Message ID | 20210218165348.4754-12-allison.henderson@oracle.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | xfs: Delayed Attributes | expand |
On Thu, Feb 18, 2021 at 09:53:37AM -0700, Allison Henderson wrote: > This patch modifies the attr remove routines to be delay ready. This > means they no longer roll or commit transactions, but instead return > -EAGAIN to have the calling routine roll and refresh the transaction. In > this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which > uses a sort of state machine like switch to keep track of where it was > when EAGAIN was returned. xfs_attr_node_removename has also been > modified to use the switch, and a new version of xfs_attr_remove_args > consists of a simple loop to refresh the transaction until the operation > is completed. A new XFS_DAC_DEFER_FINISH flag is used to finish the > transaction where ever the existing code used to. > > Calls to xfs_attr_rmtval_remove are replaced with the delay ready > version __xfs_attr_rmtval_remove. We will rename > __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are > done. > > xfs_attr_rmtval_remove itself is still in use by the set routines (used > during a rename). For reasons of preserving existing function, we > modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is > set. Similar to how xfs_attr_remove_args does here. Once we transition > the set routines to be delay ready, xfs_attr_rmtval_remove is no longer > used and will be removed. > > This patch also adds a new struct xfs_delattr_context, which we will use > to keep track of the current state of an attribute operation. The new > xfs_delattr_state enum is used to track various operations that are in > progress so that we know not to repeat them, and resume where we left > off before EAGAIN was returned to cycle out the transaction. Other > members take the place of local variables that need to retain their > values across multiple function recalls. See xfs_attr.h for a more > detailed diagram of the states. > > Signed-off-by: Allison Henderson <allison.henderson@oracle.com> > --- > fs/xfs/libxfs/xfs_attr.c | 223 +++++++++++++++++++++++++++++----------- > fs/xfs/libxfs/xfs_attr.h | 100 ++++++++++++++++++ > fs/xfs/libxfs/xfs_attr_leaf.c | 2 +- > fs/xfs/libxfs/xfs_attr_remote.c | 48 +++++---- > fs/xfs/libxfs/xfs_attr_remote.h | 2 +- > fs/xfs/xfs_attr_inactive.c | 2 +- > 6 files changed, 294 insertions(+), 83 deletions(-) > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > index 56d4b56..d46b92a 100644 > --- a/fs/xfs/libxfs/xfs_attr.c > +++ b/fs/xfs/libxfs/xfs_attr.c ... > @@ -221,6 +221,34 @@ xfs_attr_is_shortform( > ip->i_afp->if_nextents == 0); > } > > +/* > + * Checks to see if a delayed attribute transaction should be rolled. If so, > + * also checks for a defer finish. Transaction is finished and rolled as > + * needed, and returns true of false if the delayed operation should continue. > + */ > +int > +xfs_attr_trans_roll( > + struct xfs_delattr_context *dac) > +{ > + struct xfs_da_args *args = dac->da_args; > + int error; > + > + if (dac->flags & XFS_DAC_DEFER_FINISH) { > + /* > + * The caller wants us to finish all the deferred ops so that we > + * avoid pinning the log tail with a large number of deferred > + * ops. > + */ > + dac->flags &= ~XFS_DAC_DEFER_FINISH; > + error = xfs_defer_finish(&args->trans); > + if (error) > + return error; No need for the error check here. > + } else > + error = xfs_trans_roll_inode(&args->trans, args->dp); > + > + return error; > +} > + > STATIC int > xfs_attr_set_fmt( > struct xfs_da_args *args) > @@ -531,23 +559,58 @@ xfs_has_attr( > */ > int > xfs_attr_remove_args( > - struct xfs_da_args *args) > + struct xfs_da_args *args) > { > - struct xfs_inode *dp = args->dp; > - int error; > + int error; > + struct xfs_delattr_context dac = { > + .da_args = args, > + }; > + > + do { > + error = xfs_attr_remove_iter(&dac); > + if (error != -EAGAIN) > + break; > + > + error = xfs_attr_trans_roll(&dac); > + if (error) > + return error; > + > + } while (true); > + > + return error; > +} > + > +/* > + * Remove the attribute specified in @args. > + * > + * This function may return -EAGAIN to signal that the transaction needs to be > + * rolled. Callers should continue calling this function until they receive a > + * return value other than -EAGAIN. > + */ > +int > +xfs_attr_remove_iter( > + struct xfs_delattr_context *dac) > +{ > + struct xfs_da_args *args = dac->da_args; > + struct xfs_inode *dp = args->dp; > > - if (!xfs_inode_hasattr(dp)) { > - error = -ENOATTR; > - } else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) { > + /* If we are shrinking a node, resume shrink */ > + if (dac->dela_state == XFS_DAS_RM_SHRINK) > + goto node; > + > + if (!xfs_inode_hasattr(dp)) > + return -ENOATTR; > + > + if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) { > ASSERT(dp->i_afp->if_flags & XFS_IFINLINE); > - error = xfs_attr_shortform_remove(args); > - } else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { > - error = xfs_attr_leaf_removename(args); > - } else { > - error = xfs_attr_node_removename(args); > + return xfs_attr_shortform_remove(args); > } > > - return error; > + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) > + return xfs_attr_leaf_removename(args); > +node: > + /* If we are not short form or leaf, then proceed to remove node */ > + return xfs_attr_node_removename_iter(dac); Extra whitespace before the function name; Also, can we lift xfs_attr_node_removename_iter() into this function, form the current implementation of xfs_attr_remove_iter() into the UNINIT state, and move the current UNINIT state into a new RMT_BLK state to support reentry? ISTM that would condense everything to a single switch statement that can live inside xfs_attr_remove_iter(). IOW, we can kill off the 'node:' level and multi-layer state management here. Hm? > } > > /* ... > @@ -1207,22 +1272,28 @@ int xfs_attr_node_removename_setup( > if (args->rmtblkno > 0) { > error = xfs_attr_leaf_mark_incomplete(args, *state); > if (error) > - return error; > + goto out; > > - return xfs_attr_rmtval_invalidate(args); > + error = xfs_attr_rmtval_invalidate(args); > } > +out: > + if (error) > + xfs_da_state_free(*state); > > return 0; > } > > STATIC int > -xfs_attr_node_remove_rmt( > - struct xfs_da_args *args, > - struct xfs_da_state *state) > +xfs_attr_node_remove_rmt ( Extra whitespace before the opening brace. > + struct xfs_delattr_context *dac, > + struct xfs_da_state *state) > { > - int error = 0; > + int error = 0; > > - error = xfs_attr_rmtval_remove(args); > + /* > + * May return -EAGAIN to request that the caller recall this function > + */ > + error = __xfs_attr_rmtval_remove(dac); > if (error) > return error; > ... > @@ -1285,51 +1365,74 @@ xfs_attr_node_remove_step( > * > * This routine will find the blocks of the name to remove, remove them and > * shrink the tree if needed. > + * > + * This routine is meant to function as either an inline or delayed operation, > + * and may return -EAGAIN when the transaction needs to be rolled. Calling > + * functions will need to handle this, and recall the function until a > + * successful error code is returned. > */ > STATIC int > -xfs_attr_node_removename( > - struct xfs_da_args *args) > +xfs_attr_node_removename_iter( > + struct xfs_delattr_context *dac) > { > - struct xfs_da_state *state = NULL; > - int retval, error; > - struct xfs_inode *dp = args->dp; > + struct xfs_da_args *args = dac->da_args; > + struct xfs_da_state *state = NULL; > + int retval, error; > + struct xfs_inode *dp = args->dp; > > trace_xfs_attr_node_removename(args); > > - error = xfs_attr_node_removename_setup(args, &state); > - if (error) > - goto out; > - > - error = xfs_attr_node_remove_step(args, state); > - if (error) > - goto out; > - > - retval = xfs_attr_node_remove_cleanup(args, state); > - > - /* > - * Check to see if the tree needs to be collapsed. > - */ > - if (retval && (state->path.active > 1)) { > - error = xfs_da3_join(state); > - if (error) > - goto out; > - error = xfs_defer_finish(&args->trans); > + if (!dac->da_state) { > + error = xfs_attr_node_removename_setup(dac); > if (error) > goto out; > + } > + state = dac->da_state; > + > + switch (dac->dela_state) { > + case XFS_DAS_UNINIT: > /* > - * Commit the Btree join operation and start a new trans. > + * repeatedly remove remote blocks, remove the entry and join. > + * returns -EAGAIN or 0 for completion of the step. > */ > - error = xfs_trans_roll_inode(&args->trans, dp); > + error = xfs_attr_node_remove_step(dac); > if (error) > - goto out; > - } > + break; Hmm.. so re: my comment further down on xfs_attr_rmtval_remove(), wouldn't that change semantics here? I.e., once remote blocks are removed this would previously carry on with a clean transaction. Now it looks like we'd carry on with the dirty transaction that removed the last remote extent. This suggests that perhaps we should return once more and fall into a new state to remove the name..? Also, it would be nice to remove the several seemingly unnecessary layers of indirection here. For example, something like the following (also considering my comment above wrt to xfs_attr_remove_iter() and UNINIT): case UNINIT: ... /* fallthrough */ case RMTBLK: if (args->rmtblkno > 0) { dac->dela_state = RMTBLK; error = __xfs_attr_rmtval_remove(dac); if (error) break; ASSERT(args->rmtblkno == 0); xfs_attr_refillstate(state); dac->flags |= XFS_DAC_DEFER_FINISH; dac->dela_state = RMNAME; return -EAGAIN; } /* fallthrough */ case RMNAME: ... ... > > - /* > - * If the result is small enough, push it all into the inode. > - */ > - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) > - error = xfs_attr_node_shrink(args, state); > + retval = xfs_attr_node_remove_cleanup(args, state); > ... > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c > index 48d8e9c..f09820c 100644 > --- a/fs/xfs/libxfs/xfs_attr_remote.c > +++ b/fs/xfs/libxfs/xfs_attr_remote.c ... > @@ -685,31 +687,29 @@ xfs_attr_rmtval_remove( > * Keep de-allocating extents until the remote-value region is gone. > */ > do { > - retval = __xfs_attr_rmtval_remove(args); > - if (retval && retval != -EAGAIN) > - return retval; > + error = __xfs_attr_rmtval_remove(&dac); > + if (error != -EAGAIN) > + break; Previously this would roll once and exit the loop on retval == 0. Now it looks like we break out of the loop immediately. Why the change? Brian > > - /* > - * Close out trans and start the next one in the chain. > - */ > - error = xfs_trans_roll_inode(&args->trans, args->dp); > + error = xfs_attr_trans_roll(&dac); > if (error) > return error; > - } while (retval == -EAGAIN); > + } while (true); > > - return 0; > + return error; > } > > /* > * 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 > + * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the > * transaction and re-call the function > */ > int > __xfs_attr_rmtval_remove( > - struct xfs_da_args *args) > + struct xfs_delattr_context *dac) > { > - int error, done; > + struct xfs_da_args *args = dac->da_args; > + int error, done; > > /* > * Unmap value blocks for this attr. > @@ -719,12 +719,20 @@ __xfs_attr_rmtval_remove( > if (error) > return error; > > - error = xfs_defer_finish(&args->trans); > - if (error) > - return error; > - > - if (!done) > + /* > + * We dont need an explicit state here to pick up where we left off. We > + * can figure it out using the !done return code. Calling function only > + * needs to keep recalling this routine until we indicate to stop by > + * returning anything other than -EAGAIN. The actual value of > + * attr->xattri_dela_state may be some value reminicent of the calling > + * function, but it's value is irrelevant with in the context of this > + * function. Once we are done here, the next state is set as needed > + * by the parent > + */ > + if (!done) { > + dac->flags |= XFS_DAC_DEFER_FINISH; > return -EAGAIN; > + } > > return error; > } > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h > index 9eee615..002fd30 100644 > --- a/fs/xfs/libxfs/xfs_attr_remote.h > +++ b/fs/xfs/libxfs/xfs_attr_remote.h > @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args); > int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map, > xfs_buf_flags_t incore_flags); > int xfs_attr_rmtval_invalidate(struct xfs_da_args *args); > -int __xfs_attr_rmtval_remove(struct xfs_da_args *args); > +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac); > #endif /* __XFS_ATTR_REMOTE_H__ */ > diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c > index bfad669..aaa7e66 100644 > --- a/fs/xfs/xfs_attr_inactive.c > +++ b/fs/xfs/xfs_attr_inactive.c > @@ -15,10 +15,10 @@ > #include "xfs_da_format.h" > #include "xfs_da_btree.h" > #include "xfs_inode.h" > +#include "xfs_attr.h" > #include "xfs_attr_remote.h" > #include "xfs_trans.h" > #include "xfs_bmap.h" > -#include "xfs_attr.h" > #include "xfs_attr_leaf.h" > #include "xfs_quota.h" > #include "xfs_dir2.h" > -- > 2.7.4 >
On 2/24/21 11:45 AM, Brian Foster wrote: > On Thu, Feb 18, 2021 at 09:53:37AM -0700, Allison Henderson wrote: >> This patch modifies the attr remove routines to be delay ready. This >> means they no longer roll or commit transactions, but instead return >> -EAGAIN to have the calling routine roll and refresh the transaction. In >> this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which >> uses a sort of state machine like switch to keep track of where it was >> when EAGAIN was returned. xfs_attr_node_removename has also been >> modified to use the switch, and a new version of xfs_attr_remove_args >> consists of a simple loop to refresh the transaction until the operation >> is completed. A new XFS_DAC_DEFER_FINISH flag is used to finish the >> transaction where ever the existing code used to. >> >> Calls to xfs_attr_rmtval_remove are replaced with the delay ready >> version __xfs_attr_rmtval_remove. We will rename >> __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are >> done. >> >> xfs_attr_rmtval_remove itself is still in use by the set routines (used >> during a rename). For reasons of preserving existing function, we >> modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is >> set. Similar to how xfs_attr_remove_args does here. Once we transition >> the set routines to be delay ready, xfs_attr_rmtval_remove is no longer >> used and will be removed. >> >> This patch also adds a new struct xfs_delattr_context, which we will use >> to keep track of the current state of an attribute operation. The new >> xfs_delattr_state enum is used to track various operations that are in >> progress so that we know not to repeat them, and resume where we left >> off before EAGAIN was returned to cycle out the transaction. Other >> members take the place of local variables that need to retain their >> values across multiple function recalls. See xfs_attr.h for a more >> detailed diagram of the states. >> >> Signed-off-by: Allison Henderson <allison.henderson@oracle.com> >> --- >> fs/xfs/libxfs/xfs_attr.c | 223 +++++++++++++++++++++++++++++----------- >> fs/xfs/libxfs/xfs_attr.h | 100 ++++++++++++++++++ >> fs/xfs/libxfs/xfs_attr_leaf.c | 2 +- >> fs/xfs/libxfs/xfs_attr_remote.c | 48 +++++---- >> fs/xfs/libxfs/xfs_attr_remote.h | 2 +- >> fs/xfs/xfs_attr_inactive.c | 2 +- >> 6 files changed, 294 insertions(+), 83 deletions(-) >> >> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c >> index 56d4b56..d46b92a 100644 >> --- a/fs/xfs/libxfs/xfs_attr.c >> +++ b/fs/xfs/libxfs/xfs_attr.c > ... >> @@ -221,6 +221,34 @@ xfs_attr_is_shortform( >> ip->i_afp->if_nextents == 0); >> } >> >> +/* >> + * Checks to see if a delayed attribute transaction should be rolled. If so, >> + * also checks for a defer finish. Transaction is finished and rolled as >> + * needed, and returns true of false if the delayed operation should continue. >> + */ >> +int >> +xfs_attr_trans_roll( >> + struct xfs_delattr_context *dac) >> +{ >> + struct xfs_da_args *args = dac->da_args; >> + int error; >> + >> + if (dac->flags & XFS_DAC_DEFER_FINISH) { >> + /* >> + * The caller wants us to finish all the deferred ops so that we >> + * avoid pinning the log tail with a large number of deferred >> + * ops. >> + */ >> + dac->flags &= ~XFS_DAC_DEFER_FINISH; >> + error = xfs_defer_finish(&args->trans); >> + if (error) >> + return error; > > No need for the error check here. Sure, will clean up > >> + } else >> + error = xfs_trans_roll_inode(&args->trans, args->dp); >> + >> + return error; >> +} >> + >> STATIC int >> xfs_attr_set_fmt( >> struct xfs_da_args *args) >> @@ -531,23 +559,58 @@ xfs_has_attr( >> */ >> int >> xfs_attr_remove_args( >> - struct xfs_da_args *args) >> + struct xfs_da_args *args) >> { >> - struct xfs_inode *dp = args->dp; >> - int error; >> + int error; >> + struct xfs_delattr_context dac = { >> + .da_args = args, >> + }; >> + >> + do { >> + error = xfs_attr_remove_iter(&dac); >> + if (error != -EAGAIN) >> + break; >> + >> + error = xfs_attr_trans_roll(&dac); >> + if (error) >> + return error; >> + >> + } while (true); >> + >> + return error; >> +} >> + >> +/* >> + * Remove the attribute specified in @args. >> + * >> + * This function may return -EAGAIN to signal that the transaction needs to be >> + * rolled. Callers should continue calling this function until they receive a >> + * return value other than -EAGAIN. >> + */ >> +int >> +xfs_attr_remove_iter( >> + struct xfs_delattr_context *dac) >> +{ >> + struct xfs_da_args *args = dac->da_args; >> + struct xfs_inode *dp = args->dp; >> >> - if (!xfs_inode_hasattr(dp)) { >> - error = -ENOATTR; >> - } else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) { >> + /* If we are shrinking a node, resume shrink */ >> + if (dac->dela_state == XFS_DAS_RM_SHRINK) >> + goto node; >> + >> + if (!xfs_inode_hasattr(dp)) >> + return -ENOATTR; >> + >> + if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) { >> ASSERT(dp->i_afp->if_flags & XFS_IFINLINE); >> - error = xfs_attr_shortform_remove(args); >> - } else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { >> - error = xfs_attr_leaf_removename(args); >> - } else { >> - error = xfs_attr_node_removename(args); >> + return xfs_attr_shortform_remove(args); >> } >> >> - return error; >> + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) >> + return xfs_attr_leaf_removename(args); >> +node: >> + /* If we are not short form or leaf, then proceed to remove node */ >> + return xfs_attr_node_removename_iter(dac); > > Extra whitespace before the function name; Ok, will fix > > Also, can we lift xfs_attr_node_removename_iter() into this function, > form the current implementation of xfs_attr_remove_iter() into the > UNINIT state, and move the current UNINIT state into a new RMT_BLK state > to support reentry? ISTM that would condense everything to a single > switch statement that can live inside xfs_attr_remove_iter(). IOW, we > can kill off the 'node:' level and multi-layer state management here. > Hm? Sure, I think the lift should be fine, though I dont think we'll need the extra state. See below.... > >> } >> >> /* > ... >> @@ -1207,22 +1272,28 @@ int xfs_attr_node_removename_setup( >> if (args->rmtblkno > 0) { >> error = xfs_attr_leaf_mark_incomplete(args, *state); >> if (error) >> - return error; >> + goto out; >> >> - return xfs_attr_rmtval_invalidate(args); >> + error = xfs_attr_rmtval_invalidate(args); >> } >> +out: >> + if (error) >> + xfs_da_state_free(*state); >> >> return 0; >> } >> >> STATIC int >> -xfs_attr_node_remove_rmt( >> - struct xfs_da_args *args, >> - struct xfs_da_state *state) >> +xfs_attr_node_remove_rmt ( > > Extra whitespace before the opening brace. ok, will fix > >> + struct xfs_delattr_context *dac, >> + struct xfs_da_state *state) >> { >> - int error = 0; >> + int error = 0; >> >> - error = xfs_attr_rmtval_remove(args); >> + /* >> + * May return -EAGAIN to request that the caller recall this function >> + */ >> + error = __xfs_attr_rmtval_remove(dac); >> if (error) >> return error; >> > ... >> @@ -1285,51 +1365,74 @@ xfs_attr_node_remove_step( >> * >> * This routine will find the blocks of the name to remove, remove them and >> * shrink the tree if needed. >> + * >> + * This routine is meant to function as either an inline or delayed operation, >> + * and may return -EAGAIN when the transaction needs to be rolled. Calling >> + * functions will need to handle this, and recall the function until a >> + * successful error code is returned. >> */ >> STATIC int >> -xfs_attr_node_removename( >> - struct xfs_da_args *args) >> +xfs_attr_node_removename_iter( >> + struct xfs_delattr_context *dac) >> { >> - struct xfs_da_state *state = NULL; >> - int retval, error; >> - struct xfs_inode *dp = args->dp; >> + struct xfs_da_args *args = dac->da_args; >> + struct xfs_da_state *state = NULL; >> + int retval, error; >> + struct xfs_inode *dp = args->dp; >> >> trace_xfs_attr_node_removename(args); >> >> - error = xfs_attr_node_removename_setup(args, &state); >> - if (error) >> - goto out; >> - >> - error = xfs_attr_node_remove_step(args, state); >> - if (error) >> - goto out; >> - >> - retval = xfs_attr_node_remove_cleanup(args, state); >> - >> - /* >> - * Check to see if the tree needs to be collapsed. >> - */ >> - if (retval && (state->path.active > 1)) { >> - error = xfs_da3_join(state); >> - if (error) >> - goto out; >> - error = xfs_defer_finish(&args->trans); >> + if (!dac->da_state) { >> + error = xfs_attr_node_removename_setup(dac); >> if (error) >> goto out; >> + } >> + state = dac->da_state; >> + >> + switch (dac->dela_state) { >> + case XFS_DAS_UNINIT: >> /* >> - * Commit the Btree join operation and start a new trans. >> + * repeatedly remove remote blocks, remove the entry and join. >> + * returns -EAGAIN or 0 for completion of the step. >> */ >> - error = xfs_trans_roll_inode(&args->trans, dp); >> + error = xfs_attr_node_remove_step(dac); >> if (error) >> - goto out; >> - } >> + break; > > Hmm.. so re: my comment further down on xfs_attr_rmtval_remove(), > wouldn't that change semantics here? I.e., once remote blocks are > removed this would previously carry on with a clean transaction. Now it > looks like we'd carry on with the dirty transaction that removed the > last remote extent. This suggests that perhaps we should return once > more and fall into a new state to remove the name..? I suspect the diff might be making this a bit difficult to see. The roll that you see being removed here belongs to the transaction we hoisted up in patch 3 which happens after the clean up below, and we have the corresponding EAGAIN fot that one. I think the diff gets things a little interlaced here because the switch adds another level of indentation. some times i do like to I use a graphical diffviewer like diffuse when patches get weird like this. Something like this: git config --global diff.tool diffuse git difftool 3c53e49 e201c09 You'd need to download the branch and also the diffuse tool, but sometimes i think it makes some of these diffs a bit easier to see Also, it would be > nice to remove the several seemingly unnecessary layers of indirection > here. For example, something like the following (also considering my > comment above wrt to xfs_attr_remove_iter() and UNINIT): > > case UNINIT: > ... > /* fallthrough */ > case RMTBLK: > if (args->rmtblkno > 0) { > dac->dela_state = RMTBLK; > error = __xfs_attr_rmtval_remove(dac); > if (error) > break; > > ASSERT(args->rmtblkno == 0); > xfs_attr_refillstate(state); > dac->flags |= XFS_DAC_DEFER_FINISH; > dac->dela_state = RMNAME; > return -EAGAIN; > } Ok, this looks to me like we've hoisted both xfs_attr_node_remove_rmt and xfs_attr_node_remove_step into this scope, but I still think this adds an extra roll where non previously was. With out that extra EAGAIN, I think we are fine to have all that just under the UNINIT case. I also think it's also worth noteing here that this is kind of a reverse of patch 1, which I think we put in for reasons of trying to modularize the higher level functions as much as possible. I suspect some of where you were going with this may have been influenced by the earlier diff confusion too. Maybe take a second look there before we go too much down this change.... > /* fallthrough */ > case RMNAME: > ... > ... > >> >> - /* >> - * If the result is small enough, push it all into the inode. >> - */ >> - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) >> - error = xfs_attr_node_shrink(args, state); >> + retval = xfs_attr_node_remove_cleanup(args, state); >> > ... I think the overlooked EAGAIN was in this area that got clipped out..... >> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c >> index 48d8e9c..f09820c 100644 >> --- a/fs/xfs/libxfs/xfs_attr_remote.c >> +++ b/fs/xfs/libxfs/xfs_attr_remote.c > ... >> @@ -685,31 +687,29 @@ c( >> * Keep de-allocating extents until the remote-value region is gone. >> */ >> do { >> - retval = __xfs_attr_rmtval_remove(args); >> - if (retval && retval != -EAGAIN) >> - return retval; >> + error = __xfs_attr_rmtval_remove(&dac); >> + if (error != -EAGAIN) >> + break; > > Previously this would roll once and exit the loop on retval == 0. Now it > looks like we break out of the loop immediately. Why the change? Gosh, I think sometime in reviewing v9, we had come up with a "xfs_attr_roll_again" helper that took the error code as a paramater and decided whether or not to roll. And then in v10 i think people thought that was weird and we turned it into xfs_attr_trans_roll. I think I likley forgot to restore the orginal retval handling here. This whole function disappears in the next patch, but the original error handling should be restored to keep things consistent. Thx for the catch! Thx for the reviews!! I know it's complicated! I've chased my tail many times with it myself :-) Allison > > Brian > >> >> - /* >> - * Close out trans and start the next one in the chain. >> - */ >> - error = xfs_trans_roll_inode(&args->trans, args->dp); >> + error = xfs_attr_trans_roll(&dac); >> if (error) >> return error; >> - } while (retval == -EAGAIN); >> + } while (true); >> >> - return 0; >> + return error; >> } >> >> /* >> * 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 >> + * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the >> * transaction and re-call the function >> */ >> int >> __xfs_attr_rmtval_remove( >> - struct xfs_da_args *args) >> + struct xfs_delattr_context *dac) >> { >> - int error, done; >> + struct xfs_da_args *args = dac->da_args; >> + int error, done; >> >> /* >> * Unmap value blocks for this attr. >> @@ -719,12 +719,20 @@ __xfs_attr_rmtval_remove( >> if (error) >> return error; >> >> - error = xfs_defer_finish(&args->trans); >> - if (error) >> - return error; >> - >> - if (!done) >> + /* >> + * We dont need an explicit state here to pick up where we left off. We >> + * can figure it out using the !done return code. Calling function only >> + * needs to keep recalling this routine until we indicate to stop by >> + * returning anything other than -EAGAIN. The actual value of >> + * attr->xattri_dela_state may be some value reminicent of the calling >> + * function, but it's value is irrelevant with in the context of this >> + * function. Once we are done here, the next state is set as needed >> + * by the parent >> + */ >> + if (!done) { >> + dac->flags |= XFS_DAC_DEFER_FINISH; >> return -EAGAIN; >> + } >> >> return error; >> } >> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h >> index 9eee615..002fd30 100644 >> --- a/fs/xfs/libxfs/xfs_attr_remote.h >> +++ b/fs/xfs/libxfs/xfs_attr_remote.h >> @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args); >> int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map, >> xfs_buf_flags_t incore_flags); >> int xfs_attr_rmtval_invalidate(struct xfs_da_args *args); >> -int __xfs_attr_rmtval_remove(struct xfs_da_args *args); >> +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac); >> #endif /* __XFS_ATTR_REMOTE_H__ */ >> diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c >> index bfad669..aaa7e66 100644 >> --- a/fs/xfs/xfs_attr_inactive.c >> +++ b/fs/xfs/xfs_attr_inactive.c >> @@ -15,10 +15,10 @@ >> #include "xfs_da_format.h" >> #include "xfs_da_btree.h" >> #include "xfs_inode.h" >> +#include "xfs_attr.h" >> #include "xfs_attr_remote.h" >> #include "xfs_trans.h" >> #include "xfs_bmap.h" >> -#include "xfs_attr.h" >> #include "xfs_attr_leaf.h" >> #include "xfs_quota.h" >> #include "xfs_dir2.h" >> -- >> 2.7.4 >> >
On Thu, Feb 25, 2021 at 12:01:10AM -0700, Allison Henderson wrote: > > > On 2/24/21 11:45 AM, Brian Foster wrote: > > On Thu, Feb 18, 2021 at 09:53:37AM -0700, Allison Henderson wrote: > > > This patch modifies the attr remove routines to be delay ready. This > > > means they no longer roll or commit transactions, but instead return > > > -EAGAIN to have the calling routine roll and refresh the transaction. In > > > this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which > > > uses a sort of state machine like switch to keep track of where it was > > > when EAGAIN was returned. xfs_attr_node_removename has also been > > > modified to use the switch, and a new version of xfs_attr_remove_args > > > consists of a simple loop to refresh the transaction until the operation > > > is completed. A new XFS_DAC_DEFER_FINISH flag is used to finish the > > > transaction where ever the existing code used to. > > > > > > Calls to xfs_attr_rmtval_remove are replaced with the delay ready > > > version __xfs_attr_rmtval_remove. We will rename > > > __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are > > > done. > > > > > > xfs_attr_rmtval_remove itself is still in use by the set routines (used > > > during a rename). For reasons of preserving existing function, we > > > modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is > > > set. Similar to how xfs_attr_remove_args does here. Once we transition > > > the set routines to be delay ready, xfs_attr_rmtval_remove is no longer > > > used and will be removed. > > > > > > This patch also adds a new struct xfs_delattr_context, which we will use > > > to keep track of the current state of an attribute operation. The new > > > xfs_delattr_state enum is used to track various operations that are in > > > progress so that we know not to repeat them, and resume where we left > > > off before EAGAIN was returned to cycle out the transaction. Other > > > members take the place of local variables that need to retain their > > > values across multiple function recalls. See xfs_attr.h for a more > > > detailed diagram of the states. > > > > > > Signed-off-by: Allison Henderson <allison.henderson@oracle.com> > > > --- > > > fs/xfs/libxfs/xfs_attr.c | 223 +++++++++++++++++++++++++++++----------- > > > fs/xfs/libxfs/xfs_attr.h | 100 ++++++++++++++++++ > > > fs/xfs/libxfs/xfs_attr_leaf.c | 2 +- > > > fs/xfs/libxfs/xfs_attr_remote.c | 48 +++++---- > > > fs/xfs/libxfs/xfs_attr_remote.h | 2 +- > > > fs/xfs/xfs_attr_inactive.c | 2 +- > > > 6 files changed, 294 insertions(+), 83 deletions(-) > > > > > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > > > index 56d4b56..d46b92a 100644 > > > --- a/fs/xfs/libxfs/xfs_attr.c > > > +++ b/fs/xfs/libxfs/xfs_attr.c ... > > > @@ -1285,51 +1365,74 @@ xfs_attr_node_remove_step( > > > * > > > * This routine will find the blocks of the name to remove, remove them and > > > * shrink the tree if needed. > > > + * > > > + * This routine is meant to function as either an inline or delayed operation, > > > + * and may return -EAGAIN when the transaction needs to be rolled. Calling > > > + * functions will need to handle this, and recall the function until a > > > + * successful error code is returned. > > > */ > > > STATIC int > > > -xfs_attr_node_removename( > > > - struct xfs_da_args *args) > > > +xfs_attr_node_removename_iter( > > > + struct xfs_delattr_context *dac) > > > { > > > - struct xfs_da_state *state = NULL; > > > - int retval, error; > > > - struct xfs_inode *dp = args->dp; > > > + struct xfs_da_args *args = dac->da_args; > > > + struct xfs_da_state *state = NULL; > > > + int retval, error; > > > + struct xfs_inode *dp = args->dp; > > > trace_xfs_attr_node_removename(args); > > > - error = xfs_attr_node_removename_setup(args, &state); > > > - if (error) > > > - goto out; > > > - > > > - error = xfs_attr_node_remove_step(args, state); > > > - if (error) > > > - goto out; > > > - > > > - retval = xfs_attr_node_remove_cleanup(args, state); > > > - > > > - /* > > > - * Check to see if the tree needs to be collapsed. > > > - */ > > > - if (retval && (state->path.active > 1)) { > > > - error = xfs_da3_join(state); > > > - if (error) > > > - goto out; > > > - error = xfs_defer_finish(&args->trans); > > > + if (!dac->da_state) { > > > + error = xfs_attr_node_removename_setup(dac); > > > if (error) > > > goto out; > > > + } > > > + state = dac->da_state; > > > + > > > + switch (dac->dela_state) { > > > + case XFS_DAS_UNINIT: > > > /* > > > - * Commit the Btree join operation and start a new trans. > > > + * repeatedly remove remote blocks, remove the entry and join. > > > + * returns -EAGAIN or 0 for completion of the step. > > > */ > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > + error = xfs_attr_node_remove_step(dac); > > > if (error) > > > - goto out; > > > - } > > > + break; > > > > Hmm.. so re: my comment further down on xfs_attr_rmtval_remove(), > > wouldn't that change semantics here? I.e., once remote blocks are > > removed this would previously carry on with a clean transaction. Now it > > looks like we'd carry on with the dirty transaction that removed the > > last remote extent. This suggests that perhaps we should return once > > more and fall into a new state to remove the name..? > I suspect the diff might be making this a bit difficult to see. The roll > that you see being removed here belongs to the transaction we hoisted up in > patch 3 which happens after the clean up below, and we have the > corresponding EAGAIN fot that one. I think the diff gets things a little > interlaced here because the switch adds another level of indentation. > Hmm.. the roll in patch 3 appears to be related to the _cleanup() helper. What I'm referring to here is the state of the transaction after the final remote block is removed from the attr. I'm not sure we're talking about the same thing here.. > some times i do like to I use a graphical diffviewer like diffuse when > patches get weird like this. Something like this: > > git config --global diff.tool diffuse > git difftool 3c53e49 e201c09 > > You'd need to download the branch and also the diffuse tool, but sometimes i > think it makes some of these diffs a bit easier to see > I think it's easier just to refer to the code directly. The current upstream code flows down into: ... xfs_attr_node_removename() xfs_attr_node_remove_rmt() xfs_attr_rmtval_remove() ... which then implements the following loop: do { retval = __xfs_attr_rmtval_remove(args); if (retval && retval != -EAGAIN) return retval; /* * Close out trans and start the next one in the chain. */ error = xfs_trans_roll_inode(&args->trans, args->dp); if (error) return error; } while (retval == -EAGAIN); This rolls the transaction when retval == -EAGAIN or retval == 0, thus always returns with a clean transaction after the remote block removal completes. The code as of this patch does: ... xfs_attr_node_removename_iter() xfs_attr_node_remove_step() xfs_attr_node_remove_rmt() __xfs_attr_rmtval_remove() ... which either returns -EAGAIN (since the roll is now implemented at the very top) or 0 when done == true. The transaction might be dirty in the latter case, but xfs_attr_node_removename_iter() moves right on to xfs_attr_node_remove_cleanup() which can now do more work in that same transaction. Am I following that correctly? > Also, it would be > > nice to remove the several seemingly unnecessary layers of indirection > > here. For example, something like the following (also considering my > > comment above wrt to xfs_attr_remove_iter() and UNINIT): > > > > case UNINIT: > > ... > > /* fallthrough */ > > case RMTBLK: > > if (args->rmtblkno > 0) { > > dac->dela_state = RMTBLK; > > error = __xfs_attr_rmtval_remove(dac); > > if (error) > > break; > > > > ASSERT(args->rmtblkno == 0); > > xfs_attr_refillstate(state); > > dac->flags |= XFS_DAC_DEFER_FINISH; > > dac->dela_state = RMNAME; > > return -EAGAIN; > > } > Ok, this looks to me like we've hoisted both xfs_attr_node_remove_rmt and > xfs_attr_node_remove_step into this scope, but I still think this adds an > extra roll where non previously was. With out that extra EAGAIN, I think we > are fine to have all that just under the UNINIT case. I also think it's > also worth noteing here that this is kind of a reverse of patch 1, which I > think we put in for reasons of trying to modularize the higher level > functions as much as possible. > > I suspect some of where you were going with this may have been influenced by > the earlier diff confusion too. Maybe take a second look there before we go > too much down this change.... > I can certainly be getting lost somewhere in all the refactoring. If so, can you point out where in the flow described above? Brian > > > /* fallthrough */ > > case RMNAME: > > ... > > ... > > > > > - /* > > > - * If the result is small enough, push it all into the inode. > > > - */ > > > - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) > > > - error = xfs_attr_node_shrink(args, state); > > > + retval = xfs_attr_node_remove_cleanup(args, state); > > ... > I think the overlooked EAGAIN was in this area that got clipped out..... > > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c > > > index 48d8e9c..f09820c 100644 > > > --- a/fs/xfs/libxfs/xfs_attr_remote.c > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.c > > ... > > > @@ -685,31 +687,29 @@ c( > > > * Keep de-allocating extents until the remote-value region is gone. > > > */ > > > do { > > > - retval = __xfs_attr_rmtval_remove(args); > > > - if (retval && retval != -EAGAIN) > > > - return retval; > > > + error = __xfs_attr_rmtval_remove(&dac); > > > + if (error != -EAGAIN) > > > + break; > > > > Previously this would roll once and exit the loop on retval == 0. Now it > > looks like we break out of the loop immediately. Why the change? > > Gosh, I think sometime in reviewing v9, we had come up with a > "xfs_attr_roll_again" helper that took the error code as a paramater and > decided whether or not to roll. And then in v10 i think people thought that > was weird and we turned it into xfs_attr_trans_roll. I think I likley > forgot to restore the orginal retval handling here. This whole function > disappears in the next patch, but the original error handling should be > restored to keep things consistent. Thx for the catch! > > > Thx for the reviews!! I know it's complicated! I've chased my tail many > times with it myself :-) > > Allison > > > > > > > > Brian > > > > > - /* > > > - * Close out trans and start the next one in the chain. > > > - */ > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > + error = xfs_attr_trans_roll(&dac); > > > if (error) > > > return error; > > > - } while (retval == -EAGAIN); > > > + } while (true); > > > - return 0; > > > + return error; > > > } > > > /* > > > * 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 > > > + * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the > > > * transaction and re-call the function > > > */ > > > int > > > __xfs_attr_rmtval_remove( > > > - struct xfs_da_args *args) > > > + struct xfs_delattr_context *dac) > > > { > > > - int error, done; > > > + struct xfs_da_args *args = dac->da_args; > > > + int error, done; > > > /* > > > * Unmap value blocks for this attr. > > > @@ -719,12 +719,20 @@ __xfs_attr_rmtval_remove( > > > if (error) > > > return error; > > > - error = xfs_defer_finish(&args->trans); > > > - if (error) > > > - return error; > > > - > > > - if (!done) > > > + /* > > > + * We dont need an explicit state here to pick up where we left off. We > > > + * can figure it out using the !done return code. Calling function only > > > + * needs to keep recalling this routine until we indicate to stop by > > > + * returning anything other than -EAGAIN. The actual value of > > > + * attr->xattri_dela_state may be some value reminicent of the calling > > > + * function, but it's value is irrelevant with in the context of this > > > + * function. Once we are done here, the next state is set as needed > > > + * by the parent > > > + */ > > > + if (!done) { > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > return -EAGAIN; > > > + } > > > return error; > > > } > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h > > > index 9eee615..002fd30 100644 > > > --- a/fs/xfs/libxfs/xfs_attr_remote.h > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.h > > > @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args); > > > int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map, > > > xfs_buf_flags_t incore_flags); > > > int xfs_attr_rmtval_invalidate(struct xfs_da_args *args); > > > -int __xfs_attr_rmtval_remove(struct xfs_da_args *args); > > > +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac); > > > #endif /* __XFS_ATTR_REMOTE_H__ */ > > > diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c > > > index bfad669..aaa7e66 100644 > > > --- a/fs/xfs/xfs_attr_inactive.c > > > +++ b/fs/xfs/xfs_attr_inactive.c > > > @@ -15,10 +15,10 @@ > > > #include "xfs_da_format.h" > > > #include "xfs_da_btree.h" > > > #include "xfs_inode.h" > > > +#include "xfs_attr.h" > > > #include "xfs_attr_remote.h" > > > #include "xfs_trans.h" > > > #include "xfs_bmap.h" > > > -#include "xfs_attr.h" > > > #include "xfs_attr_leaf.h" > > > #include "xfs_quota.h" > > > #include "xfs_dir2.h" > > > -- > > > 2.7.4 > > > > > >
On 2/25/21 7:22 AM, Brian Foster wrote: > On Thu, Feb 25, 2021 at 12:01:10AM -0700, Allison Henderson wrote: >> >> >> On 2/24/21 11:45 AM, Brian Foster wrote: >>> On Thu, Feb 18, 2021 at 09:53:37AM -0700, Allison Henderson wrote: >>>> This patch modifies the attr remove routines to be delay ready. This >>>> means they no longer roll or commit transactions, but instead return >>>> -EAGAIN to have the calling routine roll and refresh the transaction. In >>>> this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which >>>> uses a sort of state machine like switch to keep track of where it was >>>> when EAGAIN was returned. xfs_attr_node_removename has also been >>>> modified to use the switch, and a new version of xfs_attr_remove_args >>>> consists of a simple loop to refresh the transaction until the operation >>>> is completed. A new XFS_DAC_DEFER_FINISH flag is used to finish the >>>> transaction where ever the existing code used to. >>>> >>>> Calls to xfs_attr_rmtval_remove are replaced with the delay ready >>>> version __xfs_attr_rmtval_remove. We will rename >>>> __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are >>>> done. >>>> >>>> xfs_attr_rmtval_remove itself is still in use by the set routines (used >>>> during a rename). For reasons of preserving existing function, we >>>> modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is >>>> set. Similar to how xfs_attr_remove_args does here. Once we transition >>>> the set routines to be delay ready, xfs_attr_rmtval_remove is no longer >>>> used and will be removed. >>>> >>>> This patch also adds a new struct xfs_delattr_context, which we will use >>>> to keep track of the current state of an attribute operation. The new >>>> xfs_delattr_state enum is used to track various operations that are in >>>> progress so that we know not to repeat them, and resume where we left >>>> off before EAGAIN was returned to cycle out the transaction. Other >>>> members take the place of local variables that need to retain their >>>> values across multiple function recalls. See xfs_attr.h for a more >>>> detailed diagram of the states. >>>> >>>> Signed-off-by: Allison Henderson <allison.henderson@oracle.com> >>>> --- >>>> fs/xfs/libxfs/xfs_attr.c | 223 +++++++++++++++++++++++++++++----------- >>>> fs/xfs/libxfs/xfs_attr.h | 100 ++++++++++++++++++ >>>> fs/xfs/libxfs/xfs_attr_leaf.c | 2 +- >>>> fs/xfs/libxfs/xfs_attr_remote.c | 48 +++++---- >>>> fs/xfs/libxfs/xfs_attr_remote.h | 2 +- >>>> fs/xfs/xfs_attr_inactive.c | 2 +- >>>> 6 files changed, 294 insertions(+), 83 deletions(-) >>>> >>>> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c >>>> index 56d4b56..d46b92a 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr.c >>>> +++ b/fs/xfs/libxfs/xfs_attr.c > ... >>>> @@ -1285,51 +1365,74 @@ xfs_attr_node_remove_step( >>>> * >>>> * This routine will find the blocks of the name to remove, remove them and >>>> * shrink the tree if needed. >>>> + * >>>> + * This routine is meant to function as either an inline or delayed operation, >>>> + * and may return -EAGAIN when the transaction needs to be rolled. Calling >>>> + * functions will need to handle this, and recall the function until a >>>> + * successful error code is returned. >>>> */ >>>> STATIC int >>>> -xfs_attr_node_removename( >>>> - struct xfs_da_args *args) >>>> +xfs_attr_node_removename_iter( >>>> + struct xfs_delattr_context *dac) >>>> { >>>> - struct xfs_da_state *state = NULL; >>>> - int retval, error; >>>> - struct xfs_inode *dp = args->dp; >>>> + struct xfs_da_args *args = dac->da_args; >>>> + struct xfs_da_state *state = NULL; >>>> + int retval, error; >>>> + struct xfs_inode *dp = args->dp; >>>> trace_xfs_attr_node_removename(args); >>>> - error = xfs_attr_node_removename_setup(args, &state); >>>> - if (error) >>>> - goto out; >>>> - >>>> - error = xfs_attr_node_remove_step(args, state); >>>> - if (error) >>>> - goto out; >>>> - >>>> - retval = xfs_attr_node_remove_cleanup(args, state); >>>> - >>>> - /* >>>> - * Check to see if the tree needs to be collapsed. >>>> - */ >>>> - if (retval && (state->path.active > 1)) { >>>> - error = xfs_da3_join(state); >>>> - if (error) >>>> - goto out; >>>> - error = xfs_defer_finish(&args->trans); >>>> + if (!dac->da_state) { >>>> + error = xfs_attr_node_removename_setup(dac); >>>> if (error) >>>> goto out; >>>> + } >>>> + state = dac->da_state; >>>> + >>>> + switch (dac->dela_state) { >>>> + case XFS_DAS_UNINIT: >>>> /* >>>> - * Commit the Btree join operation and start a new trans. >>>> + * repeatedly remove remote blocks, remove the entry and join. >>>> + * returns -EAGAIN or 0 for completion of the step. >>>> */ >>>> - error = xfs_trans_roll_inode(&args->trans, dp); >>>> + error = xfs_attr_node_remove_step(dac); >>>> if (error) >>>> - goto out; >>>> - } >>>> + break; >>> >>> Hmm.. so re: my comment further down on xfs_attr_rmtval_remove(), >>> wouldn't that change semantics here? I.e., once remote blocks are >>> removed this would previously carry on with a clean transaction. Now it >>> looks like we'd carry on with the dirty transaction that removed the >>> last remote extent. This suggests that perhaps we should return once >>> more and fall into a new state to remove the name..? >> I suspect the diff might be making this a bit difficult to see. The roll >> that you see being removed here belongs to the transaction we hoisted up in >> patch 3 which happens after the clean up below, and we have the >> corresponding EAGAIN fot that one. I think the diff gets things a little >> interlaced here because the switch adds another level of indentation. >> > > Hmm.. the roll in patch 3 appears to be related to the _cleanup() > helper. What I'm referring to here is the state of the transaction after > the final remote block is removed from the attr. I'm not sure we're > talking about the same thing here.. > >> some times i do like to I use a graphical diffviewer like diffuse when >> patches get weird like this. Something like this: >> >> git config --global diff.tool diffuse >> git difftool 3c53e49 e201c09 >> >> You'd need to download the branch and also the diffuse tool, but sometimes i >> think it makes some of these diffs a bit easier to see >> > > I think it's easier just to refer to the code directly. The current > upstream code flows down into: > > ... > xfs_attr_node_removename() > xfs_attr_node_remove_rmt() > xfs_attr_rmtval_remove() > > ... which then implements the following loop: > > do { > retval = __xfs_attr_rmtval_remove(args); > if (retval && retval != -EAGAIN) > return retval; > > /* > * Close out trans and start the next one in the chain. > */ > error = xfs_trans_roll_inode(&args->trans, args->dp); > if (error) > return error; > } while (retval == -EAGAIN); > > This rolls the transaction when retval == -EAGAIN or retval == 0, thus > always returns with a clean transaction after the remote block removal > completes. > > The code as of this patch does: > > ... > xfs_attr_node_removename_iter() > xfs_attr_node_remove_step() > xfs_attr_node_remove_rmt() > __xfs_attr_rmtval_remove() > > ... which either returns -EAGAIN (since the roll is now implemented at > the very top) or 0 when done == true. The transaction might be dirty in > the latter case, but xfs_attr_node_removename_iter() moves right on to > xfs_attr_node_remove_cleanup() which can now do more work in that same > transaction. Am I following that correctly? > >> Also, it would be >>> nice to remove the several seemingly unnecessary layers of indirection >>> here. For example, something like the following (also considering my >>> comment above wrt to xfs_attr_remove_iter() and UNINIT): >>> >>> case UNINIT: >>> ... >>> /* fallthrough */ >>> case RMTBLK: >>> if (args->rmtblkno > 0) { >>> dac->dela_state = RMTBLK; >>> error = __xfs_attr_rmtval_remove(dac); >>> if (error) >>> break; >>> >>> ASSERT(args->rmtblkno == 0); >>> xfs_attr_refillstate(state); >>> dac->flags |= XFS_DAC_DEFER_FINISH; >>> dac->dela_state = RMNAME; >>> return -EAGAIN; >>> } >> Ok, this looks to me like we've hoisted both xfs_attr_node_remove_rmt and >> xfs_attr_node_remove_step into this scope, but I still think this adds an >> extra roll where non previously was. With out that extra EAGAIN, I think we >> are fine to have all that just under the UNINIT case. I also think it's >> also worth noteing here that this is kind of a reverse of patch 1, which I >> think we put in for reasons of trying to modularize the higher level >> functions as much as possible. >> >> I suspect some of where you were going with this may have been influenced by >> the earlier diff confusion too. Maybe take a second look there before we go >> too much down this change.... >> > > I can certainly be getting lost somewhere in all the refactoring. If so, > can you point out where in the flow described above? Ok, I think see it. So basically I think this means we cant have the helpers because it's ambiguos as to if the transaction is dirty or not. I dont see that there's anything in the review history where we rationalized that away, so I think we just overlooked it. So I think what this means is that we need to reverse apply commit 72b97ea40d (which is where we added xfs_attr_node_remove_rmt), then drop patch 1 which leaves no need for patch 3, since the transaction will have not moved. Then add state RMTBLK? I think that arrives at what you have here. Allison > > Brian > >> >>> /* fallthrough */ >>> case RMNAME: >>> ... >>> ... >>> >>>> - /* >>>> - * If the result is small enough, push it all into the inode. >>>> - */ >>>> - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) >>>> - error = xfs_attr_node_shrink(args, state); >>>> + retval = xfs_attr_node_remove_cleanup(args, state); >>> ... >> I think the overlooked EAGAIN was in this area that got clipped out..... >> >>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c >>>> index 48d8e9c..f09820c 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr_remote.c >>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.c >>> ... >>>> @@ -685,31 +687,29 @@ c( >>>> * Keep de-allocating extents until the remote-value region is gone. >>>> */ >>>> do { >>>> - retval = __xfs_attr_rmtval_remove(args); >>>> - if (retval && retval != -EAGAIN) >>>> - return retval; >>>> + error = __xfs_attr_rmtval_remove(&dac); >>>> + if (error != -EAGAIN) >>>> + break; >>> >>> Previously this would roll once and exit the loop on retval == 0. Now it >>> looks like we break out of the loop immediately. Why the change? >> >> Gosh, I think sometime in reviewing v9, we had come up with a >> "xfs_attr_roll_again" helper that took the error code as a paramater and >> decided whether or not to roll. And then in v10 i think people thought that >> was weird and we turned it into xfs_attr_trans_roll. I think I likley >> forgot to restore the orginal retval handling here. This whole function >> disappears in the next patch, but the original error handling should be >> restored to keep things consistent. Thx for the catch! >> >> >> Thx for the reviews!! I know it's complicated! I've chased my tail many >> times with it myself :-) >> >> Allison >> >> >> >> >>> >>> Brian >>> >>>> - /* >>>> - * Close out trans and start the next one in the chain. >>>> - */ >>>> - error = xfs_trans_roll_inode(&args->trans, args->dp); >>>> + error = xfs_attr_trans_roll(&dac); >>>> if (error) >>>> return error; >>>> - } while (retval == -EAGAIN); >>>> + } while (true); >>>> - return 0; >>>> + return error; >>>> } >>>> /* >>>> * 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 >>>> + * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the >>>> * transaction and re-call the function >>>> */ >>>> int >>>> __xfs_attr_rmtval_remove( >>>> - struct xfs_da_args *args) >>>> + struct xfs_delattr_context *dac) >>>> { >>>> - int error, done; >>>> + struct xfs_da_args *args = dac->da_args; >>>> + int error, done; >>>> /* >>>> * Unmap value blocks for this attr. >>>> @@ -719,12 +719,20 @@ __xfs_attr_rmtval_remove( >>>> if (error) >>>> return error; >>>> - error = xfs_defer_finish(&args->trans); >>>> - if (error) >>>> - return error; >>>> - >>>> - if (!done) >>>> + /* >>>> + * We dont need an explicit state here to pick up where we left off. We >>>> + * can figure it out using the !done return code. Calling function only >>>> + * needs to keep recalling this routine until we indicate to stop by >>>> + * returning anything other than -EAGAIN. The actual value of >>>> + * attr->xattri_dela_state may be some value reminicent of the calling >>>> + * function, but it's value is irrelevant with in the context of this >>>> + * function. Once we are done here, the next state is set as needed >>>> + * by the parent >>>> + */ >>>> + if (!done) { >>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>> return -EAGAIN; >>>> + } >>>> return error; >>>> } >>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h >>>> index 9eee615..002fd30 100644 >>>> --- a/fs/xfs/libxfs/xfs_attr_remote.h >>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.h >>>> @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args); >>>> int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map, >>>> xfs_buf_flags_t incore_flags); >>>> int xfs_attr_rmtval_invalidate(struct xfs_da_args *args); >>>> -int __xfs_attr_rmtval_remove(struct xfs_da_args *args); >>>> +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac); >>>> #endif /* __XFS_ATTR_REMOTE_H__ */ >>>> diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c >>>> index bfad669..aaa7e66 100644 >>>> --- a/fs/xfs/xfs_attr_inactive.c >>>> +++ b/fs/xfs/xfs_attr_inactive.c >>>> @@ -15,10 +15,10 @@ >>>> #include "xfs_da_format.h" >>>> #include "xfs_da_btree.h" >>>> #include "xfs_inode.h" >>>> +#include "xfs_attr.h" >>>> #include "xfs_attr_remote.h" >>>> #include "xfs_trans.h" >>>> #include "xfs_bmap.h" >>>> -#include "xfs_attr.h" >>>> #include "xfs_attr_leaf.h" >>>> #include "xfs_quota.h" >>>> #include "xfs_dir2.h" >>>> -- >>>> 2.7.4 >>>> >>> >> >
On Thu, Feb 25, 2021 at 03:28:00PM -0700, Allison Henderson wrote: > > > On 2/25/21 7:22 AM, Brian Foster wrote: > > On Thu, Feb 25, 2021 at 12:01:10AM -0700, Allison Henderson wrote: > > > > > > > > > On 2/24/21 11:45 AM, Brian Foster wrote: > > > > On Thu, Feb 18, 2021 at 09:53:37AM -0700, Allison Henderson wrote: > > > > > This patch modifies the attr remove routines to be delay ready. This > > > > > means they no longer roll or commit transactions, but instead return > > > > > -EAGAIN to have the calling routine roll and refresh the transaction. In > > > > > this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which > > > > > uses a sort of state machine like switch to keep track of where it was > > > > > when EAGAIN was returned. xfs_attr_node_removename has also been > > > > > modified to use the switch, and a new version of xfs_attr_remove_args > > > > > consists of a simple loop to refresh the transaction until the operation > > > > > is completed. A new XFS_DAC_DEFER_FINISH flag is used to finish the > > > > > transaction where ever the existing code used to. > > > > > > > > > > Calls to xfs_attr_rmtval_remove are replaced with the delay ready > > > > > version __xfs_attr_rmtval_remove. We will rename > > > > > __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are > > > > > done. > > > > > > > > > > xfs_attr_rmtval_remove itself is still in use by the set routines (used > > > > > during a rename). For reasons of preserving existing function, we > > > > > modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is > > > > > set. Similar to how xfs_attr_remove_args does here. Once we transition > > > > > the set routines to be delay ready, xfs_attr_rmtval_remove is no longer > > > > > used and will be removed. > > > > > > > > > > This patch also adds a new struct xfs_delattr_context, which we will use > > > > > to keep track of the current state of an attribute operation. The new > > > > > xfs_delattr_state enum is used to track various operations that are in > > > > > progress so that we know not to repeat them, and resume where we left > > > > > off before EAGAIN was returned to cycle out the transaction. Other > > > > > members take the place of local variables that need to retain their > > > > > values across multiple function recalls. See xfs_attr.h for a more > > > > > detailed diagram of the states. > > > > > > > > > > Signed-off-by: Allison Henderson <allison.henderson@oracle.com> > > > > > --- > > > > > fs/xfs/libxfs/xfs_attr.c | 223 +++++++++++++++++++++++++++++----------- > > > > > fs/xfs/libxfs/xfs_attr.h | 100 ++++++++++++++++++ > > > > > fs/xfs/libxfs/xfs_attr_leaf.c | 2 +- > > > > > fs/xfs/libxfs/xfs_attr_remote.c | 48 +++++---- > > > > > fs/xfs/libxfs/xfs_attr_remote.h | 2 +- > > > > > fs/xfs/xfs_attr_inactive.c | 2 +- > > > > > 6 files changed, 294 insertions(+), 83 deletions(-) > > > > > > > > > > diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c > > > > > index 56d4b56..d46b92a 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr.c > > > > > +++ b/fs/xfs/libxfs/xfs_attr.c > > ... > > > > > @@ -1285,51 +1365,74 @@ xfs_attr_node_remove_step( > > > > > * > > > > > * This routine will find the blocks of the name to remove, remove them and > > > > > * shrink the tree if needed. > > > > > + * > > > > > + * This routine is meant to function as either an inline or delayed operation, > > > > > + * and may return -EAGAIN when the transaction needs to be rolled. Calling > > > > > + * functions will need to handle this, and recall the function until a > > > > > + * successful error code is returned. > > > > > */ > > > > > STATIC int > > > > > -xfs_attr_node_removename( > > > > > - struct xfs_da_args *args) > > > > > +xfs_attr_node_removename_iter( > > > > > + struct xfs_delattr_context *dac) > > > > > { > > > > > - struct xfs_da_state *state = NULL; > > > > > - int retval, error; > > > > > - struct xfs_inode *dp = args->dp; > > > > > + struct xfs_da_args *args = dac->da_args; > > > > > + struct xfs_da_state *state = NULL; > > > > > + int retval, error; > > > > > + struct xfs_inode *dp = args->dp; > > > > > trace_xfs_attr_node_removename(args); > > > > > - error = xfs_attr_node_removename_setup(args, &state); > > > > > - if (error) > > > > > - goto out; > > > > > - > > > > > - error = xfs_attr_node_remove_step(args, state); > > > > > - if (error) > > > > > - goto out; > > > > > - > > > > > - retval = xfs_attr_node_remove_cleanup(args, state); > > > > > - > > > > > - /* > > > > > - * Check to see if the tree needs to be collapsed. > > > > > - */ > > > > > - if (retval && (state->path.active > 1)) { > > > > > - error = xfs_da3_join(state); > > > > > - if (error) > > > > > - goto out; > > > > > - error = xfs_defer_finish(&args->trans); > > > > > + if (!dac->da_state) { > > > > > + error = xfs_attr_node_removename_setup(dac); > > > > > if (error) > > > > > goto out; > > > > > + } > > > > > + state = dac->da_state; > > > > > + > > > > > + switch (dac->dela_state) { > > > > > + case XFS_DAS_UNINIT: > > > > > /* > > > > > - * Commit the Btree join operation and start a new trans. > > > > > + * repeatedly remove remote blocks, remove the entry and join. > > > > > + * returns -EAGAIN or 0 for completion of the step. > > > > > */ > > > > > - error = xfs_trans_roll_inode(&args->trans, dp); > > > > > + error = xfs_attr_node_remove_step(dac); > > > > > if (error) > > > > > - goto out; > > > > > - } > > > > > + break; > > > > > > > > Hmm.. so re: my comment further down on xfs_attr_rmtval_remove(), > > > > wouldn't that change semantics here? I.e., once remote blocks are > > > > removed this would previously carry on with a clean transaction. Now it > > > > looks like we'd carry on with the dirty transaction that removed the > > > > last remote extent. This suggests that perhaps we should return once > > > > more and fall into a new state to remove the name..? > > > I suspect the diff might be making this a bit difficult to see. The roll > > > that you see being removed here belongs to the transaction we hoisted up in > > > patch 3 which happens after the clean up below, and we have the > > > corresponding EAGAIN fot that one. I think the diff gets things a little > > > interlaced here because the switch adds another level of indentation. > > > > > > > Hmm.. the roll in patch 3 appears to be related to the _cleanup() > > helper. What I'm referring to here is the state of the transaction after > > the final remote block is removed from the attr. I'm not sure we're > > talking about the same thing here.. > > > > > some times i do like to I use a graphical diffviewer like diffuse when > > > patches get weird like this. Something like this: > > > > > > git config --global diff.tool diffuse > > > git difftool 3c53e49 e201c09 > > > > > > You'd need to download the branch and also the diffuse tool, but sometimes i > > > think it makes some of these diffs a bit easier to see > > > > > > > I think it's easier just to refer to the code directly. The current > > upstream code flows down into: > > > > ... > > xfs_attr_node_removename() > > xfs_attr_node_remove_rmt() > > xfs_attr_rmtval_remove() > > > > ... which then implements the following loop: > > > > do { > > retval = __xfs_attr_rmtval_remove(args); > > if (retval && retval != -EAGAIN) > > return retval; > > > > /* > > * Close out trans and start the next one in the chain. > > */ > > error = xfs_trans_roll_inode(&args->trans, args->dp); > > if (error) > > return error; > > } while (retval == -EAGAIN); > > > > This rolls the transaction when retval == -EAGAIN or retval == 0, thus > > always returns with a clean transaction after the remote block removal > > completes. > > > > The code as of this patch does: > > > > ... > > xfs_attr_node_removename_iter() > > xfs_attr_node_remove_step() > > xfs_attr_node_remove_rmt() > > __xfs_attr_rmtval_remove() > > > > ... which either returns -EAGAIN (since the roll is now implemented at > > the very top) or 0 when done == true. The transaction might be dirty in > > the latter case, but xfs_attr_node_removename_iter() moves right on to > > xfs_attr_node_remove_cleanup() which can now do more work in that same > > transaction. Am I following that correctly? > > > > > Also, it would be > > > > nice to remove the several seemingly unnecessary layers of indirection > > > > here. For example, something like the following (also considering my > > > > comment above wrt to xfs_attr_remove_iter() and UNINIT): > > > > > > > > case UNINIT: > > > > ... > > > > /* fallthrough */ > > > > case RMTBLK: > > > > if (args->rmtblkno > 0) { > > > > dac->dela_state = RMTBLK; > > > > error = __xfs_attr_rmtval_remove(dac); > > > > if (error) > > > > break; > > > > > > > > ASSERT(args->rmtblkno == 0); > > > > xfs_attr_refillstate(state); > > > > dac->flags |= XFS_DAC_DEFER_FINISH; > > > > dac->dela_state = RMNAME; > > > > return -EAGAIN; > > > > } > > > Ok, this looks to me like we've hoisted both xfs_attr_node_remove_rmt and > > > xfs_attr_node_remove_step into this scope, but I still think this adds an > > > extra roll where non previously was. With out that extra EAGAIN, I think we > > > are fine to have all that just under the UNINIT case. I also think it's > > > also worth noteing here that this is kind of a reverse of patch 1, which I > > > think we put in for reasons of trying to modularize the higher level > > > functions as much as possible. > > > > > > I suspect some of where you were going with this may have been influenced by > > > the earlier diff confusion too. Maybe take a second look there before we go > > > too much down this change.... > > > > > > > I can certainly be getting lost somewhere in all the refactoring. If so, > > can you point out where in the flow described above? > Ok, I think see it. So basically I think this means we cant have the > helpers because it's ambiguos as to if the transaction is dirty or not. I > dont see that there's anything in the review history where we rationalized > that away, so I think we just overlooked it. So I think what this means is > that we need to reverse apply commit 72b97ea40d (which is where we added > xfs_attr_node_remove_rmt), then drop patch 1 which leaves no need for patch > 3, since the transaction will have not moved. Then add state RMTBLK? I > think that arrives at what you have here. > It's not clear to me if anything needs to change before this patch or the changes can just fold into this patch itself. You probably have a better sense of that than I do atm. From my perspective, I think we want that transaction to roll after the final remote extent removal unless we had some reason to explicitly change existing behavior. This used to be handled by the old loop that rolled the transaction down in the remote block removal code. ISTM that the proper way to maintain the same behavior in the new state machine code is to unconditionally fall out of a RMTBLKREMOVE state with an -EAGAIN from _iter(). IOW, __xfs_attr_rmtval_remove() returns -EAGAIN when it has more work to do. _iter() returns -EAGAIN when __xfs_attr_rmtval_remove() was called, finished its work, but we need to roll the transaction before the next step of the operation.. Brian > Allison > > > > > Brian > > > > > > > > > /* fallthrough */ > > > > case RMNAME: > > > > ... > > > > ... > > > > > > > > > - /* > > > > > - * If the result is small enough, push it all into the inode. > > > > > - */ > > > > > - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) > > > > > - error = xfs_attr_node_shrink(args, state); > > > > > + retval = xfs_attr_node_remove_cleanup(args, state); > > > > ... > > > I think the overlooked EAGAIN was in this area that got clipped out..... > > > > > > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c > > > > > index 48d8e9c..f09820c 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr_remote.c > > > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.c > > > > ... > > > > > @@ -685,31 +687,29 @@ c( > > > > > * Keep de-allocating extents until the remote-value region is gone. > > > > > */ > > > > > do { > > > > > - retval = __xfs_attr_rmtval_remove(args); > > > > > - if (retval && retval != -EAGAIN) > > > > > - return retval; > > > > > + error = __xfs_attr_rmtval_remove(&dac); > > > > > + if (error != -EAGAIN) > > > > > + break; > > > > > > > > Previously this would roll once and exit the loop on retval == 0. Now it > > > > looks like we break out of the loop immediately. Why the change? > > > > > > Gosh, I think sometime in reviewing v9, we had come up with a > > > "xfs_attr_roll_again" helper that took the error code as a paramater and > > > decided whether or not to roll. And then in v10 i think people thought that > > > was weird and we turned it into xfs_attr_trans_roll. I think I likley > > > forgot to restore the orginal retval handling here. This whole function > > > disappears in the next patch, but the original error handling should be > > > restored to keep things consistent. Thx for the catch! > > > > > > > > > Thx for the reviews!! I know it's complicated! I've chased my tail many > > > times with it myself :-) > > > > > > Allison > > > > > > > > > > > > > > > > > > > > Brian > > > > > > > > > - /* > > > > > - * Close out trans and start the next one in the chain. > > > > > - */ > > > > > - error = xfs_trans_roll_inode(&args->trans, args->dp); > > > > > + error = xfs_attr_trans_roll(&dac); > > > > > if (error) > > > > > return error; > > > > > - } while (retval == -EAGAIN); > > > > > + } while (true); > > > > > - return 0; > > > > > + return error; > > > > > } > > > > > /* > > > > > * 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 > > > > > + * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the > > > > > * transaction and re-call the function > > > > > */ > > > > > int > > > > > __xfs_attr_rmtval_remove( > > > > > - struct xfs_da_args *args) > > > > > + struct xfs_delattr_context *dac) > > > > > { > > > > > - int error, done; > > > > > + struct xfs_da_args *args = dac->da_args; > > > > > + int error, done; > > > > > /* > > > > > * Unmap value blocks for this attr. > > > > > @@ -719,12 +719,20 @@ __xfs_attr_rmtval_remove( > > > > > if (error) > > > > > return error; > > > > > - error = xfs_defer_finish(&args->trans); > > > > > - if (error) > > > > > - return error; > > > > > - > > > > > - if (!done) > > > > > + /* > > > > > + * We dont need an explicit state here to pick up where we left off. We > > > > > + * can figure it out using the !done return code. Calling function only > > > > > + * needs to keep recalling this routine until we indicate to stop by > > > > > + * returning anything other than -EAGAIN. The actual value of > > > > > + * attr->xattri_dela_state may be some value reminicent of the calling > > > > > + * function, but it's value is irrelevant with in the context of this > > > > > + * function. Once we are done here, the next state is set as needed > > > > > + * by the parent > > > > > + */ > > > > > + if (!done) { > > > > > + dac->flags |= XFS_DAC_DEFER_FINISH; > > > > > return -EAGAIN; > > > > > + } > > > > > return error; > > > > > } > > > > > diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h > > > > > index 9eee615..002fd30 100644 > > > > > --- a/fs/xfs/libxfs/xfs_attr_remote.h > > > > > +++ b/fs/xfs/libxfs/xfs_attr_remote.h > > > > > @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args); > > > > > int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map, > > > > > xfs_buf_flags_t incore_flags); > > > > > int xfs_attr_rmtval_invalidate(struct xfs_da_args *args); > > > > > -int __xfs_attr_rmtval_remove(struct xfs_da_args *args); > > > > > +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac); > > > > > #endif /* __XFS_ATTR_REMOTE_H__ */ > > > > > diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c > > > > > index bfad669..aaa7e66 100644 > > > > > --- a/fs/xfs/xfs_attr_inactive.c > > > > > +++ b/fs/xfs/xfs_attr_inactive.c > > > > > @@ -15,10 +15,10 @@ > > > > > #include "xfs_da_format.h" > > > > > #include "xfs_da_btree.h" > > > > > #include "xfs_inode.h" > > > > > +#include "xfs_attr.h" > > > > > #include "xfs_attr_remote.h" > > > > > #include "xfs_trans.h" > > > > > #include "xfs_bmap.h" > > > > > -#include "xfs_attr.h" > > > > > #include "xfs_attr_leaf.h" > > > > > #include "xfs_quota.h" > > > > > #include "xfs_dir2.h" > > > > > -- > > > > > 2.7.4 > > > > > > > > > > > > > > >
On 2/28/21 8:39 AM, Brian Foster wrote: > On Thu, Feb 25, 2021 at 03:28:00PM -0700, Allison Henderson wrote: >> >> >> On 2/25/21 7:22 AM, Brian Foster wrote: >>> On Thu, Feb 25, 2021 at 12:01:10AM -0700, Allison Henderson wrote: >>>> >>>> >>>> On 2/24/21 11:45 AM, Brian Foster wrote: >>>>> On Thu, Feb 18, 2021 at 09:53:37AM -0700, Allison Henderson wrote: >>>>>> This patch modifies the attr remove routines to be delay ready. This >>>>>> means they no longer roll or commit transactions, but instead return >>>>>> -EAGAIN to have the calling routine roll and refresh the transaction. In >>>>>> this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which >>>>>> uses a sort of state machine like switch to keep track of where it was >>>>>> when EAGAIN was returned. xfs_attr_node_removename has also been >>>>>> modified to use the switch, and a new version of xfs_attr_remove_args >>>>>> consists of a simple loop to refresh the transaction until the operation >>>>>> is completed. A new XFS_DAC_DEFER_FINISH flag is used to finish the >>>>>> transaction where ever the existing code used to. >>>>>> >>>>>> Calls to xfs_attr_rmtval_remove are replaced with the delay ready >>>>>> version __xfs_attr_rmtval_remove. We will rename >>>>>> __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are >>>>>> done. >>>>>> >>>>>> xfs_attr_rmtval_remove itself is still in use by the set routines (used >>>>>> during a rename). For reasons of preserving existing function, we >>>>>> modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is >>>>>> set. Similar to how xfs_attr_remove_args does here. Once we transition >>>>>> the set routines to be delay ready, xfs_attr_rmtval_remove is no longer >>>>>> used and will be removed. >>>>>> >>>>>> This patch also adds a new struct xfs_delattr_context, which we will use >>>>>> to keep track of the current state of an attribute operation. The new >>>>>> xfs_delattr_state enum is used to track various operations that are in >>>>>> progress so that we know not to repeat them, and resume where we left >>>>>> off before EAGAIN was returned to cycle out the transaction. Other >>>>>> members take the place of local variables that need to retain their >>>>>> values across multiple function recalls. See xfs_attr.h for a more >>>>>> detailed diagram of the states. >>>>>> >>>>>> Signed-off-by: Allison Henderson <allison.henderson@oracle.com> >>>>>> --- >>>>>> fs/xfs/libxfs/xfs_attr.c | 223 +++++++++++++++++++++++++++++----------- >>>>>> fs/xfs/libxfs/xfs_attr.h | 100 ++++++++++++++++++ >>>>>> fs/xfs/libxfs/xfs_attr_leaf.c | 2 +- >>>>>> fs/xfs/libxfs/xfs_attr_remote.c | 48 +++++---- >>>>>> fs/xfs/libxfs/xfs_attr_remote.h | 2 +- >>>>>> fs/xfs/xfs_attr_inactive.c | 2 +- >>>>>> 6 files changed, 294 insertions(+), 83 deletions(-) >>>>>> >>>>>> diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c >>>>>> index 56d4b56..d46b92a 100644 >>>>>> --- a/fs/xfs/libxfs/xfs_attr.c >>>>>> +++ b/fs/xfs/libxfs/xfs_attr.c >>> ... >>>>>> @@ -1285,51 +1365,74 @@ xfs_attr_node_remove_step( >>>>>> * >>>>>> * This routine will find the blocks of the name to remove, remove them and >>>>>> * shrink the tree if needed. >>>>>> + * >>>>>> + * This routine is meant to function as either an inline or delayed operation, >>>>>> + * and may return -EAGAIN when the transaction needs to be rolled. Calling >>>>>> + * functions will need to handle this, and recall the function until a >>>>>> + * successful error code is returned. >>>>>> */ >>>>>> STATIC int >>>>>> -xfs_attr_node_removename( >>>>>> - struct xfs_da_args *args) >>>>>> +xfs_attr_node_removename_iter( >>>>>> + struct xfs_delattr_context *dac) >>>>>> { >>>>>> - struct xfs_da_state *state = NULL; >>>>>> - int retval, error; >>>>>> - struct xfs_inode *dp = args->dp; >>>>>> + struct xfs_da_args *args = dac->da_args; >>>>>> + struct xfs_da_state *state = NULL; >>>>>> + int retval, error; >>>>>> + struct xfs_inode *dp = args->dp; >>>>>> trace_xfs_attr_node_removename(args); >>>>>> - error = xfs_attr_node_removename_setup(args, &state); >>>>>> - if (error) >>>>>> - goto out; >>>>>> - >>>>>> - error = xfs_attr_node_remove_step(args, state); >>>>>> - if (error) >>>>>> - goto out; >>>>>> - >>>>>> - retval = xfs_attr_node_remove_cleanup(args, state); >>>>>> - >>>>>> - /* >>>>>> - * Check to see if the tree needs to be collapsed. >>>>>> - */ >>>>>> - if (retval && (state->path.active > 1)) { >>>>>> - error = xfs_da3_join(state); >>>>>> - if (error) >>>>>> - goto out; >>>>>> - error = xfs_defer_finish(&args->trans); >>>>>> + if (!dac->da_state) { >>>>>> + error = xfs_attr_node_removename_setup(dac); >>>>>> if (error) >>>>>> goto out; >>>>>> + } >>>>>> + state = dac->da_state; >>>>>> + >>>>>> + switch (dac->dela_state) { >>>>>> + case XFS_DAS_UNINIT: >>>>>> /* >>>>>> - * Commit the Btree join operation and start a new trans. >>>>>> + * repeatedly remove remote blocks, remove the entry and join. >>>>>> + * returns -EAGAIN or 0 for completion of the step. >>>>>> */ >>>>>> - error = xfs_trans_roll_inode(&args->trans, dp); >>>>>> + error = xfs_attr_node_remove_step(dac); >>>>>> if (error) >>>>>> - goto out; >>>>>> - } >>>>>> + break; >>>>> >>>>> Hmm.. so re: my comment further down on xfs_attr_rmtval_remove(), >>>>> wouldn't that change semantics here? I.e., once remote blocks are >>>>> removed this would previously carry on with a clean transaction. Now it >>>>> looks like we'd carry on with the dirty transaction that removed the >>>>> last remote extent. This suggests that perhaps we should return once >>>>> more and fall into a new state to remove the name..? >>>> I suspect the diff might be making this a bit difficult to see. The roll >>>> that you see being removed here belongs to the transaction we hoisted up in >>>> patch 3 which happens after the clean up below, and we have the >>>> corresponding EAGAIN fot that one. I think the diff gets things a little >>>> interlaced here because the switch adds another level of indentation. >>>> >>> >>> Hmm.. the roll in patch 3 appears to be related to the _cleanup() >>> helper. What I'm referring to here is the state of the transaction after >>> the final remote block is removed from the attr. I'm not sure we're >>> talking about the same thing here.. >>> >>>> some times i do like to I use a graphical diffviewer like diffuse when >>>> patches get weird like this. Something like this: >>>> >>>> git config --global diff.tool diffuse >>>> git difftool 3c53e49 e201c09 >>>> >>>> You'd need to download the branch and also the diffuse tool, but sometimes i >>>> think it makes some of these diffs a bit easier to see >>>> >>> >>> I think it's easier just to refer to the code directly. The current >>> upstream code flows down into: >>> >>> ... >>> xfs_attr_node_removename() >>> xfs_attr_node_remove_rmt() >>> xfs_attr_rmtval_remove() >>> >>> ... which then implements the following loop: >>> >>> do { >>> retval = __xfs_attr_rmtval_remove(args); >>> if (retval && retval != -EAGAIN) >>> return retval; >>> >>> /* >>> * Close out trans and start the next one in the chain. >>> */ >>> error = xfs_trans_roll_inode(&args->trans, args->dp); >>> if (error) >>> return error; >>> } while (retval == -EAGAIN); >>> >>> This rolls the transaction when retval == -EAGAIN or retval == 0, thus >>> always returns with a clean transaction after the remote block removal >>> completes. >>> >>> The code as of this patch does: >>> >>> ... >>> xfs_attr_node_removename_iter() >>> xfs_attr_node_remove_step() >>> xfs_attr_node_remove_rmt() >>> __xfs_attr_rmtval_remove() >>> >>> ... which either returns -EAGAIN (since the roll is now implemented at >>> the very top) or 0 when done == true. The transaction might be dirty in >>> the latter case, but xfs_attr_node_removename_iter() moves right on to >>> xfs_attr_node_remove_cleanup() which can now do more work in that same >>> transaction. Am I following that correctly? >>> >>>> Also, it would be >>>>> nice to remove the several seemingly unnecessary layers of indirection >>>>> here. For example, something like the following (also considering my >>>>> comment above wrt to xfs_attr_remove_iter() and UNINIT): >>>>> >>>>> case UNINIT: >>>>> ... >>>>> /* fallthrough */ >>>>> case RMTBLK: >>>>> if (args->rmtblkno > 0) { >>>>> dac->dela_state = RMTBLK; >>>>> error = __xfs_attr_rmtval_remove(dac); >>>>> if (error) >>>>> break; >>>>> >>>>> ASSERT(args->rmtblkno == 0); >>>>> xfs_attr_refillstate(state); >>>>> dac->flags |= XFS_DAC_DEFER_FINISH; >>>>> dac->dela_state = RMNAME; >>>>> return -EAGAIN; >>>>> } >>>> Ok, this looks to me like we've hoisted both xfs_attr_node_remove_rmt and >>>> xfs_attr_node_remove_step into this scope, but I still think this adds an >>>> extra roll where non previously was. With out that extra EAGAIN, I think we >>>> are fine to have all that just under the UNINIT case. I also think it's >>>> also worth noteing here that this is kind of a reverse of patch 1, which I >>>> think we put in for reasons of trying to modularize the higher level >>>> functions as much as possible. >>>> >>>> I suspect some of where you were going with this may have been influenced by >>>> the earlier diff confusion too. Maybe take a second look there before we go >>>> too much down this change.... >>>> >>> >>> I can certainly be getting lost somewhere in all the refactoring. If so, >>> can you point out where in the flow described above? >> Ok, I think see it. So basically I think this means we cant have the >> helpers because it's ambiguos as to if the transaction is dirty or not. I >> dont see that there's anything in the review history where we rationalized >> that away, so I think we just overlooked it. So I think what this means is >> that we need to reverse apply commit 72b97ea40d (which is where we added >> xfs_attr_node_remove_rmt), then drop patch 1 which leaves no need for patch >> 3, since the transaction will have not moved. Then add state RMTBLK? I >> think that arrives at what you have here. >> > > It's not clear to me if anything needs to change before this patch or > the changes can just fold into this patch itself. You probably have a > better sense of that than I do atm. I can unfold it here in this patch too if people prefer. I'm getting the impression that others are having a hard time keeping up with the refactoring, so maybe that might be better. From my perspective, I think we want > that transaction to roll after the final remote extent removal unless we > had some reason to explicitly change existing behavior. This used to be > handled by the old loop that rolled the transaction down in the remote > block removal code. ISTM that the proper way to maintain the same > behavior in the new state machine code is to unconditionally fall out of > a RMTBLKREMOVE state with an -EAGAIN from _iter(). Sure, I think that describes what you have up there > > IOW, __xfs_attr_rmtval_remove() returns -EAGAIN when it has more work to > do. _iter() returns -EAGAIN when __xfs_attr_rmtval_remove() was called, > finished its work, but we need to roll the transaction before the next > step of the operation.. Right, ok, will make those modifcations here then. Thanks for the reviews! I know it's a lot. Allison > > Brian > >> Allison >> >>> >>> Brian >>> >>>> >>>>> /* fallthrough */ >>>>> case RMNAME: >>>>> ... >>>>> ... >>>>> >>>>>> - /* >>>>>> - * If the result is small enough, push it all into the inode. >>>>>> - */ >>>>>> - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) >>>>>> - error = xfs_attr_node_shrink(args, state); >>>>>> + retval = xfs_attr_node_remove_cleanup(args, state); >>>>> ... >>>> I think the overlooked EAGAIN was in this area that got clipped out..... >>>> >>>>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c >>>>>> index 48d8e9c..f09820c 100644 >>>>>> --- a/fs/xfs/libxfs/xfs_attr_remote.c >>>>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.c >>>>> ... >>>>>> @@ -685,31 +687,29 @@ c( >>>>>> * Keep de-allocating extents until the remote-value region is gone. >>>>>> */ >>>>>> do { >>>>>> - retval = __xfs_attr_rmtval_remove(args); >>>>>> - if (retval && retval != -EAGAIN) >>>>>> - return retval; >>>>>> + error = __xfs_attr_rmtval_remove(&dac); >>>>>> + if (error != -EAGAIN) >>>>>> + break; >>>>> >>>>> Previously this would roll once and exit the loop on retval == 0. Now it >>>>> looks like we break out of the loop immediately. Why the change? >>>> >>>> Gosh, I think sometime in reviewing v9, we had come up with a >>>> "xfs_attr_roll_again" helper that took the error code as a paramater and >>>> decided whether or not to roll. And then in v10 i think people thought that >>>> was weird and we turned it into xfs_attr_trans_roll. I think I likley >>>> forgot to restore the orginal retval handling here. This whole function >>>> disappears in the next patch, but the original error handling should be >>>> restored to keep things consistent. Thx for the catch! >>>> >>>> >>>> Thx for the reviews!! I know it's complicated! I've chased my tail many >>>> times with it myself :-) >>>> >>>> Allison >>>> >>>> >>>> >>>> >>>>> >>>>> Brian >>>>> >>>>>> - /* >>>>>> - * Close out trans and start the next one in the chain. >>>>>> - */ >>>>>> - error = xfs_trans_roll_inode(&args->trans, args->dp); >>>>>> + error = xfs_attr_trans_roll(&dac); >>>>>> if (error) >>>>>> return error; >>>>>> - } while (retval == -EAGAIN); >>>>>> + } while (true); >>>>>> - return 0; >>>>>> + return error; >>>>>> } >>>>>> /* >>>>>> * 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 >>>>>> + * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the >>>>>> * transaction and re-call the function >>>>>> */ >>>>>> int >>>>>> __xfs_attr_rmtval_remove( >>>>>> - struct xfs_da_args *args) >>>>>> + struct xfs_delattr_context *dac) >>>>>> { >>>>>> - int error, done; >>>>>> + struct xfs_da_args *args = dac->da_args; >>>>>> + int error, done; >>>>>> /* >>>>>> * Unmap value blocks for this attr. >>>>>> @@ -719,12 +719,20 @@ __xfs_attr_rmtval_remove( >>>>>> if (error) >>>>>> return error; >>>>>> - error = xfs_defer_finish(&args->trans); >>>>>> - if (error) >>>>>> - return error; >>>>>> - >>>>>> - if (!done) >>>>>> + /* >>>>>> + * We dont need an explicit state here to pick up where we left off. We >>>>>> + * can figure it out using the !done return code. Calling function only >>>>>> + * needs to keep recalling this routine until we indicate to stop by >>>>>> + * returning anything other than -EAGAIN. The actual value of >>>>>> + * attr->xattri_dela_state may be some value reminicent of the calling >>>>>> + * function, but it's value is irrelevant with in the context of this >>>>>> + * function. Once we are done here, the next state is set as needed >>>>>> + * by the parent >>>>>> + */ >>>>>> + if (!done) { >>>>>> + dac->flags |= XFS_DAC_DEFER_FINISH; >>>>>> return -EAGAIN; >>>>>> + } >>>>>> return error; >>>>>> } >>>>>> diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h >>>>>> index 9eee615..002fd30 100644 >>>>>> --- a/fs/xfs/libxfs/xfs_attr_remote.h >>>>>> +++ b/fs/xfs/libxfs/xfs_attr_remote.h >>>>>> @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args); >>>>>> int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map, >>>>>> xfs_buf_flags_t incore_flags); >>>>>> int xfs_attr_rmtval_invalidate(struct xfs_da_args *args); >>>>>> -int __xfs_attr_rmtval_remove(struct xfs_da_args *args); >>>>>> +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac); >>>>>> #endif /* __XFS_ATTR_REMOTE_H__ */ >>>>>> diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c >>>>>> index bfad669..aaa7e66 100644 >>>>>> --- a/fs/xfs/xfs_attr_inactive.c >>>>>> +++ b/fs/xfs/xfs_attr_inactive.c >>>>>> @@ -15,10 +15,10 @@ >>>>>> #include "xfs_da_format.h" >>>>>> #include "xfs_da_btree.h" >>>>>> #include "xfs_inode.h" >>>>>> +#include "xfs_attr.h" >>>>>> #include "xfs_attr_remote.h" >>>>>> #include "xfs_trans.h" >>>>>> #include "xfs_bmap.h" >>>>>> -#include "xfs_attr.h" >>>>>> #include "xfs_attr_leaf.h" >>>>>> #include "xfs_quota.h" >>>>>> #include "xfs_dir2.h" >>>>>> -- >>>>>> 2.7.4 >>>>>> >>>>> >>>> >>> >> >
diff --git a/fs/xfs/libxfs/xfs_attr.c b/fs/xfs/libxfs/xfs_attr.c index 56d4b56..d46b92a 100644 --- a/fs/xfs/libxfs/xfs_attr.c +++ b/fs/xfs/libxfs/xfs_attr.c @@ -57,8 +57,8 @@ STATIC int xfs_attr_node_addname(struct xfs_da_args *args, struct xfs_da_state *state); STATIC int xfs_attr_node_addname_find_attr(struct xfs_da_args *args, struct xfs_da_state **state); -STATIC int xfs_attr_node_removename(xfs_da_args_t *args); STATIC int xfs_attr_node_addname_work(struct xfs_da_args *args); +STATIC int xfs_attr_node_removename_iter(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); @@ -221,6 +221,34 @@ xfs_attr_is_shortform( ip->i_afp->if_nextents == 0); } +/* + * Checks to see if a delayed attribute transaction should be rolled. If so, + * also checks for a defer finish. Transaction is finished and rolled as + * needed, and returns true of false if the delayed operation should continue. + */ +int +xfs_attr_trans_roll( + struct xfs_delattr_context *dac) +{ + struct xfs_da_args *args = dac->da_args; + int error; + + if (dac->flags & XFS_DAC_DEFER_FINISH) { + /* + * The caller wants us to finish all the deferred ops so that we + * avoid pinning the log tail with a large number of deferred + * ops. + */ + dac->flags &= ~XFS_DAC_DEFER_FINISH; + error = xfs_defer_finish(&args->trans); + if (error) + return error; + } else + error = xfs_trans_roll_inode(&args->trans, args->dp); + + return error; +} + STATIC int xfs_attr_set_fmt( struct xfs_da_args *args) @@ -531,23 +559,58 @@ xfs_has_attr( */ int xfs_attr_remove_args( - struct xfs_da_args *args) + struct xfs_da_args *args) { - struct xfs_inode *dp = args->dp; - int error; + int error; + struct xfs_delattr_context dac = { + .da_args = args, + }; + + do { + error = xfs_attr_remove_iter(&dac); + if (error != -EAGAIN) + break; + + error = xfs_attr_trans_roll(&dac); + if (error) + return error; + + } while (true); + + return error; +} + +/* + * Remove the attribute specified in @args. + * + * This function may return -EAGAIN to signal that the transaction needs to be + * rolled. Callers should continue calling this function until they receive a + * return value other than -EAGAIN. + */ +int +xfs_attr_remove_iter( + struct xfs_delattr_context *dac) +{ + struct xfs_da_args *args = dac->da_args; + struct xfs_inode *dp = args->dp; - if (!xfs_inode_hasattr(dp)) { - error = -ENOATTR; - } else if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) { + /* If we are shrinking a node, resume shrink */ + if (dac->dela_state == XFS_DAS_RM_SHRINK) + goto node; + + if (!xfs_inode_hasattr(dp)) + return -ENOATTR; + + if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL) { ASSERT(dp->i_afp->if_flags & XFS_IFINLINE); - error = xfs_attr_shortform_remove(args); - } else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) { - error = xfs_attr_leaf_removename(args); - } else { - error = xfs_attr_node_removename(args); + return xfs_attr_shortform_remove(args); } - return error; + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) + return xfs_attr_leaf_removename(args); +node: + /* If we are not short form or leaf, then proceed to remove node */ + return xfs_attr_node_removename_iter(dac); } /* @@ -1191,14 +1254,16 @@ xfs_attr_leaf_mark_incomplete( */ STATIC int xfs_attr_node_removename_setup( - struct xfs_da_args *args, - struct xfs_da_state **state) + struct xfs_delattr_context *dac) { - int error; + struct xfs_da_args *args = dac->da_args; + struct xfs_da_state **state = &dac->da_state; + int error; error = xfs_attr_node_hasname(args, state); if (error != -EEXIST) return error; + error = 0; ASSERT((*state)->path.blk[(*state)->path.active - 1].bp != NULL); ASSERT((*state)->path.blk[(*state)->path.active - 1].magic == @@ -1207,22 +1272,28 @@ int xfs_attr_node_removename_setup( if (args->rmtblkno > 0) { error = xfs_attr_leaf_mark_incomplete(args, *state); if (error) - return error; + goto out; - return xfs_attr_rmtval_invalidate(args); + error = xfs_attr_rmtval_invalidate(args); } +out: + if (error) + xfs_da_state_free(*state); return 0; } STATIC int -xfs_attr_node_remove_rmt( - struct xfs_da_args *args, - struct xfs_da_state *state) +xfs_attr_node_remove_rmt ( + struct xfs_delattr_context *dac, + struct xfs_da_state *state) { - int error = 0; + int error = 0; - error = xfs_attr_rmtval_remove(args); + /* + * May return -EAGAIN to request that the caller recall this function + */ + error = __xfs_attr_rmtval_remove(dac); if (error) return error; @@ -1253,18 +1324,24 @@ xfs_attr_node_remove_cleanup( } /* - * Remove a name from a B-tree attribute list. + * Step through removeing a name from a B-tree attribute list. * * This will involve walking down the Btree, and may involve joining * leaf nodes and even joining intermediate nodes up to and including * the root node (a special case of an intermediate node). + * + * This routine is meant to function as either an inline or delayed operation, + * and may return -EAGAIN when the transaction needs to be rolled. Calling + * functions will need to handle this, and recall the function until a + * successful error code is returned. */ STATIC int xfs_attr_node_remove_step( - struct xfs_da_args *args, - struct xfs_da_state *state) + struct xfs_delattr_context *dac) { - int error = 0; + struct xfs_da_args *args = dac->da_args; + struct xfs_da_state *state = dac->da_state; + int error = 0; /* * If there is an out-of-line value, de-allocate the blocks. @@ -1272,7 +1349,10 @@ xfs_attr_node_remove_step( * overflow the maximum size of a transaction and/or hit a deadlock. */ if (args->rmtblkno > 0) { - error = xfs_attr_node_remove_rmt(args, state); + /* + * May return -EAGAIN. Remove blocks until args->rmtblkno == 0 + */ + error = xfs_attr_node_remove_rmt(dac, state); if (error) return error; } @@ -1285,51 +1365,74 @@ xfs_attr_node_remove_step( * * This routine will find the blocks of the name to remove, remove them and * shrink the tree if needed. + * + * This routine is meant to function as either an inline or delayed operation, + * and may return -EAGAIN when the transaction needs to be rolled. Calling + * functions will need to handle this, and recall the function until a + * successful error code is returned. */ STATIC int -xfs_attr_node_removename( - struct xfs_da_args *args) +xfs_attr_node_removename_iter( + struct xfs_delattr_context *dac) { - struct xfs_da_state *state = NULL; - int retval, error; - struct xfs_inode *dp = args->dp; + struct xfs_da_args *args = dac->da_args; + struct xfs_da_state *state = NULL; + int retval, error; + struct xfs_inode *dp = args->dp; trace_xfs_attr_node_removename(args); - error = xfs_attr_node_removename_setup(args, &state); - if (error) - goto out; - - error = xfs_attr_node_remove_step(args, state); - if (error) - goto out; - - retval = xfs_attr_node_remove_cleanup(args, state); - - /* - * Check to see if the tree needs to be collapsed. - */ - if (retval && (state->path.active > 1)) { - error = xfs_da3_join(state); - if (error) - goto out; - error = xfs_defer_finish(&args->trans); + if (!dac->da_state) { + error = xfs_attr_node_removename_setup(dac); if (error) goto out; + } + state = dac->da_state; + + switch (dac->dela_state) { + case XFS_DAS_UNINIT: /* - * Commit the Btree join operation and start a new trans. + * repeatedly remove remote blocks, remove the entry and join. + * returns -EAGAIN or 0 for completion of the step. */ - error = xfs_trans_roll_inode(&args->trans, dp); + error = xfs_attr_node_remove_step(dac); if (error) - goto out; - } + break; - /* - * If the result is small enough, push it all into the inode. - */ - if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) - error = xfs_attr_node_shrink(args, state); + retval = xfs_attr_node_remove_cleanup(args, state); + /* + * Check to see if the tree needs to be collapsed. Set the flag + * to indicate that the calling function needs to move the + * shrink operation + */ + if (retval && (state->path.active > 1)) { + error = xfs_da3_join(state); + if (error) + goto out; + + dac->flags |= XFS_DAC_DEFER_FINISH; + dac->dela_state = XFS_DAS_RM_SHRINK; + return -EAGAIN; + } + + /* fallthrough */ + case XFS_DAS_RM_SHRINK: + /* + * If the result is small enough, push it all into the inode. + */ + if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) + error = xfs_attr_node_shrink(args, state); + + break; + default: + ASSERT(0); + error = -EINVAL; + goto out; + } + + if (error == -EAGAIN) + return error; out: if (state) xfs_da_state_free(state); diff --git a/fs/xfs/libxfs/xfs_attr.h b/fs/xfs/libxfs/xfs_attr.h index 3e97a93..3154ef4 100644 --- a/fs/xfs/libxfs/xfs_attr.h +++ b/fs/xfs/libxfs/xfs_attr.h @@ -74,6 +74,102 @@ struct xfs_attr_list_context { }; +/* + * ======================================================================== + * Structure used to pass context around among the delayed routines. + * ======================================================================== + */ + +/* + * Below is a state machine diagram for attr remove operations. The XFS_DAS_* + * states indicate places where the function would return -EAGAIN, and then + * immediately resume from after being recalled by the calling function. States + * marked as a "subroutine state" indicate that they belong to a subroutine, and + * so the calling function needs to pass them back to that subroutine to allow + * it to finish where it left off. But they otherwise do not have a role in the + * calling function other than just passing through. + * + * xfs_attr_remove_iter() + * │ + * v + * found attr blks? ───n──┐ + * │ v + * │ find and invalidate + * y the blocks. mark + * │ attr incomplete + * ├────────────────┘ + * │ + * v + * remove a block with + * xfs_attr_node_remove_step <────┐ + * │ │ + * v │ + * still have blks ──y──> return -EAGAIN. + * to remove? re-enter with one + * │ less blk to remove + * n + * │ + * v + * remove leaf and + * update hash with + * xfs_attr_node_remove_cleanup + * │ + * v + * need to + * shrink tree? ─n─┐ + * │ │ + * y │ + * │ │ + * v │ + * join leaf │ + * │ │ + * v │ + * XFS_DAS_RM_SHRINK │ + * │ │ + * v │ + * do the shrink │ + * │ │ + * v │ + * free state <──┘ + * │ + * v + * done + * + */ + +/* + * Enum values for xfs_delattr_context.da_state + * + * These values are used by delayed attribute operations to keep track of where + * they were before they returned -EAGAIN. A return code of -EAGAIN signals the + * calling function to roll the transaction, and then recall the subroutine to + * finish the operation. The enum is then used by the subroutine to jump back + * to where it was and resume executing where it left off. + */ +enum xfs_delattr_state { + XFS_DAS_UNINIT = 0, /* No state has been set yet */ + XFS_DAS_RM_SHRINK, /* We are shrinking the tree */ +}; + +/* + * Defines for xfs_delattr_context.flags + */ +#define XFS_DAC_DEFER_FINISH 0x01 /* finish the transaction */ + +/* + * Context used for keeping track of delayed attribute operations + */ +struct xfs_delattr_context { + struct xfs_da_args *da_args; + + /* Used in xfs_attr_node_removename to roll through removing blocks */ + struct xfs_da_state *da_state; + + /* Used to keep track of current state of delayed operation */ + unsigned int flags; + enum xfs_delattr_state dela_state; +}; + /*======================================================================== * Function prototypes for the kernel. *========================================================================*/ @@ -91,6 +187,10 @@ int xfs_attr_set(struct xfs_da_args *args); int xfs_attr_set_args(struct xfs_da_args *args); int xfs_has_attr(struct xfs_da_args *args); int xfs_attr_remove_args(struct xfs_da_args *args); +int xfs_attr_remove_iter(struct xfs_delattr_context *dac); +int xfs_attr_trans_roll(struct xfs_delattr_context *dac); bool xfs_attr_namecheck(const void *name, size_t length); +void xfs_delattr_context_init(struct xfs_delattr_context *dac, + struct xfs_da_args *args); #endif /* __XFS_ATTR_H__ */ diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c index d6ef69a..3780141 100644 --- a/fs/xfs/libxfs/xfs_attr_leaf.c +++ b/fs/xfs/libxfs/xfs_attr_leaf.c @@ -19,8 +19,8 @@ #include "xfs_bmap_btree.h" #include "xfs_bmap.h" #include "xfs_attr_sf.h" -#include "xfs_attr_remote.h" #include "xfs_attr.h" +#include "xfs_attr_remote.h" #include "xfs_attr_leaf.h" #include "xfs_error.h" #include "xfs_trace.h" diff --git a/fs/xfs/libxfs/xfs_attr_remote.c b/fs/xfs/libxfs/xfs_attr_remote.c index 48d8e9c..f09820c 100644 --- a/fs/xfs/libxfs/xfs_attr_remote.c +++ b/fs/xfs/libxfs/xfs_attr_remote.c @@ -674,10 +674,12 @@ xfs_attr_rmtval_invalidate( */ int xfs_attr_rmtval_remove( - struct xfs_da_args *args) + struct xfs_da_args *args) { - int error; - int retval; + int error; + struct xfs_delattr_context dac = { + .da_args = args, + }; trace_xfs_attr_rmtval_remove(args); @@ -685,31 +687,29 @@ xfs_attr_rmtval_remove( * Keep de-allocating extents until the remote-value region is gone. */ do { - retval = __xfs_attr_rmtval_remove(args); - if (retval && retval != -EAGAIN) - return retval; + error = __xfs_attr_rmtval_remove(&dac); + if (error != -EAGAIN) + break; - /* - * Close out trans and start the next one in the chain. - */ - error = xfs_trans_roll_inode(&args->trans, args->dp); + error = xfs_attr_trans_roll(&dac); if (error) return error; - } while (retval == -EAGAIN); + } while (true); - return 0; + return error; } /* * 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 + * buffer that it is stored on. Returns -EAGAIN for the caller to refresh the * transaction and re-call the function */ int __xfs_attr_rmtval_remove( - struct xfs_da_args *args) + struct xfs_delattr_context *dac) { - int error, done; + struct xfs_da_args *args = dac->da_args; + int error, done; /* * Unmap value blocks for this attr. @@ -719,12 +719,20 @@ __xfs_attr_rmtval_remove( if (error) return error; - error = xfs_defer_finish(&args->trans); - if (error) - return error; - - if (!done) + /* + * We dont need an explicit state here to pick up where we left off. We + * can figure it out using the !done return code. Calling function only + * needs to keep recalling this routine until we indicate to stop by + * returning anything other than -EAGAIN. The actual value of + * attr->xattri_dela_state may be some value reminicent of the calling + * function, but it's value is irrelevant with in the context of this + * function. Once we are done here, the next state is set as needed + * by the parent + */ + if (!done) { + dac->flags |= XFS_DAC_DEFER_FINISH; return -EAGAIN; + } return error; } diff --git a/fs/xfs/libxfs/xfs_attr_remote.h b/fs/xfs/libxfs/xfs_attr_remote.h index 9eee615..002fd30 100644 --- a/fs/xfs/libxfs/xfs_attr_remote.h +++ b/fs/xfs/libxfs/xfs_attr_remote.h @@ -14,5 +14,5 @@ int xfs_attr_rmtval_remove(struct xfs_da_args *args); int xfs_attr_rmtval_stale(struct xfs_inode *ip, struct xfs_bmbt_irec *map, xfs_buf_flags_t incore_flags); int xfs_attr_rmtval_invalidate(struct xfs_da_args *args); -int __xfs_attr_rmtval_remove(struct xfs_da_args *args); +int __xfs_attr_rmtval_remove(struct xfs_delattr_context *dac); #endif /* __XFS_ATTR_REMOTE_H__ */ diff --git a/fs/xfs/xfs_attr_inactive.c b/fs/xfs/xfs_attr_inactive.c index bfad669..aaa7e66 100644 --- a/fs/xfs/xfs_attr_inactive.c +++ b/fs/xfs/xfs_attr_inactive.c @@ -15,10 +15,10 @@ #include "xfs_da_format.h" #include "xfs_da_btree.h" #include "xfs_inode.h" +#include "xfs_attr.h" #include "xfs_attr_remote.h" #include "xfs_trans.h" #include "xfs_bmap.h" -#include "xfs_attr.h" #include "xfs_attr_leaf.h" #include "xfs_quota.h" #include "xfs_dir2.h"
This patch modifies the attr remove routines to be delay ready. This means they no longer roll or commit transactions, but instead return -EAGAIN to have the calling routine roll and refresh the transaction. In this series, xfs_attr_remove_args has become xfs_attr_remove_iter, which uses a sort of state machine like switch to keep track of where it was when EAGAIN was returned. xfs_attr_node_removename has also been modified to use the switch, and a new version of xfs_attr_remove_args consists of a simple loop to refresh the transaction until the operation is completed. A new XFS_DAC_DEFER_FINISH flag is used to finish the transaction where ever the existing code used to. Calls to xfs_attr_rmtval_remove are replaced with the delay ready version __xfs_attr_rmtval_remove. We will rename __xfs_attr_rmtval_remove back to xfs_attr_rmtval_remove when we are done. xfs_attr_rmtval_remove itself is still in use by the set routines (used during a rename). For reasons of preserving existing function, we modify xfs_attr_rmtval_remove to call xfs_defer_finish when the flag is set. Similar to how xfs_attr_remove_args does here. Once we transition the set routines to be delay ready, xfs_attr_rmtval_remove is no longer used and will be removed. This patch also adds a new struct xfs_delattr_context, which we will use to keep track of the current state of an attribute operation. The new xfs_delattr_state enum is used to track various operations that are in progress so that we know not to repeat them, and resume where we left off before EAGAIN was returned to cycle out the transaction. Other members take the place of local variables that need to retain their values across multiple function recalls. See xfs_attr.h for a more detailed diagram of the states. Signed-off-by: Allison Henderson <allison.henderson@oracle.com> --- fs/xfs/libxfs/xfs_attr.c | 223 +++++++++++++++++++++++++++++----------- fs/xfs/libxfs/xfs_attr.h | 100 ++++++++++++++++++ fs/xfs/libxfs/xfs_attr_leaf.c | 2 +- fs/xfs/libxfs/xfs_attr_remote.c | 48 +++++---- fs/xfs/libxfs/xfs_attr_remote.h | 2 +- fs/xfs/xfs_attr_inactive.c | 2 +- 6 files changed, 294 insertions(+), 83 deletions(-)