninitalized */ XFS_DAS_RM_SHRINK = 1, /* We are shrinking the tree */ XFS_DAS_RMTVAL_REMOVE, /* We are removing remote value blocks */ + XFS_DAS_ADD_LEAF, /* We are adding a leaf attr */ + XFS_DAS_FOUND_LBLK, /* We found leaf blk for attr */ + XFS_DAS_LEAF_TO_NODE, /* Converted leaf to node */ + XFS_DAS_FOUND_NBLK, /* We found node blk for attr */ + XFS_DAS_ALLOC_LEAF, /* We are allocating leaf blocks */ + XFS_DAS_FLIP_LFLAG, /* Flipped leaf INCOMPLETE attr flag */ + XFS_DAS_RM_LBLK, /* A rename is removing leaf blocks */ + XFS_DAS_ALLOC_NODE, /* We are allocating node blocks */ + XFS_DAS_FLIP_NFLAG, /* Flipped node INCOMPLETE attr flag */ + XFS_DAS_RM_NBLK, /* A rename is removing node blocks */ }; /* @@ -105,8 +115,13 @@ enum xfs_delattr_state { */ struct xfs_delattr_context { struct xfs_da_args *da_args; + struct xfs_bmbt_irec map; + struct xfs_buf *leaf_bp; + xfs_fileoff_t lfileoff; struct xfs_da_state *da_state; struct xfs_da_state_blk *blk; + xfs_dablk_t lblkno; + int blkcnt; unsigned int flags; enum xfs_delattr_state dela_state; }; @@ -126,6 +141,7 @@ int xfs_attr_get_ilocked(struct xfs_da_args *args); int xfs_attr_get(struct xfs_da_args *args); int xfs_attr_set(struct xfs_da_args *args); int xfs_attr_set_args(struct xfs_da_args *args); +int xfs_attr_set_iter(struct xfs_delattr_context *dac, struct xfs_buf **leaf_bp); 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); diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c index f55402b..4d15f45 100644 --- a/fs/xfs/libxfs/xfs_attr_leaf.c +++ b/fs/xfs/libxfs/xfs_attr_leaf.c @@ -19,6 +19,7 @@ #include "xfs_bmap_btree.h" #include "xfs_bmap.h" #include "xfs_attr_sf.h" +#include "xfs_attr.h" #include "xfs_attr_remote.h" #include "xfs_attr.h" #include "xfs_attr_leaf.h" diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c index fd4be9d..9607fd2 100644 --- a/fs/xfs/libxfs/xfs_attr_remote.c +++ b/fs/xfs/libxfs/xfs_attr_remote.c @@ -443,7 +443,7 @@ xfs_attr_rmtval_get( * Find a "hole" in the attribute address space large enough for us to drop the * new attribute's value into */ -STATIC int +int xfs_attr_rmt_find_hole( struct xfs_da_args *args) { @@ -470,7 +470,7 @@ xfs_attr_rmt_find_hole( return 0; } -STATIC int +int xfs_attr_rmtval_set_value( struct xfs_da_args *args) { @@ -630,6 +630,71 @@ xfs_attr_rmtval_set( } /* + * Find a hole for the attr and store it in the delayed attr context. This + * initializes the context to roll through allocating an attr extent for a + * delayed attr operation + */ +int +xfs_attr_rmtval_set_init( + struct xfs_delattr_context *dac) +{ + struct xfs_da_args *args = dac->da_args; + struct xfs_bmbt_irec *map = &dac->map; + int error; + + dac->lblkno = 0; + dac->lfileoff = 0; + dac->blkcnt = 0; + args->rmtblkcnt = 0; + args->rmtblkno = 0; + memset(map, 0, sizeof(struct xfs_bmbt_irec)); + + error = xfs_attr_rmt_find_hole(args); + if (error) + return error; + + dac->blkcnt = args->rmtblkcnt; + dac->lblkno = args->rmtblkno; + + return error; +} + +/* + * Write one block of the value associated with an attribute into the + * out-of-line buffer that we have defined for it. This is similar to a subset + * of xfs_attr_rmtval_set, but records the current block to the delayed attr + * context, and leaves transaction handling to the caller. + */ +int +xfs_attr_rmtval_set_blk( + struct xfs_delattr_context *dac) +{ + struct xfs_da_args *args = dac->da_args; + struct xfs_inode *dp = args->dp; + struct xfs_bmbt_irec *map = &dac->map; + int nmap; + int error; + + nmap = 1; + error = xfs_bmapi_write(args->trans, dp, + (xfs_fileoff_t)dac->lblkno, + dac->blkcnt, XFS_BMAPI_ATTRFORK, + args->total, map, &nmap); + if (error) + return error; + + ASSERT(nmap == 1); + ASSERT((map->br_startblock != DELAYSTARTBLOCK) && + (map->br_startblock != HOLESTARTBLOCK)); + + /* roll attribute extent map forwards */ + dac->lblkno += map->br_blockcount; + dac->blkcnt -= map->br_blockcount; + + return 0; +} + +/* * Remove the value associated with an attribute by deleting the * out-of-line buffer that it is stored on. */ @@ -671,48 +736,6 @@ xfs_attr_rmtval_invalidate( } /* - * Remove the value associated with an attribute by deleting the - * out-of-line buffer that it is stored on. - */ -int -xfs_attr_rmtval_remove( - struct xfs_da_args *args) -{ - xfs_dablk_t lblkno; - int blkcnt; - int error = 0; - int done = 0; - - trace_xfs_attr_rmtval_remove(args); - - error = xfs_attr_rmtval_invalidate(args); - if (error) - return error; - /* - * Keep de-allocating extents until the remote-value region is gone. - */ - lblkno = args->rmtblkno; - blkcnt = args->rmtblkcnt; - while (!done) { - error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt, - XFS_BMAPI_ATTRFORK, 1, &done); - if (error) - return error; - error = xfs_defer_finish(&args->trans); - if (error) - return error; - - /* - * Close out trans and start the next one in the chain. - */ - error = xfs_trans_roll_inode(&args->trans, args->dp); - if (error) - return error; - } - return 0; -} - -/* * Remove the value associated with an attribute by deleting the out-of-line * buffer that it is stored on. Returns EAGAIN for the caller to refresh the * transaction and recall the function diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h index ee3337b..482dff9 100644 --- a/fs/xfs/libxfs/xfs_attr_remote.h +++ b/fs/xfs/libxfs/xfs_attr_remote.h @@ -15,4 +15,8 @@ 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_rmt_find_hole(struct xfs_da_args *args); +int xfs_attr_rmtval_set_value(struct xfs_da_args *args); +int xfs_attr_rmtval_set_blk(struct xfs_delattr_context *dac); +int xfs_attr_rmtval_set_init(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 c42f90e..3e8cec5 100644 --- a/fs/xfs/xfs_attr_inactive.c +++ b/fs/xfs/xfs_attr_inactive.c @@ -15,6 +15,7 @@ #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" diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index a4323a6..26dc8bf 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -1784,7 +1784,6 @@ DEFINE_ATTR_EVENT(xfs_attr_refillstate); DEFINE_ATTR_EVENT(xfs_attr_rmtval_get); DEFINE_ATTR_EVENT(xfs_attr_rmtval_set); -DEFINE_ATTR_EVENT(xfs_attr_rmtval_remove); #define DEFINE_DA_EVENT(name) \ DEFINE_EVENT(xfs_da_class, name, \
This patch modifies the attr set 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_set_args has become xfs_attr_set_iter, which uses a state machine like switch to keep track of where it was when EAGAIN was returned. Two new helper functions have been added: xfs_attr_rmtval_set_init and xfs_attr_rmtval_set_blk. They provide a subset of logic similar to xfs_attr_rmtval_set, but they store the current block in the delay attr context to allow the caller to roll the transaction between allocations. This helps to simplify and consolidate code used by xfs_attr_leaf_addname and xfs_attr_node_addname. xfs_attr_set_args has now become a simple loop to refresh the transaction until the operation is completed. Lastly, xfs_attr_rmtval_remove is no longer used, and is removed. Below is a state machine diagram for attr set 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_set_iter() │ v need to upgrade from sf to leaf? ──n─┐ │ │ y │ │ │ V │ XFS_DAS_ADD_LEAF │ │ │ v │ ┌──────n── fork has <──────┘ │ only 1 blk? │ │ │ y │ │ │ v │ xfs_attr_leaf_try_add() │ │ │ v │ had enough ├──────n── space? │ │ │ y │ │ │ v │ XFS_DAS_FOUND_LBLK ──┐ │ │ │ XFS_DAS_FLIP_LFLAG ──┤ │ (subroutine state) │ │ │ │ XFS_DAS_ALLOC_LEAF ──┤ │ (subroutine state) │ │ └─>xfs_attr_leaf_addname() │ │ │ v │ ┌─────n── need to │ │ alloc blks? │ │ │ │ │ y │ │ │ │ │ v │ │ ┌─>XFS_DAS_ALLOC_LEAF │ │ │ │ │ │ │ v │ │ └──y── need to alloc │ │ more blocks? │ │ │ │ │ n │ │ │ │ │ v │ │ was this │ └────────> a rename? ──n─┐ │ │ │ │ y │ │ │ │ │ v │ │ flip incomplete │ │ flag │ │ │ │ │ v │ │ XFS_DAS_FLIP_LFLAG │ │ │ │ │ v │ │ remove │ │ XFS_DAS_RM_LBLK ─> old name │ │ ^ │ │ │ │ v │ │ └──────y── more to │ │ remove │ │ │ │ │ n │ │ │ │ │ v │ │ done <──────┘ └────> XFS_DAS_LEAF_TO_NODE ─┐ │ XFS_DAS_FOUND_NBLK ──┤ (subroutine state) │ │ XFS_DAS_ALLOC_NODE ──┤ (subroutine state) │ │ XFS_DAS_FLIP_NFLAG ──┤ (subroutine state) │ │ └─>xfs_attr_node_addname() │ v find space to store attr. Split if needed │ v XFS_DAS_FOUND_NBLK │ v ┌─────n── need to │ alloc blks? │ │ │ y │ │ │ v │ ┌─>XFS_DAS_ALLOC_NODE │ │ │ │ │ v │ └──y── need to alloc │ more blocks? │ │ │ n │ │ │ v │ was this └────────> a rename? ──n─┐ │ │ y │ │ │ v │ flip incomplete │ flag │ │ │ v │ XFS_DAS_FLIP_NFLAG │ │ │ v │ remove │ XFS_DAS_RM_NBLK ─> old name │ ^ │ │ │ v │ └──────y── more to │ remove │ │ │ n │ │ │ v │ done <──────┘ Signed-off-by: Allison Collins <allison.henderson@oracle.com> --- fs/xfs/libxfs/xfs_attr.c | 384 +++++++++++++++++++++++++++------------- fs/xfs/libxfs/xfs_attr.h | 16 ++ fs/xfs/libxfs/xfs_attr_leaf.c | 1 + fs/xfs/libxfs/xfs_attr_remote.c | 111 +++++++----- fs/xfs/libxfs/xfs_attr_remote.h | 4 + fs/xfs/xfs_attr_inactive.c | 1 + fs/xfs/xfs_trace.h | 1 - 7 files changed, 351 insertions(+), 167 deletions(-)